International Crops Research Institute for the Semi-Arid Tropics

Socioeconomic Constraints to Developmentof Semi-Arid Tropical Agriculture

Proceedings

I n t e r n a t i o n a l W o r k s h o p o n

S o c i o e c o n o m i c C o n s t r a i n t s

t o D e v e l o p m e n t o f

S e m i - A r i d T r o p i c a l A g r i c u l t u r e

Hyderabad, India

1 9 - 2 3 February 1 9 7 9

International Crops Research Institute for the Semi-Arid T ropics

ICRISAT Patancheru P.O.

Andhra Pradesh, India 502 324

T h e c o r r e c t c i t a t i o n f o r t h e s e p r o c e e d i n g s i s I C R I S A T ( I n t e r n a t i o n a l

C rops R e s e a r c h I n s t i t u t e fo r t h e S e m i - A r i d T r o p i c s ) . 1980 . P r o c e e d -

i n g s o f t he I n t e r n a t i o n a l Workshop o n S o c i o e c o n o m i c C o n s t r a i n t s t o

D e v e l o p m e n t o f S e m i - A r i d T r o p i c a l A g r i c u l t u r e , 1 9 - 2 3 F e b r u a r y

1979 , H y d e r a b a d , I n d i a . ( A v a i l a b l e f rom I C R I S A T , P a t a n c h e r u ,

A . P., I n d i a 5 0 2 324 . )

Scientific Editor J.G. Ryan

Publication Editor H. L. Thompson

The International Crops Research Institute for the Semi-Arid Tropics(ICRISAT) is a nonprofit sc ient i f ic educational inst i tute receivingsupport from a variety of donors. Responsibi l i ty for the informationin th is publication rests with ICRISAT or the individual authors.Where trade names are used th is does not constitute endorsementof or discrimination against any product by the Inst i tute.

CONTENTS

F o r e w o r dPreface

Chapter 1

L. D. Sw inda leJames G. Ryan

vi iix

1

Soc ioeconomic Analysis of Existing FarmingSystems and Practices

Soc ioeconomic Ana lys is o f Exist ingFarming Sys tems and Practices inNo r the rn Niger ia

S o m e D imens ions o f Trad i t iona l FarmingSys tems in Semi -A r id Tropical India

Cropp ing Patterns, Farming Practices,and Economics of Ma jo r Crops in SelectedDry land Farming Regions of India

Discussant 's C o m m e n t sCha i rman 's S u m m a r y

Chap te r 2

G. O. I. A b a l u andB. D'Silva

N. S. Jodha

B. K. Rastogi

H. RuthenbergW. H. M. Mor r i s

3

11

25

3740

43

S o c i o e c o n o m i c s o f P rospec t i ve T e c h n o l o g i e si n S A T R e g i o n s

Soc ioeconomic Aspects Invo lved in

In t roduc ing N e w Techno logy in tothe Senegalese Rural Mi l ieu

Assessment o f Prospect ive So i l - , Water- ,and C rop -Managemen t Techno log ies fort he Semi -A r i d Trop ics o f Peninsular India

New Agr i cu l tu ra l Techno logy in theBrazi l ian Ser tao

Discussant 's C o m m e n t sThe Soc ioeconomics o f Prospect ive

Techno log ies : People and Prior i t iesCha i rman 's S u m m a r y

Chap te r 3

Moussa Fall

James G. Ryan, R. Sar in ,

and M. Pereira

J o h n H. Sanders

R. J. H. ChambersR. J. H. Chambers

J. L. Di l lon

45

52

73

8386

93

95

S o c i o e c o n o m i c F ie ld A s s e s s m e n to f P r o s p e c t i v e T e c h n o l o g i e s

Soc ioeconomic Field Assessment

of Dry Farming Techno logy inthe Indo-UK Project, Indore (India)

A. K. Chaudhar i 97

iii

H u m a n Na tu re a n d t h e Des ign o fAg r i cu l t u ra l T e c h n o l o g y

Farmer Par t ic ipa t ion a n d A c c o u n t i n g fo rt he Needs o f t he M o s t D i sadvan tagedG r o u p s : S o m e Ideas o n Par t i c ipa t ionat t he Outse t of a Research P r o g r a m

S o c i o e c o n o m i c F ie ld A s s e s s m e n t o fProspect ive T e c h n o l o g i es in t he S A Tof M a l i : A Case S tudy in an OACV Z o n e

S o c i o e c o n o m i c Research on Opera t iona lL a n d h o l d i n g s

Discussant 's C o m m e n t sCha i rman ' s S u m m a r y

C h a p t e r 4

Vic to r S. Doher ty

J a c q u e s Faye

B. T raore

M. Beno i t -Cat t in

Vic tor F. A m a n n

B. A. Krantz

107

120

125

136

140

142

145

Issues i n F o o d g r a i n M a r k e t i n g

R e f o r m i n g Gra in Ma rke t i ng Sys tems inWes t A f r i ca : A Case S tudy of M a l i

Foodg ra in M a r k e t i n g a n d Ag r i cu l t u ra lD e v e l o p m e n t in India

Discussant 's C o m m e n t sCha i rman ' s S u m m a r y

C h a p t e r 5

Ell iot Berg

M. v o n O p p e n , V . T . Raju,a n d S. L. BapnaB a l w a n t h ReddyJock R. A n d e r s o n

147

173

193195

197

S o c i o e c o n o m i c s o f I m p r o v e d A n i m a l - d r a w nI m p l e m e n t s a n d M e c h a n i z a t i o n

Observa t ions on the Economics o fT rac to rs , Bu l locks , a n d W h e e l e d T o o lCarr iers in t he S e m i - A r i d T rop ics o f India

An ima l - d ra f t Cu l t i va t ion i n F rancophoneAfr ica

S o c i o e c o n o m i c Aspec ts o f I m p r o v e d A n i m a l -

d r a w n I m p l e m e n t s and Mechan i za t i on i nS e m i - A r i d East A f r ica

Discussant 's C o m m e n t sCha i rman 's S u m m a r y

Chap te r 6

H. P. B inswange r ,R. D. Ghodake , andG. E. Th ie rs te inM. Le M o i g n e

Bruce F. J o h n s t o n

S. D. B iggsR. J. H. Chambers

199

213

221

234

237

239

A n a l y t i c a l R e v i e w s o f L i t e r a t u r eo n t h e S e m i - A r i d T r o p i c s o f W e s t A f r i c a

Farm-Leve l S tud ies i n t h e S e m i - A r i d

T rop ics o f W e s t A f r i ca

Discussant 's C o m m e n t sG o i n g Aga ins t t h e Gra in

M . N e w m a n ,I. O u e d r a o g o , andD. N o r m a nW. H. M. Mo r r i sBarbara Harr iss

241

264

265

iv

Discussant 's C o m m e n t sCha i rman 's S u m m a r y

C h a p t e r 7

N a t u r e a n d S i g n i f i c a n c e o f Risk

Nature and S ign i f i cance of Risk inthe Exp lo i ta t ion o f N e w T e c h n o l o g y

The Natu re a n d S ign i f i cance of Riskin t he S e m i - A r i d Trop ics

Risk and Uncer ta in ty in Peasant" D e c i s i o n - M a k i n g "

Discussants 's C o m m e n t sCha i rman ' s S u m m a r y

C h a p t e r 8

Rura l L a b o r M a r k e t s

Rural Labor Use and D e v e l o p m e n tSt ra teg ies in East A f r ica and India

Rural Labor Marke ts in Wes t A f r ica w i t hEmphas is on the S e m i - A r i d T rop ics

Labor Use a n d Labor Marke ts in S e m i - A r i dTrop ica l Rural V i l lages of Peninsular

IndiaDiscussant 's C o m m e n t sCha i rman ' s S u m m a r y

A p p e n d i x 1

A p p e n d i x 2 A p p e n d i x 3

El l iot BergJ. C. Davies

Jock R. A n d e r s o n

Hans P. B inswanger ,N. S. J o d h a , andB. C. Barah

Frank Cancian

J. L. D i l lonJ . G. Ryan

Janet Benson

Derek Byer lee

J a m e s G. Ryan ,

R. D. Ghodake ,and R. Sar inV. N. MisraJ . K a m p e n

French Abs t rac ts o f Engl ishPapersTexts of French Papers

List of Coun t ry PapersPart ic ipants a n d Observers

289293

295

297

303

317

326329

331

333

348

357

380

382387

398427431

V

F o r e w o r d

Since ICRISAT's earl iest days , soc ioeconomic research has p layed an im-po r tan t ro le in t he Inst i tu te 's ef for ts to i m p r o v e c rops and f a r m i n g sys tems inthe semi -a r i d t rop ics (SAT), a n d the l ives o f the peop le w h o d e p e n d on thesecrops fo r the i r basic f o o d supp ly .

Our G o v e r n i n g Board w a s aware tha t n e w techno log ies , and even seeds ,cou ld no t be d i s t r i bu ted and adop ted fo r use w i t h o u t an unders tand ing o f thesocial a n d e c o n o m i c cond i t i ons in w h i c h t h e peop le o f t he reg ion l i ved . Thef o l l o w i n g ob jec t i ve w a s w r i t t en in to ICRISAT's m a n d a t e in 1974, t w o yearsafter t he Inst i tu te was s ta r ted :

To identify socioeconomic and other constraints to agricultural development inthe semi-arid tropics and to evaluate alternative means of alleviating themthrough technological and institutional changes.

The cons t ra in ts to agr icu l tu ra l d e v e l o p m e n t i n t he S A T are m a n y : harshc l ima te ; l i m i t e d , errat ic, and unpred ic tab le ra in fa l l ; p o o r so i l s ; w i d e l y va ry ingresource e n d o w m e n t s , such as an a b u n d a n c e o f labor and scarc i ty o f land a n dcapi ta l in India, an abundance of land and scarci ty o f labor and capital in A f r i ca ,and — especia l ly in A f r ica — l im i t ed marke t faci l i t ies. Mos t of t he 600 m i l l i o npeop le popu la t i ng t he SAT reg ions of 49 count r ies l ive at a subs is tence leve l ,a n d t h e crops researched by ICRISAT ( s o r g h u m , pear l mi l le t , p igeonpea ,ch ickpea, and g r o u n d n u t ) are a m o n g the i r mos t i m p o r t a n t f o o d c rops .

The pr inc ipa l ta rge t o f ICRISAT's research is t he sma l l f a rmer of l im i tedm e a n s , w h o o f ten has poores t access to t e c h n o l o g y and the marke ts tha t canhelp h i m ef fect ive ly benef i t f r o m it. ICRISAT's economis t s and social an-t h r o p o l o g i s t s are w o r k i n g c losely w i t h b io log ica l and phys ica l sc ient is ts tobet ter ident i fy t he needs o f t h e sma l l f a rmer and dev ise i m p r o v e d crops andf a r m i n g sys tems tha t w i l l enhance his we l f a re and that o f t h e labor hee m p l o y s , as we l l as t hose w h o d e p e n d on h i m for f o o d supp l ies .

As par t o f ou r e f for t t o learn m o r e a b o u t t he soc i oeconom ic p rob lemsaf fec t ing th is r e g i o n , ICRISAT sponso red the In ternat iona l W o r k s h o p onS o c i o e c o n o m i c Cons t ra in ts to Deve lopmen t o f Semi -A r i d Trop ica l Ag r i cu l t u reat Hyde rabad , 19 to 23 February 1979. Th is con fe rence b r o u g h t toge there c o n o m i s t s a n d social sc ient is ts f r o m 16 count r ies , al l w i t h an interest in andexper ience w i t h t h e p r o b l e m s o f t he SAT.

The i r c o n t r i b u t i o n s a n d t h e resul ts o f the i r de l ibera t ions w i l l be f o u n d on thef o l l o w i n g pages. We a t ICRISAT have f o u n d these con t r i bu t ions to be va luab lein o u r w o r k ; we be l ieve tha t y o u a lso w i l l f i nd t h e m in teres t ing a n d usefu l .

L. D. SwindaleDirector General

VII

P r e f a c e

The primary objective of the Workshop was toconsider ways and means of overcoming thevarious socioeconomic constraints identifiedduring the course of the participants' researchin the semi-arid tropics. In particular we askedthem to focus their papers and discussions onthe role that new technologies and/or policiescould play in alleviating the constraints to thedevelopment process in the SAT.

The Workshop comprised three basic seg-ments : (1) subject matter sessions, (2) coun-try reports, and (3) field trips.

The subject matter sessions addressed issuesparticularly relevant to the SAT and on whichresearch is currently under way, both atICRISAT and in other institutions. Topics chosenincluded the analysis of existing farming sys-tems and practices, socioeconomics of pro-spective technologies, field assessment of pro-spective technologies, issues in foodgrain mar-keting, nature and significance of risk, rurallabor markets, and the economics of improvedanimal-drawn implements and mechanization.Papers were invited from colleagues workingon these topics in other institutions in SAT re-gions as well as those in the ICRISAT EconomicsProgram. A large amount of t ime was allowedfor open discussion, fol lowing the preparedcommentary by the discussant assigned foreach session.

Country reports were presented by inviteesfrom major SAT countries of primary concern toICRISAT. The objective was to obtain a set ofmeaningful socioeconomic data on the SATregions of these countries to enable cross-country comparative analyses to be made andto better identify technological and institutionalpotentialities and priorities. These papers,listed in Appendix 2, wi l l be assembled in a companion volume to these proceedings. In themeantime, copies of the papers can be obtainedf rom the Economics Program at ICRISAT.

Workshop sessions were held in Hyderabad,since ICRISAT Center was then still under con-struction on the research farm at Patancheru, 25km northwest of the city. A field tr ip was madeto the Center to familiarize participants wi th thetwo main research programs of ICRISAT, CropImprovement and Farming Systems. Scientists

f rom the five breeding programs included inCrop Improvement — sorghum, pearl millet,p igeonpea, chickpea, and g roundnu t — discussed the objectives and progress of theirresearch to develop new cultivars wi th higheryield potential, pest and disease resistance,improved quality, etc. Scientists f rom FarmingSystems discussed and demonstrated the vari-ous approaches in cropping systems and landand water management they are investigatingin order to develop viable technologies that wi l lhelp increase and stabilize crop production inthe SAT.

A second trip was made to Aurepalle, one ofthe six selected villages in ICRISAT's socio-economic studies that have been under waysince May, 1975 in the SAT of peninsular India.This visit, together wi th the one to ICRISATCenter, served to illustrate how the work of theEconomics Program directly relates to the workof the Crop Improvement and Farming SystemsPrograms, and the extent to which close collab-oration exists between scientists of these threeprograms.

The Workshop program was designed tomaximize t ime for discussion. The rapporteursrecorded major highlights of each session, in-cluding both the formal presentations and thediscussions. These were used, along wi th thewritten versions of questions and answers thatwe asked each questioner/commentator tocomplete, to help the session chairman preparea short summary of highlights and recommen-dations on the session for presentation at theconcluding session of the Workshop. It wasrequested of the session chairmen that theirrecommendations pay particular attention toquestions of (1) technology design, (2) agricul-tural policy, (3) future socioeconomic researchboth in national programs and at ICRISAT, andthe relationship of these programs to oneanother. For the convenience of readers, thesession summaries presented by the chairmenat the concluding session are found at the end ofeach chapter in this report.

This volume includes edited versions of allsubject matter papers that were submitted inwrit ing for the Workshop, and the Discussant'scomments and Chairman's summaries of each

IX

session. We have not included invited presenta-tions made orally unless these were later sub-mit ted in wri t ing.

Five of the papers were presented in French.They appear herein wi th English translations,which have been edited. The original Frenchtext wi l l be found in Appendix 1, along wi thFrench abstracts of the papers that were pre-sented in English.

I wou ld like to thank all of my colleagues in theEconomics Program and other units of ICRISATfor-their active assistance in the planning, prep-aration, and conduct of the Workshop. Es-pecially appreciated were the efforts of Infor-mation Services in the edit ing, print ing, anddistr ibution of the large volume of papers forthe Workshop and in the preparation and print-ing of these proceedings. The editorial assis-tance of Mrs. Vrinda Kumble is also gratefullyacknowledged.

I deeply appreciate the guidance and strongsupport of Dr. R. C. McGinnis, former AssociateDirector for Cooperative Programs at ICRISAT,who at that t ime had responsibil ity for inter-national conferences and symposia.

James G. RyanWorkshop Chairman

x

Chapter 1

Socioeconomic Analysis of Existing FarmingSystems and Practices

Soc ioeconomic Analys is of Exist ing FarmingSys tems and Pract ices in Nor the rn Niger ia

G. O. I. Abalu and B. D'Silva*

Abstract

Microlevel studies of existing farming practices in northern Nigeria show a clear-cut socioeconomic rationale behind the practices adopted by farmers in the area. Mixed cropping is favored because returns per acre are high and risk is minimized. Only those technologies that encourage and support the strategies behind the farming practices in this region will have a good chance of being successful. Conventional research approaches emphasizing single-crop technologies should be discouraged; an approach that examines all crops in the farming system and considers the felt needs of farmers should be evolved. The importance of multidisciplinary work and the ex ante role of the social scientist in the research effort are emphasized.

The principal variations in agricultural produc-tion in Nigeria arise from climatic differences.Four major ecological areas with homogeneousagricultural conditions can be identified: twosavanna areas in the northern part of the coun-try (Sudan savanna and Guinea savanna), theforest savanna in the middle belt, and the rainforest areas in the southern part of the country.The general pattern of these ecological areasconsists of horizontal bands proceeding fromeast to west.

The farming systems that have emerged overthe years have been shaped by the interactionof factors such as (1) tradit ion, (2) level oftechnology (hand labor, use of oxen, etc.), (3)resource availability (land, labor, capital, man-agerial skill, etc.), (4) physical environment(temperature, water availability, etc.), and(5) economic conditions (markets, storagefacilities, transportation, etc.). Any effort tochange the levels of production and standardsof living of farmers, if it is to be successful,would have to closely study these indigenousfarming systems.

In this paper, we examine the salient charac-teristics of farming systems in northern Nigeriaand evaluate the economics of existing prac-tices and cropping patterns, as revealed by

* Department of Agr icul tural Economics and RuralSociology, Institute of Agricul tural Research,Ahmadu Bello University, Zaria, Nigeria.

microlevel studies carried out in the Depart-ment of Agricultural Economics and RuralSociology at Ahmadu Bello University, Nigeriaover the last 12 years.

These s tud ies — in te rd isc ip l inary innature — have concentrated on describing,explaining, and understanding the agriculturalenvironment of small-scale farmers in northernNigeria and in identifying the economic, social,political, and institutional factors impeding ag-ricultural and rural development in the area.Considerable work has been done anddocumentation now exists on what farmers aredoing. Emphasis is now being directed towardsstudies of alternative strategies for increasingagricultural production and rural welfare andthe economic, social, and political implicationsand consequences of these strategies.

F a r m i n g S y s t e m si n N o r t h e r n N i g e r i a

The bulk of agricultural production in northernNigeria is undertaken by small-scale farmers.Their agriculture is characterized by a situationwhere:

1. the bulk of the labor force, management,and capital come f rom the same house-hold;

2. production is either consumed on the farmand/or traded in local markets;

3

3. the decision-making process is hamperedby l imited access to marketing and polit i-cal institutions; and

4. most of the farmers do not live muchabove the culturally determined subsis-tence level. (Abalu and D'Silva 1978).

There is a marked seasonal distribution ofrainfall in northern Nigeria. The rains generallybegin in Apri l or early May and continue to theend of September or middle of October. Duringthe rainy season, upland fields are cleared,plowed, planted, weeded, and ridged. The firstcrop to be harvested is millet, in August-September. Other crops (groundnuts, cowpeas,cotton, peppers, etc.) are harvested as theymature, wi th the major grain crop, sorghum,being harvested last, around January (Sim-mons 1978).

A significant aspect of the cropping systemthat has emerged in northern Nigeria is thatcrops are grown in mixtures. Norman (1972)found 24 different crops on rainfed (gona) land,in a total of 174 different crop mixtures. Solecrops accounted for only about 17% of the totalcultivated acreage (Table 1). More recentstudies have indicated a slight shift in favor ofsole crops (Ukpabio 1978).

The crops that farmers have chosen to growin the area have depended not only on physicalfactors such as rainfall, temperature, soil fertili-ty, etc., but also on economic, social, andpolitical considerations (Baker and Norman1975). The cropping system that has emerged iscomplex and dynamic. Given the available fac-tors of production, farmers allocate or wi thdrawthese factors f rom their farming activities de-pending on their goals, management abilitiesand expectations, as well as on how the seasonunfolds. Palmer-Jones (1978), for example,quotes Mike Watts as to the description givenby some farmers of their farming system inKatsina Emirate in northern Nigeria.

There are three types of land, jigawa, uplandsoils where cereals and beans are g r o w n ; andt w o types of fadama, one has a high water tablein the rainy season and somet imes f loods, andal lows us to g r o w a tobacco crop in the dryseason; the other f loods every year but we cang r o w irr igated vegetables in the dry season on it.I f there are good early rains, most of us (farmers)plant the mil let var iety zango on the jigawa atw ide spacings, some put the guinea corn varietyjadawa in between, others wai t unt i l the nextrains. If these f o l l ow the f irst rains w i th in say 10

Table 1. Common crop combinations In theZaria area. a

Crop Specification

Sole cropsSorghumGroundnutsCottonOther crops

Subtotal

Two-crop mixturesMillet/sorghumSorghum/groundnutsCotton/cowpeasOther crop combinations

Subtotal

Three-crop mixturesMillet/sorghum/groundnutsMillet/sorghum/cowpeasCotton/cowpeas/sweet

potatoesOther crop combinations

Subtotal

Four-crop mixturesMillet/sorghum/groundnut/

cowpeasOther crop combinations

Subtotal

Five- and six-crop mixturescombinations

Total

Percent of totalcultivated acreage

8.41.83.13.3

16.6

25.82.83.99.6

42.1

5.03.9

4.310.5

23.7

5.46.7

12.1

5.5

100.0

Source: Norman (1975).a. The survey included a total of 890.7 cultivated acres of

rainfed land, (i.e., upland).

days we wi l l al l plant jadawa and later we wi l linterplant the cowpea variety dan baranda. However, if there is a d rought after the f irst rains,or i f they come late, we wi l l replant w i t h the mil letvariety dan hawa and the guinea corn variety yar

4

bazerga, at wider spacings. Also if the rains arepoor we wi l l plant more guinea corn (jadawa) onthe first type of fadema and less rice because ourguinea corn wi l l not do so well on the jigawa andbecause the fadama is less likely to f lood, so ricewil l not do so well and guinea corn wi l l do better.

Faced wi th this type of scenario every year,the farmer is forced to plan purposefully if he isto meet his production targets. In the sectionthat fol lows, we attempt to evaluate theeconomic rationale behind the existing crop-ping systems in the area.

T h e E c o n o m i c s o fE x i s t i n g P r a c t i c e s

The farmers of northern Nigeria choose from a wide range of technical possibilities the prac-tices that wi l l lead them to achieve their specificgoals. With the aid of an income model, a riskaversion model, and a minimum nutrit ionmodel, this section examines the rationalitybehind existing farm practices in the area.

Maximizing Income 1

In his study of the rationale behind mixedcropping under indigenous conditions in north-ern Nigeria, Norman found that individual

1. This section draws f rom Norman's (1975) work oncrop mixtures.

yields of sorghum, groundnuts, and cotton aredepressed when grown in mixtures rather thanin sole stands. However, this was not alwaystrue for crops that are normally grown in mix-tures, such as cowpeas and millet. Sole-standcowpea appears to be more prone to insect anddisease attack than cowpea grown in mixtures;millets, which are the first seeds planted at thebeginning of the rains and the first crop har-vested in the growing season, are unlikely tosuffer greatly f rom competit ion wi th othercrops.

Average gross and net returns f rom solecrops and crop mixtures are presented in Table2. Average gross returns per acre are about 62%higher for crop mixtures. Average net returnsper acre show the same trend. It is obvious fromthis that crop mixtures bring higher returns perunit of land than sole cropping.

Since the farmer who is interested inmaximizing his income wil l most likely achievethat goal by maximizing returns from his mostl imiting resource, return per unit of land maynot be the most relevant criterion for his allo-cation of resources. In northern Nigeria, theamount of labor available during the periodbetween June and July is of crucial importanceto the type and amount of crops that can begrown. During this period land preparation,weeding and planting are taking place simul-taneously; consequently, the amount of laborthat can be made available at this t ime wil lgreatly influence the crops that are grown.Norman (1975) found that returns to June-Julylabor were higher for all crop mixtures than for

Table 2 . Ave rage gross and net re turns f r o m sole crops and cr op mix tu res .

Variable

Gross return per acre

Net return per acreLabor:

Not costedHired costedJ u n e - J u l y costedAll costed

Solecrops

153.6±22

148.9135.2133.774.1

Twocrops

240.6+ 19

235.7213.6204.7115.5

Crop Mixtures

Three Fourcrops crops

Shillings------------------

229.8±30

220.3199.1189.0105.3

340.9±80

322.9297.4276.8184.6

Allmixtures

248.3±16

240.8218.6208.2119.8

Overall

228.5±13

221.6201.2190.2110.1

Source: Norman (1975).

5

sole crops. Using the same data, Ogunfowora(1972) also concludes that the scarcity of re-sources on traditional farms in the area wasconsistent with the preference for mixed crop-ping.

Minimizing Risk

Numerous writers have incorporated yield andprice variability into models representing thedecision framework of small-holder farmers(Wharton 1968).. Farmers in northern Nigeriaare faced wi th a l imited growing season and areconstrained by l imited resources. It is thereforelogical for them to be concerned wi th ensuring a stable income in the face of biological andeconomic occurrences whose effects they can-not exactly determine in advance. Con-sequent ly , their c ropp ing pat terns areinfluenced by subjective formulations of expec-tations about future price and production situa-tions. On the basis of long-established tra-ditional procedures, they are able to arrive atcomplex calculations concerning probable fu-ture prices and yields. On this basis, they maychoose not to cultivate single crops over t ime,even where substitution ratios and price ratioexpectations may so dictate, but to cultivatecrop mixtures in somewhat fixed proportions.The rationale behind their strategy here issimply not to "pu t all of their eggs in onebasket."

The use of crop mixtures as a diversificationprecaution against uncertainty can be ac-complished in t w o ways:

1. The amount of resources can be increasedso as to be able to produce both productsA and B, for example, or

2. The amount of resources can be heldconstant whi le part of them are diverted toother products.

Since factors of production have been shown tobe l imit ing in the agriculture of the area, it is thelatter method that seems to be more applicable.Because of the marked seasonal distribution ofrainfall, the diversification strategy has tendedto be pursued through mixed cropping ratherthan through multiple or relay cropping (Bakeret at. 1975).

The question may be raised that if the farmeris dealing wi th risk aversion, why should he, forexample, mix crop A and crop B over his entirefield rather than divide his field between crop A

and crop B? The answer lies in the fact that thefarmer, realizing that he is faced wi th a l imitedand uncertain growing season, finds it attrac-tive to grow different crop species in the sameplot — provided that the relationship betweenthe species is complementary. Factors such ascomplementary growth cycles, noncompetingrooting habits, and compatible labor demandsmay enhance the desirability of growing cropstogether in the same plot rather than on differ-ent parts of the plot.

Using a mean-variance framework, an at-tempt was made to test the hypothesis that riskaversion is a critical item in explaining farmers'decisions concerning choice of cropping pat-tern.2 Information was obtained on the averagevalue, range, and standard deviation of incomederivable from a unit of agricultural land in thearea over a 5-year period, 1971 to 1975. Totalincome variance equations were calculated forpairs of identified farm enterprises includingvariances and covariances. These equationsthen provided a basis for calculating the opti-mum proportion of resources, P, that shouldtheoretically be allocated to each crop enter-prise in the pair so as to enable minimizing thevariability of income over the period. Theseproportions were then compared with the actu-al observed proportions of resources devotedto the pairs of crop enterprises. The results arepresented in Table 3. Paired comparisons on thetwo series using a t-test revealed no statisticaldifference between the model and the observedproportions at the 95% confidence level.3

The results of the analysis indicate that thecropping system in the area reflected a risk-aversion strategy.

Meet ing Protein and CalorieRequirements

Malnutrition has very serious repercussions for thesmall-holder farmer. There is always a balancebetween the nutr i t ional resources avail-able to h im (which depend largely on how muchfood was produced during the preceding har-vest) and the nutrit ional requirements to main-tain the existing and future l ivelihood of his

2. Details of the methodo logy employed can be found

In Aba lu (1976).3. It Is realized that a t-test of differences in propor-

t ions is not str ict ly val id.

6

Tab le 3 . Propor t ion o f resources a l loca ted to pairs o f c rop enterpr ises. a

A

SorghumSorghumSorghumSorghumSorghum

Sorghum/GroundnutsSorghum/GroundnutsSorghum/GroundnutsSorghum/Mil letSorghum/Mi l letGroundnuts

GroundnutsGroundnutsGroundnutsGroundnuts/CowpeasGroundnuts/CowpeasGroundnuts/Cowpeas

Crop enterprises

B

Mil let /SorghumGroundnuts/SorghumMil let/Groundnuts/SorghumMil let/Cowpeas/SorghumGroundnuts/Cowpeas/Sorghum

Mil let/Sorghum/GroundnutsCowpeas/Sorghum/GroundnutsMil let/Cowpeas/SorghumGroundnuts/Sorghum/Mil letMil let/Cowpeas/SorghumGroundnuts/Sorghum

Groundnuts/CowpeasMil let /Groundnuts/SorghumGroundnuts/Cowpeas/SorghumGroundnuts/Cowpeas/SorghumMil let/Groundnuts/SorghumMil let/Cowpeas/Sorghum

Allocation of resourcesto crop Ab

Model solut ion

0.271.000.700.480.97

0.260.180.691.000.530.66

0.480.370.200.110.660.53

Observed

0.250.750.620.68n.a.c

0.39n.a.

0.710.840.860.40

n.a.0.27n.a.n.a.n.a.n.a.

a. Proportion is computed as the quantity of land devoted to an enterprise relative to the total land devoted to the pair ofenterprises.

b. Refers to the proportion of resources allocated to crop enterprise A. The proportion of resources allocated to crop enterprise B is one minus the proportion shown here.

c. Not available.

family. This ratio of available resources torequired inputs wil l vary depending on thecrops that the farmer chooses to grow on hisl imited land base.

In the face of threats of malnutrition to him andhis family, the farmer would be expected toadjust and reallocate agricultural resourcesavailable to him and adopt or discard land usepractices. The beginning of the rains witnessesan increase in agricultural activities associatedwith land preparation, planting, weeding, andharvesting. This is also the period when manypeople are in a negative nutritional balancebecause of the shortage of food. It has beensuggested that some of the most serious de-bilitating diseases, such as malaria, diarrhea,guinea wo rm, and infections of the skin, peakduring this t ime as well (Longhurst 1978). Coin-ciding wi th a peak labor demand — when fai-lure to cultivate, plant, weed, or harvest maycritically affect future income and foodsupplies — these infectious diseases increase

the risk and vulnerability of small-holder far-mers in the area (Longhurst 1978). Hill (1972)has suggested that the ability of the farmer tosurvive this period determines to a large extentthe welfare of the family during the rest of theyear.

With the aid of a linear programming model,we attempted to find out whether the croppingsystem that the farmer has come to prefer isconsistent wi th his body requirements and thenutritional requirements of his family. Themodel had as an objective the maximization offarm income in the face of a min imum nutrit ionconstraint. Constraints were also imposed onavailable farm labor and farm land. Nine differ-ent cropping activities, reflecting those mostcommon in the area, were considered. The dataused in the analysis are derived f rom work bySimmons (1976) and Norman (1972).

Yearly requirements of protein and caloriesfor an average family in the Zaria area areshown in Table 4. The program was analyzed.

7

Table 4. Calorie and protein requirements for an average fa mily, Zaria, Nigeria.

Age category

Under 7 yearsBetween 7-14 yearsFemale over 14 yearsMale over 14 years

Average number ofpersons per family

2.181.542.622.15

Annual requirements

Calories('000 kilo cal.)

1252131319132119

Protein(kg)

13.8916.9233.9430.64

Total 6597 95.39

using farm sizes of 3, 5, 10, 12, and 18 acres(average farm size in the area is 7.9 acres,Norman 1972). The results thus obtained wouldshow a relationship between farm size, laboruse, and nutrit ional balance.

The results of the analysis are presented inTable 5. A farm size of 3 acres resulted in aninfeasible solution. All the other farm sizes,however, were able to meet the nutrit ionalconstraints. In fact, a surplus of both protein andcalories was found to exist in all cases. Over70% of the protein and 40% of the caloriessupplied by the crop activities in the plan werein surplus. We have not here compared thesebalances wi th the actual balance as revealedfrom the existing farm plans. This wou ld be a valuable exercise.

Of significant interest in Table 5 is the fact thatall crop activities in the optimal plans for all

farm sizes are crop mixtures. This is all the morestriking since the model provided for nonpro-tein and noncalorie-generating crop activities toenter the optimal plan, if they generatedenough income to purchase the min imum pro-tein and calorie requirements.

The analysis also showed that a strong re-lationship existed between income, farm size,and the hiring of labor. Thus, as farm sizeincreased and if labor was available to be hired,incomes for farmers would also increase and allfarms, except those less than 3 acres in size, metthe nutritional requirements of the work force. Itshould be emphasized that the major constrainthere would be labor availability.

It is also interesting to note that the farm sizeof 3 acres resulted in an infeasible solut ion,suggesting that only a larger acreage is capableof meeting the protein and calorie require-

Table 5. Solutions of program for different land sizes.

Landconstraint(acres)

35

10

12

18

Crop activities in the plan

InfeasibleMillet/Sorghum/CowpeasMillet/Sorghum/G'nuts (5.99 acres)Millet/Sorghum/Cowpeas (4.01 acres)Millet/Sorghum/G'nuts (9.64 acres)Millet/Sorghum/Cowpeas (2.36 acres)Millet/Sorghum/G'nuts (15.99 acres)Millet/Sorghum/Cowpeas (2.56 acres)

Production of nutrients

Calories('000 kilo cal.)

69595700558191743660

15217

Protein(kg)

220148177238116395

Net farmincome

(£ N)

NS -------------855

1382

1540

1868

Month oflaborhiring

MayMay, JuneJuly, Aug.May, June, JulyAug, NovApr, May, JuneJuly, AugOct, Nov

NS = No solution obtained.

8

ments. Consequently, a farmer confronted wi tha l imited land base would have to take alterna-tive action if he is to feed his family adequately.One such action is for h im, knowing that theland available to him is too small to meet themin imum nutritional requirements of his familyif he grew food crops, to choose to grow themost profitable crop compatible wi th his lowincome. In most cases, this would be a cashcrop. Revenue received from the sale of the cropwould thus al low a higher level of food con-sumption (as well as meet the minimum non-food cash needs) than if the entire land base hadbeen allocated to less profitable food crops.

Al though this paper does not further analyzethe protein strategy of farmers who own 3 acresor less, another study near the area does pro-vide some information on the nutritionalstrategy of land-scarce farmers. Matlon (1977)finds that groundnut production was givenrelatively greater emphasis by low-income far-mers. Their food production was inadequate tomeet their min imum nutritional requirements,yet groundnut (a cash crop) constituted a dis-proportionately large component of the cropsthey produced. The opt imum plans from thelinear programming exercise would thereforeappear to support the hypothesis that the em-phasis on crop mixtures in the agriculture of thearea is consistent wi th the nutritional require-ments of the farmer's family.

Improv ing Cropp ing Sys temsin Nor the rn Niger ia

What emerges f rom the foregoing analysis isthe fact that existing farming systems andpractices in northern Nigeria have a clear-cutrationale behind them. This does not imply thatfarmers are organizing their resources in thebest manner and making the best use of theirland. But it is obvious that efforts to improveupon the existing farming systems should aimat (1) examining the rationale behind existingpractices, (2) avoiding actions that weaken thestrategies underlying these practices, and(3) encouraging measures that support andstrengthen these strategies.

To this end, suitable technologies have to bedeveloped, accompanied by the creation andmaintenance of relevant infrastructural supportsystems and the undertaking of honest researchon farming systems.

Most of the technologies that have beendeveloped in northern Nigeria so far have beendesigned to improve single crops rather thanthe complete farming system. This is because ofthe long-standing tendency to associate pro-gressive agriculture wi th sole cropping and theselfish quest to promote the development ofcertain crops. Consequently, efforts at applyingnew technologies have resulted only in isolatedand haphazard introduction of single-cropmodern technology. This has resulted in a fewisolated public and private farms using moderntechnology while the vast majori ty of farms insurrounding villages continue to use primitiveand crude implements.

A systems approach to the development ofnew technologies appears to be needed to in-crease agricultural production and the welfareof farmers in the area. For example, it would benecessary to be aware of the effect of any newpractice on the entire farming system and, assuch, to balance the effect in such a way that thenew farming system that emerges does not loseits logic and rationali ty. New technologies,therefore, would be judged not only by theirtechnical feasibility but also by their technicalcompatibility. These technologies would be ex-pected not only to be able to increase productiv-ity given the prevailing technical environmentbut also to be compatible with the technicalenvironment in which other technologies oper-ate or are expected to operate. Whi le thetechnologies must return fair profi t to thefarmer, they must also exhibit a level of risk thefarmer can accept and meet the nutritional re-quirements of the farming family. The newtechnologies would also need to be compatiblewith existing structures and norms, as well asthe prevailing or possible levels of infrastruc-tural support.

Considerable research on farming systemswould be needed to ensure that appropriatefarming systems evolve in the future. Thisresearch must emphasize the present status offarming systems in the area and their potentialfor improvement. The research effort must dealwith the dynamic nature of the farming systemand be done f rom the "bot tom-up." This under-scores the need for the research to be multidis-ciplinary and for the social scientist to play notonly an ex post but also — perhaps moreimportantly — an ex ante role in the researcheffort. Above al l , research on existing farming

9

systems and how to improve them should

reflect a sincere concern for the welfare of the

masses in the area rather than the personal

and — as experience has shown — selfish

goals of researchers and sponsoring agencies.

S u m m a r y a n d C o n c l u s i o n

This paper has attempted to evaluate the

economics of existing farming practices in

northern Nigeria as revealed by microlevel

studies of the area. We have tr ied to show that

there is a clear-cut socioeconomic rationale

behind the practices adopted by farmers in the

area. This rationale and logic would suggest

that only technologies and institutions that

encourage and support the strategies behind

the farming practices would have a good

chance of being successful. We suggest that

conventional research approaches that em-

phasize single-crop technologies should be dis-

couraged and, in their place, a research ap-

proach that examines all crops in the farming

system and takes into account the felt needs of

farmers should be evolved. The importance of

multidisciplinary work and the ex ante role of

the social scientist in the research effort are also

emphasized.

References

A B A L U , G. O. I. 1976. A note on crop mixtures underindigenous condi t ions in nor thern Nigeria. Journalof Development Studies 12:214.

A B A L U , G.O.I., and D 'S ILVA, B. 1978a. Nigeria's foodsi tuat ion: problems and prospects. Zaria, Niger ia:Department of Agr icul tural Economics and RuralSociology, Ahmadu Bello University. (Mimeog-raphed.)

ABALU, G . O . I . , and D 'S ILVA, B. 1978b. Crop mixturesas a land use strategy in nor thern Nigeria. Presentedat Sympos ium on Land Use and Development inAf r ica , Sou th of the Sahara: Smal l -holders ' logicand Technical Rationality, 4 - 8 December 1978,ORSTOM-CVRS, Ouagadougou, Upper Volta.

A N O N Y M O U S . 1974. Agr icul tural deve lopment In

Nigeria, 1973-1975. Lagos, Nigeria: Federal Min is-t ry of Agr icul ture and Natural Resources.

BAKER, E. F. I., and N O R M A N , D. W., 1975. Cropping

Systems in northern Nigeria. Presented at the Southand East Asia Cropping Systems Network Work-shop, 18-20 March 1975, International Rice Re-search Institute, Los Bands, Phil ippines.

HILL, P.1972. Rural Hausa: a v i l lage and a sett ing.Cambridge, UK: Cambridge University Press.

LONG HURST, R. 1978. Review of Conference on Sea-

sonal Dimensions to Rural Poverty. 3 - 6 Ju ly 1978,Institute of Development Studies and Ross Schoolof Tropical Hygiene, Universi ty of Sussex, Br ighton,UK.

M A T L O N , P. 1977. The size dist r ibut ion, structure anddeterminants of personal income among farmers inthe nor th of Nigeria. Ph.D. thesis. Cornell University.Ithaca, New York, USA.

N O R M A N , D. W. 1972. An economic study of threevil lages in Zaria province. Samaru miscel laneouspaper 37, Ahmadu Bello University, Zaria, Nigeria.

N O R M A N , D. W. 1975. Rationalizing mixed croppingunder indigenous condi t ions: the example of nor th-ern Nigeria. Journal of Development Studies 11:1.

OGUNFOWORA, 0. 1972. Derived resource demand ,product supply and farm policy in the Nor th CentralState of Nigeria. Ph.D. thesis, Iowa State University,Ames, Iowa.

PALMER-JONES, R. W. 1978. Linear p rogramming andthe study of peasant fa rming - Rejoinder. Depart-ment of Agricul tural Economics and Rural Sociol-ogy, Ahmadu Bello University, Zaria, Nigeria.(Mimeographed.)

S I M M O N S , E. B. 1976. Calorie and protein intakes inthree vil lages of Zaria province. Samaru miscel-laneous paper 55, Ahmadu Bello Universi ty, Zaria,Nigeria.

S I M M O N S , E. B. 1978. A case study of seasonal varia-t ion in food and agr icul ture: northern Nigeria.Presented at the Conference on Seasonal D imen-sions to Rural Poverty, 3 - 6 Ju ly 1978, Institute forDevelopment Studies, Universi ty of Sussex,Br ighton, UK.

UKPABIO, G. E. 1978. Product ivi ty differentials bet-ween smal l and large scale farmers: a case study ofthe Funtua agricul tural project area. M.Sc. thesis.Department of Agr icul tural Economics and RuralSoc io logy , A h m a d u Bel lo Univers i ty , Zar ia ,Nigeria.

W H A R T O N , C. R. 1968. Risk, uncertainty and the subsis-tence farmer : technological innovat ion and resis-tance to change in the context of survival . Presentedat the jo int session of the Amer ican EconomicAssociat ion and the Associat ion for Comparat iveEconomics, Chicago.

10

Some Dimensions of Traditional Farming Systemsin Semi-Arid Tropical India

N. S. Jodha*

Abstract

This paper summarizes some results of village-level studies conducted since 1975 by ICRISAT in six villages in three agroclimatic zones of peninsular India. Results that are of direct relevance to the research strategy for generating new technology for SAT areas are discussed. The paper analyzes the rationale behind the practices of monsoon fallowing of deep Vertisols and intercropping in rainfed agriculture. Constraints on the spread of prospective watershed technology are also discussed. Because small farms have a relatively greater extent of monsoon fallowing and intercropping, any low cost technological advance in these research areas may help less-endowed farmers. Under the existing pattern of land distribution and utilization, prospective watershed technol-ogy is likely to face severe institutional constraints.

This discussion of traditional farming systemsin semi-arid tropical (SAT) areas of peninsularIndia is based on data f rom village-level studiesundertaken by ICRISAT in three agroclimaticzones since May 1975. The principal objective ofthe village-level studies was to understand theconstraints and potentials of traditional farmingsystems and to use this understanding as aninput in the generation of new technology forSAT agriculture. Guided by this consideration,the paper addresses itself to a few key aspectsof traditional farming systems that are of directand immediate relevance f rom the standpointof technology generation.

* Previously Senior Economist, Economics Program,ICRISAT; now Agricul tural Economist, Interna-t ional Institute of Tropical Agr icul ture, l longa, Tan-zania.

1. For sampl ing procedure and other methodolog i -cal details of ICRISAT vi l lage studies see Jodha etal. (1977). During the per iod of 3 years somehouseholds belonging to the sample of laborhouseholds (10 in each vil lage) acquired land.However, they have not been included in thepresent analysis. In keeping w i th the differentland: man ratios prevai l ing in the vi l lages, dif fe-rent ranges of operational landhold ing (in hec-

The paper makes use of data for 3 agriculturalyears (1975-76 to 1977-78) collected regularlyat intervals of about 20 days from 30 samplefarms (10 small, 10 medium, and 10 large) f romeach of the six villages.1

Table 1 provides some information about theselected villages and the sample farms and alsobroadly differentiates the three agroclimaticzones. The three zones considerably differ interms of soil types, rainfall and extent of irri-gation. The differences influence the farm levelavailability as well as pattern of resource use inthese regions.2

tares) were used to define smal l , med ium, andlarge farms as indicated below (for details seeGhodake and Asokan 1978).

Village

KanzaraKinkhedaKalmanShirapur

Aurepal leDokur

Small

0.21-2.250.21-3.000.21-6.000.21-2.500.21-2.500.21-1.00

Medium

2.26- 5.603 . 0 1 - 5.606.01-10.752 . 5 1 - 6.002 . 5 1 - 5.251 .01- 3.00

Large

> 5.60> 5.60>10.75> 6.00> 5.25> 3.00

2. This paper deals wi th some aspects of the usepattern of land and water resources and theirimpl icat ions. For a discussion of the labor re-source and its use see Ryan et al. (1979a).

11

Tab le 1 . De ta i l s o f opera t iona l l and hold ing a n d i ts use pa t te rn on samp le fa rms in six S A T Indianvi l lages f r o m 1 9 7 5 - 7 6 t o 1 9 7 7 - 7 8 . a

Village(Location, soils, annual rainfall)

KANZARA

(Akola Dist.; med ium deep Vert isols;820 mm)

KINKHEDA

(Akola Dist.; med ium deep Vert isols;820 mm)

KALMAN

Sholapur Dist.; deep and med ium-deep Vert isols; 690 mm)

SHIRAPUR(Sholapur Dist.; deep and med ium-

deep Vert isols; 690 m m )

AUREPALLE(Mahbubnagar Dist.; shal low and

medium-deep Al f isols; 710 mm)

DOKUH(Mahbubnagar Dist.; shal low and

medium-deep Al f isols; 710 mm)

Farmsize

group b

(ha)

Smal lLargeTotalh

Smal lLargeTotal

Smal lLargeTotal

Smal lLargeTotal

Smal lLargeTotal

Smal lLargeTotal

Averagesize ofland

holding

(ha)

1.414.2

6.5

1.613.26.7

3.513.88.5

1.711.36.5

1.412.05.6

0.78.23.7

Landuse

inten-si tyc

(%)

999797

979293

989598

1009192

937076

809082

Croppinginten-

sityd

(%)

103103103

104106106

105108108

108114114

104119114

120113113

Irri-gatedareae

(%)

655

454

71110

221113

52521

745960

Rainy season

Fallow f

(%)

222

233

665961

777168

575

81918

Cropped

areag

( % )

989898

989797

344139

212932

959395

928182

a. Based on details from 180 sample farms in six villages. Village-level studies have been conducted in these villages since May 1975 (Jodha et al. 1977).

b. For the details of farm size grouping see text (footnote 1). All tables in this paper exclude the labor households even if theyhave acquired land during the reference period.

Net cropped area c. Land-use intensity x 100

Total operational areaGross cropped area

d. Cropping Intensity x 100Net cropped area

a. Gross irrigated area as proportion of gross cropped area.1. Proportion of total net sown area fallowed during rainy season and planted in postrainy season. In Dokur and Aurepalle the

rainy-reason fallow lands consist of tank beds where runoff collection is druring rainy season. These areas are available forplanting crops only in the postrainy season once tank water is used up for irrigation.

g. Proportion of gross cropped area. It covers all area planted during rainy season including the double-cropped area.h. Totals include medium farms besides small and large farms in each case.

Intensity of Land Useand Cropping

From Table 1 it is apparent that irrigation is

associated wi th reduction in the average size of

operational landholding and an increase in the

cropping intensity. Except in the highly irrigated

village Dokur, and to some extent Aurepalle, the

intensity of land use during the reference period

was found to be very high (exceeding 90%).

Land-use intensity was higher on small farms

than on large farms. This is quite understand-

able as the smaller the land holding, the less the

12

scope for resting land through periodical fal low-ing.3 In all villages except Dokur, however, thecropping intensity was higher on large farms.This indicates that resource-rich farmers, in-stead of spreading their nonland resourcesthinly oyer all their land, try to concentrate theircropping efforts only on part of the total hold-ing. Furthermore, taking farms as a whole, thegap between intensity of land use and intensityof cropping widened wi th the extent of irri-gation available. The Mahbubnagar villages(particularly Dokur) clearly demonstrate thisphenomenon.4 Since the profits f rom irrigatedland are much higher than on nonirrigated land,farmers prefer to keep part of the dryland fal lowand concentrate their attention on irrigatedland. The low land-use intensity in Dokur, es-pecially on small farms, is thus largelyexplained by the extensive irrigation in thevillage. This has implications for rainfed agricul-ture. Because of the low and uncertain profitcharacterizing rainfed agriculture, less attentionis paid to it and resources are, if possible,diverted to irrigated farming.5 More about thislater.

R a i n y - S e a s o n F a l l o w s

Another feature of traditional farming is theseasonal distribution of cropping (Table 1). In

3. This has been observed at macrolevel also in thecase of the arid region of Rajasthan, where theextent of periodical fa l lowing has decl ined w i th a decrease in the farm size. Such increased landuse intensity, unaccompanied by measures toprotect and conserve the submarginal lands hasaccelerated the process of desert i f ication in theregion (Jodha 1977b).

4. The extent of i r r igat ion reported in Table 1 doesnot take into account the intensity of irr igation. Ifthis is done, the extent of irr igation in Mahbub-nagar villages wi l l increase substantial ly. SeeTable 8.

5. ft has been observed in the study vil lages andelsewhere that several farmers usually ignoresuch operations as weeding, intercultur ing, etc.,at crucial t imes - on dry land crops and opt forwage employment on irr igated farms at t imes -by temporary outmigrat ion. In such c i rcum-stances at least a part of the low product iv i tyof rainfed agriculture could be attr ibuted to theattract ion of neighbor ing irr igated farming.

the two Sholapur villages that have a highproportion of deep Vertisols and a bimodalpattern of monsoon rains, 61 and 68% of the netcropped area was kept fal low during the rainyseason and planted in the rabi, or postrainy,season.6 This practice, known as kharif (rainyseason) fallowing and rabi (postrainy season)cropping, is widespread in the deep Vertisolregion of SAT India.7

The important reasons advanced by farmersfor fal lowing the deep Vertisols during the rainyseason and then planting them in the postrainyseason were as fol lows:

1. In the absence of good soaking rains, thedeep Vertisols are too hard to work; oncesubstantial rains begin, it is difficult to entersuch fields.

2. Even if some crops are dry sown in deepVertisols prior to rains, the management of thecrop during the subsequent wet period isdifficult. Weeds may ruin the crop before thefields are dry enough to permit entry of labor.

3. The rains received during the early phaseof the monsoon are less dependable than thosereceived during the later phase. According tothe farmers' experience and meteorologicaldata, early rains are inadequate to fully saturatethe profile of deep Vertisols. The crops plantedduring the first phase of the monsoon areexposed to the risk of drought in a prolongedmidseason dry spell, and to waterlogging aswell as increased disease incidence caused bycontinuous rains in the second phase of themonsoon when they are at the f lowering orripening stages.

At present, farmers — not aware of crop va-rieties or land management practices that can

6. In Dokur, the rainy season fa l lowing was morethan 18%. But this represented a different situa-t ion f rom the Sholapur vil lages. The rainy seasonfal low areas in Dokur and Aurepal le largely con-sisted of tank beds where runoff collection tookplace dur ing the rainy season. Once the waterwas used for i r r igat ion, these tank beds becameavailable for cult ivat ion. This incidentally, re-duces the land lost due to tradit ional runoffcollection tanks.

7. It is est imated that nearly 18 mi l l ion hectares, ormore than 24% of the net sown area in SAT areasof India, is fal lowed dur ing the rainy season, to beplanted dur ing the postrainy season. [ J . G. Ryan,personal communicat ion, using distr ictwide datafrom Malone (1974)).

13

reduce these hazards of rainy-season croppingin the deep Vertisols — continue to fol low thetraditional practice of fallowing land in themonsoon season. Given the hazards of rainy-season cropping and the nonavailability of via-ble technology to counter them, the farmerprobably makes a rational choice in leaving thedeep Vertisols fal low during the monsoon. Theirrationality of rainy-season fallow can be de-monstrated only by providing a viable alterna-tive, and this constitutes the challenge for ag-ricultural research.8

Even if one ignores the benefits of reducedsoil erosion when Vertisols are planted in therainy season (Kampen et al. 1974), the potentialpayoff f rom a breakthrough in technology formonsoon-fallow areas, facilitating the raising ofkharif crops as well as rabi crops, wil l increasethe gross cropped area by nearly one-fourth ofthe current net sown area in SAT India.

Furthermore, as shown by Table 1, sincesmall farmers leave a higher proportion of theirland fallowed during the monsoon than dolarge farmers, the prospective low-costtechnology for such areas may help small far-mers more than large ones. This indicates onepossible direction for achieving egalitariangoals through technological as opposed toinstitutional means in SAT areas.

I n t e r c r o p p i n g 9

Intercropping, or growing crops in mixtures, isan important feature of traditional farming inSAT areas of India and elsewhere. The superi-ority of intercropping, in terms of higher grossreturns as well as higher and more evenlyspread employment of labor when compared tosolecropping, has been documented by Mathur(1963) and Norman (1974, 1978). Additional

8. The Al l India Coordinated Research Project forDryland Agricul ture, ICRISAT, and several otherinstitutions are currently engaged in developingtechnology for tradit ional ly monsoon- fa l lowareas. Besides the work at experimental stations,ICRISAT initiated in 1977-78 diagnostic experi-ments on farmers' f ields in three of the vil lages towh ich this study refers.

9. This discussion on intercropping draws heavilyf r om the discussion presented elsewhere; seeJodha (1979b).

reasons for this are given below.As shown by Table 2, the extent of intercrop-

ping as a proportion of gross, cropped area(average of 3 years) varied from 18 to 83% in thesix villages under study.

Factors that explain the differences in extentof intercropping in these villages were theamounts of irrigation, postrainy-season crop-ping, and extent of HYVs, as well as the extent ofsome specific crops such as paddy, castorbean,and sugarcane (rarely grown as mixed crops inthese villages) (Table 2). Table 3 illustrates thatincreases in the above factors lead to greateremphasis on sole cropping.

Diminution of intercropping with an increasein irrigation, observed also in the commandareas of new irrigation projects such as theChambal canal (Bapna 1973), is not difficult tounderstand. To the extent that intercropping is a strategy against weather-induced risk, the av-ailability of irrigation reduces this need (Jodha1977b). The same reasoning applies to thesituation where the postrainy season discour-ages intercropping. Unlike rainy season crops,postrainy crops, are grown on residual mois-ture stored in the soil profile largely in deepVertisol nonirrigated areas. Planting of crops insuch situations begins with a known state of soilmoisture; hence the need for intercropping toadjust to eventual fluctuations in the soil mois-ture situation becomes less important.

To the extent that HYVs and intercropping arenot incompatible, the limited use of HYVs inintercropping systems may hopefully be re-garded as a transitional phase. 10But the realissue is not the technical suitability of HYVs forintercropping." The farmers' decision is largelyguided by economic costs involved. From thestandpoint of the majority of SAT farmers inIndia, the HYV technology can be a high-costtechnology as it consists of costly inputs

10. This has been amply demonstrated by experi-mental evidence. See papers presented at anIntercropping Workshop, ICRISAT (1979).

11. Of course there is one technical possibil i ty ofHYVs discouraging intercropping. HYVs are gen-erally earlier matur ing crops that facilitate se-quential or relay cropping, i.e. planting a secondcrop after harvesting the first crop. Thus anyintercropping that may obstruct sequential orrelay cropping may be rejected by the HYV-grower (J. G. Ryan, personal communicat ion).

14

Table 2 . Extant o f in tercropping and re la ted detai ls in six S A T Indian vi l lages f r o m 1 9 7 5 - 7 6 to1 9 7 7 - 7 8 . a

Village

Kanzara

Kinkheda

Kalman

Shirapur

Aurepalle

Dokur

Farmsize

groups

SmallLargeTota l '

SmallLargeTotal

SmallLargeTotal

SmallLargeTotal

SmallLargeTotal

SmallLargeTotal

Proportion of gross cropped area devoted to

Intercroppingb

(%)

877073

918283

604147

111918

443435

52121

Irrigatedcrops

(%)

655

454

71110

221013

52521

745960

HYVsc

(%)

131616

477

_

11

_

1

-

31512

774344

Specificcropsd

(%)

_

22

222

354

577

415754

824550

Postrainyseason

croppinge

(%)

222

233

665961

777168

575

81918

a. Based on details from 180 sample farms in six villages. Village-level studies have been conducted In these villages since May1975 (Jodha et al. 1977).

b. The small and large farm differences in the extent of intercropping were found to be statistically significant at the one percentlevel.

c. High-yielding varieties mainly Include hybrid sorghum and hybrid cotton in Kanzara and Kinkheda, and HYV paddy inAurepalle and Dokur.

d. Includes crops like paddy, castor bean, and sugarcane, more than 90% of which are grown only as sole crops.e. Net area of postrainy season (rabi) cropping as a proportion of total net sown area.f. Total includes medium farms besides small and large farms In each case. For the basis of farm size classification, see footnote

1 of the text.

such as fertilizers, pesticides, and better man-

agement of the crop. The farmer may not want

to divert costly inputs meant for HYVs to inter-

cropped, non-HYV crops.12 The same consider-

ation tends to discourage the mixing of other

crops wi th high water-requiring, high payoff

crops like paddy and sugarcane. Furthermore,

crops like paddy, castorbean, and sugarcane

may lack strong technical complementarity

with other crops. Villages wi th a high propor-

tion of these crops correspondingly had a lower

extent of intercropping (Table 2). On the other

hand, villages (particularly Kanzara and

Kinkheda) with substantial area under crops

such as pigeonpea, cotton, or rainy-season

sorghum (largely grown as intercrops) had a

higher extent of intercropping.1312. The dif f iculty of incorporating HYVs into inter-cropping systems could be one of the factorsresponsible for a l imited spread of HYVs region-ally as well as among small farmers who practiceextensive intercropping (Table 2).

13. For details of major sole crops and crop combina-tions in mixed crops see Jodha (1979b).

15

Table 3 . Propor t ions o f a rea devo ted to so le cropping in si x S A T Indian vi l lages f r o m 1 9 7 5 - 7 6 to1 9 7 7 - 7 8 . a

Village

KanzaraKinkhedaKalmanShirapurAurepal leDokur

Proport ions of sole cropping in the total of:

Postrainy season Gross irr igatedNSA area

( % ) ( % )

99 100100 7465 8379 90

100 9499 100

HYVs area

(%)

777361

100100100

Specific cropsb

(%)

_

-939195

100

a. Based on details from 180 sample farms in six villages. Village-level studies have been conducted in these villages since May 1975 (Jodha et al. 1977).

b. Specific crops are sugarcane in Kalman and Shirapur; paddy in Dokur, and paddy and castor bean in Aurepalle.

Farm Size and Intercropping

Largely because of its risk-reducing potential,intercropping is a more popular system amongsmall farmers.14 Both because of their poorcapacity to take risk and the paucity of land tosow sole crops in different plots, small farmersmore often resort to intercropping as a defenseagainst risk than do large farmers.15 The pre-l iminary results on this aspect reported earlier(Jodha 1977b) are confirmed by data for threecrop years (Jodha 1979b); small farmers consis-tently used intercropping to a higher extentthan large farmers in all the villages exceptDokur and Shirapur. The difference betweenproportions of intercropping on small and largefarms (average of 3 years) ranged from 9 to 18%in different villages and was found statisticallysignificant. Greater use of sole cropping thanintercropping on small farms in Dokur andShirapur was highly significant because smallfarmers in these two villages had more irri-gation and postrainy-season cropping, which,

14.

15.

Analysis of data for precise quantif ication of theextent to wh ich intercropping reduce risk is still inprogress.Another reason for a higher proport ion of inter-cropping on smal l farms is that smal l farmers t ryto satisfy their prof i t , subsistence, and security-oriented needs from the same small piece of land.Intercropping serves this purpose better.

for the reasons discussed earlier, discouragedintercropping.

This suggests that any advance in intercrop-ping technology may benefit less well-endowedfarmers (and areas) more than the relativelybetter-endowed ones. It offers additional op-portunity to explicitly incorporate equity con-siderations into an agricultural researchstrategy by means of greater resource allo-cation to this area of research.

Complexity of TraditionalIntercropping Systems

Complexity and diversity are features of tradi-tional cropping systems (Table 4). If sole cropsand number of crop combinations in crop mix-tures are considered together, their numbersrange from 27 to 118 in different villages.Two-crop mixtures were popular in most vil-lages, but mixtures involving five to eight cropswere not uncommon.16 There were consider-able interregional differences in terms of theimportance of major intercrops. For instance, in

16. Mathur (1963) recorded more than 100 crop combi-nations in fields of crop mixtures in the Vidarbharegion of India; Norman (1978) recorded 230different crop combinat ions used in intercrop-ping in vil lages of northern Nigeria. This indicatesthat complexi ty of tradit ional intercropping is a general phenomenon.

16

T a b l e 4 . S o l a c r o p s a n d c r o p c o m b i n a t i o n s i n c r o p m i x t u r e s a n d t h e i r (%) s h a r e I n g r o s s c r o p p e darea i n s i x S A T I nd ian v i l l ages f r o m 1 9 7 5 - 7 6 t o 1 9 7 7 - 7 8 . *

Vil lage

Kanzara

Kinkheda

Kalman

Shirapur

Aurepal le

Dokur

Sole crops(No.)

22(27)b

19(17)34

(53)44

(82)21

(65)17

(79)

2 crops(No.)

17(26)15

(24)40

(24)23

(15)4

(6)4

(5)

Intercrops w i th mixtures of

3 crops(No.)

13(24)14

(41)28

(15)3

(2)2

(10)3

(2)

4 crops(No.)

11

(19)11

(17)13

(16)1

(1)-

2(7)

5 - 8 crops(No.)

4(4)1

(1)3

(2)-

11

(19)1

(7)

Total(No.)

67(100)

60(100)118

(100)71

(100)38

(100)27

(100)

a. Based on details from 180 sample farms in six villages. Village-level studies have been conducted in these villages since May 1975 (Jodha et al. 1977).

b. Figures in parentheses indicate the percentage share of crop/crop combination in gross cropped area during the 3-yearperiod.

Kanzara and Kinkheda (Akola District) cotton-based and sorghum-based intercrops weredominant. In other villages (except Dokur wheregroundnut -based intercrops dominated)sorghum-led intercrops were more important.Pigeonpea was an important component ofmixtures in all villages (Jodha 1979a).

The complexity of traditional intercroppingsystems is partly an outcome of farmers' infor-mal experimentation with crops that satisfytheir requirements and also fit the agriculturalenvironment of the region. The farmer is en-gaged in agriculture with multiple objectivesrelated to subsistence and employment of hisfamily and cattle, profit f rom farming, adjust-ment to drought risk, as well as potential andlimitations of his land.

To illustrate the points mentioned above,crop mixtures observed in the villages wereclassified into six categories on the basis of thespecific characteristics of the crops included ineach intercrop combination. The categorieswere defined on the basis of objectives theycould fulfil l and are briefly described underTable 5, which presents the proportions ofintercropped area covered by mixtures satisfy-ing different objectives. Since a given mixture

may satisfy several objectives, the six cropmixture categories are not mutually exclusiveand the percentages in Table 5 do not add to100.17

In the six villages the most importantcategories of crop mixtures (indicated by theirshare in total area of intercrops) were as fol-lows:C. Mixtures with different maturity lengths -The aim being to evenly distribute the laborrequirements of cropping and making fuller useof the environment (e.g. sorghum andpigeonpea).D. Mixtures of drought-sensitive and drought-resistant crops - Some examples are pearl mil-let with groundnut, or pigeonpea with cotton, toguard against moisture risk without foregoingthe benefits of good rains.E. Cash crop - food crop mixtures - Someexamples are cotton with sorghum or pearlmillet with groundnut, in order to simultane-ously satisfy cash and subsistence require-ments.

17. It is recognized that even if crops are not g rown asmixtures but as separate sole crops, some of theobjectives listed in Table 5 can still be satisfied.

17

Tab le 5. Propor t ions of d i f fe ren t categor ise of c rop mix t u res In t h e to ta l area of in tercropping Ins i x S A T Indian vi l lages (average o f 1 9 7 5 - 7 6 t o 1 9 7 7 - 7 8 ) . a

Crop mixture categoriesb

A. Special situationB. Self-provisioningC. Different matur i ty periodsD. Drought-sensit ive-

drought-resistantE. Cash crop-food cropF. Legume-nonlegume

Proport ion of different categories of crop mixtures in

Kanzara

(%)

29

58

727388

total area of intercropping in

Kinkheda

(%)

31184

815977

Kalman Shirapur( % ) ( % )

15 1218 1446 32

18C 2544 6159 40

Aurepalle

(%)

33671

13c

5384

Dokur

(%)

22979

415038d

a. Based on details from 180 sample farms in six villages. Village-level studies have been conducted in these villages since May 197S (Jodha et al. 1977).

b. The crop-mixture categories ere not mutually exclusive. Therefore the percentages are not additive. The basis of crop-mixturecategorization is as follows:Category A - Special situation: Mixture resulting from adding to the main crop of the plot a few other crops in order to adjust

to physical factors such as patches with salinity, depressions, infertile, gravelly soil, etc. (e.g., paddy combinedwith sorghum or pigeonpea).

Category B - Selt-provisioning: Mixtures having in addition to main crops of the mixtures, some crops like seasonalvegetables, tobacco, fiber crops, etc., seldom grown for the purpose of final harvests. They are harvested asand when family "self-provisioning" demands.

Category C - Different maturity lengths: Mixtures involving crops with different growth periods facilitating spread of peak(harvest) period labor requirement (e.g., sorghum and pigeonpea).

Category D - Drought-sensitive and Drought-resistant crops: Mixtures involving drought-resistant and drought-sensitive(or less drought-resistant) crops (e.g., groundnut and pearl millet).

Category E - Cash crop-food crop: Mixtures Involving cash crops and foodgraln crops (e.g., sorghum and cotton, castorbean and pigeonpea).

Category F - Legume-nonlegume: Mixtures involving legumes and nonlegumes (e.g., sorghum, pigeonpea, or greengram).

c. Bulk of the other mixtures consisted of only drought-resistant crops.d. Bulk of the other mixtures consisted of only legumes.

F. Legume-nonlegume mixtures - Examples

are sorghum or pearl millet wi th mung bean or

pigeonpea to fulfi l l fertil ity and rotation re-

quirements without sacrificing nonlegume

crops and for balancing the diet.

The other two mixture categories (A and B),

induced by self-provisioning requirements and

the need for adjustment to problems of soils,

are relatively less important in these villages.18

18. Cropping patterns in vil lages are not r igidly fixed.Depending on the quantum and t im ing of rainsand rotat ion requirements farmers do adjust theircrops and crop combinat ions. An analysis of therelat ionship between rainfall and cropping deci-sions using vi l lage level dai ly rainfall record andcropping patterns on sample fa rms, is currentlyunder progress.

The above picture demonstrates that tradi-

tional intercropping systems are complex and

diverse because they involve a conscious and

rational attempt by the farmer to adjust his

cropping pattern according to his need and

resource base. An important implication for

research on intercropping fol lows. While trying

to generate new, simple, and more productive

intercropping systems, considerations indi-

cated by mixture categories C, D, E, and F

should not be completely ignored. This would

mean ignoring the very clients for whom inter-

cropping technology is being generated.

W a t e r R e s o u r c e :T h e K e y V a r i a b l e

Without belittl ing the rationale of farmers' wis-

18

dom underlying the traditional system of farm-ing in SAT areas, it is clear the system seems tooperate more as an adjustment mechanismagainst factors causing low and unstable pro-duction rather than a dynamic enterprise show-ing possibilities for sustained growth. The cir-cumstantial evidence suggesting an assetdepletion/replenishment cycle (Jodha 1978)and aversion to risk associated with investment(Binswanger 1978) indicates the possibility ofpermanent underinvestment in SAT agricul-ture. The scope for dynamizing SAT agricultureis l imited for want of viable technological op-tions. Fertilizer-responsive HYVs have been thenew element that led to a production rise intraditional agriculture in recent years. ButHYV-based technology also works best whencomplemented by the requisite amount of soilmoisture. This brings us to a key physical factorthat can make traditional agriculture in SATareas more dynamic. Farmers, scientists, andpolicy-makers, of course, are not unaware ofthis.

During British Colonial rule, the traditionallydrought-prone areas in India received priority interms of irrigation projects, largely based onimport of water from other catchments.19 Thesubstantial irrigation investment in SAT areassince the early twentieth century did not helpbeyond creating a few pockets of prosperitywithin SAT regions where irrigation schemes,intended to irrigate and protect rainfed cropsagainst drought, ended up being used to irrigatesugarcane and paddy. Realization of the limitson ' imported water' as a solution to the prob-lems of low rainfall areas induced a search fortechnologies that would ensure maximum con-servation and efficient use of available mois-ture. A limited research effort of the 1930sgenerated what is commonly known as theBombay Dry Farming Technology. More con-centrated efforts were initiated during the early1970s when organizations such as the All IndiaCoordinated Research Project for Dryland Ag-riculture and ICRISAT came into existence togenerate relevant technological options for SATfarmers. Of the several approaches being tried,the principal one heavily emphasized by

19. For a detailed review of various technologicalapproaches and policies to reduce instability andensure growth of rainfed agriculture in India see

Jodha (1979a).

ICRISAT is management of soil and water on a small watershed basis.

W a t e r s h e d - B a s e d S y s t e mo f F a r m i n g

The basic philosophy behind the resource-centered (as against crop-centered) approach totechnology research is that the resource use inSAT agriculture should be environment-basedrather than individual holding-based. For thispurpose ICRISAT considered a watershed orcatchment to be the appropriate unit of re-source management and utilization (Krantz etal. 1977). To ensure conservation and effectiveutilization of wa te r—the scarcest of naturalresources for agriculture in SAT areas — a va-riety of measures are considered. Dependingupon soil type and slope, these measuresbroadly include necessary land smoothing;semipermanent graded broadbeds and furrowsto ensure full penetration of moisture, reduceerosion, and regulate runoff; grassed water-ways for improved drainage; and small dams ortanks to collect runoff water to be used forsupplementary irrigation during midseasondroughts, or for irrigating crops in the postrainyseason. The land- and water-managementmeasures are complemented by improved ag-ronomic inputs (Krantz et al. 1977).

There are three key elements of thewatershed-based system of farming relevant tothe present discussion. First, if the land man-agement is attempted on a watershed basis itcan ensure greater availability of moisturewhich, when complemented with improvedagronomy, can ensure higher and more stableproduction from the same land resources.20

Second, since water is the most l imiting naturalfactor in SAT agriculture, it should be used mostefficiently — i.e., on crops that do not requirelarge amounts of it — s o that a maximum areacan be covered with the limited water availablefrom runoff collection tanks in the watersheds.

The full-scale watershed technology is yet tobe tried in the villages. However, juxtapositionof some factors characterizing traditional ag-riculture and the elements of prospectivewatershed technology can give some idea of

20. For economic analysis of field-scale watershedexperiments see Ryan et al. (1979b).

19

the potential constraints on the prospectivetechnology.21

P r o b l e m o f G r o u p A c t i o n

The first problem arises f rom the existence ofseveral plots owned/operated by different own-ers on every miniwatershed in the villages(Table 6). The watersheds actually surveyed inICRISAT study villages, as well as by theDrought Prone Area Programme (DPAP) in dif-ferent states, suggest that the number of far-mers involved in a single miniwatershed canrange f rom 7 to 93. There is also considerablevariation in the size of holdings within thewatersheds. Similar variation could be ex-pected in the resource positions of the farmersinvolved in each watershed. The disappointing

2 1 . For a detailed discussion of these issues seeJodha (1975).

experience wi th cooperatives in India suggeststhat obtaining the agreement of all farmers in a watershed to its common and integrated useposes a question of group action among thefarmers.

Adoption of full watershed-based technologyon an individual farmer's land, as against thecontiguous plots owned by several farmersinvolving group action, may be difficult. Thereasons are the indivisibility of full watershed-based technology, paucity of individuallyowned plots or land fragments that are largeenough to form a miniwatershed, and thedifficulties involved in land consolidation.22

22. Of course, some components of prospectivewatershed technology could be adopted in parts.However the total impact in terms of resourceproductivi ty and conservation can be realizedonly if the who le package of technology isadopted (Krantz et al. 1977).

Table 6 . Deta i ls o f landhold lngs on smal l wa te rsheds in e igh t S A T Indian vi l lages. a

Village

Kanzara

Shirapur

Darphal

District(State)

Akola(Maharashtra)Sholapur(Maharashtra)Sholapur(Maharashtra)

Khanderajani Sangli

Krishnapurand Takli

G. R. Halli

Bayanapalle

Aurepal le

(Maharashtra)Dharwar(Karnataka)Chitradurga(Karnataka)Mahbubnagar(Andhra Pradesh)Mahbubnagar(Andhra Pradesh)

Totalarea(ha)

19.9

16.9

70.5

35.4

43.4

116.0

20.0

26.7

Farmsrepre-sented

onwater-shed

(No.)

13

13

30

10

29

93

30

7

Average Rangesize of of

plot ploton size

given onwater- water-shed shed(ha) (ha)

1.5 0.4-4.5

1.3 0.2-3.9

2.4 N.A.

3.5 N.A.

1.5 0.3-4.0

1.3 0.4-6.0

0.7 N.A.

3.8 1.3-10.0

Farmhold-ings

smallerthan

average

(%)

69

77

N.A.

N.A.

N.A.

47

N.A.

57

Soiltypes

onwater-s h e d '

MDV

DV

MDV, SV

MDV, SV

MDV

DA, SA

DA, SA

DA, SA

Annualrain-fall

(mm)

819

636

600

425

606

612

710

710

a. Detaits summarized from Sharma and Kampen (1977) and unpublished reports prepared during Training Program tor DPAPOfficers, organized jointly by All India Coordinated Research Program for Dryland Agriculture, Central Soils and WaterConservation Research and Training Institute and ICRISAT, April 10-17, 1977.

b. Soil types: MDV - Medium Deep Vertlsols; DV - Deep Vertlsols; DA - Deep Alfisols;SA = Shallow Alfisols; SV - Shallow Vertisols.

20

In order to justify a water harvesting andsupplementary irrigation technology, Ryan etal. (1979b) have estimated that the opt imumeconomic size of catchment or watershed inparts of peninsular India would seem to bebetween 8 to 16 ha. The size distribution ofindividually owned land parcels or fragments(Table 7) in ICRISAT's six study villages indi-cates that there are literally no plots with anyfarmer that could satisfy the economic (8 to 16ha) requirements of a miniwatershed for anindividual farmer. Even if the size requirementis reduced to 4 to 6 ha, in four out of six villagesone does not find more than 7% of individuallyowned land parcels that could, at least on anarea basis, qualify for a miniwatershed.23 Top-ographic information about the plots mayfurther reduce the percentage of fragmentssuited for treatment on a whole watershedbasis. Furthermore, practically all the largeplots under consideration were owned by largefarmers. Hence, for the small and mediumfarmers there may be no alternative to groupaction, if the complete watershed-basedtechnology, including provision for a runoffcollection tank, is to be adopted.

A rev iew s tudy of several agr i -cultural group organizations by Doherty andJodha (1977) suggested that besides sev-

23. It is not area alone but topography of the plot aswel l that determines its suitabil i ty as an integ-rated miniwatershed. However, such informat ionabout land parcels (Table 7) is not available atpresent.

eral other factors an easily perceivable, clear-cut, and high economic payoff is one conditionthat can induce farmers to participate in groupaction. It seems from the analyses performedon the research watersheds at ICRISAT Centersince 1975-76 that the new watershed technol-ogy does offer considerable additional profits,particularly on deep Vertisols.24

A l l o c a t i o n o f W a t e r R e s o u r c e s

As Table 8 shows, in the villages of Dokur andAurepalle, where traditional runoff collectiontanks are a principal source of irrigation, aroundthree-quarters of their gross irrigated area wasoccupied by paddy. Even in the villages wherethe extent of irrigation was only 5 to 13% ofgross cropped area, the bulk of the irrigationwas devoted to high water-consuming cropssuch as wheat, sugarcane, cotton, vegetables,etc. If the case of Kanzara, where hybrid sor-ghum was irrigated, is excluded, Kalman is theonly village where sorghum received a substan-tial proportion (30%) of the irrigation. This wasbecause wells did not have sufficient rechargeto support paddy or sugarcane. Intercrops didnot receive more than 10% of irrigation in anyvillage, once again confirming the results men-tioned earlier.

If the extent of irrigation is defined in terms ofintensity of irrigation (area irrigated multiplied

24. See Ryan et al. (1979b) for economicswatershed technology.

of

Table 7. Dis t r ibut ion of land f ragments by size on t h e sa mple farms in six S A T Indian vi l lagesdur ing 1 9 7 6 - 7 6 . a

Village

KanzaraKinkhedaKalmanShirapurAurepalleDokur

Totalfragments/land

parcels(No.)

10071

2161128783

% distr ibution of fragments in the ranges (ha) of

0.80(%)

282148533768

0.81-2.80(%)

574450394026

2.81-4.10(%)

915

2494

4.11-6.10(%)

620-4

142

a. Based on detai ls f r o m 180 sample-farms in six v i l lages. Vi l lage-level studies have been conducted in these vi l lages since May

1975 (Jodha et al . 1977).

21

Tab le 8. Percen tage share of d i f fe ren t crops in t h e gross i r r iga ted area in s ix S A T Ind ian v i l lages(average o f 1 9 7 5 - 7 6 a n d 1 9 7 6 - 7 7 ) . a

Crops

SorghumWheatPaddyGroundnutsPulsesf

VegetablesSugarcaneCotton/Castor beang

Other Sole Cropsh

Al l M ixed Crops i

.TotalTotal weighted irr ig.

area as % of GCA

Kanzara

28 c

56d

2654

----

100

6

(26)(58)

(2)(5)(4)(4)----

(100)

(22)

Proportion of different crops in gross irrigated area inb

Kinkheda

- -

44d (45)1 (1)

10 (9)25 (27)

2 (1)- -

9 (13)-

9 (4)100 (100)

3 (6)

Kalman

301964973

-

1310

100

8

(28)(23)

(5)(4)(5)

(11)(6)-

(13)(7)

(100)

(21)

Shirapur

915

1102

1222-

236

100

13

(3)(14)

(1)(10)

(1)(13)(39)

-

(17)(3)

(100)

(42)

Aurepal le

63

73--

5-

5-10

100

19

(6)(2) .

(78)e

--

(6)-

(2)-

(6)(100)

(66)

Dokur

_

-79e

20-

1--

1-

100

60

_

-(74)(24)

-

(1)--

(1)-

(100)

(307)

a. Based on deta i ls f r o m 180 samp le fa rms in six v i l lages. Vi l lage-level studies have been conduc ted in these vi l leges since May1975 (Jodha et a l . 1977). The sources of i r r igat ion are tanks and wel ls In Mahbubnaga r and Aurepa l le and we l ls In otherv i l lages.

b. Figures in parentheses ind icate t h e p ropo r t i on of each c rop in t h e gross i r r igated area recalculated us ing t h e intensi ty o fi r r iga t ion . The recalculated we igh ted gross Irr igated area is based on area i r r igated mu l t i p l i ed by number of I rr igat ions g ivento t h e same (who le) p lot . A l l i r r igat ion opera t ions fo r a g i v e n p lo t taking place w i th in a 10-day per iod have been t reated as onei r r iga t ion opera t ion to avo id the possib i l i ty of part ia l coverage of a p lo t by water be ing t reated as its fu l l coverage. The part ialcoverage m a y resu l t f r o m poo r a n d s l o w recharge in the i r r iga t ion we l l and the water -spreading me thods used in paddy f ie lds.In the case of paddy , th is m e t h o d tends to underes t imate the i r r igat ion Intensi ty, because wate r ing of paddy is a lmostcont inuous a n d the f ie ld Is a lways kept wet .

c. Hybr id s o r g h u m .d. Kanzara over 60 % and Kinkheda over 60 % HYV wheat .

e. HYV-paddy over 60 a n d 9 0 % respect ively in Aurepa l le a n d Dokur.

f . M u n g bean in Kanzara a n d K inkheda; chickpea in Ka lman a n d Shi rapur .g. Hyb r id co t ton in K inkheda ; castor bean in Aurepa l le .

h . Includes maize, sun f lower , ga rden crops in Ka lman a n d Shi rapur v i l lages, e n d f inger mi l le t in Dokur.

i . Excludes a l l vegetab les , mix tures a n d a l im i ted extent of sugarcane-vegetab le m ix tu res , Inc luded w i t h respect ive ma in crops.

by number of irrigations), the tendency towardsconcentration of the water resource on highwater-consuming crops, particularly in villageswi th little irr igation, becomes more clear. Forinstance, in Shirapur the share of sugarcane inirrigation increased f rom 22 to 39% once inten-sity of water use was considered. Correspond-ingly shares of sorghum and mixed crops de-clined from 9 to 3% and 6 to 3%, respectively.The case was similar in Kalman where sugar-cane, vegetables, and wheat gained at the costof sorghum, mixed crops, etc. Among the sixvillages, Dokur had the highest extent (60%) ofgross cropped area receiving water. Once wateruse intensity was considered, the weightedgross irrigated area exceeded three times thegross cropped area. Practically all of this area

was devoted to high water-requiring crops,particulary paddy. One can expect a similarpattern in the allocation of other inputs amongdifferent crops.

This raises a basic question about priorities inresource use on irrigated crops and irrigateddry (ID) crops in SAT areas. Social gains fromwater and other resources when used on IDcrops may be higher, but it is private benefit thatguides farmers' decisions about their resourceallocation. Thus, given the circumstances,coarse grains like sorghum and pearl mil let (twoof the five ICRISAT crops) wi l l always sufferneglect unless their low-value status is coun-tered by much lower cost of production andtheir competitiveness vis-a-vis traditionallyhigh-value crops is improved (Jodha 1973). This

22

again is a challenge for researchers. Any break-through in crop technology of SAT crops,reflected through low cost and high yield as wellas their improved competitiveness wi th othercrops, besides being an achievement in itself,could serve as a means of shifting resource (e.g.water) allocation in favour of these crops. How-ever, to the extent the gap between profitabilityof high-value and low-value crops persists(partly because of equal or stronger researchefforts on the former) one may not be left withany alternative to some form of social control ofwater to ensure its allocation to low-value cropsto better serve equity and efficiency goals.

Conclusions

The results, based on data f rom 3 crop yearsfrom six villages in three agroclimatic zones ofpeninsular India, have revealed the rationalebehind some of the traditional farming prac-tices.

In the deep Vertisol areas the practice offallowing land during the rainy season andplanting it in the postrainy season is a veryimportant practice, more so in the case of smallfarmers than large ones. Hence any technologi-cal advance facilitating rainy-season crops inmonsoon-fallow tracts can probably help smallfarmers proportionately more than large ones,as well as substantially adding to the areadouble cropped.

In the rainfed regions that do not have exten-sive deep Vertisol areas the traditional practiceof intercropping covers 35 to 73% of their grosscropped areas. The extent of intercroppingdeclines wi th increases in irrigation in villages.The small farmers again have a significantlyhigher extent of intercropping than large far-mers. This indicates that generation of a low-cost new technology for intercropping may helpless-endowed areas and farmers more than therelatively well-endowed ones. This is one of thefew opportunities where egalitarian objectivesin SAT areas could be achieved by technologi-cal means as opposed to institutional means,and it has significant implications for researchresource allocation.

As revealed by the number of crop combi-nations (as high as 84 in a single village), tra-ditional intercropping is highly complex. This ispartly an outcome of farmers' informal ex-

perimentation with crops that could satisfy theirrequirements and also fit the agricultural envi-ronment of the region. The multiple objectivesof the farmer such as security, profitability,employment and subsistence requirements ofhis family members and cattle etc., should betaken into account when evolving new inter-cropping technology.

The juxtaposition of requirements of pros-pective watershed-based technology and thefeatures of the t rad i t iona l system offarming — particularly the land ownership andusage pattern — gives an idea of the institu-tional constraints the technology is likely toface. Because integrated watershed-basedtechnology is indiv is ib le and becauseindividually-owned land parcels large enoughto constitute a composite miniwatershed arenot available, there seems no alternative togroup action that can ensure management ofland for higher productivity and conservationon a watershed basis. In order to induce groupaction among farmers, prospective watershedtechnology wil l have to be highly profitable.

A c k n o w l e d g m e n t

The author thanks J. G. Ryan, D. Jha, H. P.Binswanger, and M. von Oppen for their valuablecomments and suggestions in preparing this paper.He also thanks S. S. Badhe, V. Bhaskar Rao, M. J.Bhende, T. Balaramaiah, N. B. Dudhane, and K. G.Kshirsagar, the investigators who collected the dataused.

References

B A P N A . S . L. 1973. Economic and social implications ofgreen revolut ion: a case study of Kota District.Agro-Economic Research Center, Sardar Patel Uni-versity, Vallabh Vidyanagar, India.

BINSWANQER, H. P. 1978. Risk attitudes of rural house-holds in semi-arid tropical India. Economic andPolitical Weekly (Quarterly Review of Agriculture)13(25): A. 49 -A . 62.

23

DOHERTY, V. S., and JODHA, N. S., 1977. Condit ions for

group action among farmers. ICRISAT EconomicsProgram occasional paper 19, Patancheru A. P.India.

GHODAKE, R. D., and ASOKAN, M. 1978. Respondent

recategor izat ion for v i l lage- level studies ofICRISAT - A note. ICRISAT Economics Program,Patancheru, A. P., India. (Mimeographed.)

ICRISAT, 1979. International Workshop on Intercrop-ping, 10-13 January 1979, Patancheru, A. P. India.(Proceedings in preparation.)

JODHA, N. S. 1973. Prospects for coarse cereals:permanent constraints on jowar and bajra.Economic and Political Weekly, (Quarterly Review ofAgriculture) 8:A.145-A. 150.

JODHA, N. S. 1975. Watershed-based farm technologyand problem of group action (a research proposal).ICRISAT Economics Program, Patancheru, A. P.,India. (Mimeographed.)

JODHA, N. S. 1977. Land tenure problems andpolicies in the arid region of Rajasthan. In ICAR'sDesertif ication and its control. Presented at UNConference on Desertif ication, Nairobi, Kenya.

JODHA, N. S. 1978. Effectiveness of farmers' adjust-ment to risk. Economic and Political Weekly (Quar-terly Review of Agriculture) 18:A.38-A.48.

JODHA, N. S. 1979a. Dry farming technology:achievements and obstacles. In Agricultural de-velopment in India: policy and problems, eds. C. H.Shah and C. N. Vakil, Bombay: Orient Longman.

JODHA, N. S., 1979b. Intercropping in tradit ional fa rm-ing systems. Presented at International Intercrop-ping Workshop, 10-13 Jan 1979, ICRISAT,Hyderabad, India.

JODHA, N.S., ASOKAN, M., and RYAN, J. G. 1977. Village

study methodology and resource endowments ofthe selected vil lages in ICRISAT vil lage studies.ICRISAT Economics Program occasional paper 16,Patancheru, A. P. India.

KAMPEN, J. and associates, 1974. Soil and waterconservation and management in farming systemsresearch for semi-arid tropics. Pages 3 - 4 3 in Pro-ceedings, International Workshop on Farming Sys-tems, 18-21 November, 1979. ICRISAT Hyderabad,India.

KRANTZ, B. A., KAMPEN, J., RUSSEL, M. B., THIERSTEIN,

G. E., V IRMANI , S. M., WILLEY, R. W., and associates.

1977. The Farming Systems Research Program,ICRISAT, Hyderabad, India.

MALONE, C. CARL, 1974. Indian agriculture: progress inproduct ion and equity. New Delhi, India. Ford Foun-dat ion. (Mimeographed.)

MATHUR, P. N. 1963. Cropping pattern and employ-ment in Vidarbha. Indian Journal of Agricul turalEconomics 18:38-43.

N O R M A N , D. W. 1974. Rationalising Mixed Croppingunder Indigenous Condit ions: The Example ofNorthern Nigeria. The Journal of DevelopmentStudies 1 1 : 3 - 2 1 .

N O R M A N , D.W., and PRYOR. 1978. The Small Farmer in

Hausa Land, Michigan State University, Michigan.(Forthcoming.)

RYAN, J.G., SARIN, R., and PEREIRA, 1979b. Assessment

of prospective soi l - , water-, and crop-managementtechnologies for the semi-arid tropics of peninsularIndia. Presented at the International Workshop onthe Socioeconomic Constraints to Development ofSemi-Ar id Tropical Agriculture, 19-23 February1979, ICRISAT.

SHARMA, P. N., and KAMPEN, J. 1977. A feasibil i ty study

of improved technology for land and water man-agement under on-farm condit ions. ICRISAT Farm-ing Systems Research Program, Patancheru, A. P.,India.

24

Cropping Pat terns, Farming Pract ices,and Economics of Major Crops in Selected

Dry land Farming Regions of India

B. K. Rastogi*

Abstract

This paper presents an overview of the current level of dryland farming technology in

use by the Indian farmer, through a study of the cropping patterns, farming practices,

and economics of major crops in selected dryland farming regions of India. It is based on

the data collected under the Agro-Economic Research Scheme of the Drought-Prone

Areas Program at the All India Coordinated Research Project for Dryland Agriculture. The

information, collected from eight centers in the arid and semi-arid regions of India during

1976-77, shows that farmers in those regions operate at a very low level of technology.

In many locations, the preparatory tillage is casual and often inadequate. The use of

improved seeds, fertilizers, and plant protection chemicals is low. Tractors and threshers

are used only in some places. The animal manures available are spread too thinly; cash

crops have the first claim on them. Though the new dry-farming technology developed

at research centers has been found viable and feasible where the recommended

practices have been put to use, and though it is said to be neutral to scale, it has not yet

gained popularity with the farmers in the arid and semi-arid regions of the country.

The cropping pattern in any region is essentiallydetermined by experience gained over the yearson a given soil type, rainfall and climate, irri-gation potential, and economic considerationssuch as labor availability, prices, and marketing.In dryland areas, however, rainfall — or, moreprecisely, the amount and period of moistureavailability — overshadows other consider-ations because such areas are characteristicallysubsistence oriented. There are some excep-tions though, where cash crops that influencethe local economy are grown. These are discus-sed using farm-level data for 1976-77 derivedfrom the socioeconomic studies conducted bythe All India Coordinated Research Project forDryland Agriculture at eight of their 16 centers.

T h e S e t t i n g

Soil type and rainfall, two major determinants

* Senior Agricultural Economist, Al l India Coordi-nated Research Project for Dryland Agriculture,Hyderabad, India.

of the cropping pattern, vary greatly from loca-tion to location. Broad features of the regionsselected for analysis in this paper are given inTable 1. Their location is shown in Figure 1.

It is evident that cropping patterns in theseareas have been developed within the con-straints of low and erratic rainfall for a shortgrowing period (75 to 130 days) and soils of lowdepth and poor fertility.

Basic land statistics in Table 2 reveal that inmost of the dry farming areas cultivation hasextended to marginal and submarginal lands,raising the cultivated area from 85 to 100% ofthe total owned area. This is done to meetfamily needs of food and fodder from their ownland resources; it does not involve much cashexpenditure. The relatively low proportion ofcultivated area in Hyderabad reflects two as-pects: (1) as a rule, farmers leave a part of theotherwise cultivable area as grazing land eachyear, by rotation; (2) during the year 1976-77,some areas could not be sown because of longdry spells during the sowing period. Part of thecultivated area remained unsown at Nagaur aswell.

25

Tab le 1 . M a j o r so i l t ypes and ra in fa l l s ta t is t ics . a

Region

Ar idNagaurJodhpurNarnaulAnantapur

Semi-Ar idHyderabadBangaloreRahur i

Sholapur

Soil type

SandySandySandyRed loam/Sandy

Red chalkaRed loamBlack day /Red

laterit icBlack clay

Normalannualrainfall(mm)

310(20)366(20)428(26)583(33)

771(50)889(58)

520(31)678(41)

Normaldates ofonset of

southwestmonsoon

29 June29 June27 June

9 June

9 June5 June

10 June10 June

Normaldates of

wi thdrawal ofsouthwestmonsoon

21 Sept21 Sept23 Sept

16 Oct

10 Oct10 Nov

2 Oct2 Oct

Normal seasonalrainfall (mm)

Kharif

(Jan-Sept)

265(16)328(16)341(18)336(18)

574(37)483(32)

394(23)516(31)

Rabi

(Oct-Feb)

23(2)23(2)52(4)

175(10)

130(8)244(15)

102(6)120(7)

a. Data on ra infa l l no rma ls are based on 50 y e a r s f r o m 1901 to 1950. Dates o f onse t and w i t h d r a w a l o f sou thwes t m o n s o o n are

a lso based on 50 years f r o m 1921 to 1970. F igures in parentheses are the n u m b e r o f ra iny days.

Sou rce : Me teo ro log i ca l Un i t , D r y l a n d Research Project, Hyderabad .

Nagaur, Jodhpur, Sholapur, Anantapur, andRahuri are the regions where the irrigated areais low and the potential for further developmentis also l imited. At Narnaul, the area underirrigation appears to be relatively high, which iscontrary to the situation actually existing. Thenew minor irr igation/dug wel l schemes havecreated only a potential; sufficient water toirrigate the stipulated area is not available,leaving a part of the area nonirrigated. AtHyderabad, on the contrary, the proport ionatecropped area under irrigation during 1976-77was high because a part of the cultivable landwas either left fa l low for grazing purposes orcould not be sown for want of adequate mois-ture.

Double cropping is confined, either partiallyor whol ly, to areas receiving irrigation in therabi (postrainy) season, i.e., during October-February. Therefore, the intensity of cropping islow, ranging f rom 84% at Nagaur to 125% atHyderabad.

C r o p p i n g P a t t e r n s

Based on the experience gained w i th the land,soi l types, rainfall and irr igation facilities avail-

able, a wide variety of cropping patterns haveemerged in the different regions under review(Table 3).

Figure 1. DPAP agro-economic research loca-

tions.

REFERENCEAgro-Economic ResearchCenter

26

Tab le 2 . Ave rage v i l lage stat is t ics f r o m soc ioeconomic surveys 1 9 7 6 - 7 7 . a

Region

Ar idNagaurJodhpurNarnaulAnantapur

Semi-Ar idHyderabadBangaloreRahuriSholapur

Number o fhouseholds

994081

106

101116115

96

Totalowned area

(ha)

669184238778

358

354567611

Cultivatedarea (% tototal area)

8488

10070

60868590

Irrigatedarea (% tocult ivated

area)

2.79.8

23.38.8

27.017.311.47.1

Intensity ofcropping (%)

84105123108

125100112105

Averageholdingsize (ha)

6.764.602.945.16

3.553.054.195.74

a. Source: Farm structure studies conducted in 4 to 6 villages in each area in clusters of 2 to 3 villages each. Households arerepresented In proportion to total area.

Table 3 . Ave rage v i l lage cropping pa t te rns f r o m soc ioeco nomic surveys ( 1 9 7 6 - 7 7 ) .

Region

Ar idNagaur

Jodhpur

Narnaul

Anantapur

Semi-Ar idHyderabad

Bangalore

Rahuri

Sholapur

Totalcroppedarea (ha)

469

171

293

590

258

305

539

578

Kharif:

Rabi:

Kharif:

Rabi:

Kharif:

Rabi:

Kharif:

Rabi:

Kharif:

Rabi:

Kharif:Rabi:

Kharif:Rabi:

Kharif:

Rabi:

Percentage cropped area in different seasons(cropwise)

Pearl mil let 45, sorghum 11, moth 6, green gram 6, sesamum 13,cluster beans 6, others 2.Wheat 4, chickpea 4.

Pearl mi l let 56, green g ram 10, mo th 8, sesamum 7, cluster beans 6,others 9.Wheat 3, others 2.

Pearl mil let 30, pearl mi l let + cluster beans 13, cluster beans 6,sorghum 1.Chickpea 33, wheat 8, mustard 7, barley 1.

Paddy 6, groundnut 24, setaria 35, pearl mi l let 12, sorghum 7, horse

gram 5, others 4.Paddy 1, groundnut 3, wheat 1, sorghum 1, others 1

Paddy 16, castor 20, sorghum 9, pearl mi l let 1, sorghum + red gram22, pearl mi l let + red gram 2, sorghum + pearl mi l let 1, others 8.Paddy 10, sorghum 6, others 4.

Paddy 12, sugarcane 3, ragi 65, horse g ram 6, maize 6, others 9.No crop is possible.

Pearl mi l let 22, green gram 8, groundnut 2, others 4.Wheat 2, sorghum 6.

Paddy 4, pearl mi l let 2, maize 2, red gram 12, groundnut 13, others

6.Sorghum 58, wheat 2, chickpea 1.

27

Neutralizing the cropping pattern variationsdue either to differences in location in a givenarea or to farm sizes, it is observed that in thearid regions having sandy shallow soils (Nagaur,Jodhpur, and Narnaul) the predominant kharif (rainy season) cereal crop is pearl millet. Sor-ghum is cultivated mainly to meet fodder re-quirements. Pulses (green gram and moth) andalso sesame — an oilseed crop — are grownonly in the Nagaur and Jodhpur areas. Clusterbeans are also gaining importance in the ag-ricultural economy of the region and cover 5 to6% of the cropped area. During the rabi (post-rainy) season, chickpeas are important in theNarnaul region, covering more than one-thirdof the cropped area. Cultivation of wheat isundertaken on the wet (irrigated) land. If irri-gation water is not adequate to cover the entirearea, a part is diverted to the cultivation ofmustard, which requires in most cases only onepresowing irr igation. Other crops of economicimportance, though occupying relatively smallareas, include castor at Jodhpur and groundnutat Nagaur.

In the Anantapur area, wi th red sandy loamsoils, cultivation of minor millets is more impor-tant than pearl mil let or sorghum. The mainmillet crop is setaria, a small-seeded millet oflittle significance in other arid areas. Groundnutis of importance in the region, covering aboutone-fourth of the cropped area. Crops of pearlmil let or sorghum are usually grown as mixedcrops wi th either red gram or castor. Horsegram is normally cultivated to meet fodderrequirements. Paddy is g rown on irrigated land.

In the semi-arid Hyderabad region, theoilseed castor and sorghum are importantkharif (rainy season) crops in the red chalkasoils. Sorghum is usually grown as a mixedcrop either wi th red gram or pearl mil let, andpearl mil let is cultivated as a sole crop on a l imited area. Other crops of these drylandsinclude horse gram, ragi (finger millet), tobacco,and vegetables such as tomato, lady's f inger(okra), and brinjal (eggplant). Paddy is g rown onirrigated lands during both rainy and postrainyseasons. Sorghum is cultivated on relativelyheavier soils as a nonirrigated crop in thepostrainy season.

Ragi is the principal crop in the Bangaloreregion, covering about two-thirds of the crop-ped area, mainly on the red loamy soils. Horsegram is the important pulse crop of the area. 28

Maize is an introduced crop and is grown on a l imited area. Other dryland crops include cow-peas, castor, groundnut and vegetables. Paddyand sugarcane are irrigated crops. Croppingduring the postrainy season is not feasibleunder rainfall conditions at Bangalore.

In the black soil areas of the Rahuri andSholapur regions, postrainy-season sorghum isthe principal dryland crop. This is grown onmedium/deep soils wi th a large moisture-holding capacity. However, on the red lateriticsoils, pearl mil let and green gram are grownduring the rainy season to meet the cereal andpulse requirements in the Rahuri region. In theshallow soils of the Sholapur area, red gramand groundnut are usually cultivated as solecrops during the rainy season.

On the basis of the above, it is concluded that:1. In most dryland farming areas, double

cropping and relay cropping are not prac-ticed.

2. Rainy-season crops are more important inthe economy of Nagaur, Jodhpur, Anan-tapur, Hyderabad, and Bangalore regions.

3. Chickpea at Narnaul and sorghum atRahuri and Sholapur are the importantpostrainy-season crops.

4. The predominant foodgrain crops in eachregion vary wi th rainfall and growing sea-son, e.g. pearl mil let (Jodhpur, Nagaur,and Narnaul), sorghum (Hyderabad,Rahuri, Sholapur), setaria (Anatapur) andragi (Bangalore);

5. Intercropping is l imited to a few locationsonly and consists mostly of crops wi thdifferent durations, e.g. combinations ofpearl mil let and cluster beans at Narnauland s o r g h u m and p igeonpea a tHyderabad.

6. Cropping intensity is low; at t imes, it is notpossible to sow the entire area because oflong dry spells during the sowing period,as at Narnaul and Hyderabad.

F a r m i n g P r a c t i c e s a n dE c o n o m i c s o f M a j o r C r o p s

Discussion of the farming practices fo l lowed,cash investment made, and consequent returnsfrom irrigated lands would be beyond the scopeof this presentation. Nevertheless, it is noticedthat irrigated lands, irrespective of their share in

the cropped area, attract more attention interms of both management and investment,resulting in higher net profits.

Farming Practices

N a g a u r . The fields are first cleared of bushesand other vegetation before the onset of themonsoon, which is fol lowed by plowing, har-rowing, and planking with a tractor-drawn cul-tivator when the soil is saturated by rains.Sowing is also done wi th a tractor-drawn cul-tivator through a polythene/rubber pipe withiron funnels attached to til lers. This is a recentchange and has been accepted widely in theregion. Cattle manure is applied to drylandcrops but the average rate (400 kg/ha) is verylow. Interculture is carried out by manual laborwi th the help of kasies (blade harrows) but thereare also instances of running a bullock-drawnplow between the rows. Hardly any plant pro-tection measures are used in the case of rainfedcrops. The earheads are picked first andthreshed by tractors or cattle and the straw iscut later at leisure.J o d h p u r . Preparatory til lage is carried outmore commonly wi th tractor plows than wi thbullock-drawn plows f rom mid-June to earlyJuly for rainy-season crops and f rom late Sep-tember to the end of November for postrainy-season crops. Local varieties are still predomi-nant but, during the last few years, the use ofimproved varieties of pearl millet (BJ-104) andgreen gram (S-8) has been increasing. Farm-yard manure in small quantities is applied todryland crops, but the use of inorganic fertilizeris restricted to a few progressive farmers only.One hand weeding and one hoeing are done inthe case of peart millet, but only hand weedingis done in the case of green gram, moth andcluster beans. The common tool used for inter-culture is known as kasola. Kharif (rainy season)crops are harvested f rom mid-September tomid-October and rabi (postrainy season) cropsfrom the end of March to mid-Apri l . Manuallabor picks the ears/pods from standing cropsfirst and then harvests the remaining parts ofthe plants for fodder. Threshing is usually car-ried out by tractor-operated threshers.N a r n a u l . Postharvest ti l lage is uncommon inthe region. The preparatory til lage starts afterthe onset of the monsoon. Most of the farmersuse a wooden desi (indigenous) plow for pre-

paratory tillage. Some use tractor harrows also.Sowing of pearl millet — either pure or mixed

with cluster beans and green gram — begins inmid-June and continues into early July. Chick-pea, the important dryland crop in the rabi season, is sown during late September andearly October. Local varieties of seed are mostlyused. Sowing is done in almost all the crops bythe pora method, using a country seed drill wi tha desi plow.

No farmers use fertilizers on any drylandcrops. Farmyard manure is applied, however, toa small area under pearl millet and mustardcrops. Interculturing is done commonly in thearea for pearl millet and pearl millet mixed withclusterbeans and green gram, but rarely in thecase of cluster beans and chickpea. An ironkasola is the implement used. In all cropsharvesting is done using sickles, with threshingdone by human and cattle power.A n a n t a p u r . Preparatory tillage consists ofplowing and harrowing after the onset of themonsoon. In this region, some farmers plow theland every year while others plow only inalternate years. The regularity of plowing de-pends on the depth of the soil, crop to be grown,and availability of bullocks. Local varietiesdominate and sowings are done with indige-nous seed drills (mostly in August during1976-77 because of late onset of the rains; in a normal season, sowings occur during June-July). Farmyard manure (FYM) and sheep-penning are the traditional sources of plantnutrients. Groundnut receives preference overother crops in the application of FYM. Fertilizeris rarely used for dryland crops. Interculturingwith blade harrows is practiced by most far-mers. Handweeding is also practiced in the caseof groundnut. Grain crops are harvested withsickles and groundnut wi th blade harrows.Crops suffer heavily due to moisture stress andthis also makes harvesting of groundnutdifficult. Threshing is carried out with humanand bullock labor.

H y d e r a b a d . Preparatory tillage for sorghumand pearl millet, whether sown as sole crops oras base crops in mixed cropping systems,consists of two to three blade harrowings dur-ing premonsoon showers, while castor and rabi sorghum receive thorough preparatory tillagewith two or three plowings. Sorghum and pearlmillet are sown immediately after a good rain,f rom late May to mid-June. Local varieties of

29

seed for sorghum and pearl mil let are common.The improved variety of castor, Aruna, is be-coming popular wi th farmers, although thelocal variety still occupies almost 50% of thearea. Farmers use the same seed rates ofsorghum and pearl millet in mixed croppingsystems as in pure crops to maintain full plantpopulation. Castor sowings are usually done inJuly. Sowings are done both by the khera method (dibbling seeds in the furrows openedby desi plows) and by the 'pora' method (coun-try wooden seed dril l). Fertilizers are not usedand application of farmyard manure is mainlyrestricted to castor and rabi sorghum. One ortwo intercultivations in the case of sorghum andpearl mil let, and two to three in the case ofcastor, are done with blade harrows. Hand-weeding is not done for cereal crops. Aboutone-third of the area under castor is subjectedto handweeding. Generally, no plant protectionmeasures are used. Harvesting of castor is donein three to six pickings.B a n g a l o r e . Preparatory til lage for kharif crops is undertaken with the rains received fromApri l to June, using implements such as ironplow, wooden plow, heggunta (wooden imple-ment wi th two to six spring plated tines), orharrow. Use of local seed varieties is commonin the case of all the crops except maize. Theseed rate per hectare of ragi varies with themethod of sowing — 25 kg for broadcasting, 15to 20 kg for dri l l sowing, and 8 to 10 kg fortransplanting. Farmyard manure is regularlyapplied by farmers, but fertilizer use is l imited toimproved varieties and to less than 50% of therecommended levels. Thinning in ragi is widelypracticed by passing a kunte (a harrow wi th twoto four closely spaced tines) across the rows.Normally, ragi is harvested at one t ime, butwhenever improved varieties are used, harvest-ing is spread over several periods.R a h u r i . Flowing is done as a rule f rom Marchto June fol lowed by two or three blade harrow-ings for kharif as well as rabi crops. Mostly, localvarieties are used. Green gram is sown first andthen the farmers move to pearl millet. Rabi sorghum sowings are completed during thefirst fortnight of September. The green gramcrop receives more farmyard manure than pearlmillet and rabi sorghum. One hoeing is com-monly done in the case of pearl mil let and rabi sorghum and one handweeding is widely prac-t iced in the case of rabi sorghum and green

gram. Plant protection and fertilizer appli-cations are conspicuous by their absence in allcrops.

S h o l a p u r . Preparatory til lage starts in sum-mer wi th surface ti l lage using a blade harrow.Plowing is done only in fields where groundnutis to be sown, using a reversible iron plow orimproved country plows. Farmers generally uselocal varieties except in the case of pearl millet(occupying a small area) whose hybrid cultivarsare accepted by farmers. Sowing is usuallydone wi th a three-coulter local seed dri l l . Or-ganic manures are used for cash crops likegroundnut; fertilizers are not used at all. Rabi sorghum sowings are evenly distributed overthree periods, late August to early September,mid-September, and late September to earlyOctober. One or two intercultivations are donewith slitblade hoes drawn by a pair of bullocks.Handweeding is restricted to groundnut.Harvesting is done by cutting out the plantsand threshing wi th manual beating or animaltrampling.

From the above review of cropping patternsand practices, it is evident that farmers in thearid and semi-arid tropical regions are stilltradit ion-bound and are operating at a very lowlevel of technology. In many regions, prepara-tory til lage is casual and often inadequate. Theuse of improved seeds, fertilizers, and plantprotection chemicals is extremely low. Use oftractors and threshers is popular in the aridregions of Jodhpur and Nagaur. At other places,farmers depend largely on bullocks, which maybe in short supply. The animal manures avail-able are spread thinly and cash crops have thefirst claim on them.

Costs and Returns

With this background of the cultural and as-sociated cultivation practices fol lowed by far-mers in different regions, it would be of interestto examine the investment level for differentcrops/regions and the consequent returns. De-tails of average costs and returns for majordryland crops per hectare are given in Table 4.

Investment in seed, manures, fertilizers andplant protection measures is generally very low,though there are wide variations between cropsand regions. As a proport ion of working capital,the value of these cash inputs varied f rom 6 to22% at Nagaur, 20 to 30% at Jodhpur and 5 to

30

31

13% at Narnaul in the arid region. Pearl millet,the major cereal crop, accounted for 30% of thetotal working capital at Jodhpur, about 15% atNagaur and nearly 5% at Narnaul. The pro-portionate expenditure on cash inputs for clus-ter bean was also higher at Jodhpur thanNarnaul. Cash costs for sesame at Nagaur andchickpea and mustard at Narnaul accounted for22%, 14%, and 6%, respectively, of workingcost. At Anantapur, the cash investment forsetaria was nearly 7% of working cost andvaried between 32 and 58% for groundnut.

In the semi-arid region of Bangalore, the cashinvestment for ragi was nearly 25% of the totalworking cost, varying between 20 and 28% indifferent villages. In the Hyderabad region, cashexpenditures were nearly 40% for castor, 28%for sorghum, 14% for pearl millet, and nearly34% in the case of sorghum plus red gram.Interestingly, variations across locations werehigher at Hyderabad than elsewhere. Cash in-vestment varied between 26 and 55% for castor,7 and 33% for sorghum, and 20 and 59% forsorghum plus red gram. At Rahuri andSholapur, which are principally rabi sorghumareas, cash investment was nearly 9% of theworking cost at Rahuri and about 13% atSholapur for sorghum. No great variations werenoticed. Further, the expenditure was limited toonly 3% for pearl millet. It varied between 7 and28% (average 16%) for green gram. AtSholapur, cash expenditures on groundnut andred gram accounted for 50 and 15% of totalworking costs, respectively. This varied be-tween 28 and 70% for groundnut and between 9 and 29% for red gram over locations in theregion.

Considering the overall situation, cash in-vestment ranges between 39 and 45% of totalworking capital for oilseed crops such asgroundnut and castor. It ranges between 12 and16% for cereals and between 13 and 19% forpulses. It is as low as 3 to 4% for millets.Surprisingly, it is also low (5%) for mustard(oilseed crop) at Narnaul.

The investment in cash does not appear tohave any relationship to the size of the farm.However, the high-value crops of theregions — such as sesame at Nagaur, pearlmillet at Jodhpur, chickpea and mustard atNarnau l , g roundnu t a t Anantapur andSholapur, castor and red gram at Hyderabad,red gram at Hyderabad and Sholapur, and

green gram at Rahuri — were subjected to rela-tively higher cash expenditures than the othercrops. The breakup of cash investment intovarious components such as seed, manures,fertilizers, and plant protection measures alsosuggests that the high-value crops generallyreceived more manure/fertilizer and plant pro-tection measures than other crops — even thecereals — in the region (Table 5).

Thus it seems that plant protection measureswere not used for most crops and, whereveradopted, were inadequate. The amount spenton manures/fertilizers was also small whenviewed in relation to the corresponding pricesof manure and/or fertilizers. In some cases itappears that application of manure was nothingmore than an apology. At Hyderabad, cropswere better fertilized than in the other regions,though not adequately.

Under farmers' management, as reflected bythe cultivation practices fol lowed and the levelof inputs used, the yields of all crops consideredat all locations were extremely low. However, inthe arid region, yield levels obtained at Narnaulwere higher than those obtained either atNagaur or Jodhpur. Crops at Anantapur virtu-ally failed because the season was extremelyunfavorable. Further, the yields of pearl millet,though low in the arid region, were two to threetimes higher than those obtained either atHyderabad or Rahuri in the semi-arid region.But the yields of sorghum in the arid regionwere about half of those obtainable in thesemi-arid region even though these were low.

Thus it is clear that farming in these drylandareas has been designed for low investment interms of practices fol lowed and inputs used,resulting in low yields. Based on the economicrationale, such low yields hardly have a marginof profit sufficient to sustain a farmer and hisfamily.

R e c o m m e n d e d T e c h n o l o g y

The remedy for the malady of low crop yieldsand low profits in dryfarming areas lies in recenttechnological advances, which have convinc-ingly established the superiority of the recom-mended package of practices over the tra-ditional practices. Table 6 gives the details ofaverage yields obtained, costs incurred, andconsequent returns per hectare wi th the re-

32

Tab le 5 . Compos i t ion o f cash Inputs under f a rmers ' presen t pract ices.

Region/Crop

Ar idJodhpur

Pearl mil letGreen gramCluster beans

NarnaulPearl mi l letCluster beansPearl mil let + Cluster beansChickpeaMustard

Semi-AridHyderabad

SorghumCastorPearl mi l let

RahuriSorghumGreen g ram

SholapurSorghumGroundnutRed gram

Value of inputs (Rs/ha)

Seed

92113

1333235729

235532

1527

20329

21

Manure/ferti l izer

3920

9

4Nil

2Nil

6

1038637

327

334

Nil

Plant protection

NilNilNil

NilNilNilNilNil

Nil1

Nil

NilNil

1NilNil

Total value

484122

1733255735

12614269

1854

24363

21

commended practices under similar ag-roecological and socioeconomic environmentsas are considered in the preceding pages.

It is observed that, in all cases, the recom-mended practices gave yields many timeshigher than those obtained with the traditionalmethods of farming (Table 4). Even in Anan-tapur where, because of an extremely poorseason, crops under the farmers' level of man-agement almost failed, crop yields with therecommended practices were higher — thoughnot enough to cover the cash inputs or thecharges for the use of human/bullock labor, etc.

As one would expect, the share of cash inputsin total working costs with the recommendedpractices was much higher than with traditionalpractices, ranging between 35 and 67% consi-dering crops and locations together. It may alsobe relevant to point out that in the additionalcost, the share of fertilizers was considerable,ranging from 60 to 70%. The seed cost was also

higher with the recommended practices — 12 to15% of the additional cost. The share of theplant protection measures was 5 to 7% and therest of the additional cost was accounted for bythe use of extra labor needed for adequatepreparatory tillage, interculturing, harvesting,and threshing the additional output. A perusalof investment in inputs and the output obtainedreveals that, in the arid region alone, cashinputs were more intensively used at Narnaulthan at either Nagaur or Jodhpur. Conse-quently, the output per hectare was muchhigher at Narnaul. For example, the additionalinvestment in pearl millet ranged betweenRs 250 and 300 per hectare, with an averageadditional output of 650-675 kg at Nagaur/Jodhpur. The additional investment in inputs atNarnaul for pearl millet was over Rs 700/ha withan average additional yield of 1250 kg.

At Hyderabad, the additional investment ininputs was about Rs 250/ha for sorghum and

33

34

Rs 175/ha for castor and pearl millet. The addi-tional yield obtained ranged between 700 and900 kg for sorghum and was 330 kg for castor.With the recommended practices, the yield ofpearl millet increased by 1000 kg/ha. An addi-tional investment of Rs 400 in ragi at Bangaloreenabled the harvest of additional outputs rang-ing f rom 1200 to 1300 kg. Both at Rahuri andSholapur, an additional investment of overRs 170/ha was required for rabi sorghum; thisresulted in additional yields ranging between400 and 600 kg.

Thus, the cost of cultivation with recom-mended practices was high but the relative netprofits were still higher. The net profit rangedbetween Rs 450 and Rs 1100 per hectare for therecommended practices in most cases, leadingto additional net profits ranging f rom Rs 200/hato about Rs 1000 over and above the existinglevels.

One has to be aware of the inadequacy of thedata presented. They pertain to only 1 year, butthe year was about normal at most places andseveral soil and climatic regions were sampled.It may still provide an overview of the currentlevel of technology in use by most farmers andthe new dryfarming technology recently de-veloped at research centers. The recommendedtechnology is viable and feasible. It is also saidto be neutral to scale. But it is still to gainpopularity with farmers in the arid and semi-arid regions of the country. It remains to befurther diagnosed whether it is the low-resource base, socioeconomic inertia, or riskaversion that is responsible for the slownesswith which the technology is moving.

S u m m a r y a n d C o n c l u s i o n s

The land-utilization pattern in these eight re-gions suggests that though the area under culti-vation has been extended to marginal andsubmarginal lands, the intensity of croppingremains low, ranging between 84 and 125% atvarious locations. Double cropping is l imited toareas wi th irrigation during the postmonsoonseasons (October-February). In rainfed areas,cropping patterns have been essentially de-veloped within the constraints of low and erraticrainfall for a relatively short growing periodranging from 75 to 130 days. Considering thedifferent categories of farms and locations to-gether, it is observed that the principal food-

grain crop depends on the rainfall and thelength of the growing season. For example,pearl millet is important in Jodhpur, Nagaur,and Narnaul; sorghum is important in theHyderabad, Rahuri, and Sholapur regions whilesetaria at Anantapur and ragi at Bangalore areimportant.

The farmers in these arid and semi-arid areasare still tradition-bound as regards the farmingpractices they follow. They are operating at a very low level of technology.

Summer plowing is not practiced except insome cases at Rahuri and Sholapur. In manyregions preparatory tillage is casual and in-adequate. For example, in areas like Anantapur,plowing is done once in 3 to 5 years, dependingupon the soil depth, crop to be grown, and thebullock power available. The use of improvedseeds, fertilizers, and plant protection measuresis negligible. Even the farmyard manure avail-able is spread thinly and the farmers prefer touse it only for cash crops. Tractors andthreshers are becoming popular in the regionsof Jodhpur and Nagaur. Investment in cashinputs per hectare does not bear any relation-ship to the size of the farm.

The high-value crops of the region attractsignificant investment per hectare in cash in-puts compared to other crops of the region.Thus, within the constraints of low investmentin terms of practices followed and inputs used,the yields of crops in all locations are low and itis doubtful that such low yields generate a margin of profit sufficient to sustain a farmerand his family.

Recent research advances are capable ofremedying the malady of low crop yields andlow profits in rainfed farming areas but at a relatively higher investment. Undoubtedly,profits will be higher. However, despite the factthat the new technology seems viable andfeasible, farmers are reluctant to undertakeinvestments required to adopt the technology.Perhaps their low resource base and risk aver-sion are the two principal reasons responsiblefor the low adoption of the technology. How-ever, the real constraints to the adoption of thetechnology are yet to be ascertained.

Acknowledgment

My grateful thanks are due to Dr. S. L. Chowdhury,Project Director, Al l India Coordinated Research Pro-

35

ject for Dryland Agr icul ture, Hyderabad, for suggest-ing improvements in the original draft. I also thankMr. K. P. C. Rao, Agricul tural Economist, for consulta-t ions wh i le preparing the paper. The help receivedf rom Mr. Shyam Sunder and Mr. B. K. Gandhi , SeniorComputors, in the compi lat ion of data is thankful lyacknowledged.

36

D i s c u s s a n t ' s C o m m e n t s

H . R u t h e n b e r g *

Farming systems work may have different ob-jectives. Since this is an ICRISAT workshop, I assume that farming systems work seeks toenhance the relevance of biological scienceresearch work. I wi l l discuss the various paperswi th this objective in mind.

Drs. Abalu and D'Silva submitted a paper thatdeals with a relevant subject in our context.They provide evidence f rom northern Nigeriathat:

1. Mixed cropping produces higher grossand net returns than sole cropping.

2. Actual cropping patterns are rather closeto those that should prevail if the mini-mization of income variability is the ob-jective.

3. Actual cropping patterns are consistentwi th the objective of providing sufficientcalories and protein for the families.

The authors conclude that there is rationalityin what smallholders do. There is no doubt thatthey submit strong evidence to support theirconclusions. It should be considered, however,that more information would be needed toreally verify the hypothesis of an " incomevariability minimizat ion" strategy. We wouldhave to compare actual farm plans wi th optimalfarm plans under several strategies.

Also it might be questioned whether theunderlying assumptions of the linear pro-gramming models that they employ remainacceptable under the condition of rapideconomic and cultural change. In particular, I would question the idea of a labor constraint inthese farming systems where less field work isdone per family than in most other Africanfarming systems. For more than a decade I havebeen among those who emphasized that labor

* Institut fur Landwirtschaft l iche Betriebslehure,Universitat Hohenheim,Germany.

7000 Stuttgart 70, West

The editors regret to note the unt imely death ofProfessor Ruthenberg in Ju ly 1980. He made sub-stantial contr ibut ions to the understanding of tradi-t ional fa rming systems.

problems are crucial in tropical smallholdingsand this is generally accepted. The history ofscience indicates, however, that there is reasonto be cautious about lines of thought that aregenerally accepted. The word "constraint"originates from technical sciences and has re-cently been introduced into economics. Its pre-cise meaning is rather stringent. My experienceindicates that on African small farms we rarelyface labor constraints but rather elastic andoften kinked labor supply curves. A result basedon constraints should therefore not be acceptedunless there is much additional evidence thanthe results of normative linear programmingexercises.

Drs. Abalu and D'Silva not only present theirfindings but also give directives for research onthe basis of these findings. The adoption ofinnovations in northern Nigeria has been slow.Most recommendations are based on experi-ments with sole crops. They point to the ratio-nality of mixed cropping and argue that re-search should work for improved mixed crop-ping systems.

I fully agree with the criticism of the authors.Past agricultural research in the tropics ignoredthe rationality of mixed cropping. But I cannotfollow their argument that the presence ofmixed cropping and its rationality in the exist-ing setting is sufficient reason to invest researchresources in it. It is a fallacy to assume thatinformation about the existing systems pro-vides conclusive evidence about the directionplant science research work should take:

1. The fact that a certain crop predominatesin a given area is insufficient for researchresource allocation. Dr. Sanders, for in-stance, argues convincingly in his paperfor this Workshop about the BrazilianSertao that sorghum should be substi-tuted for the existing maize.

2. Mixed cropping may be the best in thetraditional setting. With new varieties solecropping may be better.

3. The fact that there is a food shortage in a given area is not sufficient reason to invest

37

in food crop research. It may be thatpromotion of cash crops produces morewelfare for the people concerned.

Certainly, no major research program in ag-riculture should be started without a fair under-standing of the existing farming systems. Farm-ing systems work as submitted by Drs. Abaluand D'Silva is indispensable. But if we only lookat what exists we may overlook the much betterchances that perhaps are offered by somethingentirely new.

It is useful in this context to distinguishbetween two basic strategies in the allocation ofresearch resources:

1. The improvement approach aims at thebetterment of existing systems. Relevantare such types of agricultural researchwhich produce innovations that can beadopted by farmers without major ad-ditional changes.

2. The transformation approach concerns a completely new set of production pat-terns. The existing is discarded as in-efficient and not wor th improving, giventhe new possibilities. The replacement ofZebu cattle by exotic cattle for dairying inKenya, and the development of high-yielding varieties are successful examplesof the transformation approach.

Clearly, both strategies have to be consider-ed. The important point in the context of ourworkshop is that farming systems analysis pro-vides information which is indispensable for thedesign of research strategies. However, theknowledge of the biological science researchworkers about possible "breakthroughs" intheir field is also indispensable. Informationabout both aspects has to go into thedecision-making process.

Let me turn to the next paper, submitted byDr. Jodha. It provides essential informationabout Indian cropping systems in areas withmonsoon fal lowing and about watersheds. Theimportant conclusions are that monsoon fal-lowing is there for definite reasons, that mixedcropping predominates on rainfed land, particu-larly on small farms; that farmers concentratelimited water resources on high water-consuming crops; and that watershed de-velopment requires group action. The researchimplications are well stated: innovations haveto be produced that would al low monsooncropping on Vertisols, and the competitiveness

of upland cropping has to be improved in orderto make the reallocation of water resources a profitable proposition.

Dr. Jodha's presentation indirectly impliesthat farming systems research in the context ofa research station has to proceed through twostages:

1. Baseline surveys to inform about the pre-sent situation before innovations from theresearch program are introduced. Thisinformation is needed for the initial re-search strategies.

2. The analysis of farming systems that un-dergo changes. This is the innovation-testing stage wi th a well organized feed-back to adjust ongoing biological-sciencework.

The latter type farming systems research israther different f rom farming system researchwork for other purposes:

1. It has to be much more specific. Totalfarms are too large a unit. Dr. Jodhaalready points to the need for plot-histories, which provide information oneach plot.

2. It has to be geared to hypothesis testingand hypothesis f inding at the same time. Itis certainly not enough to describe thefarming system wi th no specific objectivein mind. Precise hypotheses have to bestated and tested. At the same t ime re-searchers have to look for new and betterhypotheses.

3. Farming systems work of this kind re-qu i res m e a n i n g f u l c lass i f i ca t ionschemes. The vi l lage is certainly tooheterogeneous for the purpose. Farmsthat produce under almost identical con-dit ions should be grouped together inparticular as to soils, i.e. they shouldproduce w i th almost the same physicalproduction functions.

4. Such work requires permanency. It is bet-ter to fol low a few farms over a longerperiod than to produce a one-shot analysisof a great number.

The next paper submitted by Mr. Rastogidiscusses the existing cropping patterns inselected Indian dryland farming regions, thehusbandry pract ices tha t p reva i l , thetechnologies that can be introduced and thecosts and returns of existing and improvedpractices in monetary terms. Mr. Rastogi points

38

to the wide discrepancy between the actualfarming performance and the much higherreturns that are obtained with more intensivepractices experimented on large plots in far-mers fields. There is thus a wide gap betweenpresent practice and what should be economi-cally optimal from the farmer's point of view.The information provided by Mr. Rastogi is wellorganized and most valuable. Perhaps the valueof future work along these lines could be en-hanced by specifying classification that doesnot occur on a regional or village basis, but isdesigned on a farming systems basis. Also a break-down of a farm into separate fields andthe collection of plot histories might increasethe value of such work.

Of crucial importance is the explanation of the"gap " that Mr. Rastogi found. Plots husbandedby personnel f rom research stations normallyyield much higher gross and net returns thanfarmers' fields. If, in the area analyzed byMr. Rastogi, the index figure 100 is given to thegross return of the plots husbanded by researchpersonnel, then the farmer would harvest 66with the same inputs and his own management.He would harvest 33 wi th traditional inputs andhis own management (private communicationfrom Mr. Rastogi). These findings lead to a pleathat future farming systems research shouldnot only describe, but also try to explain, whythings are as they are. India, meanwhile, has a great number of farming systems studies,which inform about what now exists but whichwould greatly gain in value if more effort wouldbe directed to explaining why low-output andlow-productivity farming systems prevail inspite of the fact that most farmers know aboutinnovations that could improve their situation.

However, research institutes such as ICRISATshould remain "economic" as to socio-economic research that aims for explanationand understanding. Such research will be t imeconsuming. The situation is complex, andhuman behavior may change rapidly over t ime.We don't command an operational theory ofman, and some of the research may result inconflicts between controversial schools ofthought in social sciences.

The emphasis in farming systems researchshould therefore be put into a well organizedtrial-and-error process. The initial step clearly isa basleline survey. A number of cooperatingfarmers have to be found; there usually is a

sufficient number who are interested and pre-pared to cooperate, provided they obtain sup-port. The idea is to test innovations as early aspossible with these cooperating farmers, whoshould be well aware that they are dealing withsomething immature. Cropping and farmingsystems researchers face the problem that thereis a rapidly growing number of seemingly in-teresting combinations of innovations. It isimpossible to handle these numbers in experi-mental work. Crop and husbandry combi-nations chosen in experimental work at re-search stations are more or less shots in thedark. The involvement of the farming commun-ity in this process of trial and error allows thetesting of a much larger number of innovations.It includes the experiences and intelligence of a great number of farmers who have a materialinterest in what they do. They rarely adoptbundles of innovations as recommended; theyexperiment on their own, and this is likely togenerate ideas that would improve the rele-vance of research work provided there is a wellorganized feedback from the farming commun-ity to the biological science research workers.

This process of trial and error is not necessari-ly tied to an understanding of the farmingsystems and the human element involved. Thefarm household systems may be considered as"black boxes." We need not necessarily under-stand why farmers do what they do; but if theresults are positive in terms of objectives inresearch then we may conclude that the inno-vations are useful. If this is not so we maycontinue the process with modified or entirelydifferent innovations.

39

Chai rman 's S u m m a r y

W . H . M . M o r r i s *

As a basis for discussion we have the com-prehensive comments presented by Ruthen-berg.

Abalu presented results that indicate(1) mixed cropping produces higher grossand net returns than sole cropping, and(2) actual cropping patterns are rather closeto those that result f rom minimization ofincome variability and are consistent wi th theobjective of providing sufficient calories andprotein when the farm is large enough.The hypothesis of income variability minimi-

zation is not really verified by the methodologyused. Professor Ruthenberg pointed out thatthe term "constraint," borrowed from the tech-nical sciences, may be abused by economists.We rarely meet labor constraints on Africansmall farms, but rather elastic and often kinkedlabor supply curves. More evidence is neededbefore we can accept existance of a real laborconstraint.

Abalu points out that the rationality of mixedcropping has been ignored by agriculturalscientists in the tropics, but the presence ofmixed cropping alone may not be adequatejustification for investing research resources init.

A fair understanding of the existing farmingsystem is a necessary precursor for any majorprogram in agricultural research. However, al-though the kind of work reported by Abalu isindispensable, it should not cause us to overlookthe much better chances that perhaps, areoffered by something entirely new.

Ruthenberg pointed out that there were twobasic strategies in the allocation of researchresources: (1) the improvement approach tobetter existing systems and (2) the transfor-mation approach, which concerns a completelynew production pattern. Farming systems

* Department of Agricultural Economics,University, West Lafayette, Indiana, USA.

Purude

analysis provides information that is indispen-sable for the design of research strategies, butthe knowledge of the biological scientists onpossible breakthroughs is just as important.

Jodha presented essential information oncropping systems and watersheds in areas withmonsoon fallowing in India. His presentationimplies that farming systems research at anexperiment station needs to have two stages:(1) baseline surveys to understand the presentsituation, and (2) analysis of the changingfarming system in testing the innovation. In thelatter stage feedback should be used to adjustongoing biological research. This type of re-search differs f rom the baseline studies in thefol lowing respects: it has to be more specificand include individual plot histories, and it hasto be geared simultaneously to both hypothesistesting and formation; that is, description of thefarming system needs to be complemented bytesting of stated hypotheses and by the formu-lation of new and better hypotheses. This typeof study requires detailed classification. Thevillage is too heterogeneous a unit and farms orfields must be compared which have almost thesame physical production functions. Such a study also requires continuity. It is better tofol low a few farms over a longer period than doa single entry study of a large number of farms.

Rastogi discussed cropping patterns in theIndian SAT, as well as the husbandry practices,changes that can be observed, and the costsand returns of present and potent ia ltechnologies. There is a wide gap betweenfarmers' results and those from research inlarge plots on farmers' fields. A more detailedclassification of the units studied, down to plots,should decrease this variability and explain thisimportant discrepancy. This type of researchneeds to be directed towards explaining the"gap " as well as to pointing out its extent.

The large number of farming systems studiesin India would be more valuable if a greatereffort were made to explain why low output/lowproductivity farming systems prevail, in spite of

40

the fact that most farmers know about theinnovations that could improve their situation.However, institutions such as ICRISAT shouldbe economic in their socioeconomic researchprogram, seeking only explanation and under-standing because of the time-consuming natureof this work and the problems of studyinghuman behavior over time. The emphasis infarming systems research should therefore beput into a well-organized trial-and-error pro-cess. The initial phase is clearly baselinestudies. There are usually a sufficient number offarmers who are interested and prepared tocooperate — given adequate support — in test-ing innovations at an early stage. The farmersmay not accept a whole range of technologicalpackages but rather experiment on their own,and they are likely to generate ideas that im-prove the relevance of research, provided thereis a well-organized feedback to the biologicalscientists.

The study of this process of trial-and-errordoes not necessarily require an understandingof the farming and human elements of thesystems. If the results f rom the farmers' effortsare positive in terms of objectives of research,the innovations wil l be useful. If their efforts arenot positive then the process has to be modifiedor entirely different innovations tested.

In the light of the four criteria in the program,Dillon stressed the need for adding a macro-level appraisal of the socioeconomic-politicalenvironment to the microlevel socioeconomicanalysis. This type of appraisal could be used tohelp assess or decide on the allocation ofresearch resources among countries by inter-national agricultural research centers, on thebasis that there is likely to be little benefit fromexpending resources on farming systems re-search where the sociopolitical situation is un-favorable to the target group.

41

Chapter 2

S o c i o e c o n o m i c s o f P r o s p e c t i v e T e c h n o l o g i e si n S A T R e g i o n s

Soc ioeconomic Aspects Involved in In t roducingN e w Technology into the Senegalese Rural Mi l ieu

Moussa Fal l *

Abstract

This paper discusses physical constraints and human factors relating to the introduction of new technology in the rural areas of Senegal, particularly in the Sahel, where investigations have been in progress for 3 years. Although the profitability of certain techniques has been demonstrated, the farmer still needs to be motivated towards change. This motivation is presently based mainly on increase in monetary income and production surplus. Production systems should be intensified; the marketing system should be better organized; the working of cooperatives should be improved; credit should be made more flexible. Literacy remains a crucial problem in efforts at modernization.

Agricultural production represents about one-third of the gross domestic product in Senegal,but approximately 70%of the active populationis employed in this sector. In developmentplans, priority has been given to the moderni-zation of agriculture and to the increase offarmers' incomes through large investments,introduction of technical innovations, and in-tensification of farming systems.

In this paper, we shall discuss some aspectsof the introduction of new technology in therural areas of Senegal, particularly the northcentral zone (Sahel), where investigations havebeen carried out for the last 3 years. In selectingone zone, we are aware that only partial resultscan be obtained. However, these can be com-pared with results of similar studies undertakenin other countries and can be verified for otherregions of Senegal.

First, we shall briefly discuss problems re-lated to the interaction between productiontechniques and the enviornment and then pro-ceed to the economic evaluation of some inno-vations.

* Centre National de Recherche Agronomique deBambey Institut Senegalais de Recherches Ag-ricoles.

N O T E : This paper is an edited translation of theoriginal French text, which appears in Appen-dix 1.

A g r i c u l t u r a l P r o d u c t i o nT e c h n i q u e s a n d T h e i rI n t e g r a t i o n i n t o t h e P h y s i c a la n d H u m a n E n v i r o n m e n t

P r o d u c t i o n T e c h n i q u e sa n d t h e Phys ica l E n v i r o n m e n t

One of the main characteristics of agriculture isthe important influence of the physical en-vironment on the production process. There isample evidence of this influence in Senegal'srural areas, where the capacity of a given landarea to absorb capital and labor largely dependson a number of soil-related limiting factors. Thismay be obvious; but it seems useful to repeat itas errors are frequently committed on this pointand several failures are still attributable toinadequate adaptation of techniques to localphysical conditions.

P r o d u c t i o n T e c h n i q u e sa n d t h e H u m a n E n v i r o n m e n t

There is an obvious relationship between pro-duction techniques and the human environ-ment, but this aspect is often overlooked. For a long time it was believed that in order topromote new techniques and equipment it wassufficient to show how effective and easy to usethey were to future users. However, such mea-sures are now known to be inadequate. Promo-tion of innovations is slow and transformation

45

of farming techniques is difficult. There is needfor a new approach involving the farmer, histradit ions, and his social environment. Thisapproach has to be based on two assumptions:

1. In order to understand a system, changesshould be introduced that wil l help toperceive its functioning and the moti-vations and needs of those who make upthe system, and

2. Modernization of family landhoidings is a long-term process because it not onlyinvolves the adoption of innovations butalso requires a change in the mentality andbehavior pattern.

I n t e r d e p e n d e n c e o f T e c h n i q u e s

In the traditional agricultural system in Senegal,the interdependence of techniques is not clearlyseen because of the inadequate use of factors ofproduction and their poor combination. How-ever, the system is a complex of exchanges(equipment, manpower, land use, etc.), and themodification of any one of these can lead todisruption.

The introduction of certain innovations andagricultural intensification has revealed the re-lationship between techniques:

• Effective fertilization requires crop mainte-nance and application of certain methods,such as thinning for millet.

• Introduction of animal traction requires bet-ter food for animals and increased waterand fodder resources.

• The use of the planter has helped to expandcropped area, modify cropping plans, andincrease labor productivity.

Thus it appears that innovations should bepromoted through coherent and feasibletechnology packages that are integrated intothe existing production system.

P r o b l e m s o f T r a n s f e ro f T e c h n o l o g y i n R u r a lA r e a s o f S e n e g a l

P h y s i c a l a n d H u m a n E n v i r o n m e n t

The area involved is north-central zone which isentirely situated in the Sahelian belt. The dryseason extends over 8 to 9 months of the yearand average annual rainfall is about 500 m m .Soils are mainly Dior (poor in clay). There Is a

46

large rural population with frequent migration,especially among the young people.

The carre, which is the production unit, mayinclude a certain number of independenthouseholds that make up sub-land holdings.The head of the carre owns most of the pro-duction facilities and generally has controlover decision-making.

The main crops are millet and groundnut;fal low is starting to disappear. The expansion ofareas devoted to new crops (diversification) isquite slow.

P r o p o s e d I n n o v a t i o n s

The main objective of the proposed innovationsis to increase yields and labor productivity.They concern varieties, cropping techniques,equipment, diversification of activities, and theproduction system and are as fol lows:

1. Improved and suitable varieties are re-commended.

2. The main techniques that are promotedare:a. adequate tillage (preplanting plowing

before a cereal crop);b. heavy mineral fertilization (150 kg/ha)

and basic application of phosphate inan intensive farming system;

c. t imely planting;d. chemical weeding of groundnut and

early thinning (three plants per hill) ofmillet;

e. incorporation of organic matter, eitherin the form of manure or of residues of a millet crop.

3. The equipment recommended to thefarmer enables him to increase his laborcapacity and improve crop maintenance.

4. Diversification of activities enables thefarmer to minimize risks and to add to hissources of income. It was therefore prop-osed to integrate livestock raising withagriculture through summer fattening ofanimals and increased diversification ofcrops.

Simple and economically viable models arebeing studied.

C o n d i t i o n s f o r P r o m o t i n g I n n o v a t i o n s

An autonomous body SODEVA (Societe deDeveloppement et de Vulgarisation

Agricole — Society for Agricultural Develop-ment and Extension Services), along with otheragricultural services, is responsible for themobilization of farmers and promotion oftechniques. This organization is increasinglyinvolved in the distribution of production fac-tors as well as their use by the farmers. How-ever, the adoption of recommended techniquesis not without difficulties, and a detailedanalysis of the situation has revealed majorconstraints.

Technical Constraints

These constraints are found at all levels:• The formulation of technical recommenda-

tions made through research is not alwaysperfect and the technoeconomic referencesare sometimes inadequate.

• The development organization becomes a dead weight, impeding movement of tech-nical information from the top to lowerlevels. Links with research are limited tocertain aspects with little fol low-up exceptfor specific projects.

• The work and technical skill required of thefarmers often exceeds their capabilities andextends beyond their short cropping calen-dar (early th inning, postharvest til lage,production of manure, etc.).

Human Constraints

Within the framework of the promotionmethods adopted in the region, the social or-ganization of the family landholding impedesadequate investment or stabilization ofmanpower — a rare resource at certain criticalperiods. However, this constraint is not insur-mountable.

Farmers do not always have well-definedobjectives and these are often contrary to therecommendations made.

Economic Constraints

Agriculture in Senegal is being rapidly trans-formed into a money economy with a relativelysatisfactory market channel for groundnut. Thishas encouraged farmers to practice extensiveagriculture in the absence of substantial techni-cal progress.

The area cultivated per worker is a predomi-nant factor in determining labor returns. As the

second factor, the returns per surface unit maybe practically considered as constant. So popu-lation growth, resulting in a decrease in the areacropped per worker, has led to a decrease inlabor returns. Land availability is decreasingand rights of individual tenure do not permit themerging of landholdings. Owners of large land-holdings who have the highest incomes moreeasily adopt techniques such as herbicides,tillage, hired labor etc. There is a gap betweenthe big farmer and the small farmer, who isusually in debt and accepts with difficulty thecostly innovations or those that do not bring inimmediate returns.

The system of providing factors of productionand market channels for certain products doesnot always function without difficulties.Cooperatives are often blocked by debts whilefarmers wait for the equipment they have or-dered.

In all cases, the success of promotion efforts,even after proper study, still depends on properorganization of the economic enviornment ofthe producers.

E v a l u a t i o n o f t h e I m p a c to f P r o m o t e d T e c h n i q u e s

Because of the economic constraints, researchefforts are being directed towards study of theactual production systems and evaluation of theeffect of new techniques in rural areas. The ideais not to "measure" technical progress of thelandholdings but to evaluate their internaldynamics as well as their capacity to implementthe recommended innovations.

M e t h o d o f A p p r o a c h

Two types of approaches may be considered forthe area: studies of selected landholdings andmore generalized and flexible promotion ef-forts. In the first case, investigations are plan-ned to determine the extent to whichtechniques are assimilated; in the second case,differences between categories of landholdingsare measured according to the degree of inten-sification.

Investigations

Three types of investigations can be distin-guished:

47

1. investigations of population, farm equip-ment, and cultivated areas,

2. agricultural investigations,3. socioeconomic investigations.

Analytical Methods

It is possible to characterize landholdings byprocessing data obtained f rom the first type ofinvestigations. The other two types of investiga-tions help to evaluate the extent to whichthe techniques are assimilated and their ef-ficiency. For this purpose, we use comparativeanalysis, principal component analysis, regres-sions, and landholding counts.1

Comparat ive Analysis Matr ix

This analysis provides four types of output wi thvarious combinations of variables and theirstatistical characteristics. With this analysis itwas also possible to judge the importance ofsocial status in the adoption of techniques(1976) in three pilot villages of the zone.

Groundnut is cultivated by all those who workon the landholding — head of the carre, house-hold head, sourga (family labor), navetanes (seasonal labor), and women.

The head of the carre only cultivates 38% ofthe land on an average of two plots. The plotscultivated by women are much smaller thanthose of other members of the land-ho ld ings— an average of 0.39 ha, or one-

1. Due to space l imitat ions, details of thesemethodologies are not presented here but can befound in the papers contained in the list of refer-ences at the end of the paper.

quarter the size of the plots cultivated by thehead of the carre.

Yields in the womens' plots are lower thanthose in plots cultivated by the head of thecarre; as standard deviations are high, thedifferences are not significant, but they are thesame in the three villages and tally with otherobservations (Table 1).

On the whole, it was observed thattechniques do not differ greatly according tosocial status. The largest variations were re-corded in women's plots and to a lesser extentfor the sourgas (Table 2).

The study reveals the characteristics of theplots according to social status and the preemi-nence of the head of the carre in relation to othermembers of the landholding; they have thelargest number of plots with heavy fertilization(Table 3).

The variables of this study were selectedaccording to the objectives of the analysis:

• to determine the effect of the main criteriaon landholding economy.

• to determine the effect of variations in size.• to determine the levels of intensification of

the selected landholdings.Variance of all variables was explained by

four independent factors — size (physical andeconomic), use of factors of production, soilproductivity, and labor productivity.

The analysis brought out the fol lowing facts:1. The farming system is purely "extensive;"

size is a very important factor in incomeformation.

2. It is mainly the larger landholdings that areunderequipped. The better equipped far-mers make the best use of other inputs.

3. Soil productivity is not very important as

Table 1 . D is t r ibu t ion accord ing to s ta tus ( 1 9 7 6 v i l lage- l evel s tud ies i n t he Th ies -D iou rbe l reg ion ) .

No. ofStatus of cult ivator plots

Head of carre 50Household head 17Sourga 78Navetane 15W o m a n 86

Total 246

Averageplot area

(ha)

1.541.270.740.890.390.83

Total area under groundnut

(ha) (%)

77.07 37.921.58 10.657.91 28.513.37 6.633.47 16.4

203.40 100

Yield(kg/ha)

1309127310051103847

1077

Fertilizer(kg/ha)

603915501931

48

Tab le 2 . Cropp ing t e c h n i q u e s accord ing t o s ta tue.

Status ofcultivator

Head of carre Household headSourgaNavetaneWoman

Total

No. ofplots

5017781586

246

Percent plotssown onfirst rain

( % )

728263

1003157

Plots hoedtwo or more

times( % )

787694665875

Averageno. of

hoeings

2.342.592.462.071.912.23

Averageno. of

weedings

1.801.411.781.671.861.78

Fertilizedplots( % )

524115532129

opt imum use is not made of inputs. So,remuneration for work is reduced.

Multiple Regression Analyses

This method is used along wi th principal com-ponent analysis. As expected, the explanatoryvariables are mainly cultivated area per agricul-tural worker and groundnut production. At pre-sent, in a large part of the zone, extensificationpresents a definite advantage over intensifi-cation. However, these two notions are notopposed to each other and, in cases of shortageof land, intensification is necessary.

Although millet and groundnut productionare accounted for in the same manner, milletproduction has little effect on income forma-t ion, as yields were low in the year of investiga-t ion.

Annual Budgets

The main objective of this evaluation methodwas to study income formation. It does notinclude problems of marketing and interactionwi th extra-agricultural activities.

Cereal crops — millet and sorghum — wereincluded in the calculations because productionactivities are closely linked within the landhold-ing and because they are involved in transac-tions between villages.

Quantitative parameters (prices, expenses,units, workers, etc.) were used in all cases toobtain results for comparison and to avoidconsidering specific cases found in the area.

Two aspects were considered — calculatedbudgets and monetary budgets. The first methodis often used for analyzing the performance of

Table 3 . Use of fert i l izer accord ing to status.

Fertilizedplots

Status

Head ofcarre

Householdhead

SourgaNavetaneWoman

(%)

52

41155321

Average quantityof fertilizer perunit of all land

(kg/ha)

60

39155019

Quantity offertilizer used on

fertilized plots(kg/ha)

116

95949489

production systems and the second is used tostudy the accumulation capacity for explainingthe more or less rapid progress of the system. Inorder to compare production systems, espe-cially their productivity, margins (gross or net)or added values (gross or net) were used.Results obtained in 1975 and 1976 haveconfirmed the observations made above: onsmall holdings, despite higher groundnut yieldsand larger use of fertilizers, the gross productremains low.

An examination of the classes of landhold-ings in the sample shows that the carres varyfrom 3 ha to 27 ha in size. (The sample is notrepresentative of the entire zone if we considerthe fields of the marabouts.)

The net margin/worker is largest in the 27-hacarre size group whereas the net margin/ha islargest in the 15- to 18-ha carre. The results forfarms using bullock power have shown thatalthough bullock power helps to expand land

49

area, it should, in order to obtain better returns,be integrated in a system where all the factors ofproduction are used rationally.

R e s u l t s o f F a r m - L e v e l S t u d i e sb y E x t e n s i o n S e r v i c e s ( S O D E V A )

As part of agricultural program evaluation,SODEVA has studied 162 landholdings of thezone (1977-78 agricultural program investi-gations, Diourbel region). The criteria for land-holding selection were: landholdings that havealready been studied, possibility of access inany season, and agreement of farmers.

The study has helped to assess the differ-ences between categories of landholdings:TL - easily applicable or traditional techniques;TB — bullock power; TBFF - bullock power,heavy fertil ization.Two important facts may be emphasized:

1. The superiority of the TBFF (during theyear of study —1977)2 for millet productionbecause of the increased use of hoes andintensive fertil ization. Use of these techniquesexplains the increase in yield by 90 kg/ha on theTBFF.

2. Regression analysis helped to calculatethe advantage of extensification.

In the conditions that were studied, it wouldbe useful for the average farmer to increase hisland area/worker up to 2.2 ha for millet (averagenow 1 ha) and up to 3.5 ha for groundnut(average now 1.5 ha). In this case millet pro-duction would increase by half and the grossmargin for groundnut would increase to a total of 36 000 Frs.

O v e r a l l C o m p a r i s o n o f t h e E c o n o m i cR e s u l t s o f L a n d h o l d i n g s (TL-TB-TBFF)

A comparison between the three types of land-holdings shows a clear difference in the netmargin per worker. Using TL as the referencelevel, the fol lowing indices of net margin can beobtained.

TL = 100TB = 108TBFF = 128

The difference between TBFF and TL is due tomillet and groundnut yields, cropping pattern,

2. 1977 was a year of drought w h e n g roundnutproduct ion was part icularly affected.

50

and farm equipment.A breakdown of the net margin/worker gives

the fol lowing TBFF/TL ratios:Groundnut gross margin/worker 1.61Millet gross margin/worker 1.09Total gross margin/worker 1.29Redemption/worker 1.38Net margin/worker 1.28

C o n c l u s i o n

We have discussed an important aspect ofagricultural development — the introduction ofnew technology — in order to present the prob-lem of its evaluation and its impact in ruralareas. This analysis, though partial, has re-vealed the importance of the economic en-vironment of the farmer in the farm moderni-zation process.

In Senegal, although the profitability of cer-tain techniques has been demonstrated, thefarmer still needs to be motivated towardchange. This motivation is mainly based onincreased monetary income and productionsurplus.

For this purpose, marketing needs to be wellorganized; cooperatives should be allowed toplay their true role; and the credit systemshould be made more flexible. Literacy remainsa crucial problem for the increasing rural popu-lation.

Land shortage is beginning to be felt. Theanalysis revealed the fol lowing:

1. The farming system is purely extensiveand income is determined by size.

2. Farms are relatively underequipped.3. Soil productivity is low because pro-

duction factors are underused.Production systems need to be modif ied and

economic conditions created for profitable in-tensification.

References

BROCHIER, J. La d i f fus ion du progres technique enmi l ieu rural senegalais, France: Institut d'Etude duDeveloppement Economique et Social.

DEPARTMENT D'ECONOMIE RURALE. 1978. L 'approche

socio-economique des exploi tat ions au Senegal.Report 1.

FALL, M. 1977. Programme moyen terme Sahel: In-terpretat ion statist ique des donnees economiquesde la campagne 75-76 . Bambey. Senegal: CentreNational de Recherche Agronomique.

MINISTERE Du DEVELOPPEMENT RURAL ET HYDRAULIQUE/

SOCIETE DE DEVELOPPEMENT ET DE VULGARISATION

AGRICOLE/BESP. Enquete suivi P. A. 1977-1978. (Re-g ion Diourbel).

RAMOND, C., and FALL, M. 1976a. Economie des exploi-tat ions agr icoles des terroirs-test. Bambey,Senegal: Centre National de Recherche Ag-ronomique.

RAMOND, C., and FALL, M. 1976b. Taux de penetrationdes themes techniques et incidences sur les rende-ments des cultures dans les terroirs-test (Prog-ramme moyen terme Sahel). Bambey, Senegal:Centre National de Recherche Agronomique.

VANHAEVERBEKE, A. Remunerat ion de travail et com-merce exterieur: Terme de I'echange des produc-teurs d'arachide au Senegal. Louvain, Belg ium:CRD.

51

Assessment o f Prospect ive So i l - , Water - ,and C rop -Managemen t Techno log ies

for t h e Semi -Ar id Trop ics o f Peninsular India

J a m e s G . R y a n , R . S a r i n , a n d M . P e r e i r a *

Abstract

This paper summarizes the economic analyses of experiments conducted by ICRISAT's

Farming Systems Research Program from 1975 to 1978 to develop improved methods of

soil, water, and crop management. On the deep Vertisols at ICRISAT Center the

broadbed-and-furrow system of soil and water management was highly profitable. This

could not be shown for the medium-deep Vertisols. On the Alfisols steps-in-technology

experiments, profits from broadbeds and furrows were about 60% less than on Vertisols

but were still attractive. However, in the larger scale Alfisols watershed experiments,

broadbeds and furrows were less profitable than flat cultivation. The complete t echno lo -

gy package of improved variety, fertilization, and soil management on both Vertisols and

Alfisols at ICR ISAT Center generated maximum profits per hectare with risks that

seemed well within the revealed risk-aversion preferences of Indian SAT farmers. On

ICRISAT Center Vertisol watersheds, the maizelpigeonpea intercrop system generated

substantially higher profits, with less variability, than did the maize-plus-chickpea

sequential system, especially on the medium deep Vertisols. The likelihood that a

water-harvesting and supplementary irrigation technology will prove viable is much

higher on ICRISAT Center Alfisols than Vertisols, due to the greater runoff-generating

potential and scope for profitable crop yield responses to irrigation on Alfisols. It is

unlikely that this technology will be viable on Sholapur Vertisols. Optimum watershed

sizes would seem to be between 8 and 16 ha.

Scientists in the Farming Systems ResearchProgram (FSRP) at ICRISAT have been under-taking research for the past several years thatfocuses on the development of improvedmethods of soil, water, and crop managementthat can help increase and stabilize crop pro-duction in the semi-arid tropics (SAT). From theoutset, much of the research has been conductedon plots that have generated input-output data suitable for economic analysis. Theauthors have been working wi th their col-leagues in the FSRP on the analysis of thesedata in order to identify economic strengths andweaknesses in the evolving technology andthus enhance the process of development ofviable technologies.

* Economist and Leader, Research Technician, andformer Research Technician, Economics Program,ICRISAT.

This paper attempts to summarize the resultsof this work over the 3 crop years 1975-76 to1977-78. In the first section, we discuss the datagenerated and the experiments on which theensuing analyses are based. In the secondsection, we discuss the results f rom the steps-in-improved-technology (SIIT) experimentsconducted on Vertisols and Alfisols at ICRISATCenter at Patancheru near Hyderabad. We thenproceed to discuss the larger scale watershedexperiments at ICRISAT Center on both soiltypes. Next we describe simulation exercisesperformed to evaluate the economic potentialfor water harvesting and supplementary irri-gation in three different agroclimatic situations.We conclude wi th a summary of the majorfindings of the work.

D a t a

The input-output data for the economic

62

analyses of the steps-in-improved-technologyexperiments were generated collaboratively bystaff of the FSRP and the Economics Program(EP) of ICRISAT. The objective of these experi-ments, which began in 1975-76, was to mea-sure the individual and complementary effectsof the various components of prospective ag-ricultural technology for the SAT. They weredesigned with plot sizes ranging from 0.07 to0.1ha. These plot sizes were to enable the soil- andcrop-management treatments to express them-selves. This is not possible with the muchsmaller plot sizes generally used in agriculturalexperiments.

The four basic treatments in the SIIT experi-ments (Table 1) were (1) variety, (2) fertilization,(3) soil and crop management, and (4) watermanagement. Each of these four basic treat-ments were examined at t w o levels in a factorialexperimental design. One was aimed atsimulating local practice (L) and the other theimproved technology (I). The experiments wereconducted on both the Alfisols and Vertisols atICRISAT Center. More detailed explanation ofthe technical aspects of these experiments canbe found in the reports by the Farming Systemsstaff (various years). Details of the economicanalyses, including methodology, inputs, out-

puts, and price data, can be found in Ryan andSarin (1977, 1978).

The operational watershed experiments on a field scale at ICRISAT Center aim to test the soil-and water-management aspects of prospectivetechnology on a scale that simulates farm con-ditions. Small agricultural watersheds rangingin size from 0.3 to 11 ha on the Vertisols, and 0.3to 5 ha on the Alfisols, have been the conceptualfocus of this phase of the research of the FSRP.Input-output data on these large plots weremonitored by field supervisors who recordedthem daily in field pocket notebooks designedfor the purpose. These were checked by staff ofFSRPand the EP on a daily basis. Care was takento ensure that pure "research-related inputs,"such as moisture probes and the labor for theirinstallation, were not included for purposes ofeconomic analysis. Crop yields were derivedfrom randomized sample cuts in eachwatershed treatment as described by the Farm-ing Systems staff (various years).

The major comparisons of economic interestin these watershed experiments were (1) rela-tive performance of the graded broadbed-and-furrow method of cultivation compared with thetraditional flat techniques across the slopeand/or on the contour; (2) effects of different

Tab le 1 . Descr ip t ion o f t r e a t m e n t s u s e d i n t h e s t e p s - i n - i m p r o v e d - t e c h n o l o g y e x p e r i m e n t s (SIIT) a tICRISAT Center .

Level

Local (L)

Improved (I)

Technology steps treatments

Varieties Fertilizers

Tradit ional Farmyardlocal manure only

High-yielding Opt imumvarieties doses of

N,P, asappropriateto cropand soil

Soil and crop management

Local animal-drawnimplements; f latcul t ivat ion; delayedland preparationafter harvest; onehandweeding; min i -m u m insect control

Improved animal-drawn implements;cult ivation onbroadbeds andfur rows on gradedcontour;post-harvest landpreparat ion;improved plantprotect ion

Water management

No supplementaryirr igation

Supplementaryirrigationappl ied ifcrops arestressed, plusadded fertilizersif required.

53

graded slopes of the broadbeds and furrows;(3) effects of different types of bunds — contour, graded contour, f ie ld; (4) effects ofwater harvesting and supplementary irrigation.

Data f rom the rainfall-runoff experimentsconducted since 1973-74 at ICRISAT Center,near Hyderabad, and f rom 1951-52 at the AllIndia Coordinated Research Project for DrylandAgriculture (AICRPDA) Station, Sholapur, havebeen used to develop empirical models forprediction of runoff. The descriptions of dataand methodology are reported elsewhere (Ryanand Pereira 1978). In the present paper, weutilize historical daily rainfall data f romHyderabad and Sholapur weather stations for70 and 29 years, respectively, to simulate dailyrunoff using the models derived.

S t e p s - i n - l m p r o v e d - T e c h n o l o g yA n a l y s e s

It is important to measure the contribution ofthe various elements that constitute what iscommonly termed the "package of practices."In this way, those elements that contribute mostto the profitability of the new technologies canbe identified, as wel l as their complementarit ieswi th other elements. Because of constraints ofcapital, labor, or other resources, many farmersmay not be in a position to immediately adoptthe ful l package of practices that might generatemaximum profits per hectare; therefore, de-lineating the separate contributions of eachelement can assist in providing "ranges oftechnological opt ions" f rom which farmers canchoose according to their particular constraints(Ryan and Subramanyam 1975).

In deciding upon the approach taken inanalyzing these experiments we have drawnliberally on the excellent paper by Perrin et al.(1976). Because risk is a major factor in the SAT,we have combined the SIIT results over 2 yearsin the case of the Vertisol experiments and 3 years for the Alfisols in order to enable a firstapproximation of the variability (risk) of profitsof each treatment to be made, in addition toaverage profitability. The basic results of thisexercise are shown in the appendices to theunabridged paper presented at the workshop.

Vertisols

On the basis of both profits per hectare and

54

profits per rupee of variable cost, treatment inthe Vertisols involving improved varieties, fer-tilizers, soil and crop management, along wi thsupplementary irrigation (llll), was the mostprofitable over the 2 years 1976-77 and 1977-78. The average net benefit per hectare was inexcess of Rs 3600 and every rupee spent re-turned Rs 1.89 in net benefit, a rate of return of189% (Fig. 1). The next best treatment was thesame as the first except that it was withoutsupplementary irrigation (III). The extra variablecost of irrigation was around Rs 300/ha plusabout Rs 50 in added fertilizer, which togethergenerated an additional Rs 740 in net benefits,or Rs 2.13 per rupee of expenditure. However,not all the irrigation treatments involved the useof water f rom runoff storage on the samecatchment; in some case it was pumped f romwells and in others f rom nearby drains. The costof obtaining water f rom these sources has notbeen included, so it is not an appropriate evalu-ation of the payoffs f rom water harvesting andsupplementary irrigation.

It is evident f rom Figure 1 that expenditure ofamounts beyond about Rs 1500/ha is ex-tremely profitable. For example, shifting f romLll (local varieties, improved fertilizers and im-proved soil and crop management) to III ( im-proved varieties, improved fertilizers and im-proved soil and crop management) involvingonly an additional Rs 100 can generate anadditional Rs 1700 in net benefits. Figure 2 shows that such a shift may not only increaseprofits but also reduce risk, as measured hereby the standard deviation of net benefits per haover the 2 years analyzed. In this f igure areshown the three steps that dominate all the rest,in that there are no steps more profitable andless risky than these.1 Hence one would chooseonly among LLI, III, and llll when generatingfarmer recommendations. The change f romlocal to improved varieties under local fertilizerand local soil and crop management (LLL to ILL)increases net benefits f rom around Rs 70 tonearly Rs 900 wi th only a few rupees additionalcost.

In general, Figure 2 shows that for theseVertisols during these two years, additionalprofits were earned at the expense of additionalrisk, as measured by standard deviation. How-ever, Figure 3 shows that if we take the

1. That is to the southeast of t hem.

Figure 1. Average net benefits and variable costs per ha in steps-in-improved-technology

experiments on Vertisols at ICRISAT Center, 1976-77 and 1977-78.

55

Treatments* and variable cost (Rs/ha)

( * Refer to Table 1 for detai ls of treatments)

Average net benef i ts /ha

Standard deviation of net benef i ts /ha

coefficient of variation (CV) of net benefits asour risk criterion, and map it against cost re-quirements of each treatment, then only stepsIII and LLI are cost/risk efficient. By investing theRs 1130 and Rs 1600/ha required to implementthese steps, we can achieve the lowest CVs of 37and 20%, respectively. No other steps can offerlower CVs wi thout increasing the cost. it isinteresting to note in both Figures 2 and 3 that

the supplementary irrigation treatments alwaysinvolve more risk than the nonirrigated treat-ments.

The additional net benefits f rom the im-proved soil- and crop-management treatmentsimplemented under either local varieties andfertilizers, local varieties and improved fertiliz-ers, or improved varieties and local fertilizers isaround Rs 500/ha (Fig. 1). The estimated capital

Figure 2. Net benefitlrisk profile of steps-in-improved technology on Vertisols at ICRISAT Center, 1976-77 and 1977-78.

56

Figure 3. Variable cost/coefficient of variation

of net benefit profile of steps-in-

improved technology on Vertisols

at ICRISAT Center, 1976-77 and

1977-78.

cost of the bullock-drawn wheeled tool carrierused in the improved broadbed-and-furrowsystem, which is a major component of thisstep, is around Rs 6500.2 If this implement wereonly used for cultivation activities on 13 ha peryear, it could more than pay for itself in 1 yearunder these conditions.3 Under improved va-rieties and fertilizer levels, the improved soil-and crop-management steps can increase netbenefits per hectare by more than Rs 1000,al lowing the wheeled tool carrier to be paid forin 1 year f rom its operation on only 7 ha.

The additional net benefits f rom the changefrom local to improved fertilizer levels underlocal varieties, w i th either improved or local soiland crop management, amounted to around

2. Personal communicat ion G. E. Thierstein, Princi-pal Agr icul tural Engineer, ICRISAT. Proratedoverhead costs of this implement have beenincluded in our cost est imates.

3. Of course, it is also designed for other activitiessuch as cart ing, wh ich w o u l d further enhance itsvalue.

Rs 600/ha, f rom an added expenditure of nearlyRs 400/ha. In both cases, however, this involvedan increase in the standard deviation of netbenefits. When improved fertilizers were substi-tuted for local fertilizers wi th improved varietiesand local soil and crop management, the addednet benefits were nearly Rs 1000/ha from anadditional expenditure of Rs 500, wi th nochange in standard deviation. When thischange was effected on improved varieties wi thimproved soil and crop management, theadded net benefits were more than Rs 1500/hafrom an extra expenditure of Rs 400, represent-ing a phenomenal 375% rate of return. Thisadditional return also comes with a smallerstandard deviation than the local fertilizer step.

Alfisols

The treatment involving improved varieties,fertilizers, and soil and crop management wi th-out supplementary irrigation (III) generated thehighest average net benefits on the Alfisol SIITexperiments conducted over the 3 years 1975-76 to 1977-78 (Fig. 4). Contrary to what we wil lshow later for the watersheds, supplementaryirrigation on the Alfisols SIIT experiments wasnot profitable.4 It also increased risk, as re-flected in the standard deviation of net bene-fits.

The differences in the steps involving localvarieties and those wi th improved varieties,holding other steps constant, were not nearly asgreat as found in the Vertisol SIIT experiments.However, with one exception, those treatmentsinvolving improved varieties were better thantheir local variety counterparts. On an averageover alI the treatments, the Vertisols gave about60% higher net benefits per hectare than theAlfisols and a standard deviation that was 20%less.5

4. This is largely because low-priced and low-yielding sorghum was the crop in the SIIT exper-iments, whereas in the Alf isol watersheds to-matoes, maize, sorghum, mil let, and other cropshave been prof i tably irr igated. High-valued to-mato has been one of the most responsive cropsin the watersheds and illustrates the importanceof crop choice in determining the value of sup-plemental i rr igat ion.

5. As the Alf isol SIIT data relate to 3 years whereasthe Vertisols only relate to 2, we should not infertoo much from this.

57

Figure 4. Average net benefits and variable costs per ha in steps-in-improved-technology

experiments on Alfisols at ICRISAT Center, 1976-77 to 1977-78.

58

Treatments* and variable cost (Rs/ha)

(•Refer to Table 1 for detai ls of treatments)

Average net benef i ts /ha

Standard deviation of net benef i ts /ha

Figure 5 shows that the profit/risk dominanttreatments are III, Lll and LIL on Alfisols in these3 years, when standard deviation of profits isused as the measure of risk. No other treat-ments can give more net benefits with less risk.The line joining these three points in the figurerepresents the E-V efficiency frontier f romamong which a choice of recommendations canbe made to farmers, depending upon theirrelative preferences for profits vis-a-vis risk.Added profits can only be earned by incurringmore risk at the frontier, as was the case on theVertisols.

The general slope of the E-V curves betweenthe extremeties in Figures 5 and 2 is about thesame at around 0.4.6 Hence it seems the profit/risk tradeoffs on these t w o soil types may besimilar. The steepest segment in Figure 5 be-tween Lll and III has a slope (cotangent) of about0.75. In Figure 2 the the steepest slope is about

Figure 5. Net benefit risk profile of steps-in-

improved technology on Alfisols at

ICRISAT Center, 1975-76 to 1977-

78.

6. We have measured this as where is netbenefi t and its standard deviat ion.

0.9 between III and ll l l. Binswanger(1978), in hisinnovative study of risk attitudes of farmers insix SAT villages of Maharashtra and AndhraPradesh, found that the vast majority of themhad moderate to intermediate levels of riskaversion. This implied a slope of their E-Vindifference curves of between 2 and 3. It seemsthen that the profit/risk efficient technologiesidentified in these SIIT experiments may in-volve added risks per unit of added profits thatare well within the risk aversion preferences ofthe bulk of SAT farmers. This means that f romboth a risk and profit point of view it is reason-able to recommend the III technology in the caseof the Alfisols, and the llll technology in the caseof Vertisols.7

Figure 6 shows that the cost/risk efficientpoints using the CV of net benefits as the riskcriterion, are LIL and Lll, Ill is marginally lessefficient than Lll. For farmers unable to gainaccess to the Rs 1400/ha required for III, theanalysis shows that both LIL and Lll would beattractive alternatives, the latter probably moreso than the former.

When the improved soil- and crop-manage-ment step was implemented on theseAlfisols under local fertilizers wi th either local orimproved varieties, the added net benefits wereless than Rs 300/ha from an extra expenditure ofaround Rs 150/ha. When the same step wasimplemented under an improved fertilizer andlocal variety regime, it resulted in an added netbenefit in excess of Rs 800/ha from an expendi-ture less than Rs 100. The largest increment ofall came f rom the improved soil- and crop-management step introduced on improved var-ieties grown with improved fertilizers. Net be-nefits per hectare increased more than Rs 1200from the expenditure of less than Rs 100. Theselatter two steps, however, both came at theexpense of additional variability in net benefits.

The effect on net benefits of adding improvedfertilizers on the Alfisols when local varietiesand local soil and crop management wereemployed was about the same as for the Ver-tisols, around Rs 600/ha. The effect of addingfertilizers when local varieties were grown, butwi th improved soil and crop management, wasmuch higher on the Alfisols than the

7. The caution noted earlier about the underestima-t ion of the costs of i rr igat ion in the case of l l l l inthe Vertisols must be repeated here.

59

Vertisols — around Rs 1100 compared wi thRs 600. The reverse was true of adding fertilizersunder a situation of improved varieties but w i thlocal soil and crop management. On theAlfisols, this step earned just over Rs 400additional net benefits per hectare whereas onVertisols the f igure was about Rs 1000. Addingfertilizers to improved varieties grown underimproved soil and crop management generated

about the same added net benefits of nearRs 1500/ha on both Alfisols and Vertisols.

It was significant that on the Alfisols theimproved fertilizer step always reduced thevariability of net returns at the same t ime itincreased the average net returns. This was trueregardless of the other four treatments onwhich it was introduced on Alfisols. This wasonly true for one of the four cases on the

Figure 6. Variable cost/coefficient of variation of net benefit profile of steps-in-improved technol-

ogy on Alfisols at ICRISAT Center, 1975-76 to 1977-78.

60

Vertisols: the application of fertilizers underimproved varieties and soil and crop manage-ment.

W a t e r s h e d E x p e r i m e n t s

The Vertisol watersheds on which maize wasintercropped wi th pigeonpea in 1976-77 and1977-78 were, on the average, 73% moreprofitable than the maize plus sequential chick-pea cropping pattern (Table 2). The intercropsystem also was 34% more stable than thesequential crop system. The intercrop profitadvantage over the sequential was much grea-ter on the medium-deep Vertisol watershedsthan on the deep Vertisols (151% better vs52%). However, the reduction in variability fromthe intercrop system compared with the se-quential on the deep Vertisols (39%) was grea-ter than the reduction on the medium-deepVertisols (29%).8 These results would supportthe generally accepted view that intercroppingoffers advantages over sequential (sole) crop-ping, not only in terms of profitability but also interms of risk.

8. It is realized that we have combined both t ime andcross-sectional variabil i t ies here in order to ob-tain an assessment of relative riskiness wi thoutthe appropriate separation of variance compo-nents. Wi th such a l imited data set this seems thebest approximat ion we can make at this stage.

The deep Vertisols seem much more profita-ble than the medium-deep Vertisols but areequally unstable. This is particularly true in thecase of the maize plus sequential chickpeacropping system where the profits per hectarewere more than 120% higher on the deepVertisols. With the intercrop system, thesuperiority of the deep Vertisols was only 40%.

It seems from the experience of these 2 yearsthat on the deep Vertisols the broadbed-and-furrow soil- and crop-management system isabout 20%/ha more profitable under an inter-crop situation than the flat method of cultivationimplemented with the same HYVs and fertilizerlevels.9 It seems about one-third more variablethan the flat system (Table 3). With sequentialcropping the broadbed-and-furrow system wasonly 10% more profitable than the flat systembut was 17% less variable.

9. The difference between the broadbed and fu r rowand the flat system of cult ivation relate to the soilmanagement, where the former has broadbedswi th furrows formed 150 cms apart, comparedw i th no beds in the case of the flat system(fertilizers and varieties are kept the same in both).Analyses of variance of pooled gross returnsf rom the cropping treatments, using replicationvariabil ity as the measure of error, showed thaton deep Vertisols beds and furrows had a highlysignificant superiority over the flat method. Onmedium-deep Vertisols there was no significantdifference (B. A. Krantz, personal communication).

Table 2 . Compar ison o f pro f i tab i l i ty and var iabi l i ty o f t w o cropping systems on Ver t iso l wa te r -sheds a t ICRISAT Center . 1 9 7 6 - 7 7 and 1 9 7 7 - 7 8 .

Watersheds

Deep VertisolsMedium-deep Vertisols

Al l Vertisols

Maize plus sequential chickpea Maize/pigeonpea intercrop

Weighted average Standard deviation Weighted average Standard deviationgross prof i ts of gross profits gross profits of gross profits

per NCHa per NCHb per NCH per NCH(Rs/ha) (Rs/ha) (Rs/ha) (Rs/ha)

2167 1510 3302 921948 1310 2383 927

1712 1499 2958 984

a. NCH = net c ropped hectare. Gross prof i ts are def ined as the tota l gross revenue f r o m sale of gra in and fodder less the costs of

seed, fert i l izer, pest ic ides, insect ic ides, h u m a n labor (charged at ICRISAT wage rates of Rs 4.5 per day), bul lock labor at actual

cost, a n d pro-rated overhead costs o f an ima l -d rawn imp lements . For these crop compar isons no account is taken of land lost

due to def ferent ia ls in areas of bunds , wa te rways etc.

b. S tandard dev iat ions are calculated f r o m u n w e i g h t e d per ha gross prof i ts fo r each of the 2 years taken separately across a l l

wa te rsheds In the compar i son . There are 12 observat ions on the deep Vert iso ls and 10 on the med ium-deep Vert isols.

61

On the medium-deep Vertlsol watersheds(Table 3) the broadbed-and-furrow soil- andcrop-management system was slightly inferiorto the flat-cultivation soil- and crop-manage-ment system. This was true for both the inter-crop and the sequential crop system; there wasnot much difference in the variability of the two.It thus seems the improved broadbed-and-furrow technology does best only on the deepVerstisols. On the basis of these comparisons ofnet profit it cannot be recommended for themedium-deep and shallow Vertisols. If there aresufficient advantages f rom reduction of soilerosion under the broadbed-and-furrow systemthis, of course, could compensate for its lowerprofitabil ity in these soils. However, the evi-dence for reduced soil erosion under the systemis mixed.

The broadbed-and- fu r row system im-plemented on a 0.4% graded contour on thedeep Vertisois generated 20% higher profitsthan those watersheds where the system wasimplemented on a 0.6% grade (Fig. 7).10 Thoseadditional profits came at the expense of addedvariability, particularly in the intercrop system.

The deep Vertisol watershed on which thebroadbed system was implemented within theexisting field border bunds was slightly moreprofitable than when implemented wi th gradedbunds (Fig. 8). The same was true under the flatsystem of cultivation. To the extent this is truegenerally, it suggests that achievement of thepotential benefits of the broadbed-and-furrowtechnology may not require that traditional fieldborder bunds be removed and graded bundsput in their place. It seems that the broadbedscan perform equally wel l wi th in the farmers'own field borders. This may mean that it maynot be necessary to convince contiguousgroups of farmers on a watershed to agree toignore their traditional land boundaries in orderto effectively implement this essential ingre-

10. As is t rue of all the comparisons discussed in th is

sect ion, one cannot attr ibute all of the differencesobserved between the various watersheds con-st i tut ing a g iven system to the system per se. Thisis because other th ings also differ between thesystems, such as soi l depth , drainage, etc., wh ichmay confound the comparisons. The FSRP iscurrent ly developing plans for replicated experi-ments to make more critical and rel iable evalua-t ions of these treatments.

62

Figure 7. Comparison of profitability ( ) and

variability ( ) of two graded slopes

(0.6% and 0.4%) of broadbed-and-

furrow soil and crop management

systems under two cropping sys-

tems on deep Vertisols at ICRISAT

Center, 1976-77 and 1977-78.

client of the prospective SAT technology. How-ever, more research on this issue is requiredbefore a f i rm conclusion can be reached. It ishoped that the watershed research currentlyunder way in ICRISAT's Village-Level StudiesPhase II in Maharashtra and Andhra Pradeshwi l l th row more l ight on this question.

Less research has been conducted on theAlfisol watersheds at ICRISAT in the past thanon the Vertisols; 1977-78 was the first year inwhich we have been able to undertake somecomparisons of the broadbed-and-furrow sys-tems wi th the flat-cultivation system.

In 1977-78, the flat-cultivation watershedsearned on the average almost Rs 2100 (in grossprofits) per ha compared to Rs 1900 in the watersheds wi th broadbeds and furrows. The cropgrown was sorghum fol lowed by sequentialsafflower, wi th adequate fertilizers in bothcases.

Table 3 . Compar ison o f prof i tab i l i ty a n d var iabi l i ty o f t w o cropping systems under t w o soi l- andc rop -management systems on Ver t iso l watersheds a t ICRI SAT Center , 1 9 7 6 - 7 7 and1 9 7 7 - 7 8 . a

Soi l - and crop-managementsytem

Deep VertisolsBroadbeds and furrowsFlat cult ivat ion

Medium-deep VertisolsBroadbeds and furrowsRat cult ivation

Cropping system

Maize plus sequential chickpea

Net benefits per hab

Weighted Standardaverage deviat ion(Rs/ha) (Rs/ha)

2204 14842038 1790

887 14511077 1303

Maize/pigeonpea intercrop

Net benefits per hab

Weightedaverage(Rs/ha)

33952770

23162522

Standarddeviat ion

(Rs/ha)

983729

9521027

a. Calculat ions are on a net c ropped hectare basis.

b. Nat benef i ts are calculated by subt rac t ing overhead land deve lopment costs f r o m gross prof i ts.

In the case of the pearl millet/pigeonpeaintercrop system, fol lowed by tomatoes andsafflower in the postrainy season (becausepigeonpea failed due to caterpillar attack), thegross profits were about the same for thebroadbed-and-furrow and the flat-cultivationsystems, at around Rs 1000 per ha.

A key element in the prospective technologyfor the SAT is the concept of water harvestingand supplementary irrigation in order to in-crease and stabilize crop production (reports ofthe Farming Systems staff, various years). Sever-al of the watersheds at ICRISAT Center havetanks (water storage reservoirs) built on themto collect excess runoff f rom the rainfall for usein supplementary irrigation. We now have 3 years' data available on the performance of thiscomponent, f rom 1974-75 to 1976-77 (Table4). The analysis shows that the response tosupplementary irrigation in these 3 years hasbeen more profitable on the Alfisol than on theVertisol watersheds. Additional net benefits ofalmost Rs 400/ha have been achieved on theformer while losses of around Rs 150/ha haveoccurred on the latter. The returns have beenmost spectacular on the tomatoes grown on theAlfisols, where wi th two irrigations the addedprofits exceeded Rs 4800/ha. Irrigated rainy

season maize and sorghum on Alfisols were thenext most profitable.

Profitable responses to supplementary irri-gation on the Alfisol watersheds have occurredabout 50% of the time, whereas on the Vertisolsthis has been true 30% of the t ime over these 3 years. These data suggest that the potential forwater harvesting and supplementary irrigationtechnology may be higher on the Alfisols thanon the Vertisols. This is also supported by thework of von Oppen and Binswanger (1977), whofind that traditional paddy irrigation tanks inIndia are more densely concentrated wheregranitic substrata exist. These substrata aremostly restricted to the Alfisol regions ofpeninsular India. In another unpublished study,von Oppen shows that these traditional tanksare also more profitable on the Alfisols than onthe Vertisols.

To adequately measure the long-run poten-tial for a water-harvesting and supplementaryirrigation technology aimed at supporting up-land crops — as opposed to the traditional tanksystems used for irr igation of lowlandpaddy — it is necessary to assess the prob-abilities of being able to have sufficient excessrunoff stored and available in small tanks on theuplands at the t ime when crops can profitably

63

Figure 8. Comparison of profitability and

variability of two types of b u n d -

ing under two soil and crop m a n -

agement systems on deep Vertisols

with maize|pigeonpea intercrop at

ICRISAT Center, 1976-77 and

1977-78.

respond to irrigation. The work that has beenunder way at ICRISAT on this is discussed in thenext section.

E v a l u a t i o n o f W a t e r H a r v e s t i n ga n d S u p p l e m e n t a r y I r r i g a t i o nU s i n g S i m u l a t i o n

The extensive rainfall-runoff data available atICRISAT Center and the AICRPDA Station atSholapur have been used by Ryan and Pereira(1978) to derive empirical models for the predi-ction of runoff in these two areas. As the detailsof the data, methodology, and selected modelsare provided in that paper, we wi l l not attemptto duplicate them here. The range of variablesemployed in the regression equations selectedto predict runoff by Ryan and Pereira were:

64

(1) Daily runoff = f (daily rainfall, vegetativecover, area of water-shed|p lot , r idge andfurrow dummies, flatcultivation dummy, typesof bunding dummies,antecedent rainfall, veg-etative cover index, rain-fall intensity, soil type,soi l d e p t h , age ofwatershed).

The derived empirical prediction equationshave been incorporated into simulation modelsthat use historical daily rainfall distributionsand generate daily runoff estimates for speci-fied levels of the deterministic variables such assoil type, area of catchment, size of tank, type ofcultivation and vegetative cover. Evaporationlosses f rom the tank surface are determined byusing monthly open-pan averages converted toa daily basis, wi th separate values for rainy andnonrainy days. The tank design has been as-sumed to be basically cylindrical in order tocalculate surface evaporation losses in storage.For these studies, a tank of 20 ha/cm has beenassumed and catchment sizes of 3, 8, 12; 16, and20 ha simulated, using 70 years of daily rainfalldata f rom the Hyderabad weather station since1901 and 29 years f rom the Sholapur stationsince 1945.

We have chosen to present the results of thesimulations for week 43; that is, the end ofOctober. This is because, particularly for theVertisols wi th their better moisture-holdingcapacities, it is possible that most of the payofff rom water harvesting may derive from increas-ing the chances and yields of sequential post-rainy season crops. We are in the process ofconstructing a model that wi l l enable evaluationof the relative benefits of utilizing the harvestedrunoff water on rainy season crop versus savingit for postrainy season crop irrigation. However,as it is not yet completed, we thought it useful torestrict ourselves for the present to the post-rainy season option.

Table 5 shows that under cropped conditionson the broadbed-and-furrow system at ICRISATCenter, the Alfisols have about double therunoff potential of the Vertisols. Under croppedflat conditions the Alfisols have only about 25%more potential. The Hyderabad Alfisols havealmost three t imes the runoff potential of theSholapur Vertisols under a cropped ridge-and-

Table 4 . Add i t i ona l ne t benef i ts f rom supplementary i r r ig at ion f r o m runof f s to rage tanks a tICRISAT Center , 1 9 7 4 - 7 5 t o 1 9 7 6 - 7 7 .

Crops irrigated

Rainy-season maizeRainy-season sorghumPostrainy-season sorghumRatoon sorghum

Tomatoes (ridges and furrows)- one irr igat ion- two irr igation

Tomatoes (flat)- one irr igation- two irr igation

Pearl mil letSaff lowersunf lowerChikpea

Setaria

Overall average net benefit/ha

Overall standard deviat ion of net benefitlha

Weighted average additional net benefits f rom irr igat iona

(Rs/ha)

Vertisolsb

- 4 4 75050

- 7 2 580

- 154

660

Alfisols

27812455

117- 7 6

30534861

12402375

1054- 2 9 4

70< 0

< 0

380d

1180d

a. Weighted by the area irrigated.b. About 85% of the Vertisols were deep where the irrigation was applied; the others were medium deep.c. Comparisons are on the basis of increased yields over and above zero irrigation.d. Using single irrigation for tomatoes only.

furrow system. Under a flat-cropped system,they can generate about 50% more runoff thanthe Sholapur Vertisols.

The simulations clearly show that thecoefficients of variation of net cumulative runoffat the end of October are inversely related to themean levels of cumulative runoff at that t ime.This is true whether stored runoff changes dueto changes in vegetative cover or catchmentsize. This suggests that in situations such as theSholapur Vertisols, which have limited poten-tial for runoff, there wil l be a much greatervariability in the availability of runoff forirrigation — and hence more risk to the use ofthis sys tem—than in situations where runoffpotential is high, such as on the HyderabadAlfisols. The CVs reach as high as 130% in theSholapur Vertisols on a 3 ha catchment wi thcropped ridges and furrows, whereas on a similarly treated and sized Hyderabad Alfisol

catchment the CV can be as low as 28%.The effect of increased catchment size on

runoff generation is not proportional, due to thenegative signs on the area variables in theprediction equations (Ryan and Pereira 1977,pp 17, 22, 27).11 For example, increasing thecatchment sizes f rom 3 to 8 ha (170%) increasednet cumulative runoff by around 70% on theHyderabad and Sholapur Vertisols and by only40% on Hyderabad Alfisols under cropped

11. The Sholapur runoff plots used to derive theequations were extremely small (0.002 to 0.007ha) compared w i th the small watersheds used atHyderabad (0.33 to 11 ha). Hence, to adjust theSholapur equations to enable predict ion forcatchments up to 20 ha, the max imum plot sizefor Sholapur was Inserted into that equation andthe coefficient on area f rom the Hyderabad Ver-tisol equation was also incorporated.

65

ridges and furrows (Table 5). In addit ion, evap-oration losses increase wi th larger catchmentsas they more often have runoff in storage tanksthan do smaller catchments.

The amounts of excess runoff which the 20ha/cm storage tank could not capture are posi-tively related to the mean net cumulative runofftotals at the end of October. Hence, the morerunoff that is harvested on average, the moreone could harvest if the runoff was profitable touse. This is where we now turn in our analysis.

What is the value of the stored runoff at theend of October? To determine this, we haveadopted the approach of calculating the break-even crop yield increments required to recoupthe costs of construction and operation of thetank, pump and pipes required for the waterharvesting and supplementary irrigation sys-tem.

The basic equation for calculating the break-even yield assuming all costs wi l l be borne by a single group of watershed farmers was derivedas:

(2)where = weighted average break-even

yield over i years (i = 1, 2, 3 (N= price of crop j (net of harvesting,

threshing and marketing costs)= averagenet cumulative runoff at

week 43 over i yearsIf we assume that one pump and a set of pipes

can be successfully shared by farmers f romthree catchments instead of the one assumed inequation (1) we obtain:

(2a)Both (2) and (2a) generate estimates of the

average break-even yields which must beachieved in every one of i years. However, therewil l undoubtedly be years when R I wi l l be zero.Thus a more useful measure of the economicviabil ity of the technology is obtained by deriv-ing the break-even yields in years when onlynon-zero runoff is available at week 43. This iscalculated as:

66

We have derived these break-even yield esti-mates for three agroclimatic situations wi th a range of catchment sizes, vegetative covers,and types of cultivation (Tables 6, 7, 8). Al l havebeen calculated assuming a crop priced at Rs 90per 100 kg and use of irrigation. They show thatthe economic potential for water harvesting andsupplementary irrigation in the postrainy sea-son is best on Alfisols at ICRISAT Center, fol-lowed by Vertisols at the same location, andthen on Vertisols at Sholapur. The superiority ofthe Hyderabad Vertisols and Alfisols over theSholapur Vertisols increases as we move from a flat cultivation to a ridge-and-furrow system.For example, the break-even yields required innon-zero runoff years on Vertisols at Hyderabadunder a flat-cropped situation are 28 to 66% lessthan on Sholapur Vertisols, depending oncatchment size, whereas under a ridge-and-furrow cropped system, they are 50 to 63%. ForAlfisols at Hyderabad, the comparable figuresare 43 to 60% and 77 to 80%, respectively.

It would seem that, unless there are goodreasons for believing that crop responses to thesame amounts of irrigation are going to bevastly superior in the Sholapur Vertisols, it wi l lbe more profitable to concentrate on situationssuch as one has on the Hyderabad Alfisols andVertisols than on Sholapur Vertisols. Whetherwater-harvesting and supplementary irrigationtechnology is going to be profitable depends onthe ability of crops to generate yield incrementsin excess of the break-even figures derived inTables 6, 7, and 8.

The likelihood that water-harvesting andsupplementary irrigation technology based onpostrainy season crop irrigation is going to beeconomically feasible if implemented onHyderabad Vertisols under a cropped ridge-and-furrow regime is doubtful according to ouranalysis. Under the optimistic assumption thatthe costs of a pump and pipes can be shared bythree groups of watershed farmers, yields ofSAT cereal foodgrain crops such as sorghumand maize would have to increase by more than1800 kg/ha in the years when runoff is available(77% of the t ime) on a catchment size of 3 ha(Table 6). Yield increases on a 16-ha catchmentwould have to exceed 800 kg/ha in the 97% ofyears when runoff wou ld be available at the endof October for supplementary irrigation. If itproves infeasible to share the pump and pipeswi th any but a single group of watershed

(3) Yj =

in the case o

(3a) Yj =

in the case

N

Pj ( N - Q )

f ( 2 ) a n d :N

P j(N - Q)of (2a), whenumber of

(88 + 10470/R)

(88 + 7500/R)

reyears when Ri = 0 Q =

67

Tab le 6 . B r e a k - w e n c rop y ie lds requi red to cover coats o f w a t e r harvest ing and supp lementaryi r r iga t ion sys tem on Ver t i so ls a t ICR ISAT Center . a

Catchmentsize(ha)

38

121620

38

121620

38

121620

38

121620

Type ofvegetative

cover

Fallow

Cropped

Fallow

Cropped

Type ofcult ivation

Flat

Flat

Ridges andfur rows0.4%grade

Ridges andfur rows0.4%grade

Addit ional break-evenyield required in N

yearsb on

1 Watershed 3 Watersheds

(kg/ha)

1140810760770770

1220830790780790

171011301000

970960

18801180104010201040

850610570580580

1000620590590590

1260840740720710

1380880770760770

% of yearsw i th zerorunoff atend Oct(100Q/N)

( % )

44000

60000

266933

236639

Addit ional break-even yield requiredin ( N - Q ) yearsb on

1 Watershed 3 Watersheds

(kg/ha)

1190850760770770

1290830790780790

2310120010901000990

24401260111010501140

880630570580580

950620590590590

1690890810740740

1790930820780850

a. A crop va lued at a P j Rs 90 per 100 kg is assumed here such as s o r g h u m , maize or pearl mi l le t . If a c rop va lued at Rs 240 is

assumed, such as p igeonpeas or chickpeas, then a l l t h e break-even y ie lds here w o u l d decrease by about two- th i rds .

b. N = tota l number of years used in the s i m u l a t i o n ; Q = n u m b e r of the N years w h e n net cumula t i ve run off at week 43 is zero.

farmers, the break-even yields on a 3-ha catch-ment would increase to more than 2400 kg/haand those on a 16-ha catchment to almost 1100kg/ha. These break-even figures do not de-crease greatly even if a fal low is maintainedinstead of a crop.

For the Sholapur Vertisol situation, it wouldnot seem that water harvesting and supplemen-tary irrigation based on postrainy season cropirrigation could be profitable, except perhapsunder a flat fal low system. On a 3-ha croppedridge and-furrow watershed, yield incrementsin excess of 3600 kg/ha would be required topay the costs of the technology, assuming threegroups are sharing pump and pipe costs (Table7). On a similar 16-ha catchment more than 2200

kg/ha would still be required. On a flat fal lowcatchment of 3 ha, the comparable break-evenyields are about 1200 and 900 kg/ha, respec-tively.

Achievement of any of the above yield incre-ments may be quite difficult in the Vertisols, whichhave a much higher moisture-holding capacitythan the Alfisols. In years of good rainfall(and hence runoff) soil moisture profiles wi l l berelatively ful l and crop response to irrigationmay be small. Also, even though the prob-abilities of a further " w e t " week after week 43 inHyderabad are less than 32% and in Sholapurless than 43% (Virmani et al. 1978; p 12), therewi l l be some occasions when rain falls after theapplication of irrigation, thus negating the ef-

68

Table 7 . Break-even crop yields requi red to cover coats o f wa te r harvest ing and supp lementary

Catchmentsize

(ha)

38

121620

38

121620

38

121620

38

121620

i r r igat ion sys tem w i t h pump and pipes shared b e t w e e n fa r mers on Ver t iso ls a t A I C R P D As ta t i on , Sholapur. a

Type ofvegetative

cover

Fallow

Cropped

Fallow

Cropped

Type ofcult ivation

Flat

Flat

Ridges andFurrows

Ridges andFurrows

Addit ional break-evenyield required in N

yearsb on

1 watershed 3 watersheds

(kg/ha) -

1280950980

10201090

14901070111011501220

21001430143014601540

25501540152015401600

940710730760790

1100800820860900

15301050106010701080

18501130112011301180

% of yearswi th zerorunoff atend Oct

(100 Q/N)(%)

1714171724

1714243131

4134414448

4845484848

Addit ional break-even yield requiredin (N -Q) b years on

1 watershed 3 watersheds

(kg/ha)

15401100119012301430

18001240146016701780

36002180245026402980

49202780294029703100

1140820880910

1050

1320920

108012401310

26101610180019402080

35802040216021802270

a. A c rop va lued at a P j of Rs 90 per 100 kg is assumed here such as s o r g h u m , maize or pearl mi l let . If a crop va lued at Rs 240 is

assumed, such es p igeonpeas or chickpeas then all the break-even yields here w o u l d decrease by about two- th r ids .

b. N = tota l number of years used in the s imu la t i on ; Q = number of the N years when net cumulat ive runof f at week 43 is zero.

feet of irrigation and probably not producingcompensatory runoff. All these factors act tosuggest that these break-even yield estimatesare absolute min imum levels required to justifyinvestment in the system.

The Hyderabad Alfisol situation seems quitedifferent. Yield increments of less than 1000kg/ha are required to justify investment in waterharvesting and supplementary irrigation ofpostrainy season crops. It can be as low as 600kg on catchments exceeding 8 ha (Table 8). Withthe experience we have had at ICRISAT withyield increments on Alfisols wi th irrigation, itseems these figures should be achievable.

As mentioned earlier, the present analysishas assumed all runoff is saved for irrigation of

postrainy season crops and not used for irri-gation of rainy season crops when they becomestressed. To the extent the latter is a moreprofitable use of stored runoff, then the break-even yields reported here may be overesti-mates. To adequately answer this requiresmore extensive simulation of alternative irri-gation decision rules employing crop yield-irrigation response functions. Work is presentlyunder way on this. However, based on experi-ence to date and on a priori grounds, it isunlikely that rainy-season supplementary irri-gation on Vertisols wil l be profitable. B. A.Krantz (personal communication) suggests thatthe mere availability of stored runoff in the rainyseason can act as an insurance for the farmer,

69

Tab le 8 . Break-even crop y ie lds requ i red to cover cost s o f w a t e r harvest ing and supp lementa ry

Catchmentsize(ha)

38

121620

38

121620

38

121620

i r r igat ion sys tem o f A l f iso ls a t ICRISAT Center . a

Type ofvegetative

cover

Cropped

Cropped

Cropped

Type ofcult ivat ion

Flat

Ridges andFurrows0.4%Grade

Ridges andFurrows0.4%Grade

Addi t ional break-evenyield required in N

years6 on

1 Watershed 3 Watersheds

(kg/ha)

1010730720720730

1020750730730730

800730720720720

750550540550550

760560550550550

600550550540540

% of yearsw i th zerorunoff atend Oct

(100 Q/N)( % )

10000

30000

30000

Addi t ional break-even yield requiredin ( N - Q ) yearsb on

1 Watershed 3 Watersheds

(kg/ha)

1020730720720730

1050750730730730

820730720720720

760550540550550

780560550550550

620550550540540

a. A c rop va lued at a P j of Rs 90 per 100 kg Is assumed here such as s o r g h u m , maize or pearl mi l le t . If a c rop va lued at Rs 240 Is

assumed , such as p igeonpeas or chickpeas t h e n all the break-even y ie lds here w o u l d decrease by about two - th i rds .

b. As there were no f a l l ow t rea tments in the runof f data i t was not poss ib le to der ive runof f or break-even y ie ld est imates fo r

f a l l ow s i tuat ions. In these l o w mo is tu re -ho ld ing capaci ty soi ls post - ra iny season c ropp ing after fa l lows is p robab ly n o t

w o r t h w h i l e anyway .

c. N = to ta l number of years used in the s i m u l a t i o n ; Q = number of the N years w h e n net cumu la t i ve runo f f at week 43 is zero.

leading him to invest more in HYVs, fertilizers,plant protection, and other inputs than hewould wi thout a tank. However, any such in-surance effect can only operate over the poten-tially irrigable parts of the catchment. Witha cropped-and-ridged catchment of 8 ha forexample on Hyderabad Vertisols, only 25% ofthe catchment can be irrigated on average. OnHyderabad Alfisols the figure is 44% and onSholapur Vertisols 20%. The fact also remainsthat the insurance value must be real; whenthere is a dry spell in the rainy season theremust be viable amounts of stored runoff in thetank to save the crops. In the dry years when theinsurance is most needed, the probabil ity thatthere wi l l be insufficient runoff wi l l also in-crease. The answer to this requires simulationof the historical data as we are currently doing.

It seems that a m in imum catchment size

might be about 8 ha. Break-even yields risesubstantially for catchments less than this. Italso seems that once catchments exceed about16 ha the break-even yields begin to rise due tothe negative effects of area and evaporation onrunoff.

If instead of a crop worth Rs SO per 100 kg,higher-valued crops of say Rs 240 are irrigated,then the break-even yields in Tables 6 - 8 wouldall fall by a factor of two-thirds.

Conclusions

In the Vertisol steps-in-improved-technologyexperiments at ICRISAT Center, the completepackage of technology generated the highestprofits of Rs 3600/ha, which represented a 190%rate of return on variable costs of Rs 1900/ha.The t w o most profit/risk efficient treatments

70

were local or improved varieties with improvedfertilizers along wi th improved soil and cropmanagement. Irrigation applied to either ofthese increased profits but also substantiallyincreased variability of profits.

On Vertisols, the implementation of im-proved broadbed-and-furrow systems underlocal fertilizers with or without HYVs, or underlocal varieties with improved levels of fertilizergenerated Rs 500 additional profits per hectare.If implemented under HYVs and improved fer-tilizers, this rose to Rs 1000. If these small fieldresults are replicable on larger watersheds,added profits could pay for the extra Rs 6500capital costs of a bullock-drawn wheeled toolcarrier in about 1 year if it is used on 7 ha.

For an added investment of about Rs 400 perha on both the Vertisols and the Alfisols, theincrease in profits f rom the addition of fertilizersranged between Rs 600 and Rs 1500, dependingon the varieties and soil- and crop-managementtreatments employed.

In the Alfisol steps-in-improved-technologyexperiments, profits of Rs 2000/ha f rom the fullpackage were 60% Iess than those generated onthe Vertisols. Variability, however, was 20%less.

The local variety, improved fertilizers andimproved soil- and crop-management treat-ment, along with the local variety, improvedfertilizer, and local soil- and crop-managementtreatment, were the most profit/risk superior ofall treatments on the Alfisols. The levels of theprofit/risk tradeoff implied by these two treat-ments, as well as the two efficient ones on theVertisols, are well within the risk aversion limitsrevealed by the sample of farmers in the IndianSAT studied by Binswanger (1978). Hence, onewould be confident in recommending these tofarmers.

The implementation of improved broadbed-and-furrow systems on Alfisols wi th local fer-tilizers and local varieties or HYVs increasedprofits by about Rs 300 ha. With local varietiesand improved fertilizers, this increased toRs 900, while with HYVs and improved fertilizerthe increased profits amounted to Rs 1200/ha.

In the larger-scale watershed experiments onVertisols at ICRISAT Center, the maize/pigeonpea intercrop system, on the averageover 2 years, generated 73% higher profits thanthe maize-plus-chickpea sequential crop sys-tem. The intercrop system also had 34% less

variable profits. This intercrop advantage wasgreater on the medium-deep Vertisols than onthe deep Vertisols.

Deep Vertisol watersheds in general were 40to 120% more profitable than those onmedium-deep Vertisols. Broadbed-and-furrowsystems with intercrop maize/pigeonpeas onthe deep Vertisol watersheds were about 20%more profitable than the flat system of inter-cropping. This advantage is much less thanshown in the steps-in-improved technologyexperiments and adds a caution on makinginferences from small-field experiments towatersheds that can be more than ten times thesize of experimental fields, particularly wheresoil-management systems are being used.Under sequential maize and chickpea the ad-vantage of the broadbed-and-furrow system ondeep Vertisols was reduced to 10%. Broadbed-and-furrow systems were less profitable thanflat systems on medium-deep Vertisols.

The next step in this work is to incorporate thevarious steps in technology into activityanalysis models that include both profit and riskconsiderations. Programming optimal technolo-gy choices under various constraint situationscan determine the place of this new technologyvis-a-vis traditional technology.

The profits from broadbed-and-furrow sys-tems implemented within traditional farm fieldbunds on deep Vertisols were comparable tothose implemented on a complete watershedwith graded "gu ide" bunds. If this result isreplicable, it suggests that the broadbed-and-furrow system may not require implementationon a full watershed, with all the potential prob-lems of group action this may entail, to have itsadvantages expressed. It may be possible toachieve the benefits on an individual holdingbasis.

On the Alfisol watersheds preliminary indi-cations are that the flat methods are more profit-able than the broadbeds and furrows. The aver-age profitability of supplementary irrigationfrom excess runoff harvested on the water-sheds has proved to be much greater on theAlfisols at around Rs 400/ha. It has been up toRs 5000/ha in the case of irrigated tomatoes. Onthe Vertisols, losses from irrigation have aver-aged Rs 150/ha.

Simulation models, using derived rainfall-runoff relationships to evaluate the economicpotential of postrainy season supplementary

71

irrigation f rom harvested runoff water, showedthat ICRISAT Center Alfisols have double therunoff-producing potential of the Vertisolsunder cropped ridges and furrows. The ICRISATCenter Alfisols have three t imes the runoffpotential of the Sholapur Vertisols.

The coefficients of variation of annual netcumulative runoff available at the end of Oc-tober using historical rainfall distributions wereinversely related to the mean level of annual netcumulative runoff. Hence those areas whererunoff generation potential is least wi l l also bethe ones where runoff availability wi l l be mostvariable.

It is doubtful that water-harvesting andsupp lementa ry - i r r iga t ion techno logy forpostrainy-season crops can be profitable onHyderabad Vertisols under a cropped ridge-and-furrow system; it is more likely under a flat-fallow system where more runoff is pro-duced. It seems more or less certain that thistechnology wi l l not be economically viable onthe Sholapur Vertisols. The Alfisols offer muchbetter possibilities for this type of technology.More research is required to determine therelative merits of utilizing stored runoff forirrigation of rainy season versus postrainy sea-son crops. Only the latter alternative has beenexamined here.

The opt imum economic size of catchment inthese three areas would seem to be between 8 and 16 ha.

References

BINSWANGER, H. P. 1978. Risk att i tudes of ruralhouseholds in semi-arid tropical India, Economicand Political Weekly (Review of Agr icul ture)13(25):A49-A62.

FARMING S Y S T E M S STAFF. Various years. Annua l Re-

por t of the Farming Systems Research Program (Anim fo rma l report), ICRISAT, Patancheru, A.P., India.

PERRIN, R. K., W I N K E L M A N N , D. L., MOSCARDI , E. R. and

ANDERSON, J. R. 1976. From agronomic data tofarmer recommendat ions: an economics t ra in ingmanual . In format ion Bullet in 27, Centre Interna-cional de Mejoramiento de Maize Y Tr igo (CIMMYT),Mexico City, Mex ico.

72

R Y A N , J. G. and SUBRAHMANYAM, K. V. 1975. Package of

practices approach in adopt ion of high-yieldingvarieties: an appraisal. Economic and PoliticalWeekly (Review of Agriculture) 10(52): A101-A110.

R Y A N , J . G., and S A R I N , R. 1977. Economic analyses of

steps in improved technology exper iments .ICRISAT Economics Program, Patancheru, A.P., In-dia. (Mimeographed.)

R Y A N , J. G., and S A R I N , R., 1978. Economic analyses of

steps in improved technology exper iments, IC-RISAT Economics Program, Patancheru, A.P., India.(Mimeographed.)

R Y A N , J. G., and PEREIRA, M. 1978. Derivat ion of

empir ical models for the predict ion of runof f onsmal l agricultural watersheds in the semi-aridtropics. Presented at the International Workshop onthe Agrocl imatological Research Needs of theSemi-Ar id Tropics, ICRISAT, 22-25 Nov 1978,Hyderabad, India. (Mimeographed.) Avai lable f romICRISAT, Patancheru P.O. A. P. 502 324, India.

VIRMANI , S. M. SIVAKUMAR, M. V. K., and REDDY, S. J.

1978. Rainfall probabi l i ty estimates for selectedlocations of semi-ar id India. ICRISAT Research Re-port 1, Patancheru, A. P., India.

VON OPPEN, M., and BINSWANGER, H. P. 1977. Inst i tu-

t ional and physical factors affecting tank irr igat iondensity in India. Presented at the Seventh Annua lConference of the Indian Econometric Society, Nov1977, T r i vandrum, India.

N e w Agr icu l tu ra l Techno logyin t he Brazi l ian Ser tao

J o h n H . S a n d e r s *

Abstract

This paper describes the relevant characteristics of the Brazilian Sertao with regard to new agricultural technology. It then reviews the research effort on one new technology — the introduction of sorghum — drawing some useful lessons. In the given region, labor is not merely a seasonally limiting factor of production; laborers also suffer from chronic nutritional problems. Thus, the effects of new technology on the distribution of income of producers have efficiency as well as equity considerations. The potential for increasing yields of basic foods depends on fertilizer response and soil conditions, both of which tend to be poor at present. It is necessary to first identify plants with drought-tolerance potential and then select varieties and undertake basic ag-ronomic research. As the agronomy results studied show that sorghum outyielded corn, and the modeling and farm trials show that the introduction of sorghum fits into the farmer's other activities and is profitable, policy intervention in favor of grain sorghum seems called for. Several steps are suggested for such intervention.

The Brazilian Northeast has the largest concen-tration of poverty in Latin America wi th a percapita income of approximately $200 in 1970.Fifty-two percent of the land and 4 1 % of thepopulation (11.5 mill ion) are in the semi-aridregion, the Sertao (Table 1). Out-migration f romthe rural areas has been occurring at a stagger-ing rate. Nearly 2 mil l ion left the rural areas ofthe Northeast in the 1958 drought (Ward andSanders, 1978). Meanwhile, population growthis continuing at a 2.5% rate, further aggravatingthe present inadequacies of nutri t ion, publichealth, and human capital formation in a regionwith inadequate, irregular rainfall.

What can new agricultural technology do inthis environment? What are the appropriatetypes of agricultural technology? How does thisrelate to ICRISAT? This paper wi l l attempt to atleast provide some background to respond tothese questions. In the first section some rele-vant characteristics of the region will be brieflydescribed. Then some proposed and potentialtechnologies wi l l be considered. Some more

* Centro Internacional Agr icul tura Tropical (CIAT),Call, Colombia.

specific results of modeling and field testing ofthe introduction of sorghum into the region wil lbe analyzed. A final section wil l attempt togeneralize the process of ex ante technologyevaluation.

C r o p s , D r o u g h t ,a n d C o n s t r a i n t si n t h e T a r g e t A r e a

Cattle raising was the principal reason for settl-ing the Sertao as land utilized for sugarcane andsubsistence crops in the Zona da Mata (humid/coastal strip) became too valuable to producecattle in the late eighteenth century. Settlementincreased along the rivers, and later dams, asthe vaqueiros (cowboys) acquired their owncattle and became independent. In thenineteenth century the wor ld demand for cottonsubstantially increased, especially during theU.S. Civil War. This hastened the developmentof the northern Sertao northwest of the Zona daMata, i.e. in the states of Piaui, Ceara, and RioGrande do Norte, wi th the diversification intotree cotton production combined wi th the live-stock activity. Sertao land distribution evolved

73

Tab le 1 . T h e popu la t ion and area o f t h e d i f f e ren t eco log ic subregions o f t h e Braz i l ian Northeast?

Subregions

Humid coastal str ip("Zona da Mata")

Transi t ion zone("Agreste")

Drought Polygon("Sertao")

Transit ion to AmazonHumid valleys and highlandsTransit ion to central plateau

("Cerrados")Total northeast

Populat ion (1970)

No. (mil l ion)

7.7

4.1

11.5

3.41.30.2

28.1

%

27.3

14.5

41.0

12.o4.50.7

100.0

Area%

7.2

10.8

52.4

22.42.44.8

100.0

a. Source: Unpublished data from SUDENE.

into two principal types: large cattle estates,often wi th tree cotton tended by sharecroppers,and small subsistence units. Over t ime highpopulation growth and the need for all Sertaofarms to be located near a water source resultedin these small farms acquiring their presentlong, narrow, rectangiriar shapes, w i th one endof the rectangle on the river.

As in most cattle zones, land distr ibution inthe Northeast became extremely concentrated.In 1970, 68% of the farms in the Northeast wereless than 10 ha but only occupied 5.6% of theland area. Ninety-four percent of the farms wereless than 100 ha, occupying only 30% of thearea (Censo Agropecuario Brasil, 1970). TheNortheast had the largest number of farm prop-erties of all the Brazilian regions wi th 2.2 mil l ionin 1970. There were 873 000 farms less than10 ha — t h e largest concentration of small hold-ings in Brazil (almost 59% of the total in thecountry). At the same t ime, there were 3572properties wi th more than 10 000 ha in theNortheast. These 3572 properties had a largerland area (3.6 mil l ion ha) than that of the entiresmall holding group (2.7 mi l l ion ha).1 The landdistr ibution problem is further complicated bythe lack of fertile areas except for isolatedregions of h igh ferti l i ty generally on the coastand in some valleys and highlands

1. Data f r o m the Prel iminary 1970 Census cited inPaiva et a l . (1973, pp 372-3) .

The predominant crop activities of the north-ern "Ser tao" are tree cotton-corn-cowpeas inassociation (or mixtures). In the first year of thefive-year economic cycle of the tree cotton, cornand cowpeas are planted between the cottonrows by the sharecroppers and the small far-mers. After the first year the small farmergenerally prunes the tree cotton back and plantscorn and cowpeas again whereas the share-cropper goes on to another area of the same oranother farm to repeat the cycle of obtaining hissubsistence crops while establishing the land-lord's investment crops, cotton and pasture.2 Inthe Sertao south of Paraiba tree cotton becomesless important; instead there is a greater con-centration on cattle and supporting activitiessuch as "pa lma, " a spineless opuntia cactuswi th a high water content which is often com-bined wi th cottonseed meal as cattle feed. Fieldbeans, cassava, castor beans, upland rice, sisal,and tobacco are also important in particularregions. Livestock is a method of capital for-mation for small and medium farmers and isthe principal activity of most larger farmers.The tree cotton leaves serve as an emergencypasture reserve in inadequate rainfall years.

The farmer's approach to drought is to plantlong-term crops wi th good drought resistancesuch as tree cotton, castor beans, and cassava.His food crops — cowpeas and corn — are

2. For fur ther detai l see Johnson (1971).

74

planted in the 3- to 5-month rainy season.Pasture, fruits, sweet potatoes, and rice arecultivated on the small areas of the farm withwater. Unfortunately, rainfall availability for thebasic food crops (corn and cowpeas) is veryerratic during the growing season.

Most of the Sertao receives from 500 to 800mm of rainfall concentrated in a 4-monthperiod. Approximate probabilit ies of in-sufficient rainfall within the growing seasonare:3

• Full-scale drought (i.e., no rain dur ing mos t ofg row ing season) — 12% (last one in 1970).

• Min i -drought (i.e., no rain for 2 to 6 weeks ingrowing season) — 3 0 % (last one in 1976).

Cultural practices also revolve around thisrainfall irregularity. In Irece, a large-farmer beanarea in the Sertao of Bahia state, two to threeplantings are often necessary. In small-farmerregions, farmers plant a mixture of varieties ofthe basic food crops and plant at several t imesto minimize risk. Fertilizer use is generally zero,because of little response wi th present varietiesand irregular rainfall conditions.4 The mostobvious pressing constraint is the lack of water;thus drought tolerance and better methods to

3. These subjective estimates were obtained f r omresearchers in the agricultural school of the Fed-eral University of Ceara. Substantial rainfall dataare available. See Hargreaves (1973).

4. For a review of fertilizer studies on tradit ionalNortheastern crops see the references cited inSanders and de Hol landa, (1979, p 119).

take advantage of available moisture are themost important new technologies.

After water-saving technologies, there issome sketchy evidence that labor is more con-straining than land. Land prices have been in-creasing slower than labor prices have been inBrazil and in the Northeast (de AlbuquerqueLima and Sanders 1978 p 101). Moreover, inten-sity of land utilization decreased dramatically asfarm size increased slightly (Table 2|. Slightlylarger small farmers cultivated less than 20% oftheir area, whereas farmers having 10 ha or lesscultivated almost 60%. With the lack of chemi-cal fertilizer inputs and the low initial fertil ity ofmost of the soils in the Sertao, long rotationperiods are practiced to give a slight fertilityboost to the low yields of the basic crops.

If farmers take the risks inherent in semi-aridagriculture, there is additional land available inspite of the high concentration. Sharecroppersmove between farms, and seasonal laborshortages — especially for the necessary weed-ing after the intense rains — are important indetermining cotton yields and area cultivated.5

By overcoming these constraints of seasonallabor availability a reasonable internal rate ofreturn of 35% is attained wi th the purchase of ananimal cultivator (de Albuquerque Lima andSanders 1978, pp 107-110). This inexpensiveimplement, useful for both cultivation and landpreparation on lighter soils and manufactured

5. See A. W. Johnson, (1971) for more detail onsharecroppers and Albuquerque Lima and San-ders, (1978, pp 110-112) for more informat ion onseasonal labor problems.

Tab le 2. Land use in tens i ty for crops in a samp le of smal l fa rmers in Caninde Ceara , 1 9 7 3 .

Indicators

Mean area cult ivated (ha)Land use intensitya

Number of observations

Avai lable land area (ha)

Up to 10 10 -20

3.8 5.70.59 0.49

10 16

2 0 - 3 0

8.50.369

3 0 - 4 0

8.80.277

4 0 - 5 0

9.40.248

50 -60

12.20.244

60 -70 7 0 - 8 0

10.3 11.10.17 0.187 5

a . Measured a s (A c /A t - A N A ) , w h e r e

A c : cu l t iva ted area in crops

A t : to ta l area

A N A : a r e a not approp r ia te for agr icu l tu re accord ing to the farmer .

Source : de A l b u q u e r q u e L ima a n d Sanders (1978, p 103).

75

in Sao Paulo, is presently being adopted bysmall farmers in the Northeast.

Labor is not only a seasonally l imit ing factorof production, but is also affected by chronicnutrit ional problems. In a sample of Ceara smallfarmers and sharecroppers, 46% of the familieshad inadequate caloric levels. Protein deficitswere less because of the cowpeas and small-scale livestock production. Surprisingly, thenutrit ional situation of the low income urbaninhabitant was even worse (Ward and Sanders1978, pp. 8, 17). An inadequate investment innutritional and related human needs of low-income rural inhabitants implies that theyprobably wi l l not be very successful as migrantsto the city. If the rural migrants are unable tomove out of their low income position in theurban area, their nutrit ional situation becomesworse. Thus, the effects of new agriculturaltechnology on the distribution of income ofproducers has efficiency as well as equity con-siderations, given the present human capitaldeterioration in the Sertao.

A g r i c u l t u r a l T e c h n o l o g yA l t e r n a t i v e s i n t h e S e r t a oa n d P u b l i c P o l i c y

Since the Sertao was developed as a cattleregion and most of the area is concentrated inlarge holdings, it is not surprising that mostregional effort in research as well as agriculturalcredit has been for cattle. One theory of Sertaodevelopment has been to foster out-migrationso the minifundia could be consolidated intolarger units more efficiently producing higherprice and income elasticity products such ascattle. In the late 1950s the central governmentbegan promoting industrialization in the urbanNortheast through income tax credits and otherdirect subsidies. Since 1960 the urban popu-lation of the Northeast has been increasingat a 4.6% annual rate compared wi th only 1.2%for the rural populat ion (Ward and Sanders,1978, pp 17-18).6

In addit ion to rural-urban migrat ion, rural-rural migration has been encouraged throughcolonization projects. Two weeks after person-ally witnessing some effects of the drought of

6. For further detai l see Moura (1971).

1970 President Medici implemented the PIN(Programa de integragao Nacional) program ofnational highway construction to penetrate un-settled areas in the Amazon and northern MatoGrosso (Sanders 1973). The Transamazonhighway was regarded as a method of trans-ferring a rural population f rom a region of ex-cessive density to one of inadequate density.Unfortunately, the road construction was in-adequately supported by agricultural research,and little migration actually took place.7

In the early 1970s, the central governmentalso implemented a program concerned wi thagricultural productivity and income distribu-t ion wi thin the Northeast, the PROTERRA (Prog-rama de Redistribuicao de Terras e de Estimulo a Agro-lndustria de Norte e Nordeste) program.(Venezian 1972, p. 33). The productivity of thebasic food crops of the Northeast had beenfall ing, wi th total production being maintainedby area expansion (Table 3). For example, cornacreage in the Northeast increased more than2.5 t imes f rom 1950 to 1975, whereas yieldsslightly decreased. (Johnson 1977, p. 6).

In various international centers there hasbeen research on the basic food crops of theNortheast: corn, f ield beans, cowpeas, cassava,and rice. However, only a small proport ion ofthis research has been on drought tolerance.Within the Northeast some traditional ag-ronomic research has been done, such as ferti l-ization, spacing, and variety testing. Somebreeding improvements have been made, asin field beans, and new seeds of improved se-lections have been distributed.

In general, the potential for increasing yieldsof these basic food crops does not appear verypromising. Why? First, yields at present areextremely low. Second, developing cultivarswi th good fertilizer response under irregular

7. The target in the organized colonization was100 000 famil ies or approximately one-half mi l l -ion people over the period 1971-1974. By 1974INCRA (the Brazilian land reform and colonizationagency) reported that only 5717 famil ies hadbeen sett led. There was also spontaneous set-t l e m e n t but much of th is sett lement was tempor-ary and f ew data are available. In 1974 and 1975the central government changed its strategy forthe Amazon to p romot ing g rowth poles and largeenterprises. See W o o d and Schmink (1978, pp 8,19, 20).

76

Table 3 . Trends in corn and bean product ion in t h e Brazi l ian Nor theast . a

Crop Year

1967/681968/691969/701970/71

1971/721972/731973/741974/75

Corn

Production('000 t)

17031597871

1695

1535151214181580

Area Yields('000 ha) (kg/ha)

2128 8002062 7741700 5122159 785

2192 7002199 6872186 6782450 645

Beansb

Production('000 t)

870818448961

861868601794

Area('000 ha)

1517150711991593

1640168114991705

Yields(kg/ha)

573543374603

525517401466

a. Source : Unpub l i shed data f r o m SUPLAN, Min is te r io da Agr icu l tu re , Brasi l ia, 1976.

b. Includes f ie ld beans and cowpeas. In Bahia approx imate ly 95% of the beans are f ie ld beans and in Pernambuco 50%. The

p ropo r t i on of f ie ld beans is lower for most of the rest of the Northeast , being only 5% in Ceara.

rainfall conditions appears unlikely given theexperience with present and improved varietiesin the poor soil conditions of the Northeast.Hence, it appears to be necessary first to iden-tify plants wi th drought-tolerance potential andthen select varieties and undertake basic ag-ronomic research.

Subsistence cropping has always been a supplementary activity for vaqueiros, parceiros (sharecroppers), and small farmers; hence, fewinputs have been utilized, little research hasbeen done, and absolute yields have remainedlow. There is a further complication: the aver-age annual rainfall in the Sertao is not ex-tremely low compared to other semi-arid reg-ions of the wor ld, but it is extremely irregular,and the probability of inadequate rainfall in thegrowing season is high. Corn is notoriouslysensitive to lack of rainfall during the criticalstages of growth. "Corn yields are reduced by50% or more if there is water stress during thecritical periods regardless of the level of rainfallfor the rest of the growth cycle."B Perhaps cornis an inappropriate crop for the region, and a new crop wi th drought tolerance is necessary.

Al though there are an estimated 50 mil l ion hasown to sorghum in the wor ld, placing it fourth

8. Cited f rom Goodwin et al. (1980, p 7.) See alsoDenmead and Shaw (1960) and Robins andDomingo (1953).

in grain production behind wheat, rice, andcorn, very little grain sorghum is planted in theBrazilian Northeast.9 In addition to its droughttolerance, sorghum can tolerate light sandysoils, salinity, and low fertility (Purseglove 1972,pp. 270, 286). It seems to be the perfect crop forthe irregular rainfall, low fertility conditions ofthe Sertao. Given the physical production

9. It appears that grain sorghum was introducedf rom Africa into Brazil in the early slave trade.However, the slaves principally stayed in theZona da Mata where adequate rainfall makescorn superior. Only in the last century wi thgreater populat ion pressure in the Sertao havesubsistence crops become important there. Grainsorghum thus has no tradit ion of production orhuman consumpt ion except in the Acu Valley ofRio Grande do Norte, where a type of cornbread("cuzcuz") is tradit ionally made f rom sorghum.On the introduction of sorghum see 0. Johnson,1978, pp 5, 6).In semi-arid East Africa where sorghum probablyoriginated and where it is an important food crop,the exact opposite substi tut ion is occurring.Maize is generally preferred for food, hence has a higher market price, requires less labor, does nothave bird problems, and is less attacked bystorage insects than sorghum. Moreover, newearlier varieties of maize are being introduced.Wi th this subst i tut ion, the potential for disasterincreases when rainfall is irregular. See Purse-glove, (1972, pp 259, 262).

77

characteristics of corn and sorghum, cornseems to be a historical accident; it does poorlyunder both drought and the adverse soil con-ditions cited.

The only available estimate of grain sorghumavailability in the Northeast is 4274 tonnes in1973, of which 3000 tonnes were producedthere, 1000 were imported from the Braziliansouth and the rest f rom outside the country. Themajor use was for chicken feed (Nobre andKasprzykowsky 1975, pp 46).

How does sorghum yield in the Northeast?Across regions f rom 1973-1975, under experi-

mental conditions, sorghum consistently out-yielded corn and gave reasonable absoluteyields of 3 to 4 t/ha (Table 4). Rainfall conditionswere normal to above average during thisperiod. The real advantage of sorghum wasillustrated in a mini-drought year when the cornharvest practically failed but selected sorghumvarieties still yielded around 2 t/ha wi th fertilizerand 1.6 t/ha without fertilizer (Table 5).

How does the sorghum do in farm trials andwhat are the other constraints to its intro-duction? The next section attempts to answerthese questions.

Tab le 4 . Y ie lds o f reg ional t r ia ls o f IPA so rghum l ines an d co rn , 1973—1975 .

Identif icationnumber of IPAcultivar

73002017300206730095873011547301183Corn

Rainfall range dur ingthe crop cycle (mm)

Mean of 2 regional Mean of 6 regionaltr ials 1973 trials 1974

(kg/ha) (kg/ha)

3330 31132674 3184

Not planted 34383402 35743107 30742188 2476

250-361 295-748

Mean of 12 regionaltrials 1975

(kg/ha)

436839293861419240033353

508-941

IPA s tands fo r the I n s t i t u t o de Pesquisas A g r o n o m i c a s of Recife, Pernambuco , Brazil.

Fert i l izat ion levels of 90 kg of N, 90 of P2O5 and GO of K2O w e r e ut i l ized.

Aztec var ie ty (Corn) w a s ut i l ized in 1973 and w a s replaced by Cent ra lmex in 1974 -1976 .

a. Source: Faris a n d de Lira (1977, p 7).

Table 5 . Resul ts o f so rghum reg iona l t r ia ls in Paraiba a n d Pe rnambuco w i t h a n d w i t h o u t fer t i l i za-t i on In a m in i -d rough t year , 1 9 7 6 . a

IPA-identification number

73002017300206730095873011547301183Corn

Wi th ferti l izerb

(Mean yields over 7 (kg/ha)

25072169174818652359

740

Wi thout fertil izerregional trials)

(kg/ha)

17901750172415461584627

Source: Faris a n d de Lira (1977, p 7.).

a . The ra infa l l du r i ng the c rop season ranged f r o m 147 to 322 m m .

6. 90 kg of N, 90 kg of P 2O 5 , a n d 60 kg of K2O.

78

N e w T e c h n o l o g y i n t h e S e r t a o :M o d e l i n g , F a r m T r i a l s ,a n d M a r k e t i n g

A new crop combination of tree cotton andsorghum more than doubled farm income in a MOTAD (Minimization of Total Absolute Devi-ations) risk activity analysis for one region of theSertao. Farmers were shown to be strong riskavoiders but highly efficient in their choice ofcrop systems (Sanders and de Hollanda, 1979,pp 109-115). These model results were testedon farms in Rio Grande de Norte in 1976. There,sorghum yielded 1154 kg/ha without fertilizerand 1785 kg/ha with a low fertilizer applica-t ion.1 0 The corn crop was almost a completefailure, w i th a 70 to 80% yield reduction in thismini-drought year. As in the sorghum yield dataof Table 5, there was a physical response tofertilization in the farm trials in spite of therainfall deficit and it was profitable to usefertilizer.11 With a better rainfall distribution thereturns to fertilizer on sorghum are expected tobe even higher. This provides some supportingevidence for the hypothesis that drought toler-ance must first be obtained before attacking thefertility problem. Sorghum has now performedvery well in the Northeast in comparison wi thother alternatives, not only in agronomic trialsbut also in the profitability-risk and farm con-straint evaluation context. Moreover, the modelresults were confirmed in farm trials.

Further refinements of the model and dataindicated that even with the extremely high riskavoidance preference characteristics of Sertaofarmers incorporated, the introduction of sor-ghum resulted in a 25% increase in farm income(the "net margin") . The sorghum activity en-tered principally because of the substantialreduction in the standard deviation of the opti-mum farm plan. With this risk reduction f romthe availability of sorghum technology, therewas an expansion of the area cultivated (Good-win et al. 1980, p 17). At present risk-aversionlevels, unfertilized sorghum was preferred.

10.

11.

Twenty kg/ha of N and 60 kg/ha of P2O5 wereuti l ized. Birds were not a serious problem in thesefarm trials nor was the so rghum midge. Barbosaet al. (1976, p 16).Profits were almost doubled by fert i l ization inspite of the inadequate rainfal l .

With lower risk-avoidance coefficients, higherlevels of chemical fertilizer were utilized inmodel solutions. This appears to indicate theimportance of field estimation of farmers' riskcoefficients as an input into research design. Inspite of the large emphasis in Brazil on subsi-dized credit this instrument had no effect uponthe introduction of new technology in the mod-eling results because, at these high levels of riskavoidance, farmers use few purchased inputs.The model results indicated that a farmer,knowing about the grain sorghum activity butexaggerating the riskiness of this new activityand even having a high coefficient of riskaversion, wi l l adopt sorghum on a small area ofhis farm.12 After observing field results of thisinnovation, the accuracy of the farmer's per-ception of its riskiness will increase. Moreover,increased earnings wil l in t ime reduce farmer'srisk aversion and encourage him to use higherlevels of purchased inputs, especially fertil-izer.13 Once the farmer is prepared to use morepurchased inputs, the credit instrument be-comes potentially more effective. In summary,farm-level testing is expected to be a muchmore efficient instrument than credit subsidiesin the early stages of technology introduction.

Extending the data series and introducingmore fertilizer technologies into the model,especially for sorghum and other crops on thesmall area wi th Vazante (water), the preferredcrop choice was to continue the typical consor-cio (cotton-corn-cowpeas) for the subsistencerequirement but also introduce sorghum in a monoculture without fertilization. If the farmeris prepared to take more risk, he can earn a higher income. With these crop plans, sorghumforces the traditional consorcio into the poorersoil areas until it disappears from the productmix on small farms.14 With the introduction of

12.

13.

14.

A farmer is expected to exaggerate the riskinessof new technology unti l he has gradually ac-quired his own empirical knowledge of the cropon his farm or in his region.Increased income over t ime should lead to grea-ter wealth. At higher wealth levels individuals cantake more risks because they can survive badyears.Mean farm yields of sorghum over a 12-yearperiod were estimated to be 1290 kg/ha wi thoutfertil izer and 1958 kg/ha w i th fertil izer of 80 kg/haof N (Barbosa, 1977).

79

sorghum, farm income can be doubled as in theearlier results. With medium size farms, thehigh prices for corn and cowpeas, and theincreased labor requirements for sorghum en-able the cotton-corn-cowpea activity wi th im-proved technology to force out sorghum at thehighest risk levels. Thus, the model resultsappear to show a comparative advantage forsmaller farms in the production of sorghum.Fertilizer subsidies had no effect upon the farmplans chosen, except at the highest income-riskposition.15 By shifting the income-risk curveupward, agricultural insurance plans resulted innew technology being introduced much soonerat lower risk levels. The risk insurance alsostimulated a more intensive land use.16

The agronomic results show that sorghumoutyielded corn, and the modeling and farmtrials show that the introduction of sorghum fitsinto the farmer's other activities and is profit-able. Now who wil l buy the sorghum?

There are many potential uses of sorghum. Inthe short run it is unlikely that it wi l l be used forhuman consumption in spite of the sorghum"cuzcuz" eaten in the Vale do Acu. Human foodhabits in the absence of extreme pressures — as from a natural calamity — appear to be fairlyinflexible in the short run. However, even onsmall farms in the Sertao animals are impor-tant, and the sorghum can be easily substitutedfor corn. Very preliminary estimates indicatethat up to one half of the corn utilized on smallfarms in the Sertao is for animal feed. There isan increasing demand for poultry in the North-east with the rapid introduction of poultrytechnology and the resulting reduced costs ofproduction. Ration-mixers in the Northeastimport the concentrate and, when rainfall isadequate, mix it w i th locally purchased corn.When rainfall is inadequate, corn must be im-ported from the south. Sorghum would in-crease the probabil i ty of local availability of a cheaper feed source by being less likely to failand it would eliminate the high transportation

15. The 40% fertil izer price subsidy available in Brazilpr ior to 1977 was util ized in the model .

16. This agricultural insurance was not the officialp rogram, PROAGRO, wh i ch operates onlythrough credit and had no effect according tomodel results. Rather t w o programs were consi-dered guaranteeing 6 0 % and 7 5 % of the incomeof the basic consorcio w i t h improved technology.

cost of import ing corn f rom southern Brazil.Utilizing linear programming of chicken ra-

tions and the projection of the total demand forcorn in the Northeast at different relative pricesof sorghum and corn, potential demand curvesfor sorghum and corn were constructed byCampos et al. (1978, p 113). By 1980, 400 000 to514 000 tons of sorghum could be utilized inchicken rations.17

The risk here for the small farmer is the shiftf rom home consumption of com and cowpeasto sale of sorghum to an oligopoly of feedfactories. Governmental intervention or thecreation of cooperatives could overcome thisrisk. Policy intervention in marketing seemsmuch more feasible than the search for cornvarieties wi th sufficient drought resistance to becompetit ive wi th sorghum. There are manyother potential uses of sorghum, such as a partial substitute for wheat in bread, but thesemay require even more governmental inter-vention.

Given the large population in the Sertao, thefailure of various types of policies to eliminatethis high density of small farmers and share-croppers, the efficiency loss to society f romallowing human capital deterioration of thisgroup, and the lack of interest of most largefarmers in the Sertao in other activities besidescattle, Brazilian policy makers may wellmaximize the social benefits of new sorghumtechnology by designing it specifically for smallfarmers. There are a series of policy options todo this:

17. In 1974, there were 25 factories producing ba-lanced rations in the Northeast w i th an installedcapacity to produce 276 740 tons. Ninetyf ive per-cent of th is rat ion produced was for chickens w i th3% and 2% for swine and catt le, respectively. SeeNobre and Kasprzykowsky, (1975, p 75).

1. Concentrate research on grain sorghumonly, not forages.

2. Develop varieties, not hybrids. Small far-mers wil l not be interested in purchasingnew seed each year and private com-panies selling hybrids are generally moreinterested in large farmers.

3. Give no subsidies for mechanical harvest-ers. Wi th the large labor requirements ofbird control, sorghum harvesting, and in-creased labor for weeding, small farmersmay have a comparative advantage in

80

sorghum production unless the govern-ment offsets this advantage through sub-sidies for machinery purchase.

4. Use governmental intervention if neces-sary to enforce advance contracts betweenfeed mixers and farmer cooperatives.

5. Finance regional farm trials of sorghum sothat region-specific problems can be identi-fied and farmers can become familiar withsorghum technology.

There are many technologies which could bedeveloped for the Northeast. It appears thatthe most viable for rapid introduction now issorghum. With regional testing on farms andgovernmental intervention to facilitate market-ing, the introduction of sorghum could progressrapidly in the Sertao. Corn acreage has beenincreasing over the past 25 years in manyregions not appropriate for corn. The researchappears to be available now for a governmentinitiative to substitute sorghum for corn in theSertao and thereby concentrate corn in zoneswith higher rainfall where yields can be in-creased wi th higher levels of inputs.

E x a n t e E c o n o m i c A n a l y s i sa n d T e c h n o l o g y D e s i g n

This paper has attempted to describe the rele-vant characteristics of the target region and newtechnology and then to document the researcheffort on one new technology, the introductionof sorghum. Suggested stages of economicsupport of the research design process areoutlined in the f ive steps below.

1. Identification of the crop. The benefits ofcrop research depend upon:

a. the potential area of the crop in the region;b. the ability of researchers to correctly iden-

tify the principal restrictions to increasedyields;

c. the potential of the scientists to f ind a technical solution to the (se) restriction(s);

d. the economic conditions required to im-plement the technical recommendationsand market the product.

In international centers the crop is designatedby external funding agencies or previous com-mitments so that concern can be focused onitems b and d above. National centers wil l haveto make the difficult allocative decision betweencrops.

Here sorghum was chosen over corn because

of its greater tolerance of drought and lowfertility. The marketing problems appeared tobe superable, wi th some governmental inter-vention.

2. Identification of the clientele. Researchersoften claim that their technology is scale neutralbut assume that their client farmers wil l usehigh input levels and have sufficient wateravailable at critical times of crop growth. Theseare critical assumptions because there is fre-quently a trade-off between varieties withmaximum yields wi th adequate water and highfertilization and varieties with drought toler-ance and some response to lower fertility con-ditions.18 Maximizing the yield under optimisticprojections about water availability and fertil-izer use could then lead to the rejection ofvarieties with good performance under moreadverse conditions. Many regions will not beaffected by irrigation or innovations in water-saving technology. Moreover, small farmerswithout water control are not expected to usehigh levels of inputs because of strong risk-avoidance characteristics. If the technology is tobe oriented towards small farmers, it may benecessary to make the more pessimistic as-sumptions.

For various reasons discussed previously,small farmers are expected to be the principalclientele of the new sorghum technology. Thenew varieties, which are not as impressive onthe experiment station, may have much morepotential for diffusion. Subjective projectionsabout potential diffusion of various types ofmaterial may be as important as the biologists'estimation of the potential yield increases re-sulting from various breeding strategies.

3. Ex ante modeling of the farm level effect ofnew technology. Using all available data andsynthesizing data when there were gaps, thefarm level potential effects of new technologyintroduction were evaluated. Sorghum was anexcellent potential activity according to theseresults. Farm testing confirmed the model re-sults.

4. Feedback. In 1976 IPA began doing re-search on unfertilized sorghum (reported inTable 5).19 Physical data gaps on the yields ofsorghum wi th different states of nature — i.e.,

18.19.

J. Ryan called this trade-off to my attent ion.The first sorghum ex ante model ing activitieswere presented in 1975.

81

climate and pests — and information require-

ments about farmer's risk attitudes were iden-

tif ied as necessary information for future re-

search.

5. Results. In 1976 the State agricultural

planning agency (CEPA) of Rio Grande do

Norte, w i th World Bank f inancing, began ex-

tending credit for animal-powered mechani-

zation and undertaking farm-level testing of

sorghum in "consorcio," w i th and without

fertilizer, in the Rural Norte Project.

Postscript:

Unfortunately, no direct Brazilian central

government intervention to facilitate grain sor-

ghum marketing has as yet occurred. Moreover,

much of the research focus on sorghum in the

Northeast has been shifting to forage sorghum

as a result of strong pressures f rom the cattle

producers.

References

de ALBUQUERQUE L I M A , D. M., e SANDERS, J. H. 197S.

Selecao e avaliacao de nova tecnologia para osp e q u e n o s a g r i c u l t o r e s do Se r tao , RevistaEconomics do Nordeste 9(1).

BARBOSA, A. R., SANDERS, J. H., e A B D O N de LYRA, J. H.

1976. Opcoes tecnologicas para a regigo semi-aridade Rio Grande de Norte (Comissao Estadual dePlanejamento Agr ico la : Natal , Rio Grande do Norte:novembro de 1976).

BARBOSA, S. 1977. Inovacoes tecnoldgicas agricolas e polit icas para reduzir os zeus riscos. Estudo em umamicro-regiao semi-aride do Nordeste: O ser ido doRio Grande do Nor te, M.S. thesis, Departamento deEconomia Agr ico la, Universadade Federal do Cearta.

CAMPOS MESQUITA, T., S ILVA, P.R., e SANDERS, J. H.

1978. Procura potencial para o sorgo granl ' fero noNordeste Brasileiro. Revista Economica do Nor-deste 7(1).

CENSO AGROPECUARIO BRASIL, VIII Recenseamento

Gerai 1970. 1975. Rio de Janeiro: Fundacao Institute-Brasileiro de Geografia e Estatistica.

DENMEAD, O. T., and S H A W , R. H. 1960. The effects of

soi l moisture stress at dif ferent stages of g row th onthe development and y ie ld of corn. Ag ronomyJournal 52: 272-274.

FARIS, M. A. d de LIRA, A. M., 1977. Aval iacao da

produt iv idade de cult ivares de sorgo grani fero emrelacao ao mi lho nos estados ha Paraiba e Pernam-buco, Brasil (1973-1976). Relatorio parcial No.34/75, Secretaria de Agr icul tura, I n s t i t u t o de Pes-qu iaas A g r o n o m i c a s , Rec i fe , P e r n a m b u c o .(Mimeographed.)

G O O D W I N , J. B., SANDERS, J. H., and de HOLLANDA, A. D.

1980. Ex ante appraisal of new technology: sor-g h u m in northeast Brazil. Amer ican Journal ofAgr icul tural Economics. (Mimeographed.)

HARGREAVES, G. H., 1973. Monthly precipitation prob-abil i t ies for northeast Brazil. Department of Agr icul-tural and Irr igation Engineering, Utah State Univer-sity, USA.

J O H N S O N , A. W. 1971. Sharecroppers of the Sertao:Economics and dependence on a Brazilian planta-t ion . Stanford, Cali fornia, USA: Stanford UniversityPress.

J O H N S O N , D., 1978. Sorgo grani fero no nordeste doBrasil. Revesta Economica do Nordeste 9(4) :483-500.

M O U R A , H. A U G U S T O , 1971. Crescimento demograf ico

dos estados do nordeste, 1940-1970 (Depar-tamento de Estudos Economicos do Nordeste,Banco do Nordeste do Brasil.

NOBRE, J. M. E., e KASPRZYKOWSKY, J. W. A. 1975.

Mercado Potencial para o Sorgo no Nordeste. De-partamento de Estudos Economicos do Nordeste,Banco do Nordeste do Brasil.

PAIVA, R. M., SCHATTAN, S., and TRENCH de FREITAS, C.

F. 1973. Brazil's agricultural sector, economic be-havior, problems and possibil i t ies. Rio de Janeiro:Graphos.

PURSEGLOVE, J. W. 1972. Tropical Crops: Monocoty le-dons I. New York, USA: Wiley.

ROBINS, J. S.,and DOMINGO, C. E. 1953. Some effects ofsevere soi l moisture deficits at specific g row thstages in corn. Ag ronomy Journal 45 :618 -621 .

SANDERS, T. G., 1973. The Northeast and Amazonianintegrat ion. Amer ican Universit ies Field Staff Re-ports, East Coast South America Series, 17(3).

SANDERS, J. H., and de HOLLANDA, A. D. 1979. Technol-

ogy design for semi-arid northeast Brazil. Pages102-121. in Economics and the design of smal lfarmer technology, eds. A. Valdes, G. M. Scobie, andJ. L. Di l lon Ames , Iowa: Iowa State Universi ty Press.

VENEZIAN, E. L., 1972. Agr icul tural product iv i ty andproduct ion technology research in northeast Brazil.Presented at a Ford Foundation seminar, November1972.

W A R D , J. O. and SANDERS, J. H. 1980. Nutr i t ional

determinants and Migrat ion in the Brazilian Nor th-east: A Case Study of Rural and Urban Ceara,Economic Development and Cultural Change. 56 p.(Forthcoming.)

W O O D , C. H., and SCHMINK, M. 1978. B laming the

v ic t im: smal l farmer product ion in an Amazoncolonizat ion project. Presented at the Amer icanAssociat ion for the Advancement of Science,Wash ington, D. C., February 1978.

82

D i s c u s s a n t ' s C o m m e n t s

R . J . H . C h a m b e r s *

Fall's succinct paper discusses physical con-straints and human factors relating to the intro-duction of new technology in the rural areas ofSenegal generally and the Sahel in particular.He presents some findings f rom a study ofactual production systems and the effect of newtechniques in rural areas. Many questionsmight be asked about the methodology of thestudy and about the findings. Of particularinterest is the finding that women and sourga (family labor) have the smallest plots, the low-est rates of fertilizer application, and the lowestgroundnut yields. It is difficult to interpret thesignificance of this without more information.

The author stresses the value of individualcase studies of farmers; these might provideinsight into farmers' constraints and rationality.However, one wonders whether there are notdangers of a certain dirigisme when, for exam-ple, he writes: " In Senegal, although theprofitability of certain techniques has beendemonstrated, the farmer still needs to bemotivated toward change." There may benuances here that are lost or added in transla-t ion f rom the French. But it may be askedwhether it is not often more important forresearchers to learn more about farmers' situa-tions and views rather than for farmers to bemotivated. There is, indeed, a hint of this pointof view when the author says, "The foundationof technical recommendations made throughresearch is not always perfect." The key ques-t ion, not addressed in this paper, is not how tomotivate farmers, but how to improve theresearch and development (R & D) process.

The paper by Ryan, Sarin, and Pereira sum-marizes the economic analyses that have beenmade of the experiments conducted byICRISAT's Farming Systems Research Pro-gram f rom 1975 to 1978. These have beendesigned to develop improved methods of soil,

* Institute of Development Studies, University

Sussex, England.

of

water, and crop management that can helpincrease and stabilize crop production in thesemi-arid tropics.

There can be no doubt about the value of thissort of economic analysis. The "steps-in-improved-technology" (SIIT) approach com-pares the costs, benefits, and risks of variouscombinations of four innovations concerningvariety, fertilizer application, soil and crop man-agement, and supplementary irrigation, on twotypes of soil. This provides a tool for screeningout lines of research that are unlikely to beuseful to farmers. It is possible that the detail ofthe analysis represents an overkill, when deci-sions about research priorities wil l be based oncomparative orders of magnitude. But on theother hand, detail and rigor are often necessaryin order not merely to be right, but to be seen tobe right.

A point of interest in the paper is the sugges-tion (attributed to Krantz) that the mere avail-ability of stored runoff wil l act as assurance to a farmer, encouraging him to invest more inHYVs, fertilizers, plant protection and otherinputs, even if in a given year he does not usethe water.

In the various combinations of treatments onthe Vertisols and Alfisols, it is surprising to notethe relative significance for profitability of im-proved fertilizers compared wi th supplemen-tary water. One wonders whether, in the realworld of small farming, the relative importanceof these two factors might not be different forthe fol lowing reasons:

1. Access and the real price of fertilizer. Theactual price of fertilizer to farmers is oftenhigher than the list price because of the incen-tive payments that have to be made to officials.There is also evidence that smaller, less influen-tial farmers have to pay more than larger, moreinfluential farmers, not only in cash terms but inthe t ime they have to devote to negotiations.The risks of not being able to obtain fertilizermay also be greater for smaller and less influen-tial farmers. As official prices were used in thecalculations there may be some over-

83

estimation of profits f rom fertilizers, thoughprobably not sufficient to alter the basic conclu-sions.

2. The cost of water. The experiments havebeen structured by the scale of watershedchosen for the tanks, the size of tank, thepumping technology (diesel), and the size ofpump (5 HP). A conclusion is drawn that:

A minimum catchment size might be about 8 ha.Break-even yields rise substantially for catch-ments less than this. It also seems that once catchments exceed about 16 ha the break-evenyields begin to rise due to the negative effects ofarea and evaporation on runoff.

There are two lines of argument here, leadingin opposite directions:

First, wi th these techniques and this scale, itmight be that for the smaller farmers (thosewho were less influential) and those higher upon the catchment, the stored water would beless profitable than in the calculations. Thelarger and more influential farmers would belikely to get water first, in a more t imely mannerand in larger quantities, and those lower downin the catchment could argue that their watercost less because it required less diesel topump, so that those higher up should pay more.

Second, and on the other hand, if the tank andpumping technology had a smaller scale andwere cheaper, then the profitability of water forthe smaller, less influential farmers and thosehigher up the catchment could be greater.Obviously there are many considerations here,and ICRISAT has canvassed many ideas aboutalternative systems. But in real village situ-ations, it might be that the combinations oftechnology developed on the research stationwould have been screened out as impracticableat a very early stage, and the R & D processmight have fol lowed a quite different courseleading to micro-storage of water on individualfarmers' fields and animal or human power forwater lift for high-priced crops.

The thrust of these points about relativeaccess to and costs of improved fertilizer and ofsupplementary irrigation is that the value ofeconomic analysis is l imited by the location andbuilt-in rigidities of the research itself. It is muchbetter to have economic analysis than not tohave it, as this meticulous and thorough paperdemonstrates. But it is better still if r ight f romthe start of research, farmers and social scien-tists are involved and if the balance of location

for research of this sort is shifted to villages.This would eliminate lines of development thatfarmers would not accept, shortcut some of theneeds for sophisticated analysis, and identify atan early stage what is so much more difficult togauge from a research station: who is likely tobenefit and who to lose f rom a new bundle oftechniques in the real wor ld of the village.

The paper by Sanders describes the relevantcharacteristics of the Brazilian Sertao with re-gard to new agricultural technology and re-views the research effort concerning one suchtechnology: introduction of sorghum, a newcrop in the region.

There are many points of interest and insightin the paper. These include the relationshipbetween nutrit ion and efficiency (although thephrase "human capital deter iorat ion" isperhaps unnecessarily clinical); the irrelevanceof fertilizer and credit when farmers have highrisk-avoidance; the value of field evaluation offarmers' risk coefficients as an input into re-search design; the observation that farmerscarry out small-scale trials on their own fields,and the idea that R & D can be designed spe-cifically for small farmers to give them a com-parative advantage, as wi th sorghum wi th itshigh labor requirements.

Al though it does not overstress them, thepaper raises crucial issues about governmentpolicy and political economy. These issues havegeneral application but are all too often notdiscussed. The question here is whether mea-sures wil l be taken that wi l l benefit the smallerand less influential farmers. This questionapplies at three levels of decision and action:

1. Pricing policy and policy for mechanicaltechnology. Sanders writes:

With the large labor requirements of bird control,sorghum harvesting, and increased labor forweeding, small farmers may have a comparativeadvantage in sorghum production unless thegovernment offsets this advantage through sub-sidies for machinery purchase.

This issue may be taken even further througha policy for technology as India has dona byprohibit ing the import or manufacture of com-bine harvesters. But this can only be done wi th a resolute political wi l l to offset the powerfulinterests that benefit f rom such technologies tothe detriment of the poorer people.

2. Marketing. The author writes, "Unfor tu-nately, no direct Brazilian central government

intervention to facilitate grain sorghum market-ing has as yet occurred." One may speculatethat such marketing organization would servethe ma jo r interests — the large catt lefarmers — of the region.

3. Research priorities. The interests of thesmaller farmers do not appear to have beenreflected in research priorities. Only a smallproportion of past research was on droughttolerance; and much of the research focus onsorghum in the Northeast has been shifting toforage sorghum due to strong pressures fromcattle producers. This is the point at which thepaper ends, and another subject starts. Thedetermination of research priorities, whether bydesign or default, is political since it determineswho wil l benefit. If these priorities are influ-enced by large-farmer interests then counter-vailing research is needed to offset them. It ishere that ICRISAT, through its choice of cropsand priorities, has a major role to play.

85

The Soc ioeconomics o f Prospect iveTechno log ies : People and Pr ior i t ies

R . J . H . C h a m b e r s *

Abstract

This paper is concerned with the decisions and processes that generate mechanical,

biological, and chemical technologies that have an impact on life in the semi-arid tropics

(SAT). The author makes these assumptions: that such technologies, separately or

combined, influence social relations and the distribution of benefits within societies; that

in much of the SAT, rural populations are increasing and will continue to do so for

decades to come; and that there is room for maneuver in setting research and

development priorities and in decisions taken during the research and development (R &

D) process. The central issue is how to optimize decisions and action that affect and are

part of R & D. It is contended that modes of thought, values, and criteria need to be

re-examined. In much of the literature of agricultural development, including ag r i cu l -

tural economics, people are treated as resources rather than users of resources — a s

means rather than ends. And thinking about research priorities often starts with a crop or

a farm system or a mechanical technology rather than with the poorer people in a rural

environment and their interests and future. The author suggests that decision-making

and research might be improved through expanding environment-specific research,

conducting more of it in collaboration with rural people, developing cost-effective

methods for rural appraisal, changing professional reward systems, enabling p ro fes -

sionals to become individually more multidisciplinary, and learning from the true

multidisciplinarians, the rural people themselves.

"If everybody minded their own business," saidthe Duchess in a hoarse growl, "the world wouldgo round a deal faster than it does." LewisCarroll, Alice in Wonderland."When the rockets go up, who cares where theycome down? That's not my department..." TomLehrer song.

In this paper "socioeconomics" is taken to

include the concerns of the social sciences

general ly, and not just sociology and

economics. It would be artificial, restrictive, and

* Institute of Development Studies, University ofSussex, England

EDITORS' NOTE: In preparing for the workshop,

Dr. Chambers wrote this thought fu l paper on thesubject of the session. A l though he also was askedto open discussion it was felt this paper meri tedInclusion as a separate contr ibut ion to fo l l ow hiscomments on the papers formal ly presented.

unhelpful to make the scope narrower. "Pro-

spective technologies" are similarly interpreted

widely to include technologies that are "in

prospect" in the sense of being actively de-

veloped, and also others that are not being

developed but might be developed if accorded

priority. "Technologies" include mechanical,

biological, and chemical technologies. The

Ft & D discussed is the formal R & D of organiza-

tions and not the informal R & D of rural people,

important though the latter is.

Analysis of the socioeconomics of prospec-

tive technologies can focus on the receiving

environments or on the processes that generate

the technologies, or on both. Though consider-

ing both, this paper is primarily concerned wi th

the decisions and processes that generate

technologies, since this is where many impor-

tant choices lie. It is not concerned with the

diffusion of technologies that already exist. The

central issue is how to optimize decisions and

action that affect and are part of R & D proces-

ses.

86

In approaching this issue, three assumptionsare made.

First, in the semi-arid tropics (SAT), mechani-cal, biological, and chemical technologies,separately or combined, influence social re-lations and the distribution of benefits withinsocieties. Prospective technologies can be ex-pected to continue to exert such influences.Decisions about research priorities and deci-sions within the R & D process are thereforepolitical and value-laden in their implications,however technical they may appear. In receiv-ing environments, the "talents effect" as An-drew Pearse(1977) has called it after the biblicalparable, is widely prevalent, so that those whohave more get more and those who have lessmay lose even the little that they have. But

technologies vary widely in the ease or difficultywith which they can be captured and used bydifferent categories of people, and they can bedesigned with target groups in mind, includingpoorer people. Whether deliberately or by de-fault, a social policy is built into new technol-ogy.

Second, in much of the SAT, proportions asbetween population, land and water are chang-ing and will continue to change. In almost allareas, rural populations are increasing and canbe expected to continue to increase. Short ofsome demographic disaster, and in spite of highlevels projected for rural-urban migration, ruralpopulations in most countries wil l continue togrow rapidly for decades. Table 1 indicatesorders of magnitude: rural populations in most

Table 1. Rural and urban popu la t ion project ions for some countr ies who l ly or part ly In t h esemi-ar id t rop ics . a

AngolaBoliviaBotswanaChadEthiopia

IndiaKenyaMal iMexicoMozambique

NigerNigeriaPakistanParaguayRhodesia

SenegalSudanTanzaniaThai landUpper VoltaZambia

1975(mil l ions)

rural

5.23.40.63.5

24.8

481.511.84.9

21.88.7

4.251.551.6

1.75.0

3.215.914.435.1

5.53.2

urban

1.22.00.10.63.1

131.81.50.8

37.40.6

0.411.419.01.01.2

1.32.41.07.00.51.8

2000(mill ions)

rural urban

8.1 4.44.8 5.41.0 0.44.9 2.0

42.2 11.4

717.3 342.024.6 6.48.7 2.6

28.7 103.615.3 2.3

7.9 1.694.0 40.984.6 62.32.6 2.6

10.1 5.0

4.7 3.530.0 8.929.8 4.2562.2 23.49.2 174.6 7.0

% increases1975-2000

rural urban

55 27843 16965 45942 25470 265

49 160109 32876 23732 17777 300

90 28382 25964 22859 167

100 309

48 17989 271

107 30777 23667 24444 280

a. Source: FAO, based on data a few years o ld . Mo re recent f igures w o u l d probab ly genera l ly s h o w s l igh t ly lower percentage

Increases, but w i t hou t af fect ing orders of magn i tude . Percentages are based on the or ig ina l f igures wh i ch were in thousands ,

a n d w h i c h have here been rounded to mi l l ions to one dec ima l place. Brazil has not been Included because much less than half

Its area is in the SAT. Its percentage increases fo r the count ry as a who le are, however , es t imated at only 13% rural (44.5

m i l l i on to 90.3) a n d 149% urban (65.3 m i l l i on to 162.2).

87

of the SAT countries are projected by the FAO toincrease in average of about 58% between 1975and 2000. Percentage rates of increase of ruralpopulations in the SAT are estimated to begenerally lowest in Central and South America(Mexico 32%, Bolivia 43%), wi th Asia higher(India 49%, Pakistan 64%, Thailand 77%), andWest and East Africa highest (Nigeria 82%,Ethiopia 70%, Tanzania 107%, Kenya 109%).Since these are national averages, some localand regional increases wil l be higher. Locally,too, resources wil l be augmented (as throughirrigation, forestry, imported inputs such aschemical fertilizer) or diminished (as throughdepletion of groundwater, removal of forests,declining soil fertil ity, and soil erosion). Theremay be a widespread phenomenon of poorerpeople, as in parts of Kenya (Mbithi and Barnes1975; Johnston 1979), migrating into lowerrainfall and fragile environments, which theythen degrade irreversibly with the possibility ofa later backwash of refugees returning to den-sely populated higher potential areas.

The third assumption is that there is room formaneuver in setting R & D priorities and indecisions taken during the R & D process, andthat more room could be made. This is a controversial point. On the one hand, we havethe experience of market and institutionalforces combining to give priority and resourcesto research that benefits the better-off and thathas often concentrated on commodities forexport. We have examples of pressures f romcommercial producers for research that favorstheir interests but may not favor the smallerfarmers (e.g. Sanders 1979). We have the theoryof induced innovation that sees innovation as a response, albeit sometimes lagged, to factorproportions and prices (Ruttan 1977). But on theother hand, research institutions in the publicsector are to varying degrees insulated f rompolitical pressures. The international researchinstitutes, more than most, have discretion ormandate to give priority to the interests of thoserural people who are without voice and who areunable to make demands upon research. Theyare also well-placed, through their prestige andtheir training programs, to strengthen newprofessional values and to influence the incen-tives arid rewards in national research systemsthat induce priorities and affect behavioramong other scientists. Further, in the "post-green revo lu t ion" period, there is now

heightened awareness of the social andeconomic implications of research decisions.This should extend the room for maneuver inusing R & D as a means for conscious socialengineering.

These three assumptions bear on the ques-tion of how to optimize decisions and action inR & D . Questions can be asked about threeclusters of points:

• modes of thought, values and criteria• future orientation• professionalism and priorities.

M o d e s o f T h o u g h t , V a l u e s ,a n d C r i t e r i a

Introspection is a first step. How have we beenprogrammed to think about research? Whatpoints of departure and modes of thought aredominant? The words used in discussion reflectimplicit priorities and direct attention in somedirections rather than others. For example,what meanings are likely to be attributed to theword "development" when it is set in thephrase "socioeconomic constraints to de-velopment of semi-arid tropical agriculture?" Isit rural development as defined in the WorldBank Sector Paper on Rural Development?

Rural development is a strategy designed toimprove the economic and social life of a specificgroup of people — the rural poor. It involvesextending the benefits of development to thepoorest among those who seek a livelihood inthe rural areas. The group indudes small-scalefarmers, tenants and the landless. (World Bank1975 : my emphasis)

Or is it increasing agricultural production? Ordoes it mean different things to differentpeople, or different things to the same people atdifferent t imes, in different contexts?

Much of the literature, especially in agricul-tural economics, tends to equate developmentwi th agricultural production. People are treatedas resources rather than users of resources, asconstraints rather than purposes, as meansrather than ends. Discussing poverty we thentalk clinically about the low productivity ofhuman resources, underutilization of labor re-sources in rural areas (Ruttan 1977, pp213, 215),and even the efficiency loss to society of humancapital deterioration (Sanders 1979, p 16).People are quantified as labor, as facelessfigures in factor proportions. All this has its

88

uses; without careful numerate analysis gooddecisions would be harder. But the dangers arealso great. Those most in need-the poorest-maybe treated as a residual rather than as a primaryfocus, and technical euphemisms such as "un-derutilized labor resources" and "human capi-tal deterioration" may hide acutely deprivedpeople.

If people are the starting point, then theyshould come first in lists of criteria for ex-anteappraisal of prospective technologies. There is,of course, much scope for argument here aboutdirect and indirect benefits to the poorer people.The issues are not simple. But in terms of theWorld Bank definit ion, there is a case for ex-anteanalysis that starts with all the people in a givenenvironment, including and especially thepoorest and the landless. It can be misleading tosuggest that " the farmers with tiny landhold-ings" are " the poorest of the poor."1 They maybe; but often they are not.

Starting with people, a key priority is thecreation of adequate livelihoods. Where theseare not already adequate, a criterion becomesthe net livelihood-intensity of prospectivetechnology in a receiving environment.Livelihood-intensity here is the extent to whicha technology would generate or sustain liveli-hoods at or above an acceptable level. It is notthe same as labor-intensity. The livelihood-intensity of a technology is not constant; it isspecific to an environment and is sensitive toseasonalities. For example, a technology thatprovided food or income flows for poor people

1. Quotation f rom statement prepared by Dr. SterlingWor tman, Vice-President, Rockefeller Foundationfor two Subcommittees of the US House of Rep-resentatives, 23 September 1975, quoted inSprague 1976. The full relevant part of the quota-t ion in Sprague is: "The bulk of the basic foodsupplies of the agrarian nations are produced bythe many farmers w i th t iny landholdings, often inremote and isolated areas, plus those people incoastal areas who depend upon near-shorefisheries and aquaculture for a l ivel ihood. For themost part, the gains in productivi ty and income ofthese rural p e o p l e - t h e poorest o f the p o o r - w i l lrequire the development for and use by manyfarmers of new high-yielding, science-based cropand animal product ion systems tai lored to theunique combinat ion of so i l , cl imate, biological ,and economic condit ions of every locality in everynat ion."

during lean periods of the year so as to pushthem above a minimum for the whole yearwould, in the environment concerned, have a high livelihood-intensity. The net l ivelihood-intensity is specified because a new technologyusually displaces an old one that was alreadygenerating and sustaining livelihoods.

In practice, however, where does thinkingabout research priorities start? How often, andin what circumstances, does it start with thepoorer people in a rural environment and theirinterests and their futures? Does such thinkingin practice start from other points and ask otherquestions, starting with a crop, or a farm sys-tem, or a mechanical technology, or a problem(a pest, a water deficit, salinity, a "constraint"),and proceed through the avenues of technicalanalysis, or technical plus economic analysis?Does it approach actual people only through theanalysis of factor proportions and prices andthe underutilization and low productivity ofhuman resources? Do such formulations runthe risk of making things worse for some of thepoorer people? Are opportunities missed toimprove the lots of poorer people?

Future Orientation

In the theory of induced innovation, researchresponds to changing factor proportions. Thereis a process of dynamic adjustment to changingrelative factor prices (Ruttan 1977). In this ad-justment there may be t ime lags, sometimes ofdecades. Thus factor proportions and priceschange first, and priorities fol low later. Koppel,however, has argued that assessment oftechnology should be future-oriented (1978, pp6-7). By this he means that it should anticipatethe consequences of prospective decisions,rather than be an exercise in forecasting. Thismay not go far enough.

Forecasting has been discredited to the pointthat the word " futurology" is used wi th disdain.Perhaps this is in part because the arrogance offuturologists has been matched so often by theenormity of their errors. It has, however, beenpracticed mainly in fields of complex and rapidchange and of high uncertainty, and mainly inthe rich countries; in Third World countries ithas concentrated on urban and industrial de-velopment, again areas of relatively high uncer-tainty. But are changes in population, in relativefactor proportions, and in technology in rural

89

areas in the SATs as uncertain? Might a guess atthe rural population in a SAT region in the year2000, say, inspire more confidence than a guessat the ant i -an t i . . . missile-missile technology ofthe same year? And might it be made wi th muchless expensive research? If so, it may not be toodifficult to be ahead of the game. Environmentby environment, future endowments and pro-portions could be estimated and, at the veryleast, directions of change identified. Researchpriorities might then be reviewed, bearing inmind probable gestation periods for setting upinstitutions, recruiting staff, conducting R & D ,and then enabling and allowing diffusion. In-stead of waiting for factor proportions tochange, the approach would be to anticipatetheir change, and thereby to help generateadequate future livelihoods.

Professionalism and Priorities

It is easy to add research objectives to lists ofcriteria. Technical scientists cannot be uni-formly pleased wi th their social science col-leagues for having made decisions more com-plicated. Suggesting that adequate livelihoodsshould be placed at the center of researchobjectives for many environments, and thatpriorities should fol low f rom environment-specific and future-oriented analysis, may ap-pear to be the last straws. Agricultural scientistsand engineers may sometimes despair at thefailure of some social scientists to understandthe nature of their work. Social scientists, it mayseem, have compulsions to multiply criteria; nosooner is one set met than another is added, sothat the technical scientists can never win. Insuggesting any new criteria, there is an obliga-t ion to see if, at the same t ime, decision-makingand research can be made more manageable.Four complementary suggestions fol low. Allwould bring scientists closer, in thinking andunderstanding, to rural people and rural en-vironments.

First, research might generally move physi-cally closer to and into rural environments andbe carried out in closer collaboration wi th ruralpeople. Some activities are best carried outunder controlled conditions on research sta-t ions; others are most efficiently carried outwi th farmers as collaborators and evaluators(Hildebrand 1977, pp 14-15). Biological re-search can be viewed along a spectrum from

very basic research in supportive sciences, suchas genetics, to very applied operational re-search on farmers' fields (Binswanger and Ryan1977, p 221). There are institutional, profes-sional, and personal reasons of conveniencethat tend to concentrate personnel and re-sources at the basic research and researchstation end of the spectrum. The high returns toagricultural research suggest that resourcesdevoted to it should be expanded. Much recentexperience suggests that that expansion shouldcoincide wi th and reinforce more decentral-ization of research, moving a higher proportionof scientists closer to the rural environmentsto which their research should relate, and inwhich much of it should be conducted.

Second, cost-effective rural appraisal mightbe developed much more systematically as a subject. Rapid appraisal of rural situations iswidely practiced but not much written about.2

Authors are coy about describing the way theyfind out about rural situations when t ime isshort. Yet rapid rural appraisal is very widelyundertaken and its methods are continuouslybeing improved by practitioners; but they lackthe respectability of elephantine surveys orprofound participant observation. They are de-scribed apologetically as "quick and di r ty"when in practice much conventional rural re-search is inefficient — " long and di r ty" — andmuch rapid work is "quick and clean."Ladejinsky, for example, identified the adversedistributional effects of the "green revolut ion"on brief field visits in India (Ladejinsky 1969a,1969b, cited in Clay 1978), years before expen-sive, extensive surveys came up with the samefindings to two (spurious) decimal points. Thereis much potential here for new methods oflearning from rural people, for example throughrepertory grid techniques (Richards 1979) andappropriate methods of quantification (Barket1979). Armed wi th cost-effective "quick andclean" methods of learning f rom rural people

2. See, however, the papers of the Workshop onRapid Rural Appraisal , Insti tute of DevelopmentStudies, University of Sussex, November 1978.See also the work of Belshaw (1976), Biggs (1978),Coll inson (CIMMYT 1977, 1978) and Hildebrand(1978), and for thcoming f rom CIMMYT, "PlanningTechnologies Appropr ia te to Farmers' Cir-cumstances: A Manual of Procedures."

90

and about rural situations, technical and socialscientists engaged in R & D should be able toshort-circuit long channels of noncommuni-cation and should be able to experience andlearn for themselves much more directly.

Third, changes might be made in professionalreward systems for researchers, relating theseto desired behavior. Some of the most serioussocioeconomic constraints to development arein ourselves, the elite professionals who writepapers and conduct research. What are thethings we do not see, understand or do becauseof our condit ioning, motives, and lifestyles? It iseasier to ask this question than to answer it; andthe writer makes no pretense to any virtue onthis score. But perhaps more could be donethrough rewards to those scientists whopioneer new methods that truly serve thepoorer people, who work on subjects of lowprestige, and who work in creative partnershipwith rural people.

Finally, there is the question of who doeswhat. If war is too important to be left to thegenerals, as Clemenceau believed, social scien-tists have also been right in arguing that agricul-tural and mechanical R & D is too important tobe left to technical scientists. The most com-mon, and desirable, response has been to addsocial scientists to the staffs of research insti-tutes and stations, and sometimes they areinvolved in research decisions. But the socialand economic implications of prospectivetechnologies seem too important to be left justto the social scientists. Multidisciplinary collab-oration has its uses and its place. It is especiallyvital in determining research priorities and theallocation of research resources. But it also hasits well-known difficulties and costs. TheDuchess in Alice in Wonderland might havebeen speaking for many who feel that theinvolvement of many disciplines slows up deci-sions and action. At the same t ime, the socialand economic implications of new technologyare so great that technical scientists cannotwash their hands of them: it is the irresponsibil i-ty, in part, to see "where the rockets comedown. " Perhaps solutions can be sought in allprofessionals becoming more truly multidisci-plinary, in social scientists learning to think morelike technical scientists, in technical scientistslearning to think more like social scientists, and inall alike working with and learning from those truemultidisciplinarians, the rural people themselves.

References

BARKER, DAVID. 1979. Appropriate methodology: an

example using a tradit ional African board game tomeasure farmers' attitudes and environmental im-ages. Institute of Development Studies Bulletin10(2).

BELSHAW, DERYKE. 1976. Appropriate methodology for

development research, Institute of DevelopmentStudies Bulletin 8(1).

BIGGS, STEPHEN D. 1978. On-farm and vil lage-level

research: an approach to the development of ag-ricultural and rural technologies. CIMMYT Asian Report 6, available f rom ICRISAT Liaison Office, 23Golf Links, New Delhi, 110003, India.

BINSWANGER, HANS P., and RYAN, JAMES G. 1977.

Efficiency and equity issues in ex ante al location ofresearch resources. Indian Journal of AgriculturalEconomics 32(3).

BINSWANGER, HANS P., JODHA, N. S., RYAN, J. G., and

von OPPEN, M. 1977. Approach and hypotheses forthe village-level study of the International CropsResearch Institute for the Semi-Arid Tropics.ICRISAT Economics Program occasional paper 15,Patancheru, A. P., India.

CHAMBERS, ROBERT. 1978. Towards rural futures: anapproach through the planning of technologies.Institute of Development Studies, discussion paper134, University of Sussex, Br ighton, UK.

CIMMYT (Centro Internacional de Mejoramiento deMaiz y Trigo) 1977. Demonstrations of an interdis-ciplinary farming systems approach to planningadaptive agricultural research programmes, Part ofSiaya District, Nyanza Province, Kenya, CIMMYTEastern Africa Economics Program Report 1, P.O.Box 25171, Nairobi, Kenya.

CIMMYT (Centro Internacional de Mejoramiento deMaiz y Trigo) 1978. Demonstrations of an interdis-ciplinary farming systems approach to planningadaptive agricultural research programmes, Part ofSerenje District, Central Province, Zambia, CIMMYTEastern Africa Economics Program, P.O. Box 25171,Nairobi, Kenya.

CLAY, EDWARD. 1978. Direct and indirect methods ofobservation in rapid rural appraisal. Presented at theWorkshop on Rapid Rural Appraisal , University ofSussex, Institute of Development Studies, Br ighton,UK.

HILDEBRAND, PETER E. 1977. Generating smal l- farm

technology: an integrated mult idiscipl inary system.Presented at the 12th West Indian AgriculturalEconomics Conference, Caribbean Agro-EconomicSociety, 24-30 Apr i l 1977, Ant igua.

HILDEBRAND, PETER E. 1978. Generating technology for

tradit ional farmers: a mult idiscipl inary methodol-

91

ogy. CIMMYT Asian Report 8, available f r o m LiaisonOffice, 23 Golf Links, New Delhi 110 003, India.

JOHNSTON, BRUCE F. 1979. Socioeconomic aspects of

improved animal-drawn implements and mechani-zation in semi-arid East Afr ica. Presented at theInternational Workshop on Socioeconomic Con-straints to Development of Semi-Ar id Tropical Ag-r iculture, 19-23 Feb 1979, ICRISAT, Hyderabad,India.

KOPPEL, BRUCE. 1978. Technology assessment and

research management. SEARCA Bulletin 4. SouthEast Asian Research Center for Graduate Study andResearch in Agr icul ture, Laguna, Calif, USA.

LADEJINSKY, W O L F , 1969a. The green revolut ion inPunjab: a field t r ip . Economic and Political Weekly4(23).

LADEJINSKY, W O L F . 1969b. The green revolut ion inB i h a r - t h e Kosi area: a field t r ip . Economic andPolitical Weekly 4(39).

M B I T H I , PHILIP, and BARNES, CAROLYN. 1975. The spon-

taneous sett lement problem in Kenya. Nairobi ,Kenya: East Afr ican Literature Bureau.

PEARSE, ANDREW, 1977. Technology and peasant pro-

duct ion: reflection on a global study. Developmentand Change 8(2).

RICHARDS, PAUL. 1979. Communi ty environmental

knowledge in Afr ican rural development. Insti tute ofDevelopment Studies Bullet in 10 (2).

RUTTAN, VERNON. 1977. Induced innovat ion and ag-

r icultural development. Food Policy, August p 196-216.

SANDERS, J O H N H. 1979. New agricultural technology

in the Brazilian Sertao. Presented at the Interna-t ional Workshop on Socioeconomic Constraints toDevelopment of Semi-Ar id Tropical Agr icul ture,19-23 Feb 1979, ICRISAT, Hyderabad, India.

SPRAGUE, HOWARD B. 1976. Combined crop-livestock

farming systems for developing countries of thetropics and sub-tropics, Technical Series Bullet in 19,Technical Assistance Bureau, Office of Agr icul ture,Agency for International Development. WashingtonDC, USA.

World Bank. 1975. Rural Development: Sector PolicyPaper. International Bank for Reconstruction andDevelopment, Washington, DC, USA.

92

Chai rman 's Summary

J . L . D i l l o n *

Three papers were presented in the session.The paper by Fall pertaining to Senegal em-phasized that whi le some new techniques wereavailable, their acceptance by farmers was low.Lack of adequate marketing and credit facilitieswere identified as critical constraints. Ryan et al.presented an analysis of improved soil andwater management and watershed techno-logies being researched at the ICRISAT Center.Analysis of steps-in-improved technology,runoff s imulat ions, and watershed-basedsupplementary irrigation systems yielded someuseful guidelines about research strategies andalso the cost-effectiveness of new techniques.The paper by Sanders examined the rationale,possibilities of adoption, potential economicimpact, likely constraints in adoption, and con-sequent policy needs of improved sorghumtechnology in Northeast Brazil.

The discussion covered four broad areas.First, several methodological issues wereraised. For example, it was pointed out thateconomic evaluation of prospective techno-logies should be based on a range of possibleinput-output prices and not on current orsingle-valued prices only. The importance ofconducting sensit ivity analysis was em-phasized. Similarly, since risk is such an impor-tant characteristic of the SAT environment,economic evaluation should incorporate thisfact. The usefulness of activity analysis andprogramming techniques in ex ante evaluationwas also pointed out.

The other point repeatedly emphasized wasthe crucial importance of testing on farmers'fields and village-level trials. While .there was nodifference of opinion regarding this need, someparticipants expressed the fear that unless pro-fessional biases were overcome and the rewardand incentive systems for scientists underwenta drastic change, the real purpose of such" tes t ing" on farmers' fields would not beserved.

The importance of incorporating needed poli-

* Universi ty of New England, Armidale, NSW,

tralia.

Aus-

cy and institutional changes in economic evalu-ation of prospective technologies was alsostrongly emphasized. It was felt that eventhough these often involve political consider-ations, it would be useful to spell out constraintsin this area. Price policies for output and inputs,as well as marketing and credit institutions,were repeatedly mentioned as important in thiscontext.

The need to analyze likely effects oftechnological change on income distributionwas stressed by several participants. Thedifficulties in such analyses were pointed outbut the group felt that, to the extent possible,this should be attempted.

It was also pointed out that the evaluationwork should also include the realities of "differ-ential access" to factor markets of farmers indifferent size categories. The importance ofoffering a range of improved technologicaloptions rather than a single "package" wasconsidered relevant in this context.

The main conclusions were as fol lows:1. Technology design. It was felt that there is

a need to learn f rom, and take advantage of,elements of farmers' present practices. Thisshould be used as the starting point for anyresearch. Greater stress is needed on the exante economic evaluation of prospectivetechnology.

2. Agricultural policy. Economic evaluationof prospective technology should include dis-cussion of the constraints imposed by existingpolicies being pursued by governments andalso by the existing rural institutional infrastruc-ture. Inter alia, it must indicate needed changesin these directions.

3. Future socioeconomic research in national programs. Large-scale field trials on farmers'fields were identified as a major need.

4. Future socioeconomic research at ICRISAT. Using ICRISAT's own prospective contributionsto technology as case examples, ICRISATshould aim to develop methodologies orientedto the ex ante design and evaluation of newtechnologies so that these methodologies maybe available to national programs.

93

Chapter 3

S o c i o e c o n o m i c F i e l d A s s e s s m e n t o f P r o s p e c t i v eT e c h n o l o g i e s

Soc ioeconomic Field Assessment o f Dry FarmingTechno logy in t h e Indo-UK Pro jec t , Indore ( India)

A s h o k K . C h a u d h a r i *

Abstract

This paper describes the effects of the introduction of improved soil-, water-, crop-, and

livestock-management technology on Vertisols in the Indo-UK Project at Indore, Madhya

Pradesh, India. Since 1974, when the project began, farmers on the 2000 ha area have

markedly increased their adoption of HYVs of sorghum, wheat, and chickpea and have

begun growing soybeans and maize as new crops. The area under kharif (rainy season)

crops has risen substantially during the past 4 years from 32% of the gross cropped area

to 54% in addition to a tenfold increase in the area double cropped. Along with increased

use of fertilizers and irrigation, these changes have increased crop yields by 30 to 500%

and net returns per hectare by an average of more than 170%. The technology has been

labor and bullock intensive and this has favored small farmers who have much higher

man|land and bullock|land ratios than large farmers.

The Malwa region, in which Indore District fails,is characterized by an annual average rainfall of980 mm and heavy Vertisols (black soils). It hasnot responded very enthusiastically to theadoption of new dryfarming technology. Still invogue is single cropping, mostly in the rabi (postrainy) season on l imited stored moistureand very little use of fertilizer, coupled wi thtraditional farm practices and continuance ofoutdated implements. This is mainly because ofthe farmer's fear of greater risk in the adoptionof kharif (rainy season) cropping and the cost ofmodern inputs such as fertilizers, high qualityseed, and plant protection chemicals. His fearhas been reinforced by unsatisfactory soil- andwater-management practices previously re-commended.

The average size holding in the Indore area isabout 6.9 ha, but scattered holdings wi th lowproduction have posed management and in-vestment problems. The investment on inputssuch as fertilizer is low because of the very smallproport ion (9%) of net cultivated area receivingassured irrigation.

Research work conducted by the All IndiaCoordinated Research Project for Dryland Ag-riculture (AICRPDA) main center at Indore since

* Production Economist, Indo-UK Dry Farming Pro-ject, Collage of Agr icu l ture, Indore, India.

the rabi of 1970-71 has resulted in developmentof a dryland technology which, if applied tofarmers' fields, wi l l completely change the con-dit ion of barani (dry) farming. However, there isa large gap between the average yields ofexperimental plots and farmers' fields.

This project was started during late 1974 on a net cultivated area of some 2000 ha. Its objec-tive is to bring about development of farmingon a catchment basis after utilizing all availableresources. The project seeks to increase landproductivity by adoption of advanced technolo-gy in land improvement, soil and water conser-vation, crop husbandry, animal husbandry,exploitation of underground water, etc.

There are four facets of the actions throughwhich the project seeks to transfer the technolo-gy suited to this dry farming area:

1. land husbandry, including a soil and watermanagement program,

2. cropping program,3. livestock management program, and4. farm machinery program.The main task of the farm economics disci-

pline is to monitor agroeconomic data emergingfrom the various implemented programs andestimate the periodic changes in investmentand returns of individual farms. Socioeconomicassessment has been made of:

1. benefit to and participation of small far-mers,

97

2. adoption among farmers,3. changes in cropping patterns,4. changes in irrigated areas,5. overall indications of project-induced

changes.Before we proceed to socioeconomic as-

sessment, it is imperative to focus on what hasbeen done through various programs to im-prove the productivity of agriculture and lives-tock. Though no estimation has been made ofmonetary gains to each program, they all havecontributed significantly to better returns f romcrop and dairy enterprises.

S o i l a n d W a t e r M a n a g e m e n t

Various activities have been undertaken in soiland water management, mostly on communityland, but also on individual plots where a groupof farmers have agreed to the program.

About 6 km of storm drains were constructedto protect 390 ha of land f rom erosion. Grassedwaterways of 10 km were constructed to drainwater f rom 736 ha. In waterway bed stabiliza-t ion and gully reclamation 15 gabion structureswere constructed on waterways and haveworked successfully during the rainy season.

These measures have improved overalldrainage and minimized soil erosion over someareas and helped eliminate water stagnationalong some bunds.

C r o p p i n g P r o g r a m

Several measures have been adopted underthis program to improve the productivity offarmers' land — namely, a change in varietiesof popular crops, introduction of more re-munerative crops and replacement of kharif fal low by crops. It has been the policy of theproject to give special consideration to smalland medium-sized farms. Some examples ofthese measures are:

• Change in varieties of already popularcrops. Sorghum hybrid CSH-5 was plantedon 300 ha during 1978-79. In rabi, high-yielding and improved varieties of wheat(Kalyan sona, RR 2 1 , Narmada-4, andNarmada-112) on 364 ha and chickpea(Ujjain-24) on 365 ha were planted duringthe crop year 1977-78 on the advice ofproject scientists.

• Introduction of new crop varieties. Soy-bean (T 49, JS 2, and Ankoor) on 224 ha andmaize (Chandan 3 and Ganga 5) on 89 hawere planted during kharif 1978-79. Thesecrop varieties are new to this area and arequite remunerative.

• Replacement of kharif fal low by crops. On1800 ha of cultivated land, the area underkharif crops has increased f rom 32% ofgross cropped area in 1973-74 to 54%during 1977-78.

• Double cropping. Farmers practiced dou-ble cropping on 244 ha in 1975-76, on 420ha in 1976-77, and on 744 ha in 1977-78.Before 1975-76, the double cropped areawas only 72 ha.

• Increased use of fertilizers. The quality offertilizer nutrients applied by farmers hasincreased progressively (Table 1).

Table 1 . Use o f nu t r ien ts in Indo-UK Dry Farm-ing Pro jec t a rea .

Year

1974-751975-761976-771977-78

Nutrients used (kg)

N

4 260159311471416 630

P2O5

3 32610 63010 83210 845

K2O

712648

1090840

These measures on the part of the croppingprogram have increased the land productivityconsiderably, mainly through crop-variety sub-sti tut ion, fertilizer application, and stability inarea coverage (Table 2).

Fodder crop yields also have been increased.Vidisha 60-1 and J-69 varieties of sorghumfodder, berseem, oats, and a new foddercrop — feosinte (Euahlaena mexicana) — havebeen introduced in the project area to improvelivestock management. Al l have higher yieldpotential (Table 3). Thus the crop husbandryscientists have helped farmers to raise bettercrops and double production — especially inthe kharif, — maintain better livestock, andrealize higher returns through enhanced landproductivity.

98

Tab le 2 . Crop y ie lds be fo re and a f te r t h e Pro jec t .

Crop

SorghumMaizeSoybean

Wheat (rainfed + irrigated

Chickpea

Before Project (1973-74)

Variety

LocalSatnaKalitur(black)

Local + MexicanLocal

Average y ie ld(kg/ha)

430550550

500450

After Project(1977-78)

Variety

CSH-5Chandan-3T-49JS-2Narmada-4MalawrajUjjain-24

Range(kg/ha)

1200-43001000-2400

650-1150800-1350

N.A.a

N.A.a

Average yield(kg/ha)

20201500750820

630630

a. Not available.

Tab le 3 . Improved fodder c rop y ie lds inIndo-UK Dry Farming Project area.

Fodder Crop

Sorghum Vidisha 6 0 - 1 + CowpeaSorghum j-69Berseem + OatsLucerne + Oats

Average yieldgreen fodder

(metric tonne/ha)

41.551.8

124.381.2

L i v e s t o c k M a n a g e m e n tP r o g r a m

The project's effort has been to improve lives-tock to actively complement productivity of theland by applying more organic matter in thesoil , raising better young stock to augment farmpower, and improving milk animals. The fol low-ing three main activities were undertaken:

1. Feeding greens to yield more milk. Basedon observations of 27 cows and 56 buf-faloes, an average increase of 400 ml indaily milk production per cow (rangingf rom 1.3 to 1.5 liters) and of 650 ml perbuffalo (ranging f rom 3.2 to 3.8 liters) werenoted as a result of feeding greens.

2. Breeding. In al l , there were 92 local cowsinseminated wi th Jersey semen and 21she buffaloes wi th murrah bull semen,besides natural service of 79 she buffaloes.

3. Health coverage. Animals of the area were

protected against seasonal contagiousdiseases through vaccinations in addit ionto treatment for minor ailments.

The idea behind operating the livestock prog-ram was to minimize the risk of failure of thesingle-crop enterprise and to provide farmerswi th a stable source of income.

F a r m M a c h i n e r y P r o g r a m

The contribution of this discipline to projectfarmers has not been very significant. Only onesurvey — the field activity survey — was con-ducted to assess the use of t ime by human laborfor various field operations. The study revealedthat in most of the field operations, after every45 to 60 minutes of work farmers take 10 to 15minutes rest, reflecting the hard labor involved.This reduced the actual working hours to only 6 or 61/2 in an 8- to 10-hour working day. Besidesdrudgery, work is constrained by weather andsoil conditions. "Du fan" and "Phadak" imple-ments allow fairly rapid sowing but need two orthree laborers to operate them. Availability oftractor power is only about 0.1 hp/ha in thefarm-size groups above 4 ha, demanding use ofalternative tractor power sources.

A p p r o a c h t o E x e c u t i o no f P r o g r a m s

The project approach, especially in land hus-bandry, was to work wi th the approval of

99

farmers. When the farmer objected, we did notundertake the work; nor did we attempt to useforce (through the state law on land improve-ment). Wi th any new practice related to soil andwater management, cropping program, orlivestock management, farmers were firstshown the usefulness of the new practices andthen convinced of their suitability.

Input subsidies in various forms were givento participating farmers. Particular care wastaken to include small farmers to the extentpossible. Subsidies in the beginning on itemssuch as hybrid/improved seed, fertiliser, andplant protection chemicals were wi thdrawngradually so that farmers were able to bui ldtheir economy through their own efforts. Untilthe 1978-79 kharif, small farmers (< 2 ha) wereprovided free seed whereas for others it was onan exchange basis. This measure has de-veloped a sense of self-reliance among themajority of farmers.

B e n e f i t t o a n d P a r t i c i p a t i o no f S m a l l F a r m e r s

This project, unlike many other developmentalprojects, has especially benefited small farmersowning up to 2 ha, and the number of suchparticipants is quite high. In the beginning, landproductivity was low for small farmers, eventhough they used more human labor per hectarethan large farmers. The obvious reason wastheir very l imited investment capacity.

An attempt was made to estimate the inter-group variations in the yield level of importantcrops for the year 1976-77. About 35% of thetotal farm holdings were taken into consider-ation (Table 4). Average yield levels have aninverse relationship to the size of holding. Ingeneral, small farmers (up to 2 ha) obtained thehighest yields of crops and also raised twosequential crops on most of their holdings. Thiswas because small farmers have more humanlabor (6.8 family members per ha) and bullockpower (1 pair of bullocks for 1.8 ha of land)compared w i th large farmers (>12 ha) whohave 0.6 family members per ha and 1 pair ofbullocks for each 10.1 ha of land (Table 5). Thedryland technology's requirement of moretabor thus is better suited to small andmedium-sized farmers.

The figures in Table 6 show the number and

percentage of participating small farmers in thecrop program during 1978-79.

A d o p t i o n o f I m p r o v e d P r a c t i c e sb y F a r m e r s

This was a common program agreed to by theIndo-UK Dry Farming Project and the coordinat-ing cell of the AICRPDA in Hyderabad. Theprime aim of this study was to ascertain to whatextent farmers adopt feasible and viable re-commendations and their motivation for doingso, thus determining pre-requisites necessaryto bring needed changes in agriculture. In all, 72farm families (20% of the population) werestudied during 1975-76 and 137 farm families(38% of the population) during 1976-77. Sizegroups of holdings were made in accordancewith instructions from the Coordinating Cell ofAICRPDA as fo l lows: below 1.0 ha, 1.1-2.0 ha,2.1-4.0 ha, 4.1-8.0 ha, and in excess of 8.0 ha.Six crops — maize, soybean, sorghum, wheat,chickpea, and sugarcane — were included inthe study to estimate the extent of adoptionduring 1975-76 and all but sugarcane during1976-77.

Adoption during 1 9 7 5 - 7 6

Sampled farmers had 434 ha under crops. Ofthis, only about 43% was planted to high-yielding improved local varieties.

The max imum use of fertilizer was observedamong farmers in group 1.1 to 2.0 ha (15.6 kg/haN and 11.2 kg/ha P2O5) fol lowed by the largestsize group (14.7 kg/ha N and 10.4 kg/ha P2O5).Soils of this area are not deficient in potash;therefore, farmers seldom apply any quantity ofthis nutrient.

The analysis of the level of fertilizer useindicated that farmers are still applying im-balanced doses of fertilizer nutrients, and theyneed to be educated on this aspect. Farmerswi th holdings of more than 4.0 ha in many casesapplied between 50 and 100% of the recom-mended doses of nutrients, whereas smallerfarmers generally applied less than 50% of therecommended doses, reflecting their financialinability to invest.

Most farmers adopted only two practices:improved high-yielding seed and fertilizer. Sig-nificant adoption occurred in maize, chickpea.

100

Tab le 4 . Va r i a t i on in t h e average y ie ld levels across f a r m size groups ( 1 9 7 6 - 7 7 ) .

Size-group(ha)

Below 2.02.1-404.1-8.08.1-12.012.1 and above

All Groups

Size-group(ha)

Below 2.02.1-4.04.1-8.08.1-12.012.1 and above

All Groups

Size-group(ha)

Below 2.02 .1-4 .04 .1 -8 .08.1-12.012.1 and above

Phaseolus (Urd)

No. ofobs.

43

123

13

35

AverageArea yield(ha) (kg/ha)

4.35 2103.00 30

13.50 1904.75 70

39.25 130

64.85 140

Maize

No. ofobs.

109

156

15

55

AverageArea yield(ha) (kg/ha)

4.04 11104.60 4509.15 3602.19 880

12.35 460

32.33 540

Wheat (irrigated)

No. ofobs.

1113174

15

60

AverageArea yield(ha) (kg/ha)

10.50 211015.90 171031.80 13808.25 1760

49.90 1860

116.35 1720

Hybrid sorghum

No. ofobs.

23

-14

10

Area(ha)

0.903.55

-0.504.30

9.25

Averageyield

(kg/ha)

17801660

-

14001460

164

Black soybean

No. ofobs.

24728

23

Area(ha)

1.004.60

11.250.75

10.90

28.50

Averageyield

(kg/ha)

850840640670520

630

Wheat (unirrigated)

No. ofobs.

148

238

28

81

Area(ha)

9.828.15

51.8515.75

159.40

244.97

Averageyield

(kg/ha)

790730660530590

610

Local so rghum

No. of Areaobs. (ha)

11 11.4410 9.8525 35.47

6 12.7525 92.95

77 162.46

Averageyield

(kg/ha)

420380490250260

330

Groundnut

No. of Areaobs. (ha)

1 0.602 1.053 0.85

-6 3.55

12 6.05

Averageyield

(kg/ha)

8028101800

-450

1010

Chickpea

No. of Areaobs. (ha)

13 11.6014 12.8134 71.45

9 24.3028 133.35

98 253.51

Averageyie ld

(kg/ha)

430500210150190

220

and wheat. In maize, 10% of the farmersadopted high-yielding seed, covering 10 to 40%of their area; about 1 1 % fertilized their area. Inchickpea, about 13% planted improved seedand 17% fertilized their crop. Wheat was themost popular crop in terms of adoption, with30% of the farmers planting the recommendedseed variety in 10 to 40% of their cropped area.More than 54% applied fertilizer on 40 to 50% oftheir wheat area.

Reasons for nonadoption were given as: thecostly nature of seed and fertilizer input, capital

shortage, risk of possible losses, nonavailabilityof suitable seeds, poor cooking quality of theproduce, and unawareness of the methods ofhybrid seed cultivation, fertilizer application,and plant protection.

Five farmers under 25 years of age, 28 far-mers of the 25 to 50-year age group, and 18above 50 all cited lack of capital as a majorconstraint to adoption. Four farmers were not atall convinced about the new technology.

Ten illiterate, 1 literate, and 29 farmers wi thprimary school education cited lack of capital as

101

Tab le 5 . Ava i lab i l i t y o f bu l lock p o w e r ( 1 9 7 6 - 7 7 ) i n Indo-UK Pro jec t a rea .

Size-group(ha)

Below 2.02.1-4.04.1-8.08.1-12.012.1 and above

Vil lage

TotalNo. offarms

3126461628

147

Totaloperated

area(ha)

30.8981.36

277.88153.16564.33

1107.62

Average Totalsize of No. ofhold ing bullock

(ha) pairs

1.00 173.13 216.04 489.57 21

20.15 56

7.53 163

Operatedarea coveredby a pair of

bullocks(ha)

1.83.95.87.3

10.1

6.8

Bullockpairs/haof land

(5/3)

0.550.260.170.140.10

0.15

Bullockpairs/farm

(5/2)

0.550.811.041.312.00

1.11

Table 6 . Par t i c ipa t ion by sma l l f a r m e r s inc r o p p r o g r a m .

Sorghum CSH 5 Maize Chandan 3

Ganga 5 Soybean T 49

JS 72 -44

Smal l farmersa

(no.) (%)

40 1413 417 13

15 136 30

Other farmers

(no.) (%)

252 8619 5946 87

101 8714 70

a. Percentage of small farmers to total farmers is about 16.

the main reason for nonadoption, whereas 3 literate and 3 primary school-educated farmerswere not satisfied about the superiority of thetechnology. Four primary school educatedfarmers said diff iculty in marketing their pro-duce was the reason for nonadoption.

A d o p t i o n d u r i n g 1 9 7 6 — 7 7

Sample farms covered 729 ha of the net culti-vated area. Double cropping was done on 197 ha,indicating a cropping intensity of 127%.

Seed. Of the five crops included in the study,chickpea and wheat alone covered two-thirds ofthe area under recommended seed varieties.Chickpea covered the largest cropped area(17.6%), fo l lowed by wheat (13.9%) (Table 7).Among the five size-groups, the highest per-centage adoption was in the 1.1 to 2.0-hagroup; adoption declined wi th increased sizeof holding.

102

Fertilizer. Of the 927 ha cropped area, only350 ha (37.8%) was fertilized. The percentageshare of various crops in the fertilized area was:wheat, HYV/improved, 40 .1 ; wheat, local, 13.0;sorghum HYV 10.3; soybean, 9.6; chickpea, 8.7;and maize, HYV 7.2. The rest of the 11.1%fertilized area was in local sorghum and othercrops not included in the study (Table 8).

Highest adoption of fertilizer was observed inthe case of the smallest size-group (48.2%cropped area fertilized) fol lowed by size-group5 (44.7%), size-group 3 (38.4%) and size-group 2 (32.1%). Percentage of fertilized area was high-est in the smallest size-group because they havemuch less area to cultivate and fertilize (Table9). This confirms our earlier contention thatsmaller farmers are adopting dry farmingtechnology in increasing numbers.

Consumption of N per hectare was foundhighest (18 kg) in size-groups 1 and 3, fo l lowedby 14.7 kg in size-group 4 (Table 9). Applicationof P2O5 was highest in size-group 1 (12.1 kg),fo l lowed by size-group 3 (10.0 kg).

Among the six crops, nitrogen applicationwas highest on sugarcane (81.0 kg/ha), fol lowedby vegetables (70.8 kg/ha), wheat HYV (50.8kg/ha), and sorghum HYV (33.5 kg/ha).1 P2O5

application was highest on sorghum HYV (50.1kg/ha), fol lowed by improved chickpea (26.8kg/ha). Almost every farmer indicated lack ofcapital as the main constraint to use of fertilizer.About one-third expressed their lack of aware-ness of suitable crop varieties. A few farmers

1. Gross cropped area.

Tab le 7 . A d o p t i o n o f r ecommended t e e d , 1 9 7 6 - 7 7 . a

Crop

Sorghum

Maize

Soybean

Wheat

Chickpea

Total

Size-group (ha)

Below 1.01

1.33(7.9)a

1.27(7.5)0.18(1.1)2.75

(16.3)2.82

(16.7)

8.35(49.5)

1.1-2.02

1.85(3.5)1.17(2.2)5.39

(10.1)10.18(19.1)12.89(24.2)

31.48(59.1)

2.1-4.03

4.33(4.0)2.43(2.2)

11.22(10.3)26.48(24.3)13.12(12.1)

57.58(52.9)

4.1-8.04

6.95(3.3)3.35(1.6)

24.97(11.9)37.17(17.6)36.87(17.5)

109.31(51.9)

8.1 and above5

26.01(4.8)7.07(1.3)

33.00(6.1)

52.50(9.8)

97.2018.1)

215.78(40.1)

Al l Groups

40.47(4.4)

15.29(1.6)

74.76(8.1)

129.08(13.9)

162.90(17.6)

422.50(45.6)

a. Figures in parentheses indicate percentage to total cropped area.

expressed their doubts about the superiority ofsome improved HYVs.

One significant fact observed in the case offertilizer consumption was that quantity of nutri-ents applied per hectare has increased over theyear 1975-76. This led to better crop yields in1976-77.

Change in Cropping Pattern

It has been the main emphasis of the soil- andwater -management program to provide

Tab le 8 . A d o p t i o n o f fer t i l i zer , 1 9 7 6 - 7 7 .

Crop

Sorghum HYVSorghum localMaize HYVMaize local

SoybeanWheat HYV/IMPWheat localChickpea

Total

Areas o w n<ha)

40131

1528

75129203231

852

Areaferti l ized

(ha)

36161318

341214630

314

Percentagecropped area

ferti l ized

(%)

89128863

45932313

37

maximum soil cover through kharif cropping tominimize soil erosion and save the upper fertilestratum of the soil. Through project innova-t ions, the percentage of kharif-cropped area hasincreased from 30.8% of total cropped area in1974-75 to 49.4% in 1977-78. The increase inrainy season crop hectarage that this representsis over 200% (Table 10). More kharif croppinghas increased not only the average net earningsof farmers, but also helped in building a morefertile upper soil stratum. Some of the increased(nonirrigated) kharif crop area has been at theexpense of a reduction in the area of (nonirri-gated) rabi crop area, which fell by 18%.

Change in Irrigated Area

Various activities in soil and water managementto improve the drainage conditions and in thecropping program to g row better crop varietiessuitable for rainfed conditions have temptedfarmers to invest in constructing open wells andlift equipment to provide at least one or twolight irrigations for rabi crops such as wheat andvegetables, for which moisture supply is consi-dered to be very critical. The number of suchwells has increased from 130 in 1973-74 to 186during 1977-78, increasing the area under as-sured irrigation f rom 172 ha to over 254 ha. Thisis an increase of about 48%. Of the increasedirrigated hectarage, more than 80% has gone

103

104

into wheat alone, and the rest into vegetablesand other rabi crops.

F a r m B u s i n e s s A n a l y s i s

As revealed by Table 11, significant changeshave occurred in operational charges, grossincome, and net income. The level of invest-ment and return declined during 1974-75 be-cause of adverse climatic conditions and a poorharvest. The changes in input and operationalcosts per hectare may not appear substantial

since the cropped area has increased consider-ably and the total quantum of investment wouldtherefore be far more than the 1973-74 level.

For want of an appropriate price index for thearea, deflation of cost and return data could notbe determined. However, if we allow for 5%annual inflation, the net return per hectare forthe years 1974-75 to 1976-77 would be Rs. 163,Rs. 386, and Rs. 445, respectively (Table 11).These figures represent an increase of 137% in1975-76 and 173% in 1976-77 over the netreturn per hectare of 1973-74. We may not f ind

Tab le 1 0 . Cropp ing p a t t e r n I n t h e Pro jec t a rea .

Crop

SorghumSorghum mixturesMaizeSoybean

Other kharifa ( irrigated)

Other kharif (nonirr igated)

Total kharifa (nonirr igated)

Total Kharifb

Wheat (irrigated)Wheat (nonirr igated)

Chickpea (irrigated)Chickpea (nonirr igated)

Other rabi (irrigated)Other rabi (nonirr igated)

Total rabi c (irrigated)

Total rabi (nonirr igated)

Total rabib

Total Area Cropped

Net Area Cropped

Year

1974-75

161189

21132

16(2.4)

151

654(97.6)

670(30.8)

115582

10583

35183

160

1348(89.4)

1508(69.4)

2178

1946

1975-76

173210

6485

25(3.2)

199

730(96.8)

755(34.2)

126546

16628

4489

186

1263(87.2)

1449(65.8)

2204

1960

1976-77ha

169238

80127

40(4.5)

222

836(95.4)

875(36.6)

174543

29670

3761

240

1274(84.2)

1513(63.3)

2389

1969

1977-78

320260115414

24(1.8)

215

1324(98.2)

1348(49.4)

197508

21570

5132

270

1109(80.4)

1378

(50.6)

2726

1981

Note : Figures in parentheses ind ica te percentage a rea—

a. to khar i f c ropped area,

b. to to ta l c ropped area,

c . to rab i c ropped ares .

105

Tab le 1 1 . Fa rm business analysis o f samp le fa rmer . a

Particulars

No. of farmersNet cult ivated area (ha)Gross cult ivated area (ha)Intensity of cropping (%)Input costs (Rs/ha)

Operational costs (Rs/ha)

Interest on work ing capital , land revenueand taxes (Rs/ha)

Total expendi ture (Rs/ha)

Gross income (Rs/ha)

Net income (Rs/ha)

Year

1973-74

107607625103233

(100)231

(100)

29

493(100)656

(100)163

(100)

1974-75

17110061088

108196(84)169(73)

26

388(79)560(85)172

(106)

1975-76

137823942114269

(116)168

(116)

39

576(117)1005(153)429

(263)

1976-77

146996

1309131257

(110)313

(136)

34

608(123)1127(172)523

(321)

a. Figures in parentheses denote percentage change, with the bate year 1973-74 taken as 100.

any actual change in net return of 1974-75, asthe project did not undertake a major soil- andwater- management and cropping program thatyear.

Final analysis of the farm-survey data for1977-78 has not been completed. However,case studies of four individual farmers revealthat cropping intensity on these farms hasincreased by one-third and in some cases isdouble that of 1973-74. Net farm income hasincreased threefold to f ivefold and net incomeper hectare f rom twofold to fourfold whencompared wi th 1973-74.

O v e r a l l A s s e s s m e n t

As previously stated, the project's main thrustwas on farmers' participation in the program towin their confidence and achieve success. Far-mers of the area now understand thephilosophy of the dry-farming technology. Thedemonstrations w i th new and improved high-yielding crop varieties on farmers' fields haverevealed their yield potential under rainfedconditions. The project's main emphasis onincreased kharif cropping to ful ly utilize theabundant moisture available in the monsoon in

these heavy black soils and to provide anefficient soil cover to prevent erosion has beenrealized. Farmers now grow more recom-mended kharif crops (also rabi crops whereverpossible), apply fertilizer, take proper care oftheir milk cattle and thus realize better yields.They are now becoming prosperous, which isevidenced by their better standard of l iving.More than half of their overdue institutionalloans have been repaid.

Evaluation studies have succeeded in identi-fying socioeconomic constraints to adoption ofscientific crop and livestock management. Im-balances in fertilizer application and other re-lated problems and profitabil ity of traditionaland recommended cropping patterns havebeen revealed. These observations have beenconveyed to concerned scientists so that theycan make appropriate adjustments whi le for-mulating plans for the future.

106

H u m a n Na tu re and t h e Designof Agr icu l tu ra l Techno logy

Vic to r S . Dohe r t y *

Abstract

Application of research results from the cross cultural study of human group size and

function suggests organizational strategies suitable for promoting widespread im-

provement of land management and for development of supplementary irrigation.

Improvement of land management may be most rapid if the necessary implements can

be owned individually by even the smallest farmers, and by small-scale entrepreneurs.

Development of supplementary irrigation may be most rapid if irrigation sources are

owned by individuals or are large enough to justify government administrative

assistance in their operation and upkeep. Further cross cultural study and field expe r i -

ments are suggested to test these hypotheses.

This paper comprises a resume of some an-thropological and sociological studies on thesizes and functions of human groups and anapplication of conclusions f rom these studies tothe improvement of land- and water-resourcemanagement in semi-arid tropical agriculturalregions. In particular there are discussions ofalternative possibilities for the ownership anduse of mult ipurpose, bullock-drawn wheeledtool carriers and for supplementary and life-saving irr igation of rainfed crops. Thetechnologies and farming systems discussedhave been studied over the last several years atICRISAT. They have recently been brought tothe stage of experimentation in village contexts.Thus the present paper also includes a consid-eration of ways to obtain meaningful socialorganizational data and analyses from suchfield experiments. It should be noted that theparticular agronomic techniques discussedwere first designed in an Indian context and arebeing tested at present under Indian conditions.Thus adaptation is clearly needed before theirapplication in other contexts. At the same t ime,

* Principal Social Anthropolog is t , Economics Prog-ram, ICRISAT.

N O T E : This is a revised vers ion of the paper pre-sented at the Workshop and incorporatesf indings f r om research conducted since then.

the principles involved, including improvedland management and supplementary irri-gation, are themselves widely applicable.

The primary anthropological questions heredeal wi th how much cooperation is to be ex-pected from different sizes of groups of farmersunder different conditions. The answers tothese questions will help to determine whethergroup ownership and control can be consideredin a particular case or whether individual own-ership and control is necessary. The an-thropological analysis is put forward on thebasis of hypotheses about human nature and isnot l imited to any particular cultural context.

A n t h r o p o l o g i c a l A n a l y s i so f G r o u p S i z e a n d F u n c t i o n

In order to contribute to a general understand-ing of the conditions under which one canexpect cooperation by farmers, an exploratoryreview of literature and of a l imited number ofcases was carried out recently by Doherty andJodha (1977). On the basis of the study severalhypotheses were formulated. One of these wasthat small groups of people form effective taskgroups for short-term cooperation but do notcooperate successfully over the long term.Another was that successful long-termcooperative groups are likely to be of large size.In the cases of both large and small groups, it is

107

assumed that cooperation is to be enforcedprimari ly by the group itself or that outsideadministrative help wil l be minimal. As we wil lsee below, if structural, administrative aid f romoutside the group itself is available then thesituation changes somewhat.

The above hypotheses are supported by workcarried out in two related sciences, anthropolo-gy and sociology, seeking to explain the basicdynamics of human groups. On the sociologicalside the research referred to is that on smallgroups, so-called. Useful summaries of thesestudies are available in a series of reviewarticles (Deutsch 1968; Hare 1968; Homans1968; Raven 1968; Sherif and Sherif 1968).Much of this research was initially formulated toanswer questions about efficiency and functionin different sorts of group. The research referredto in anthropology includes the important workof Birdsell (1968, 1973) on sizes and functions ofgroups in hunter-gatherer societies. On theaxiomatic basis of the unity of the humanspecies, it is hypothesized that the cross culturalcomparison of simple societies, such as thoseof hunter-gatherers, wi l l be instructive regard-ing the possible existence of basic tendenciesand structural units, which arise in humannature and which must be taken into accounteven in more complex societies. Among thespecific studies to be discussed below, socialorganizational and cognitive studies in strati-f ied but stil l relatively s imple societies(Goodenough 1965) and in much more com-plex, stratified societies (Mahar 1959; Hiebert1971: 58-66) confirm that basic organizationalprinciples suggested by the work of Birdsell andthe small group sociologists on the one handand by social psychological studies on the other(Miller 1956)1 have a basic constancy acrosssocial types.

Some of the basic reasons for a seemingdifference in the ability of different sized groupsto perform different functions lie in the fact thatthere can be no society without individuals orindividual domestic groups. Each domesticgroup, for example, will have continually differ-ent needs vis-a-vis neighboring households asits members move through different stages oflife and as continually varying economicchanges affect each of these differing combi-

1 . Quoted in Goodenough (1965).

nations of individuals in different ways. Suchdevelopment is inevitable in human life, even inegalitarian band societies. The inequalities ofmodern and peasant societies exacerbate thesituation. In the cases of both large and smallgroups, smoothness of operation and con-t inuity of the group itself depend upon how wellindividual or individual household needs can bedealt wi th. Sahlins (1972) shows how the essen-tial individualism of nuclear groups is main-tained in even the most egalitarian and wellintegrated society and in societies supposed tobe ruled by the communal ethics of extendedlineage ties. Similar insights regarding the im-portance of individual benefits in assuringgroup continuity are contained in the work ofMancur Olson (1971). One need not be a socialatomist to maintain such a posit ion. Manysocieties are indeed cooperative, peaceful, andvalue unagressive behavior. Yet logically itwould seem that these characteristics are pos-sible exactly because potential causes ofconflict among individuals are minimized by thestructure of the society and by its culture.

Since societies are made up largely of indi-viduals and individual households, and sinceindividual differences are inevitable, smallgroups can be expected to be less stable thanlarge groups; this is because differences ac-quire relatively more importance the fewer thenumber of individuals in a group. Large groups,similarly, wil l have a better chance of balancingout such differences, putting down factions asthey arise, and enforcing general rules. On thebasis of such reasoning we can propose ap-propriate functions for different sizes of group:short-term tasks for small groups, and long-term administrative tasks for larger ones.Where small groups do have long-term co-herence we can expect that it is because someadditional, outside structure supports them,preventing or compensating for the rise ofindividual differences, which would disruptsuch a group left on its own. Very high profits tocooperation might also bind small groups.

Support ing the view that small groups aretask groups, there is some cross cultural evi-dence that people order their environmentphysically and mentally into small sets forday-to-day operational purposes. A hunting-gathering local band is likely to contain about 25persons. On a basis of f ive persons per family orhearth, it wi l l contain about five discrete interest

108

units based on common economic alliances,including marriage. Both psychologists andanthropologists have data to indicate that se-ven, plus or minus two, is the max imumnumber of discriminations easily made by theunaided human mind on the basis of any singlecontinum (Goodenough 1965: 17-18; Miller1956; Mahar 1959). The work by Mahar, oncaste ranking in an Indian village, revealed thather informants lumped together some rituallyrelevant actions out of a list of 13 such actions;the result was to distinguish seven potentiallyoperational status ranks. When 21 village casteswere ranked on the basis of whether or not keyactions f rom the ritually relevant set were al-lowed wi th members of certain castes, theresult indicated that the same informants usedthese potential ranks to classify their fel lowvillagers into an average of 5.4 functional statusgroups. In a study wi th similar results, Hiebert1971, pp 58-66) analyzed the responses of 42adult male informants to a series of questionsabout relative caste rank among 30 castes andabout relative individual rank among 18 menknown to all the respondents. He found that, asa whole, his informants dist inguished ninecaste status groups and four individual statusgroups of statistical significance.

Taken together, these findings suggest thatas task groups, groups of small size (7 plus orminus 2 members) would be suitable f rom a soc io log ica l and socia l psycho log ica lstandpoint, whi le they would also commandadvantages of scale by combining the efforts ofseveral persons. If long-term coordination isneeded, and if this coordination is to be locallygenerated rather than administered, then itseems that much larger groups would beneeded to overcome the tendencies to fac-t ionalism that are unavoidable in small groupscombining several individuals or domesticunits whose interests wi l l inevitably divergeover all but the short term. Opt imum sizes oflarge groups are much less clear theoreticallythan are opt imum small group sizes (seeBirdsell 1973). Birdsell's modal number is 500population, or about 100 families for linguisti-cally unif ied groups among hunter-gatherers.Progressively larger sizes may be necessary associety and technology become more complex.

If hypotheses of the kind formulated here canbe substantiated, they can be quite valuable inassessing the merits in different social situ-

ations of different agricultural technologies. Thelimited generalizations made here on the op-t imum size and the generally appropriate func-tions of small groups do seem to be wellsupported by both anthropological andsociological research. These generalizationsare applied below to the solution of particularproblems. The generalization that impersonalrules are best established and enforced in thecontexts of groups much larger than the op-t imum task group size of about seven personsalso seems to be well-founded and is appliedbelow. More information is needed on bothsmall and large groups and their potential rolesin promoting agricultural development. Re-search in this area should be carried out focus-ing particularly on the possible effects of differ-ent cultural, social, and economic contexts.

P r o s p e c t s f o r I m p r o v e d L a n dM a n a g e m e n t i n t h e S e m i - A r i dT r o p i c s

The bullock-drawn wheeled tool carrier alreadymentioned provides our first example. Theimplements—especial ly plows, cultivatingblades, and simple seeders — most commonlyused by smallholder farmers of the centralIndian semi-arid tropics today have advantagesof being cheap as well as being easily manufac-tured and repaired in villages by minimallyskilled workmen. At the same t ime, these im-plements are difficult to use for deep plowing;for relatively sophisticated land-managementtechniques a imed at reducing eros ion,maximizing rainfall infiltration, and eventuallyimproving soil quality; and for precise handlingof high-potential seeds wi th greater-than-min imum doses of fertilizer. For these reasonsan improved machinery system has beensought by scientists working on these prob-lems; one choice has been a bullock-drawnimplement known by its tradename of Tropicul-tor.

Experiments using this basic mult ipurposeunit and its attached implements have beencarried out at ICRISAT with a broadbed-and-furrow system to improve water storage in the soilprofile and to reduce erosion and waterlogging.The furrows are placed at a gradient (0.4% to0.5% slope, or 40 to 50 cm drop per hundredmeters distance) sufficiently steep to assure

109

good drainage whi le leading the water off theland slowly. This allows for good infiltration ofrainwater and minimizes erosion. The widthfrom furrow to furrow is 150 cm with the furrowbeing several centimeters deep. The furrowscan also be used to apply water if supplemen-tary irrigation is planned wi th runoff watercollected f rom the catchment. Grassed water-ways are required to dispose off surplus waterfrom the graded furrows. Reorganized fieldboundaries may be desirable to make it easierto site the grassed waterways in the mostappropriate location. This is also likely to im-prove the field layout and make it more efficientfor tool carrier operations. The raised beds are a semipermanent structure and remain in placefrom season to season. The beds themselvesrequire a min imum of t i l lage; this providesfurther protection against erosion. The im-proved implement used provides additionalbenefits of faster plowing, which helps to con-serve the fleeting soil moisture. It providesmore exact seed and fertilizer placement, im-portant when high-potential seeds and more-than-minimum fertilizer doses are used. Whereindiv idual ly owned plots are relat ivelylarge — on the order of 10 to 20 ha — or wheregroup action among farmers is possible, grad-ing and readjustment of field borders and theinstallation of grassed waterways can be car-ried out on a unified watershed basis (Krantz etal. 1978). There are indications from economicanalysis, however, that profits to the improvedsystem of ti l lage within existing field bound-aries are comparable to those on comprehen-sively developed small catchments (Ryan et al.1979); so the job of improving land manage-ment may be easier than originally thought.

It has been calculated roughly that at 1979prices a wheeled tool carrier of this type — wi tha min imum set of implements of medium price,and manufactured in India — might cost ap-proximately Rs 7500 (about U.S. $930) or more.Analysis of experiments on the ICRISAT re-search station indicates that it is possible toreach a high degree of profitability in SAT India,given such costs (Ryan et al. 1979; Binswangeret al. 1979); however, further research, in on-farm situations, is still needed.2 Much dependsof course on the availability of fertilizer andcare-responsive, high-yielding cultivars of highvalue rainfed crops wi th a good local market.

In considering how to put the wheeled tool

carrier into wide use by all classes of farmers inthe countryside, one immediately encountersorganizational questions. The report byBinswanger et al. assumes that the implementwill be used for agricultural purposes on approxi-mately 15 ha of land, an area that presentlyrepresents the l imit of the machine's coverage.(Some reduction in range might be experiencedon some soils and in some climatic situations.)This specification concerning range, as well asthe cost projections already presented, seem toargue against the wide use of the implement bysmall farmers in India. Many of these small-holders use almost no purchased inputs ontheir rainfed holdings, whatever may be thecase on their irrigated plots (Bapna et al. 1979).Their holdings in India are much below the15-ha range noted above. Their overall financialsituation in one representative area (Table 1) isnot good. Their existing agricultural imple-ments are of the simplest and cheapest kind,while the greatest part of their existing assets isrepresented by their land.

Although their financial situation makes itdifficult to envisage how such an expensiveimplement could be introduced on a wide scaleamong such farmers, there are several reasonswhy an implement such as the Tropicultor is tobe preferred to some simpler implement. As a s imp le , m i n i m u m - t i l l a g e a r rangement ,broadbed-and-furrow systems or other ridgingsystems can be formed by so-called ridgingplows; this system is common in parts of WestAfrica. However, in the semi-arid tropics aselsewhere cropping patterns vary greatly f romseason to season and year to year, and the bedsformed by ridging plows are too narrow to beflexible in planting a number of crops in a variety of patterns f rom year to year wi th thewheeled tool carrier. Furthermore, plantingquickly at an exact depth whi le still arrangingfor exact seed and fertilizer placement can bedifficult wi thout a tool carrier of some kind, and

2. The wheeled too l carrier can also be used as thebase for a bullock cart, wh ich wou ld seem to havesome advantage over many tradit ional carts sinceIt wou ld have pneumatic tyres and an increasedload carrying capacity. Sufficient levels of nonag-ricultural use wou ld reduce the hectare coveragenecessary to make the wheeled tool carrier pay.More research on the returns to all these varia-t ions, part icularly in on-farm situations, is needed.

110

Table 1 . Ave rage f inancia l cond i t ion o f smal l fa rmers* in s e lected areas o f semi-ar id t rop ica lpeninsular India, 1 9 7 6 - 7 7 crop season.

Village

Aurepal le(Andhra Pradesh)(10 farmers)

Shirapur(Maharashtra)(9 farmers)

Kanzara(Maharashtra)(9 farmers)

Land Value ofoperated owned land

(ha) (Rs)b

1.4 2593

1.4 8236

1.3 4256

Value of year's gross Value of ownedagricultural output implements

(Rs) (Rs)

517 165

1341 416"

696 83

No. of pairsof bullocks

owned c

0.3

0.1

0.1

a. Data col lected in t h e Vi l lage-Level Stud ies of ICRISAT's Economics P rog ram. Sma l l fa rmers are de f ined f o r each v i l lage as t h eten farmers hav ing the smal lest operat ional landho ld ings out of a random sample of th i r ty fa rmers In each v i l lage. Furtherdetai ls are in Jodha et a l . (1977).

b. Rs 8.00 = US $1.00 in 1979.

c. The cost of a pair of wo rk ing bul locks ranges f r o m Rs 800 to Rs 4000 or more , at 1979 prices.

d. The relat ively m u c h h igher average value of o w n e d imp lements in Shi rapur is due to the fact that one fa rmer in the samp le forthat v i l lage owns an electr ical ly powered p u m p . The average value of o w n e d imp lements for the other 8 sma l l fa rmers inShi rapur is Rs 56.

agronomists see these as elements that are atleast as important as the tillage system alone.The strategy for intensive rainfed cropping insemi-arid tropical areas is to cover large areasof land quickly and exactly wi th opt imumamounts of seed and fertilizer during opt imumplanting periods of fleeting duration, so that thesoil wil l be protected by a crop and later rainfallpenetration wil l be maximized.

Under these circumstances how is one tointroduce on a wide scale an improved imple-ment that most farmers cannot afford, andwhose cost would take years to be amortized byan impoverished, individual smallholder? Theproblem is compounded even further by thefact that, for the long-term ecological health ofsuch semi-arid tropical farming regions, uni-form coverage is needed on a watershed basiswith this kind of improved implementationrather than piecemeal coverage of individualfarms. Ryan et al. (1979) reported that in somecases the improved tillage system gave similarreturns both wi thin preexisting field boundariesand within reorganized field boundaries. Stil l, inorder to minimize dangers f rom erosion andlocalized flooding for any single farmer it maybe better for all farmers to sow in the improved

way. Such uniform coverage would be desira-ble both from a community standpoint andfrom a long-term individual standpoint as itwould conserve soil, improve soil conditions onthe watershed as a whole, and improve localwater tables. Nevertheless, whi le watershedschosen for development feasibility studies inthree villages typical of the three main agro-climatic zones of semi-arid tropical peninsularIndia involve contiguous areas of only about 18,17, and 20 ha, respectively, they include the landof 5, 14, and 12 individuals (Table 2). Moreover,the farmers on these watersheds represent a range of social and economic classes and castesin their respective villages, and the prospectsfor cooperation among them are not increasedby these basic socioeconomic disparities.

Faced wi th this kind of situation, one mightargue that if the maximum amount of land is tobe covered by improved tillage it makes senseto start with large farmers. It could be arguedthat this would also have the fastest effect inraising overall yield levels. However, such a strategy is unacceptable from the viewpoint ofequity. If all farmers of alt sizes are to benefitf rom improved land management, then meansfor cooperation among them or for the supply of

111

Table 2 . D a t a a on se lec ted wa te rsheds in th ree cent ra l peninsular Ind i an locat ions ( 1 9 7 8 - 7 9 cropseason) .

Vil lage

Aurepal le Range(5 farmers) Mean(Watershed 18 ha)

Shirapur Range(14 farmers) Mean(Watershed 17 ha)

Kanzara Range(12 farmers) Mean(Watershed 20 ha)

Area ofwatershed

plot(ha)

0.5-8.03.5

0.5-3.91.4

0.4-5 .5 c

1.4

Total areaoperated

per farmer(ha)b

1.6-9.7C

5.7

0-16.97.1

0 -38 .4 c

8.5

Watershedplot as % oftotal oper-ated area

2 9 - 5 2 c

40

0 -10025

0-100c

33

No. of oiland/orelectricpumpsowned

0 - 1 c

0.8

0 - 10.3

0 - 1 c

0.2

Pairs ofdraft

bullocksowned

1-4C

2.3

0 - 41.4

0 - 1 0 C

2.2

No. offamily

members

4 - 6 c

4.8

4 - 1 07

1-14 c

7.6

No. offamily

workers

2 - 4 c

3

1-42.5

1-8C

3.5

a. The data g i ven here are based on verba l in te rv iews ; in s o m e cases a lso they are incomple te . A l t h o u g h they can be taken as

indicat ive o f the genera l s i tua t ion fo r the fa rmers to w h o m they refer, they cer ta in ly w o u l d be rev ised by m o r e t h o r o u g h

invest igat ions. Such fu l l invest igat ions o f resources and o f f a r m i n g sys tems f o l l o w e d by par t ic ipat ing wate rshed fa rmers are

p lanned. A l so no te tha t due b o t h to organizat iona l and resource constra ints the actual land and water deve lopments f inal ly

carr ied out in 1 9 7 8 - 7 9 deal t on l y w i t h por t ions of these watersheds. The exper ience of th is and subsequent seasons w i l l be

repor ted in subsequent pub l ica t ions .

b. W h e r e to ta l opera ted area is zero a l l l and o w n e d by a fa rmer is leased out inc lud ing plots on the selected wate rshed .

c. Data not avai lable fo r one or m o r e fa rmers .

low-cost improved implements to individualsmall farmers would seem to be necessary.Having identified the innovation and its pur-pose, one is faced in the case of the wheeledtool carrier wi th a problem of human organi-zation to ensure its best use for human well-being.

G r o u p S i z e S t u d i e s t o S o l v eO r g a n i z a t i o n a l P r o b l e m si n I m p r o v e d L a n d M a n a g e m e n t

For improved land management using thewheeled tool carrier, a number of means tointroduce the implement system are possible.They include individual ownership of one ormore implements and the addition of rental ofwheeled tool carriers to the already existingservices of multipurpose farmers' cooperatives.We need to know, however, how such alterna-tives rate in terms of human organizationalefficiency. We need to take account of humannature in the same way we measure agronomicand economic suitability.

112

If our reading of the implications of researchon group size is correct, the alternative ofcooperative ownership for the machine can berejected. This is because the economics of scaledo not seem to indicate that a group largeenough to be stable could be formed around themachine.3 Opt imum planting days are l imited inthe semi-arid tropical environment and, as al-ready noted, the area that the implement cancover efficiently is likely to be 15 ha or less,depending upon variables such as field layout,climate and soil condit ion, bullock power avail-able, and the task involved. Given the data inTables 1 and 2 we would expect to need roughlyfrom 5 to 15 small farmers to reach a combinedholding size equal to the range of the imple-ment. It seems to be too much to expect groupsof this size to take on the complicated long-termownership and managerial duties required by

3. The returns f r om Tropicul tor use calculated InBinswanger et al. (1979) do not seem to be largeenough to hold together a small group w i thoutconsiderable administrat ive help.

common possession and use of a farm imple-ment. Large farmers and those with uppermiddle size holdings seem to present no prob-lems, as individual ownership could be approp-riate for them economically. Cooperation on thepart of groups of small farmers does seemunlikely, judging from the analysis of group sizeand function presented above.

Although anthropologically a large groupwould seem to be more likely to be able toenforce stability, the economics of the situationseems to preclude dependence on such a group. The buying power of a hundred or morefarmers could purchase tractors as easily as itcould purchase wheeled tool carriers. Whateverthe social benefits or costs outside the coopera-tive group (Binswanger 1978), such an organi-zation would be likely to f ind tractors moreremunerative than bullock-drawn tool carriers.Thus the alternative of wheeled tool carrierownership by a multipurpose cooperative isrejected according to economic reasoning.

Having applied criteria of group size andeconomy, we are left wi th the alternatives ofindividual ownership by farmers or by entrep-reneurs. Either arrangement is anthropologi-cally acceptable according to our reasoninghere. The choice depends upon the scope forreducing the cost of the implement or forincreasing the area that one machine can coverduring the opt imum planting season. Extremelow cost would be the key to making theimplement available to individual small far-mers.4 On the other hand, assured coverage of a relatively large area — or perhaps enhancedperformance of nonagricultural work — mightbe necessary to make the implement pay, evenfor a very small-scale entrepreneur. Plans todevelop either option would be likely to benefitf rom a survey that examined existing patternsof investment and use of implements, particu-larly expensive items such as pumps, which,like the improved tool carrier, are relativelylarge capital investments.

4. Work is presently in progress at ICRISAT on such a low-cost model , based on a small wooden bullockcart; the cart itself cost Rs 550 (about US $70) In a rural market t o w n in central peninsular India in1978. The cost of the added toolbar and attach-ments is as yet unknown but could not be less than

Rs 600 ($75) w i t h 1979 prices.

P o t e n t i a l f o r D e v e l o p m e n to f I r r i g a t i o n B a s e do n R u n o f f W a t e r

The question of rainfall runoff conservation isparticularly complex (see, for example, Kampenet al. 1974; Krantz et al. 1978; Ryan et al. 1979).Climatology, soils, cropping patterns, and otherphysical and biological variables, along wi theconomics, all play a part in determining theminimum size and the proper height-to-depthratio for a tank that wil l assure efficient andprofitable storage without too much loss fromevaporation and seepage. Analysis of some ofthese criteria are in the three reports just notedand in von Oppen (1974, 1978). In a full analysisof the problem, we must also ask whether groupsize and function have human organizationalconsequences that should be considered indetermining the sort of farmer for whom sup-plemental irrigation from tanks might be ap-propriate and what sort of farmer might benefitparticularly f rom the results of different re-search strategies.

Runoff collection reservoirs, or tanks, built onexperimental watersheds at ICRISAT Centernear Hyderabad are relatively small and aredesigned to trap and hold water for the protec-tive and supplementary irrigation of drylandcrops, by gravity or pumping (Kampen et al.1974). Similar traditionally used tanks in theHyderabad area are often operated wi th thewater distributed in a more or less unregulatedflow, and almost always for rice rather than forcrops that need less water. The traditionallyoverwhelming importance of rice in this contextdeserves special notice. There is a great deal tobe said for rice as a crop, particularly because ofits agronomic qualities under prevailing condi-tions in the rainfed area around Hyderabad.Masefield (1977, pp 21-22) points out that:

There are rice fields in Asia which haveprobably been continuously under the cropfor centuries without any conscious inputof plant nutrients by the cultivators, butwhich can still be relied upon, provided thatwater is available, to produce a steady halftonne of paddy per acre (1250 kg/ha). Noother cereal can emulate this feat.

Nevertheless, Masefield (1977, p 22) adds that"a good rotational program" wi th the right

113

inputs might actually be a more efficient user ofresources. Thus development of new technology,including new varieties of high-value drylandcrops, and the development of institutions tosupport supplementary irrigation instead ofintensive irrigation do seem to hold promise forincreased food production over wide areas. Thedevelopment of supplementary irrigation mightalso lead, in the right institutional context, to theconcentration of crop production effort on areasof the greatest potential for intensive drylandcropping, thus allowing for reduction of pres-sure on marginal areas of shallow soil.

Recent analysis of runoff potential for smallcatchments in two semi-arid tropical regions ofcentral India indicates that on Alfisols (red soils)of medium depth, even under improved til lageand cropping patterns, catchment areas of bet-ween 8 and 16 ha could generate enough runoff tomake supplementary irrigation of drylandgrains and legumes economically feasible for atleast part of the catchment when used to sup-port a postrainy-season second crop (Ryan et al.1979; Ryan and Pereira 1978). Such irrigationwould also be planned to protect the catchmentfarmers against occasional droughts during therainy season; however, the economics of thisportion of the technology remains to be asses-sed. The crops envisioned include, but are notlimited to, dryland cereals such as sorghum(Sorghum bicolor) and pearl millet (Pen-nisetum americanum) and dryland legumessuch as groundnuts (Arachis hypogaea), chick-pea (Cicer arietinum) and pigeonpea (Cajanuscajan).

A p p l i c a t i o n o f G r o u p S i z eS t u d i e s t o P r o b l e m s o f R u n o f fW a t e r H a r v e s t i n g a n d U s e

In designing the experimental tanks for sup-plemental irrigation of crops consuming smallamounts of water, it was originally thought thatseveral farmers could cooperate to use thiswater. In considering such a possibility, how-ever, one immediately runs into problems ofgroup size. On Alfisol experimental watershedsat ICRISAT Center, analysis showed profits w i thsupplementary irrigation f rom small tanks onwatersheds 8 to 16 ha in catchment size.5 Table2 suggests that under central peninsular Indianconditions about 5 to 15 farmers would be

114

involved in a catchment of such a size. Sig-nificantly, such a group of farmers would befaced with very difficult management problemsif they were to cooperate on their own. Theanalysis of returns f rom supplementary irriga-t ion in the postrainy season indicates that evenin a profitable situation there would be enoughrunoff on average to provide water for onlyabout 44% of an 8-ha catchment (Ryan et al.1979, pp 30-31). The potential managementdifficulties for a cooperating group are mademore difficult by the fact that runoff variesaccording to rainfall f rom year to year so thatthe members of a group could not depend upona constant resource. The potential cooperatorswould also f ind it hard to achieve individualequity for an irrigation resource to be used bydifferent individuals on crops wi th differentwater requirements at different times.

Our analysis of the relative durability of largeand of small groups leads us to hypothesize thatsmall groups of f rom 5 to 15 farmers could notcooperate easily to maintain such irrigationtanks and to distribute the water f rom them.Such cooperation might be possible given out-side administrative monitoring by government;it might also be fostered by a very high profit-ability from supplementary irrigation. It wouldseem, however, that this would need to accrueat the same rate to each farmer, and to berealizable in no other way (Doherty and Jodha1977). Outside administrative control of thedegree needed to counteract fissiparous ten-dencies in such small groups around such a variable resource would probably be quite ex-pensive to apply widely, given the large popula-t ion and intensive agricultural exploitation ofthe Indian countryside. One might also proposethat a large group, such as a village, unite tocontrol the administration of many small tanks.

5. Economic returns f rom supplementary irr igat ionwi th collected runoff water are much less certainon Vertisols (black cotton soils) than on Alf isols.This is because the Vertisols retain moisture in thesoil prof i le rather wel l , so that particularly in thedeeper Vertisols moisture is available in the croproot zone even when the surface is dry. There maybe potential for supplementary irr igat ion on Ver-t isols; wel ls prove quite prof i table in some of theseareas, for example, and there are often signif icantamounts of runoff. However, these subjects needfurther study.

Since the resource in question under such a scheme — the tanks — would be divided, thisplan as well would seem to invite difficultiesand appropriation of the tanks by individuals(see Doherty and Jodha 1977; Olson 1971).6

If unaided cooperation among small groupsof farmers is not likely to be successful for thedistribution of supplemental irrigation water,alternative strategies to improve the use ofrunoff water are needed. Either tank size orirrigation technology (pumps, sprinklers, dripsystems), or both, might be investigated. Thesame subjects could be studied to see howlarger, potentially more stable groups could beaccommodated for purposes of supplementaryirrigation. Indian agricultural and governmentalgroups are now in fact experimenting wi th ID, or"irr igated dry," cropping management of tankcommand areas. Such groups might not have tobe large enough for self-regulation if enoughfarmers or enough land area were included toreduce the costs of government administrativeaid; this seems to be the strategy wi th the Indianexperiments. One would still be advised to beon guard against sources of factionalism. Ifirrigated dry cropping strategies seem appro-priate, much could be learned about how tosupport them by study of successful and unsuc-cessful management strategies followed bythose administering traditional rice-irrigationtanks. One cannot neglect the possibility alsothat under tank irrigation, rice-growing was or isthe best strategy, given limits placed by suchfactors as rainfall, geology and soils, evapora-t ion, and potential for response to irrigation bydifferent crop species. Even if they are to con-tinue primarily to support rice, traditional tanks

6. The possibil i ty of joint ownership and operation byan extended fami ly or by a lineage group alsopresents itself. In this case the authori ty of thefami ly substitutes for what wou ld otherwise benecessary: the authori ty of the society as a whole.Such cases do exist in India for the commonownership of wel ls, and tenurial rights seem to beadjustable by such means as one party buying outanother, when in its normal cycle of developmentthe extended fami ly becomes too attenuated. Sucha possibi l i ty might be wor th investigating w i threspect to tanks. This possibil i ty too , however,w o u l d face the problems of equity among usersand of prof i tabi l i ty relative to other possiblesources of water, particularly wel ls.

should be studied to see what managementtechniques have been successful and how theycould be improved, so that they may make theiropt imum potential contribution in the contextof semi-arid tropical farming systems. The ques-tion of wells and of sale of water by well ownersalso seems to be important. And we should alsonote that tanks often raise or make possible theyield of economic amounts of water f rom wellsdownstream; percolation tanks have been builtin India, specifically for such effects, in someareas not especially suited for tank-based gravi-ty f low irrigation.

F o r m u l a t i n g a n d F i e l d T e s t i n gB e h a v i o r a l H y p o t h e s e s

Neither technology nor an organizationaljudgment can be considered proven until it hasbeen tested and found acceptable under naturalcondi t ions—that is, in farmers' fields and insocial organizational contexts typical of theagricultural areas for which it was made. Forthis reason we require field tests of specifichypotheses derived from analyses such asthose above.

Field experiments designed to test an-thropological hypotheses cannot be carried outwith as many replications, and with the samekind of control over variables, as can biologicalexperiments on research stations. However,field tests and experimental control of variablesare possible. As in any test, it is useful toproceed on the basis of specific hypotheses. Wemay hypothesize first what would happen if itwere shown that supplementary irrigation ofsome dryland crops is profitable in particularagronomic contexts. Our hypothesis is that farmers would prefer that small-sized sources for supplementary irrigation be privately owned. If the hypothesis is correct and if theoptimum size of catchment for a small tank usedfor postrainy-season irrigation is 8 to 16 ha(Ryan et al. 1979), we can expect that only thoserelatively few farmers who happen to own a large piece of land of the appropriate sort wouldshow interest in small tanks. The rest would beexpected to show more interest in wells; or insupplementary or intensive irrigation f rompublicly-supported tanks and from canal sys-tems; or in chances to purchase water f rom wellowners where the latter might have surplusavailable to sell.

115

With respect to the wheeled tool carrier, our second hypothesis is that both large and small farmers generally will prefer individual owner-ship of such small and medium-scale cultiva-tion implements. If such is the case, we shouldexpect cooperative groups to form about thesetool carriers only if enforced and supportedorganizationally f rom outside. Such groupsmight also be promoted by a combination ofexceptionally high returns and relatively highcapital costs for the implement. Overall how-ever, if the implements were to spread widely,we would expect that it would be as a result ofreducing capital costs to the individual farmeror of systematically placing it in the hands ofmany entrepreneurs, or of both these condi-t ions.

Testing such hypotheses requires workingdirectly wi th farmers, in contrasting social andcultural environments. The aim is to separatethe influences of human nature on the onehand — the supposed basis of the hypotheseswe have advanced above — f r o m theinfluences of particular social, cultural, andeconomic contexts on the other. For this reason,and in order to reduce the volume of sheerdescriptive work necessary as background, it isdesirable to choose societies — and wherepossible, particular communities — that are al-ready well known economically and ethno-graphically. Even where such choices are possi-ble, it must be recognized that social, cultural,and economic variables can only be held con-stant or varied in the most approximate way inthe real wor ld. In order to reduce the possibilityof error, therefore, it is necessary to supplementfield tests wi th comparative studies of a largenumber of related cases drawn f rom theethnographic and economic literature. Thuscomparison is important both in the initialformulation of hypotheses and in their testing.

An experiment to test the technology discus-sed in this paper is under way on representativewatersheds in three villages of peninsular Indiain which sample farm families participate instudies by ICRISAT scientists. Data regardingthe watershed farmers and their holdings on thewatersheds were detailed in Table 2. In thesethree villages a wide range of economic data(Binswanger and Jodha 1978) has been col-lected every 3 to 4 weeks since June 1975 on thesame 40 farm households in each vil lage, andthey thus provide a particularly favorable en-

116

vironment for an experiment of this kind. Thethree villages are also desirable for such anexperiment as they are in somewhat contrast-ing cultural and social areas within India and incontrasting natural agricultural environments.

The exact plans for development of eachwatershed were laid out in March and Aprill1979 by ICRISAT staff of the Farming Systemsand the Economics Programs, staff of memberinstitutions of the Indian Council of AgriculturalResearch, and the farmers in the villages. Thefinal agronomic and organizational shape of theexperiment has depended upon the advice of a committee of these scientists on the one handand and the decisions of the watershed farmerson the other. The decisions to be made concernsuch matters as whether and how to provide forsupplementary irr igation; how to organize theuse of a l imited number of wheeled tool carriersamong a relatively large number of farmers;which cropping patterns to concentrate on;whether to reorganize existing field boundariesin order to make land management easier andto improve drainage; and other similar ques-t ions. Basic items of technology to be tested arethe use of supplementary irrigation on a dry-land cropping system, and improved land man-agement wi th the use of a wheeled tool carrier.Supplementary irrigation is part of the experi-ment on at least one watershed (in Aurepalle inan Alfisol region); depending on furtheranalysis, it may or may not be provided onVertisols. It is planned that the first, financially-aided phase of the experiment wil l run for 2 years. At the end of 2 years, financial aid will bewithdrawn. Monitoring the actions of thefarmers — both on the study watersheds and inthe rest of the village — as they decide whatfurther use if any to make of the technologiesand systems presented wil l constitute the nextphase of the experiment.

It is apparent that during the 2 years offinancial aid to farmers participating in theexperiment, not even an approximately com-plete field test wi l l be possible for the hypoth-eses advanced above. Whereas we hypo-thesized that private ownership of small-scalesupplementary irrigation resources would bemost sought after, in the one village where suchirrigation is planned so far it wil l be madeavailable f rom a single source, and the agree-ments for its use among the several watershedfarmers wil l be supported by the efforts of

ICRISAT staff. Similarly, we hypothesized thatprivate ownership of the wheeled tool carrierwould be preferred; yet cooperative use is to beenforced and managerially aided for these 2 years. This period is, however, both experimentand demonstration as regards the farmers'participation. Although much can be learned ona social organizational level f rom their reactionsduring this t ime, the main purpose must be tofamiliarize a relatively large number of farmerswith the technology so that they will be able tomake valid judgments about it in relation totheir own situations. The real test, botheconomically and organizationally, wil l come atthe end of the 2 years, when technical advicewil l be provided but financial subsidy wil l bewi thdrawn. If the various components of thepackages have proved themselves and areadopted for their own use by the farmers, itshould be possible at that stage to learn a greatdeal about the social organizational arrange-ments that are locally preferred and feasible tosupport the use of the new technology. By thatt ime also comparative studies of similar casesshould be completed and wil l provide consider-able insight as the further progress of theexperiment is measured. In al l , such a testingprogram might require 4 to 5 years. As more islearned about how to carry out such tests, andas major areas of interest requiring such testsare identified, the scope wil l increase for suchmult ipurpose field studies and for field studiesparalleling laboratory and research station ex-periments.

Conclusion

In many areas of the semi-arid tropics, andparticularly in SAT India, it seems clear that theevolution of agricultural systems today must betowards more intensive use of natural agricul-tural resources and towards the simultaneousprotection and improvement of these. The taskis complex since new technologies that aredevised must take account of human potentialas wel l as of agronomic and economic poten-tial. Applications of the comparative method inanthropology, such as that made here wi threspect to human group size and function, seemto have potential for important practical resultsif combined wi th agronomic and economicinsights. What is needed now is an expansion

and diversification of such applied studies. Fieldtests of hypotheses formed on the basis of suchstudies are also necessary and should be car-ried out in a series of contrasting environmentsthat are well known socially, economically, andagronomically, in order to come as close aspossible to a situation in which variables arecontrolled. Such field tests should be accom-panied by special, more widely comparativestudies of particular institutions and techno-logical questions which prior analysis hassuggested wil l be important so that the resultsof field tests can be applied operationally assoon as possible.

In this paper insights gained from compara-tive analysis have been applied to problems inthe use and ownership of an improvedimplement—a bullock-drawn wheeled toolcarrier — for improved land management; andto problems in the collection and use of waterfor supplementary irrigation of dryland crops.Areas likely to be of particular importance indevising eventual solutions were identified.These areas include work to make the toolcarrier inexpensive enough for individual small-holders wi th very small financial means andresearch to make the implement attractive toindividual, small-scale entrepreneurs whowould rent it out or use it to do custom work. Asone step in the testing of anthropologicalhypotheses, a survey of existing investmentand use patterns wi th such relatively largecapital items as pumpsets was also suggested.Areas of importance for special investigationwith respect to the organization of supplemen-tary irrigation include the funding, manage-ment, and administration of existing tanks de-signed for gravity irrigation or for groundwaterrecharge; they also include study of the owner-ship and management of wells. Research on allthese items is under way or planned. Fieldexperiments are also under way to introducefarmers to the concepts and technologies in-volved and to encourage them to make theirown adaptations. These experiments are beingcarried out in villages where stratified randomsamples of farm households have been studiedwith regard to economics, social organization,and farming practices for 3 years prior to theinitiation of the field agronomic experimentsdescribed here and where the same samples ofhouseholds continue to be studied. It is hopedthat this combination of field research and

117

special topic investigation wil l prove efficient indevising ways to spread improved land andwater management more quickly and effec-tively in semi-arid tropical agricultural regions.

Acknowledgment

The author wishes to thank colleagues wi th inICRISAT, an anonymous reviewer, and Frank Cancianand Richard Swanson for helpful comments on var i -ous drafts of this paper. The author himself, of course,is ful ly responsible for any deficiencies remain ing.

References

BAPNA, S. L., JHA, D. and JODHA, N. S. 1979. Ag-

roeconomic features of semi-arid tropical India. Pre-sen ted a t t he In te rna t iona l W o r k s h o p onSocioeconomic Constraints to Development ofSemi-Ar id Tropical Agr icul ture, 19-23 Feb 1979,ICRISAT, Hyderabad, India.

BINSWANGER, H. P. 1978. The economics of tractors insouth Asia: an analytical review. Agricul tural De-ve lopment Council Occasional publ icat ion, NewYork, NY, USA. Also available from ICRISATEconomics Program, Patancheru, A.P., India.

BINSWANGER, H. P., and JODHA, N. S. 1978. Manual of

instructions for economic investigators in ICRISAT'svil lage-level studies, 2nd ed. ICRISAT EconomicsProgram, Patancheru, A.P., India.

BINSWANGER, H. P., GHODAKE, R. D., and THIERSTEIN, G.

E. 1979. Observations on the economics of tractors,bul locks, and wheeled tool-carriers in the SAT ofIndia. Presented at the International Workshop onSocio-economic Constraints to Development ofSemi-Ar id Tropical Agr icul ture, 19-23 Feb 1979,ICRISAT, Hyderabad, India.

BIRDSELL, JOSEPH. 1968. Some predict ions for the

Pleistocene based on equi l ib r ium systems amongrecent hunter-gatherers. In Man the hunter, eds. R.B. Lee and I. Devore Chicago: Ald ine.

BIRDSELL, JOSEPH. 1973. A basic demographic unit.Current Anthropology 14:337-56.

DEUTSCH, M O R T O N . 1968. Group Behavior. Pages 2 6 5 -276 in articles on Groups, International encyc-lopedia of the social sciences, Vol . 6, ed. David L.Sil ls. New York, USA: Macmi l lan.

DOHERTY, V. S., and JODHA, N. S. 1977. Condit ions for

group act ion among farmers. ICRISAT EconomicsProgram occasional paper 19, Patancheru, A.P.,India.

118

GOODENOUGH, W. H. 1965. Rethinking "s ta tus" and" r o l e " : toward a general model of the culturalorganization of social relationships. In The rele-vance of models for social anthropology, ed. M.Banton. London: Tavistock.

HARE, PAUL A. 1968. Role structure. Pages 283-288 inarticles on Groups, International encyclopedia of thesocial sciences, Vol. 6, ed. David L. Sills. NewYork: Macmi l lan.

HIEBERT PAUL, G. 1971. Konduru: structure and integ-ration in a south Indian village. Minneapolis: Univer-sity of Minnesota Press.

H O M A N S , GEORGE C. 1968. The study of groups. Pages259-265 in articles on Groups, International encyc-lopedia of the social sciences, Vol . 6, ed. David L.Sil ls. New York: Macmi l lan.

JODHA, N. S., ASOKAN, M., and R Y A N , J. G. 1977. Vil lage

study methodology and resource endowments ofthe selected vil lages in ICRISAT's vil lage-levelstudies. ICRISAT Economics Program occasionalpaper 16, Patancheru, A.P., India.

KAMPEN, J., and ASSOCIATES. 1974. Soil and water

conservation and management in farming systemsresearch for the semi-arid tropics. Pages 3-43 inProceedings, International Workshop on FarmingSys tems, 1 8 - 2 1 November 1974, ICRISAT,Hyderabad, India.

KRANTZ, B. A., KAMPEN, J., and V IRMANI , S. M. 1978.

Soil and water conservat ion and uti l ization forincreased food product ion in the semi-arid tropics.ICRISAT Farming systems Research Program,Patancheru, A.P., India.

MAHAR, P. M. 1959. A mul t ip le scaling technique forcaste ranking. Man in India 39:127-47.

MASEFIELD, G. B. 1977. Food resources and produc-t ion . In Human ecology in the tropics, 2nd ed., eds. J.P. Garlick and R. W. J. Keay. London: Taylor andFrancis.

MILLER, G. A. 1956. The magical number seven, plus orminus t w o : some l imits on our capacity for proces-sing in format ion. Psychological Review 63:81-97.(Quoted in Goodenough 1965.)

O L S O N , M A N C U R 1971. The logic of collective action.Cambridge, Mass: Harvard University Press.

RAVEN, BERTRAM H. 1968. Group performance. Pages288-293 in articles on Groups, International encyc-lopedia of the social sciences, Vol . 6, ed. David L.Sil ls. New York: Macmi l lan.

R Y A N , J. G., and PEREIRA, M. 1978. Derivat ion of

empir ical models for the predict ion of runoff onsmall agricultural watersheds in the semi-aridtropics. Presented at the International Workshop onthe Agrocl imatological Research Needs of theSemi-Ar id Tropics, 22-25 Nov., Hyderabad, India.

Available f rom ICRISAT Economics Program,Patancheru, A.P., India.

R Y A N , J. G., SARIN, R., and PEREIRA, M. 1979. Assess-

ment of prospective soi l , water and crop manage-ment technologies for the semi-arid tropics ofpeninsular India. Presented at the InternationalWorkshop on Socioeconomic Constraints to Semi-Ar id Tropical Agricul ture, 19-23 Feb 1979, ICRISAT,Hyderabad, India.

SAHLINS, MARSHALL. 1972. Stone-age economics.

Chicago: Ald ine.

SHERIF, MUZAFER, and SHERIF, CAROLYN W. 1968. Group

fo rmat ion. Pages 276-263 in articles on Groups,International encyclopedia of the social sciences,Vol. 6, ed. David L. Sills New York: Macmil lan.

VON OPPEN, M. 1974. Determination of the op t imumlocation and capacity of a storage tank on a typicalwatershed. ICRISAT Economics Program occasionalpaper 3, Patancheru, A. P., India.

VONOPPEN, M. 1978. Instability of area and product ionunder tank irr igat ion in selected districts of AndhraPradesh and Tamil Nadu. ICRISAT Economics Prog-ram discussion paper 9, Patancheru, A.P., India.

119

Farmer Par t ic ipa t ion and Accoun t i ngfor t h e Needs o f t h e M o s t

D isadvan taged Groups : S o m e Ideason Par t ic ipa t ion a t t h e Outset

of a Research P rog ram

J a c q u e s F a y e *

Abstract

This paper presents some ideas on farmer participation in agricultural development

programs, based on a research program conducted in the Experimental Units of

Sine-Saloum in Senegal during 1974-76. The author finds that any project concerning

large rural populations has to be based on an analysis of social structures and

stratification. He finds also that the most favored groups and categories develop

strategies whereby any changes introduced in the system invariably maintain or

strengthen their position in the hierarchy. He argues that the research or development

worker cannot be neutral or indifferent to the fact that innovations are always rejected or

distorted by a group or social category to the disadvantage of others. Often, it has to be

remembered that the poorest strata are poor not because they are backward, but

because they are involved in a series of unequal relationships. It is hoped that this

microlevel study can be extended to a national or international scale.

We take this opportunity offered by thisICRISAT workshop to present some ideas aboutfarmer participation in agricultural develop-ment programs. This text is not final and atpresent is based only on a research programconducted in the Experimental Units of Sine-Saloum' in Senegal f rom February 1974 to July1976.2 We expect to gradually extend this re-search to other programs within and outside theExperimental Unit project. This explains why

* Institut Senegalais de Recherches Agr ico le,Senegal.

NOTE: This is an edited translat ion of the or iginalFrench text , wh ich appears in Appendix 1.

1. The papers for th is workshop by Moussa Fall andM. Benoit-Cattin are also on the experimental unitproject.

2. Madicke Niang, geographer, and various scientistsf r om the Senegal Institute for Agr icul tural Re-search (ISRA) part icipated in this program. For anoveral l report of the research program see Fayeand Niang (1977).

the text is not properly structured as well as whyit is summary in nature. The examples are givento illustrate rather than demonstrate the re-search that has been conducted.

The research program on which this paper isbased is entitled Land-Tenure Systems Prog-ram. The objectives of the program are:

• to study traditional land-tenure laws, espe-cially the changes fol lowing the legislationon land-tenure systems passed in 1964 inSenegal but not implemented, especiallyin the Experimental Unit villages and thoseneighboring them.

• to define and experiment with a methodolo-gy for developing and consolidating villagelands through inexpensive techniques andmethods that can be used by farmers andrequire wide farmer participation.

• to propose practical ways of implementingland-tenure laws, taking into account theeffects of new technologies and of thetraditional land-tenure laws on the land-tenure system.

This involves research-action or research-development interaction. The study was notl imited to the land-tenure system of a given

120

farming community but, with the consent andparticipation of most of the villages in the Unit,we introduced radical changes in their ways ofusing land and soil.3

The research-action character of the Land-tenure Systems Program arises mainly f romtechnical considerations, and it can be summa-rized as fol lows:

During experiments wi th new techniques andcropping operations, agronomists identifiedthe division and distribution of cultivated plotsand the unreliable character of land-tenurerights of a large number of farmers, amongother constraints to the adoption of new crop-ping practices. The shape of these plots wasunsuitable for animal-draft cultivation.

During the development of the research pro-gram for studying and alleviating these con-straints, the essentially technical approach waswidened and also became a land-tenure systemapproach.

F a r m e r P a r t i c i p a t i o ni n t h e R e s e a r c h Program

The program involves the problem of farmerparticipation in two ways:

• explicitly: because one of the main objec-tives was a return to simple and inexpen-sive land development and consolidationtechniques that would enable participationby farmers and would possibly be imple-mented by them. Similarly, in the prop-osals for implementing land-tenure laws,the mode and structures for land man-agement had to be entrusted to the far-mers.

• implicitly: because such an important in-tervention cannot in any way leave thefarmers indifferent.

Their involvement may undoubtedly rangefrom open hostility with actions or violentreactions to collaboration. Noninvolvementand indifference on the part of the farmers, forwhatever reason, was not permitted.

The etymological meaning of the word " toparticipate" is "take part in something" and "Le

3. In three vil lages land was developed and consoli-dated: one in the first year and two in the secondyear; a total of 3000 ha involving 250 "masters ofl and . "

Robert" dictionary defines the word "participa-t ion" as "adherence, collaboration, contribu-t ion . "

We can also hold that the attitude of thefarmer who remains indifferent to technicalproposals made to him is not neutral, and thatthis attitude is in itself another way of participat-ing.

In the first stage at least we prefer to adopt theabove definition and we should try to examine,in the light of the Land-tenure Systems Pro-gram, what the other conditions are that enablefarmer participation in an appropriate researchprogram.

Conditions for Participation

Taking into Account the Needsand Priorities of Farmers

Today this condition is unanimously accepted.Projects and research programs developed re-cently go even further, as they aim at "meetingthe needs of the poorest farmers."

At the outset of the program, a negativeattitude could be expected from farmers whoseopposition to the land-tenure laws was general.The law on national territory in force for 10years (1964-1974) was only applied occasion-ally.4

The rural people have retained the fol lowingideas regarding the legislative laws:

The State holds the lands and recognizesthe right of cultivation of those who actuallycultivate the land; the lands that theirformer masters could not cultivate may betaken and given to other cultivators. Thus,the State not only deprived landowners oftheir land but also forbade land loans,transferred former loans to the be-neficiaries, who were for the most part"foreigners" brought in and settled by thelandowners; finally, the State wanted toredistribute their land, preventing themfrom transferring it to their descendants.

However, these aspects of the law only con-cerned a few farmers of this region, only a few

4. The farmers a p p l i e d — a n d cont inue to apply — the tradit ional law, only in the case of serious

conflict or when the confl ict ing parties decide toappeal to the administrat ive authorit ies is theexist ing law used to settle the conflict.

121

big maltres de terres (masters of land), andsome groups of peulh livestock raisers.5 On theother hand, this law, which also envisagedconsolidation of lands, carried positive aspectsfor most of the people.

The first phase of research work was topresent the aims of our program to the farmersto enlist their support. During our meetings wi thfarmers, organized in every quarter, we foundthat their attitudes were mainly hostile to theprogram and ranged f rom blind distrust to openhostility. One of the meetings even ended ininsults. Farmers seemed to be unanimous intheir rejection of the law and therefore refusedany research aiming at its implementation. Onthe basis of these reactions, we tried to analyzeand understand the motivation and concerns ofthe people.

Fortunately for this analysis, we were able touse the studies conducted during the past 2 years by an anthropologist, L. B. Venema, onsocial structures and stratification of the ruralpopulation. We also benefited f rom the detailedinformation that the extension workers hadobtained in the quarters where they worked.Collective and individual discussions were sys-tematically carried out to identify farmers' at-titudes and to relate them to the categories andsocial groups in Venema's study.

Rather than immediately formulate and pre-sent to farmers proposals for developing andconsolidating their lands, we put forward thefol lowing contractual agreement:

• The research would be divided into severalphases, each being preceded by collectivemeetings organized in every quarter. Theresearch team would present the workdone by them and this would be discussedsubsequently. The action planned in thenext phase would be presented and dis-cussed. After each phase the team wouldundertake to continue wi th the programonly if the farmers approved of the workdone in the earlier phase and gave theirconsent for the next one.

• In the first phase, in addition to studies onthe land-tenure laws, land surveys wouldalso be carried out. The farmers had to

S. The paulhs are a race whose mainlivestock raising. This l aw under

occupation iswh ich lands

wou ld also be consol idated had posit ive aspectsfor most of the farmers.

guide us over the area and settle litigationsaccording to the land-tenure law or thetraditional law.

During the first phase, we progressively tookinto account the concerns and interests mani-fested by the different groups and developedproposals for development and consolidationof the land. The proposals were presented to thefarmers individually and in small groups tostudy their reactions. Once the distrust towardsthe research team disappeared, we were suc-cessful in relying on a number of influentialfarmers who were leaders of their groups forrelaying proposals or for making counter pro-posals to us.

After establishing the land survey, we wereable to present and obtain the consent of mostfarmers f rom all quarters, except the peulh quarter, for a set of coherent proposals fordevelopment and consolidation of their lands.The farmers undertook to participate in actuallyapplying these proposals.

The tentative conclusion of this part of theLand-tenure Systems Program is that in a pro-ject concerning the entire rural population, it isindispensable to start w i th an analysis of thestructures and social stratification of the con-cerned rural population. The needs andpriorities that are identified should be related tothe social categories and groups that make upthe rural population. It is only on the basis of thisanalysis that a program can be built and objec-tives determined that wil l aim to meet the needsof the poorest categories.

Once this analysis has been carried out andwe have identified the poorest categories thatgenerally make up the majority of the farmingcommunity in the Third World (these categoriesbeing viewed in their relation to the other socialcategories with which they have dependenceand domination relationships), we can build a program and determine the objectives for meet-ing their needs. This condition appears to be a necessary one if farmers are to participate in theproject and not be passive towards it. Participa-t ion is not an abstract notion, it is the adherence,the collaboration of groups whose needs theproject aims to satisfy.

H e l p i n g t h e P o o r e s t F a r m e r st o M e e t T h e i r R e q u i r e m e n t s

In a community, any modification of the land-

122

tenure system concerns all the people: thosewho do not cultivate any longer because theywish to transfer their land to their descendants;those who cultivate; those who are not cultivat-ing as yet because they are the future users ofthis land.

To the extent that the program aims to assistthe most disadvantaged categories, it modifiesthe relationship between the different groupsand the social categories. It should be expectedthat the most favored groups and categorieswil l develop strategies whereby these changesmaintain or strengthen their position in thefamily or social hierarchy or, alternatively re-fuse these changes.

Through relationships of authority, clientele,domination, and dependence, certain people arebetter placed to defend their interests or use theproject to their benefit. These are men in rela-t ion to women; adults in relation to the youngerpeople; rich farmers (farmer-traders, farmer-transporters) in relation to poor farmers;families long settled in the area in relation tonewly settled families.

A few examples illustrate this phenomenon:• Information on conflicts within certain car-

res (compounds) proves that heads oflandholdings or of carres have tried withthe consolidation of land to oust thewomen or young people to poor-qualityplots or to reduce the size of their plots;others have tr ied to take the fields man-aged and cultivated by them, but belong-ing to nephews or younger people who arestill their dependents. On the other hand,certain young people have tried to takeadvantage by forcing the adults who man-aged their lands to divide the family landsinto equal parts, which is against the tradi-tional law that gives certain privileges toadults.

• During the official land surveys the maltres de terres have tried to recover the landsthat they had loaned to families who hadcome and settled in the village sometimesmore than 5 years ago, and which overt ime would certainly become gifts of land.

• The people living in quarters neighboringthose of the peulhs, although opposed toany idea of land reallocation affectingthem, constantly asked us to reallocatethe fields that the peulhs possessed but didnot cultivate.

• In both the Experimental Unit villages wecame up against the category of rich far-mers.6 Without being openly opposed toproposals for land development and con-solidation, they systematically refused toexchange fields, arguing that their landswere already of a good size or more fertile.On at least three occasions they involvedclientele, family, and Iineage solidarities toraise the farmers against us, and obligedus to stop operations of dividing fields andto initiate discussions with the commun-ity.7

Although the research or developmentworker does not put himself in the place ofgroups or social categories whose require-ments the project aims to satisfy, it is notpossible to be indifferent to this aspect and toconfine the program to the proposal of innova-tions. These innovations can always be rejectedor distorted by one group or social category tothe disadvantage of others. Therefore, one ofthe tasks is to help the groups whose livingconditions the project aims to improve by put-ting them in a situation where they developtheir own strategies to modify their socialstatus.

We approached this problem in the fol lowingways:

• first, by participating in the life of thevillage — long stays, participation in theirceremonies, visits to carr6s, discussions inthe village, meals with farmers, and vari-ous small services. Thus we deliberatelycreated a situation for a dialogue with thefarmers so that they could express theiropinions spontaneously.

• by considering the problems raised by thegroups and their proposals in order toclarify them and to relay them to othergroups for their opinions.

6. Farmers call them borom barke (literally " thosewho have the baraka"). Generally, they possesslarge areas of land; their oxen manure the land,which is sown to cereals and loaned at usuriousrates to poor farmers between cropping seasons.Their activities are usually nonagricultural — mostly trade and transport — and control of farm-ing cooperatives.

7. However, one of them succeeded in postponingallocations in the Thysse Kayemore vi l lage for 1 year.

123

• by organizing assemblies in the quartersthrough informal discussions wi th leadersand different members of the groups. Ex-perience has shown us that without thesepreliminary discussions the groups andcategories situated at the bottom of thesocial hierarchy are rarely able to expresstheir interests because of the order ofspeakers, relationship of authority, orderof b i r th, age, sex, etc.

• by assisting the farmers in clarifying andtranslating their requirements into pro-posals for action in this sensitization work,the research team had to be careful not somuch to remain neutral but to avoid beingthe tool of any group. This is all the moredifficult as the poor farmers have a ten-dency to adopt a passive attitude and to letthe project team oppose the other farmersthe moment a proposal conflicts wi th theirinterests. At first it may be surprising to seefarmers urge us to appeal to the admin-istrative authorities or to compel one oftheir people to accept a land exchange ormodification of the boundaries of hisfields, whereas they themselves refuse topublicly take any position or exert pressureon him. In one case, for example, where a rich head of a carr6 refused to modify theboundaries of his fields under the pretextthat it was harmful to his interests, itrequired 2 hours of heated discussionswith the villagers to make them agree toappoint a committee to verify the merits ofhis arguments. Without the intervention ofthe young people present, we would nothave been successful. In fact, in this case,he was one of the usurers of the village.

Conclusion

Our experience shows that t w o conditions arerequired for farmer participation:

• The first is to know whom the project aimsto help — which categories of people (themen and/or women and/or the youngpeople), which farmers (the poor, and/ormiddle-class, and/or rich farmers), in orderto determine if the objectives of the pro-gram correspond to the needs of the peopleconcerned. For this purpose, as we havementioned earlier, it is necessary to start

124

from an analysis of the structure of therural population in order to identify thedifferent groups and the relationships bet-ween them.

• The second is that it is not sufficient thatthe program be based on such an analysisif the decision is to help the poorest strataof the rural population. It must be kept inmind that these people are not poor be-cause they are technically backward, butbecause they are also involved in a seriesof unequal relationships that prevent themfrom adopting the proposed innovationsor, when they do adopt them, to benefitf rom these innovations.

If research-action aims to obtain the adher-ence and collaboration of the most disadvan-taged strata of society, it should help them towork on the unequal relationships in order tochange them. Studies of these relationships atthe village-level in developing countries arerare and should be increased. These should notbe restricted to internal village structural rela-tionships but relationship also between villagesthemselves and the national and internationalsystems of which they are a part.

Reference

FAYE, J . , and N I A N G , M. 1977. Une experience de

restructurat ion agraire et d 'amenagement de l'es-pace rurale senegalaise. Environnement Afr icain2(4) and 3(1).

Soc ioeconomic Field Assessment of Prospect iveTechnolog ies in t he S A T of Ma l i : A Case S tudy

in an O A C V Zone

B. Traore*

Abstract

This paper presents a model of the peasant household economy, taking the area under

an O A C V + in Mafias a case and condensing its basic features into a linear programming

algorithm. This model is then used to simulate the impact on small holders of various

government rural development strategies. Based on the findings of the study, the

following policy implications are derived: (1) although the intermediate modern

technology being introduced is considered to be cheap, it is still too costly for the

subsistence farmer and an even cheaper technology is needed; (2) increased farm credit

and output price incentives would be the most effective tools for adoption of new

technology by farmers in the area; (3) extension institutions which play an important

role should be well organized and made truly operational; (4) all possible effects must

be considered and the measures recommended must be within an all-embracing,

cohesive agricultural development plan. The paper also lists the limitations under which

its findings must be viewed and makes suggestions for further research.

Fully recognizing the Importance of rising ag-ricultural productivity in the growth and de-velopment of the economy (the rural sectorcontributed approximately 64% to Mali's1971-72 GDP), the Malian Government em-barked upon a vast scheme of agriculturaldevelopment in its 1974-78 plan. The centralthrust of this scheme was:

1. to develop a set of crops most adequate forthe ecology and the socioeconomiccharacteristics of each zone;

2. to develop the production set (livestock,fishery, handicrafts, etc.) that wil l as-sociate most effectively with agriculture;

3. to develop activities or aspects of sociallife other than only economic poten-

* Head, Planning Cell, Institute of Rural Economy(IER) of Mali and Research Scholar, Department ofAgricul tural Economics, University of Ibadan,

Nigeria.+ "Operat ion Arachide et Cultures Vivr ieres" (OACV)

is an extension scheme for the development ofgroundnut and cereals in a specific zone, spon-sored by the Government of Mal i , the Wor ld Bank

and the French FAC.

tialities, particularly the management skillof the farmers.

To foster these policies, the Government op-erated through extension institutions called OP-ERATIONS. These schemes, in collaborationwith research institutions, have developed pack-ages of new technology that are being intro-duced to farmers. The problem is, however, thatvery little is known about the socioeconomicorganization and decision parameters of thefarmers who are expected to adopt this newtechnological package. Hence, interventionsinto the peasants' farming systems are beingeffected with only rudimentary knowledgeabout the constraints faced by farmers and thelikely impact of these interventions on ruralhouseholds.

As Low's (1974) study in Ghana indicated,peasant farmers tend to behave in ways thatoptimize their objectives, given the constraintswithin which they operate, and it is only byunderstanding why and how they come to theirproduction decisions that it would be possibleto pursue policy objectives consistent wi th far-mers' aims. Thus, for effective planning andpolicy formulat ion, situational economic andtechnical farm data are required. We need

125

reasonable information on the physical re-sources, farm income levels, net revenue ofdifferent enterprises, and the potential increasein income and production that can result f romthe optimal utilization of the existing and poten-tial farm resources.

Following the compilation of such informa-t ion, what is necessary to answer questionsinvolved wi th interventions in the rural sector isa structural-form representation of the peasanthousehold economy. The basic features of sucha representation can be condensed into a linearprogramming algori thm, which can then beused to simulate the impacts of various gov-ernment rural development strategies onsmall-holders. This is the conceptual procedurefol lowed in this study using an OACV area inMali as an example.

O b j e c t i v e s o f t h e S t u d y

The objectives of this study were:1. To formulate a simulation model of

small-holding farmers in the area understudy, using a linear programming algo-r i thm.

2. To use this simulation model to test theimpact of government rural developmentpolicies on the OACV area.

The aim was to identify the production poten-tial and determine the expansion path consis-tent with increased producer prices, increasedresource levels, and improved technology.

T h e S t u d y A r e a

OACV covers an area of 138 000 km2 , encom-passing three administrative regions — Kayes,Bamako, and Segou (see Maps 1 and 2). Fifteenpercent of the total area is located in theGuinean climatic zone, where annual rainfall ismore than 1000 mm. The remainder is locatedin the Sudanian and Sudano-Sahelian climaticzones, where the rainfall ranges between 450and 1000 mm. The OACV zone is completelyincluded in the area defined as semi-arid t rop i -cal by ICRISAT. Both climatic zones noted aboveare characterized by a long dry season (7 to 8 months) and a short rainy season (4 to 5 months). The first useful rains occur at the endof May and early June. July, August, andSeptember are the wettest months.

126

Map 1. Republic of Mali. Scale 1/1.000.000

The total population of the area was esti-mated in 1976 at 1.1 mil l ion inhabitants, corres-ponding to a density of 8 inhabitants per km2

(for the entire country the density is 4 inhabit-ants per km2). Ninety percent of the activepopulation is employed in agriculture. Bam-baras and Malinkes are the most importantethnic groups in the area, and they are known tobe hard-working farmers, receptive to inno-vations (n'Daiye 1970).

The OACV is divided into ten operationalsectors. This study is based on data collected ineight villages of the subsector of Sirakorola, oneof the three subsectors of the Koulikoro sector(see Map 3). The most important crops grown inthe area are sorghum, groundnut, millet, suna,fonio, cowpea, and bambara nut. Some farmersuse ox-plows, but most of the farmers still useonly hand tools.

E m p i r i c a l A p p r o a c h

The analytical technique consisted of tables andcharts and a linear programming model. First,an "average" farm was computed for twocategories of farmers: the wealthier, relativelywel l-equipped farmers, and the poorer,nonequipped smallholder. Second, a prog-ramming analysis of the average farms was

carried out to determine the optimal croppingprogram, given the existing technical relation-ships, average yields, simulated expectedyields, prices, resource base, and other l imi-tations specified by the households' planningenvironment. Third, solutions were obtained:first, with varying levels of resources, keepingthe technological coefficients and output pricesconstant; second, with varying output priceskeeping the levels of resources and technologi-cal coefficients constant; and third, w i th im-proved technology, increased levels of re-sources and output prices. Finally, static norma-tive supply curves were derived from the pro-gramming results of varying output prices.

Programming Modelsfor the Study Area

Restrictions in the Models

Two kinds of restrictions were considered in themodels: resource restrictions and subjective

127

Map 2. OACV Zone in Mali.

State Boundaries

Region Boundaries

Circle Boundaries

State Capital

Main City of Region

Main City of Circle

Operation

Non-covered Zone

Extension 1974

Map 3. OACV Sector of Koulikoro.

Scale 1:1.500.000

800 mm IsohyeteSector Boundaries

River

Main Road

Railways

Secondary Road

Vil lageKoulikoro Main city of Sector

restraints. Land, labor and capital were the mostimportant resource restrictions whi le subjectiverestraints included consumption habits, familyliving expenditures, etc. Farm sizes, farm labor,labor devoted to nonfarm activities, non-durable capital available for production, wereall restricted to their average amounts observedduring the survey.

Activities in the Models

Five groups of activities were defined in thestudy:

1. farm production and off-farm actvities.2. cash holding and family living costs ac-

tivities.3. capital borrowing, labor hiring, and social

labor-using activities.4. food inventory and consumption activities.5. buying and selling activities.

Technology and Technical Coefficientsof Production

As stated, we considered two groups of far-mers: the nonequipped farmers whose prac-tices are more traditional (Group I), and equip-ped farmers who are more involved wi th mod-ern practices (Group II). The improved prac-tices, which are not yet generally adoptedexactly as advocated by research and extensioninstitutions, were simulated and included in thebasic models for investigation.

The improved practices being introduced in-clude better land preparation, early seeding,seeding density, improved seed varieties, useof fungicides and fertilizers, and ox-drawnplowing practices. Information on improvedtechnology was drawn f rom a report of theInstitute of Rural Economy (1973). Results f romexperimentation in research stations were cor-rected for normal farm conditions to informa-t ion concerning three crops grown under im-proved technology (IT): groundnut, sorghumand millet.

Technical coefficients of production were ob-tained by computing the average of the farmingunits in each group.

Analytical Model

In terms of our problem, the linear program-

128

S o u r c e s o f D a t a

The data collection process and the way inwhich the data were used in the model are fullydiscussed in the main study. Some of the datacollected are presented in Tables 1 to 9 at theend of this paper.

The analysis is based on data obtained fromthree sources:

Farm Survey

The author carried out a farm survey over a 6-month period (August 1977 through February1978) in the study area. Twenty-four house-holds were chosen to represent the farmingsystems in the area (22 were retained for theanalysis) and input-output data were collectedfrom them. Twice-weekly visits were paid toeach of them by enumerators throughout thesurvey period. In combination wi th the costroute method, we used the direct measurementtechnique and the group farm interview. A structured interview schedule was used. All 161farm plots of the sample were carefully mea-sured wi th field compasses and tapes. Produc-tion and daily household food consumptionwere, as far as possible, directly measured.Market prices of agricultural products were alsocollected throughout the survey period.

Personal Interviews

Personal interviews took place with the staff ofthe Evaluation Unit of the Institute of RuralEconomy, OACV field demonstrators and otherstaff members, and the local administrativeauthorities of Sirakorola.

Reports and Documents

Relevant government reports and documentswere consulted.

Table 1 . Evolut ion o f t o t a l , ru ra l , and agr icu l tura l popu lat ions o f Ma l i f r o m 1 9 6 9 to 1 9 7 3 .

Years

19691970197119721973

Average growthrates %

Totalpopulat ion

('000)

49294980514252575376

2.19

Rural populat ion

Total('000)

45374646477548714969

2.29

%

9293939392

Agr i . populat ion

Total('000) %

4171 854198 844224 824299 824337 81

0.97

Source: Service de la Statistique Generale de la Compatabilite Nationale et de la Mecanographie (1976).

129

Tab le 3 . In t roduc t ion o f equ ipmen t In t h eO A C V .

Equipment

Mult i funct ion p lowsSeedersCarts

Source: OACV (1976).

Before Increase1974 1974-75 Total

2425 905 33301941 887 28281353 180 1533

Tab le 4 . Cred i t s i t ua t i on i n O A C V ( 1 9 6 7 /6 8 - 1 9 7 4 / 7 5 ) (mi l l ion M.F. ) .

Kind of credit

SeedsFertilizerEquipment

Total

Source: OACV (1976).

Amountlent

568546212

1326

A m o u n trepaid

413415

90918

Balance tobe paid

156131122409

Table 5 . Character is t ics o f hold ings in t h es tudy area ( 1 9 7 7 / 7 8 ) .

Group 1

Average farm size (ha) 5.34Average number of plots 5.55Average size of plots (ha) 0.91Average number of collective plots 3.33Average size of collective plots (ha) 1.27Average number of individual plots 3.33Average size of individual plots (ha) 0.58

Group II

14.198.101.634.852.295.251.07

Source: Derived from survey data.

Tab le 6 . Compos i t ion a n d average s ize o fhouseholds in the survey v i l lages,Sep tember 1 9 7 7

Male adultsFemale adultsChildren (8-14)

Total

Group I

2.93.42.08.3

Group II

5.36.83.4

15.5

Source: Derived from survey data.

130

Table 7 . Ave rage number o f tools and equip-m e n t per household ( 1 9 7 7 / 7 8 ) .

Tools and equipment

Hand toolsPlowsSeedersCartsPlowing-oxenPulling-donkeys

Group 1

6.22-----

Group II

10.311.690.851.083.231.15

Source: Derived from survey data.

Tab le 8 . Ave rage yields observed for d i f fe -rent c rop enterpr ises in t h e s tudyvi l lages 1 9 7 7 / 7 8 ( k g / h a ) .

Crop

GroundnutSorghumMilletSunaFonio

CowpeaBambara nutSweet potatoSO/CP Sorghum

CowpeaGN/BN Groundnut

Bambara nut

GN/SO GroundnutSorghum

MI/CP Mil letCowpea

SU/SO SunaSorghum

M/C/B Mil letCowpeaBambara nut

Group 1

218518269

1038916

-184

9300674

2914045

41692

--

804408---

Group II

295580527910691

0245

-355

549631

-

-321328

--

8075

16

Source: Derived from survey date.

S u m m a r y o f F i n d i n g s

Table 10 shows some results of the model withimproved technology and varying resourceslevels. Columns B2 through B6 are results f romhypothetical increases in resources levels. Theoverall results of the study can be summarized

in words as fol lows:1. The analysis of production factors in the

study area shows that for future development ofagricultural production in the area, introductionof appropriate improved technology shouldcontinue for increased productivity and farmsize.

2. The analysis shows also that mixed crop-ping is relatively unimportant in the study area;it occupied only about 12% of the average farmland for group I (nonequipped farmers), and 6%for group II (equipped farmers).

3. Three types of labor were observed duringthe study: family labor, social labor includingcommunal and exchange labor, and hired labor.Group I farmers spent about 46% of total familylabor on off-farm activities, and group II farmersspent about 33%. These activities generatedapproximately 55% of the total income ofhouseholds.

4. It was discovered that, in terms of totalquantity of rainfall, the survey year was amongthe worst seasons experienced in the area.

5. It has been demonstrated that, in abnor-mal conditions, the use of expected — ratherthan actual — crop yields was more appropri-ate in the study of farmer decision-making.

6. Optimum farm income is very low andthere is very little scope for improvement, giventhe existing level of resources, technology, andfarm prices. However, this opt imum is 40%higher than the actual farm income observed inthe area.

7. With existing practices, the relativeprofitability of groundnut decreases as re-source level is increased.

8. Groundnut was produced under improvedtechnology only at high resource levels.

9. Introduction of new technology resulted inhigher gross income and higher gross returnsrelative to resources, and it engendered largerfarm size.

10. Adoption of new technology would in-crease wi th higher resources levels.

11. For the adoption of new technology, capi-tal is the most l imiting factor for group I farmers,and the September-October labor shortage isthe most l imiting for group II farmers.

12. Labor productivity increased wi th theintroduction of new technology.

13. Supply responses to groundnut andsorghum prices are higher when improvedtechnology is introduced in the model.

131

Table 9. Average i ncome in t h e s tudy vi l lages, M.F. per household ( 1 9 7 7 / 7 8 ) .

Gross farm incomeVariable costs

Gross marginFixed costs

Net farm income

At official prices

Group 1

99 94811 317

88 6313 317

85 314

Group II

233 73228 706

205 02644 609

160417

At market prices

Group I

208 75011 317

197 4333 317

194 116

Group II

488 62628 706

459 92044 609

415311

Source: Derived from survey data.

P o l i c y I m p l i c a t i o n s

1. A l though the intermediate moderntechnology reckoned to be cheap is the typebeing introduced, it is sti l l too costly for thesubsistence farmers. Therefore, research in-stitutions should continue searching forcheaper technology.

2. Cereals are actually harvested manually;an improvement in this area by research institu-tions could contribute to relaxation of the laborconstraint.

3. Increased farm credit would be the mosteffective policy for technology adoption in thestudy area; this would enable farmers to ac-quire improved farm inputs and to hire casuallabor at peak farming periods. It shoud, how-ever, be supplemented by an output price in-centive for increased farm income that wouldenable farmers to pay the credit back.

4. Extension institutions have a very impor-tant role to play in the process; these institu-tions bring government decisions to farmers,train them for improved skills, and help them totake ful l advantage of such decisions. There-fore, these institutions should be well organizedand operational.

However, in planning agricultural develop-ment schemes aimed at farmers, it is alwaysnecessary to consider both direct and indirecteffects of the policies involved. For example: a farm output expansion policy should be ac-companied by adequate arrangements for stor-age and marketing of the increasing quantitiesof products to avoid a cost-price squeeze; an

increased groundnut price could reduce thecompetitive position of groundnut on the inter-national market and cut down exports; risingcereal prices could lead to a price-wage spiral;etc. Therefore, whatever policy is chosen has tobe within an all-embracing cohesive agricul-tural development plan.

L i m i t a t i o n s a n d S u g g e s t i o n sf o r F u r t h e r S t u d i e s

1. The major l imitation of this study is that itdetermined the most profitable path of adjust-ment and supply responses to improvedtechnology, credit and price policies, but notnecessarily the most probable ones. For manyreasons — mostly social, technical and weatherconstraints — it generally takes a long t ime forfarmers to fol low the lines suggested in theresults. Intensive research, education, and ex-tension activities could go a long way to reducethe lag in adjustment.

2. The analysis was not extended to deriveelasticities that would indicate the magnitudesby which suggested policy variables should bemanipulated; this could be part of furtherstudies.

3. Other issues such as integration of live-stock enterprises, or influence of weather un-certainty on farming decisions, were not in-cluded in this study and would be desirablefields for further research work.

4. More study and empirical work, coveringother agricultural products and other parts of

132

133

134

the country, remains to be carried out to give a

complete picture of the problems encountered

by the farmers, their possible solutions, and the

consequences of these solutions. If this kind of

study, or a variant of it, is conducted in other

parts of the country, it could be of particular

interest for the formulation of agricultural de-

velopment policies, the orientation and im-

plementation of research and extension institu-

tions.

References

CNPER. Rapport final de la CNPER. Premiere partie.Situation de I'economie rurale malienne en 1972.

Bamako, Mal i .

INSTITUT D'ECONOMIE RURALE. 1973. Rapport de factibi-

lite de I'Operation Arachide et Cultures Vivrieres.Vol. 1.

Low, A. R. C. 1974. Linear programming as a too l ininnovation selection for farms in South-East Ghana.In Planning agriculture in low income countries.University of Reading development study 14.

N'DIAYE, B. 1970. Groupes ethniques du Mal i . Collec-t ion Hier. Bamako, Mal i .

OACV (OPERATION ARACHIDE ET CULTURES VIVRIERES).

1976. Compte rendu de la campagne agricole1974-75.

SERVICE DE LA STATISTIQUE GENERALE DE LA COMPTABI-

LITE NATIONALE ET DE LA MECANOGRAPHIE. Rapport de

I'enquete agricole 1973-74. Mal i .

UNITE D'EVALUATION/INSTITUT D 'ECONOMIE RURALE.

1976. Resultats d'une enquete descriptive de la zoned'intervention de I'OACV en 1976.

135

Soc ioeconomic Research on Opera t iona lLandhold ings

M . B e n o i t - C a t t i n *

Abstract

This paper explores the methodology of on-farm research. The method is based on a

rational choice of specific situations. It concentrates on the farm as a place where

decisions on agricultural production are taken. The author holds these decisions are a

result of the dialectic relationship between economic and social systems and ecosys-

tems and a given historical context. Topics covered include selection of investigation

locations, village-level studies, the residential unit as a unit of investigation, i den t i f i ca -

tion and analysis of the farm, the cultivated plot as an observational unit, study of

livestock raising, method for analyzing observations, interdisciplinary work, and the

integration of analysis and action.

This paper puts forward several proposals con-cerning the methodology of on-farm researchconducted within an agricultural research pro-gram. These proposals are based mainly on theauthor's experience in Senegal within the Ex-perimental Units project and earlier in the IvoryCoast (interdisciplinary study in a forest region).

The general objective of this research is tounderstand the dynamics of farms in order toexplain their present status and to gain a betteridea of their future potential. This is neither a review of a development project nor is it a statistical study. Farm dynamics are related totechnical innovations or proposals developedthrough agricultural research. It is therefore a useful link between research and development;it is also useful for technical research in terms ofident i f i ca t ion , se lect ion, and des ign ingmethodology to study problems (feedback).

The method is based on a rational choice ofspecific situations. A l imited number of situa-t ions are selected to enable a better understand-ing. It focuses on the farm as a decision-makingcenter for agricultural production. These deci-sions are a result of the dialectic relationshipbetween economic and social systems andecosystems and a given historical context.

* Charge de recherches, INRA-GERDAT AgronomyDivision of IRAT, Montpel l ier , Cedex, France.

N O T E : This is an edi ted t ranslat ion of the originalFrench text, wh ich appears in Appendix 1.

P r i m a r y C o n c e r no f t h e I n v e s t i g a t o r

The investigator should first draw up a generalbibliography. Using theses and other literaturehe should collect information on the settle-ment process and important events inagr icu l tu re — wars , co lon izat ion, marketeconomy, etc. From development projects andtheir reviews, he can gain information on past,present, and future development efforts.

S e l e c t i o n o f I n v e s t i g a t i o nL o c a t i o n s

The place for investigations wil l be the village.In practice, investigations can be conducted inthree or four villages with one or two investi-gators per village. Beyond that, it will be difficultto handle information, especially in the begin-ning.

Generally, the village is a definite administra-tive unit but, in practice, the study can beconducted in a village made up of severalpossibly widespread quarters or in several vil-lages having certain relationships within thesame cooperative or area. These quarters orvillages may correspond to different lineages,ethnic groups, or castes.

The selection of these villages is very impor-tant and should be made in consultation wi th allthe people w h o know the region —

136

administrators, researchers, development of-ficials, etc. (collaboration with developmentofficials will avoid the possibility of disputesover the "representativeness" of the results).The selection criteria should be related to theecosystem (rainfall, soil, vegetation, etc.) andthe economic and social system — t ime anddensity of settlement (land saturation), differentethnic groups, proximity to a town, etc.

V i l l a g e - L e v e l S t u d i e s

A few preliminary investigations wil l enable a better verification and characterization of thevillages selected and will provide a basis forfurther investigation.

The investigator should study:• the history of the village and its quarters.• the land-tenure system and land use in-

cluding livestock raising.• the collective equipment.• the cooperative.• the relations wi th outside areas (neighbor-

ing villages, migrations, exchange of pro-duce, transfer of income, etc.).

• the cropped areas and other structures inrelation to development projects.

The information can be collected throughdiscussions with certain people (village heads,leading citizens, development officials, etc.). Anexhaustive demographic census may be carriedout after the cens us of residential units. This wil lprovide more accurate data on family struc-tures, migrations, extra-agricultural activities,and income, etc.

T h e R e s i d e n t i a l U n i t a s a U n i t o f I n v e s t i g a t i o n

Without forming any preconceived ideas onwhether the farm (economic unit) correspondsto the compound or physical residential unit,the latter wil l be retained as the primary investi-gation unit because it is easily identified, so-cially recognized and organized, relatively sta-ble, etc. In Senegal the residential unit is thecarre ( "compound" in English1) although sub-sequent investigations may reveal that a carre

1. A lso called " k e u r " in Wolof and " m ' b i n d "Serere.

in

may include several farms or economic (pro-duction) units.

An identification method is established andused on all schedules —year, village, quarter,residential unit, household.

A rational choice can be made for detailedinvestigations based on village-level studies,the residential units, and different situations(ethnic group, caste, lineage, land availability,t ime of settlement, farming systems, system oflivestock raising, internal complexity, etc.).Each investigator should be able to study aboutten residential units.

T h e F a r m : I t s I d e n t i f i c a t i o na n d A n a l y s i s

Analyses should be based on the farm since it is,by definit ion, the place where economic andsocial systems interact wi th the ecosystemthrough agricultural practices.

The farm is a system. The study of this systemand how it functions requires the identificationof its factors and the analysis of the relationshipbetween these factors in relation to the ob-jectives of the system.

The agricultural system has several objec-tives in a transitional agricultural economy(from subsistence to market economy). When,as for any other system, its primary objective isits perpetuation, it seeks to be self-sufficient infood and to maintain the land-tenure system.Most of the other objectives (these vary withindividuals, their status, etc.), increasingly gen-erate monetary requirements.

The factors of the system may be classifiedinto: anthropic, biotic, and abiotic having re-ciprocal relationships:

The relationships between these threecategories of factors are identified and analyzedwithin and between the residential units.

Types of relationships include:• relationships between anthropic factors:

family structures, households, socialstatus and hierarchy, organization ofwork, monetary exchanges, debts, etc.,

137

• relationship between biotic factors: croprotations, cropping patterns, composi-t ion of herds, livestock-raising systems,

• relationship between abiotic factors:sets of farm equipment.

• relationship between anthropic and bio-tic factors: distribution of crops andanimals according to individuals andtheir status, consumption of agriculturalproduce.

• relationship between anthropic and abio-tic factors: distribution of land, of imple-ments, etc. according to individuals, theirstatus, and mode of acquisition.

• relationship between biotic and abioticfactors: yield factors, animal traction.

These relationships occur mainly within theorganization of agricultural work (in a broadsense): who does what, wi th whom, on whoseplot, etc. It is this complex interdependencewithin the organization of agricultural work thatenables the identification of the farm.

T h e C u l t i v a t e d P l o ta s a n O b s e r v a t i o n U n i t

It is in the cultivated plot that these factors,mainly their relationships, can be observed:ownership of the plot, organization of work, useof draft animals and equipment, cultural prac-tices, etc.

To complement these observations, studiesare required of the residential unit and farmsonce they have been identif ied: population,migrations, extra-agricultural activities and in-come, inventory of sets of farm equipment(draft animals and implements), monetary ex-penditure (debts, etc.), consumption of agricul-tural produce, etc.

S t u d y o f L i v e s t o c k S y s t e m s

This study fol lows the same general approach.The animals constitute the biotic factors to berelated to the anthropic factors (owners, mana-gers) and abiotic factors (pastures, harvest by-products, organic manure for plots). Thisapplies to relationships wi thin the residentialunit or vil lage, and possibly between villages(transbumance or nomadism).

It is important to identify the interactions

138

between livestock and agricultural systems.These systems supplement and/or competewi th each other and these relationships have a determining influence on their future.

M e t h o d f o r A n a l y z i n gO b s e r v a t i o n s

The quantification of observations is not an endin itself but a means of providing a betteridentification of the relationships between thefactors. Thus a study of operation schedules isof little use in itself; moreover, it is tedious andoften disappointing. On the other hand, havingobtained information on the organization ofwork, it would be useful to specify the nature ofcertain exchanges (reciprocation, imbalance)by evaluating their importance. An estimationof half days of work is generally sufficient forthis purpose. The same observations can bemade for budget investigations.

For recollection, the methods for statisticalanalysis are — factorial, correspondence, re-gression etc. These are the means available toinvestigators and should be used to studycertain problems, not vice versa — data shouldnot be collected to apply existing methods.

The relationships between factors (whetherquantified or not) can be presented in a verypractical way by using a matrix and this can bedone at all stages of the investigation (primaryanalysis to the final presentation). Using a double-entry matrix of this type, it is possible toquantify, aggregate, classify, and check(cross-checking, blank squares) data.

Mathematical farm models are a simplif iedway of representing certain types of farms andare generally used for computerized simulation.But it should be noted that this model is onlypossible, and in any case relevant, if properinformation is available on the mechanism ofthe farm — that is, if these analyses aresufficiently advanced and include innovationsmade within the production systems. A modelrequires the use of normative data compiled byusing a specific methodology.

R e m a r k s o n I n t e r d i s c i p l i n a r yW o r k

The type of factors within the farm system and

their relationships belong to several scientificdisciplines. However, all these analyses can becarried out by an investigator with training inagriculture wi th some knowledge of socio-economics. There is, however, a risk that hemay concentrate on the biotic and abiotic fac-tors.

On the other hand, a useful collaboration maybe set up between two or three scientistsspecializing in sociology, agronomy, andeconomics.

The areas where the circles of each disci-pline overlap shows interaction between theapproaches.

As the same scientist often knows agrono-my and economics, the area of study is widenedby including a specialist in geography whowould have a better spatial perception of thephenomenon and expert knowledge of map-ping techniques.

In these two cases of interdisciplinary workthe common problem may be extended to thedynamics of the rural society. The common ground between the two wil l be the selectedvillages as shown earlier.

Such truly interdisciplinary work seemsdifficult when more than three scientists areinvolved because it requires that they worktogether at the location (and in the office) andthat they should have certain affinities and a certain common philosophy.

R e s e a r c h a n d A c t i o n

The originality and fruitfulness of the workcarried out for more than 10 years in theExperimental Units project in Senegal is due tothe integration of analysis and action. Theanalysis of the situations and their changesprovides guidelines for action, and the results ofaction enrich the dynamic analyses.

Initially, the primary objective was to prom-ote as quickly as possible in the rural areastechnical innovations developed through ag-ricultural research. This required a certainnumber of corollary activities concerning lives-tock raising, cooperation, marketing of pro-duce, extension activities, land-tenure system,farm management, etc. Analysis and actionwere combined for all these aspects.

For farm management, much work was doneand its analysis has provided the mainmethodological proposals given in this paper.Work on farms has been progressively or-ganized and structured so that it has beenpossible to develop a method of recommenda-tions on farm management. This method hasimproved with increasing information on farmdynamics.

At present, the recommendations on long-term management focus on promotion oftechniques (for agriculture and livestock) andprovide basic material for analyses on farmdynamics.

Research on farm dynamics becomes morerelevant as it is conducted over a long eventfulperiod: a single still does not make a movie.

139

sociology(anthropic)

agriculture(biotic)

economics(abiotic)

sociology(anthropicl

geography(biotic)

agriculturaleconomics(abiotic)

Discussant 's Comments

V i c t o r F . A m a n n *

The five papers presented in this session covermany subjects — f rom the wel l-documentedfield assessment presented by Dr. Chaudhari tothe theoretical discussion of proposed researchby Benoit-Cattin. Unfortunately, I was askedonly one hour ago to discuss these papers be-cause the discussant was unable to attend.

Dr. Chaudhari discusses the adoption of dry-land farming technology in the Malwa region ofMadhya Pradesh. His findings show the gapbetween farm and experimental results. This ismainly due to low investment in seed and fer-tilizer, which is not an unusual result. He found,however, that the project results showed sig-nificant yield increases through adoption ofnew varieties, which is quite significant in a 5-year period. It seems to demonstrate thatfarmers are wil l ing to adopt new ideas that in-crease income if labor requirements do notchange significantly.

Mention was made of improved animal hus-bandry and of a machinery program. However,no data was presented on the use and distribu-tion of draft power. These two factors are im-portant in improving farm output in Africa and I wonder if the same problems are important inIndia.

The analysis shows that small farmers wi thless than 2 ha of land were the major be-neficiaries of the scheme, although the authorlater discusses a general imbalance in the use offertilizer. It would be of interest to know how thesmal l group received greater benefi ts. Forexample, was it better weeding or increasedlabor use? The paper suggests that intensifi-cation of labor use has high marginal returns onsmall units.

The last point, on the improvement in farmerresponse to farm business analysis after theywere fed back some data, is similar to experi-ence in other countries. It seems logical that

* Divis ion of Planning and Statistics, MinistryAgr icul ture, Goborone, Botswana.

of

when farmers see the results of the researchafter the endless questions they are asked andthe cooperation requested of them by field re-searchers, they are more responsive. Thereseems to be a lesson here for researchers tohave a system of information feedback to theirrespondents.

The second paper, by Dr. Doherty, discussesthe sociological problems of introduction of twoseparate technologies. The introduction of thewheeled toolbar does seem to create some spe-cial problems in human relationships, espe-cially, as the paper suggests, when the cost ofthe machine is beyond the purchasing capabi-lity of a single farmer. Group action for purch-ase of such an implement would undoubtedlycause human re lat ions p rob lems becausetimely plowing is so important for ensuring a crop. The paper suggests tha t despite thetechnological possibilities of the machine inimproving output, reducing power needs, andincreasing the speed of plowing, it must bemade available at a lower price if adoption is tobe expected. The idea of group action in thepurchase and use of a machine as costly as thewheeled tool carrier seems remote. I would liketo know how the Tropicultor in West Africa re-ceived such widespread usage. Perhaps some-thing can be learned from that experience foruse in both India and southern Africa.

The second innovation discussed is thewatershed technology for supplemental irriga-t ion. This seems to be an old practice in parts ofIndia, but more widespread use is suggested.The problem raised for discussion by the authorrelates to management of the organization andthe question of the opt imum size of a group tomake this innovation workable. This samequestion is now under discussion in Botswanawhere dams are built at the instigation ofgroups for watering cattle. The control of live-stock numbers for ecological and range preser-vation reasons is part of the package, but inarable agriculture water use may be moredifficult to control.

Field tests designed to show possible

140

socially-related preferences for different op-tions in the application of new technology, suchas the studies under way at ICRISAT and de-scribed in Doherty's paper, are quite important.These should be carried out in a variety of socialenvironments, as well as in a variety of agro-climatic environments as at present. In India, itseems necessary to identify relatively egalita-rian SAT farming societies for this purpose,which might reveal contrasting socially-basedpreferences regarding such new technology, ascompared with the highly stratified societies ofthe three villages in which such research ispresently under way.

ICRISAT scientists, and trainees and scien-tists who visit ICRISAT, need to know muchmore about different types of social organi-zation of agriculture in SAT areas. These diffe-rent organizations need to be assessed for theireffects on productivity, investment, conserva-t ion, and welfare. The results of such assess-ment need to be taken into account f rom thebeginning in the design of new agriculturaltechnology. It is suggested that there is anurgent need to add to our knowledge of thecomparative social organization, especially forland use, land tenure, water use, water tenure,labor, and related topics.

I shall discuss the papers by Messrs. Faye andBenoit-Cattin together as they both deal withtheoretical approaches to landholding re-search. The paper by Mr. Faye discussed therelationships between the wealthier and thedisadvantaged groups in society. His thesis isthat changes usually do not have any appreci-able effect on the relative positions of groupswithin a village or society. The poorest stratausually get these by chance and remain poordue to a series of unfortunate circumstances.Research is required at the microlevel to see ifthese circumstances can be understood and ifunequal relationships can be changed either byinternal or external forces. The author did notsuggest that differential aversion to risk may bethe chief determinant of the relative position ofpeople in the socioeconomic hierarchy.

Mr. Benoit-Cattin presented a theoreticalpaper on the methodology of research on oper-ational landholdings in the West African con-text. His basic concern is the farm unit or family.He points out that a family in a compound orvillage is affected by other interrelationshipsthat are not family-related. The relationships

between sociology, economics, agronomy andgeography are all important in operationallandholding research.

One point he raised that was given very littleattention at this workshop was the interactionbetween livestock and cropping. Understand-ing the competitive and complementary rela-tionships is important in determining a moresatisfactory landholding situation and can ef-fect interpersonal relationships within a village,group, or family. This wil l require some mi-crolevel research.

Both papers concluded that land tenure mustinvolve every family in the village of compoundand that social structures and land tenure rela-tionships require closer examination. The paperby Faye was mostly concerned with land tenureand its effects on the various economic strata inthe society and how they are affected by landtenure and social change. It seems that a fullerunderstanding of the land tenure/social justicerelationships would be useful for policy makers.

The last paper I want to discuss was pre-sented by Mr. Traore of Mali, who presented a very well-organized analysis. Methodologieswere clearly identified which aimed at testingfarmer acceptance of the package of newtechnologies within a linear programmingframework. The results showed that larger far-mers were the earlier and more frequent adop-ters, which is different f rom what Chaudharifound in his research in India. The paper pointsout that adoption of a relatively simple inter-mediate technology is slow because its costoften seems to be too high for early acceptance.If price and credit were closely tied these factorscould lead to larger output, provided farmerswere assured of adequate credit at reasonableterms and product prices were consistentlyhigh. Unfortunately the author should haveemphasized that in the real wor ld, these condi-tions cannot usually be met, especially forexport crops. Even though the paper showsresults subject to the constraints in the model , itdid not produce any surprising new discoveries.The methodology used is very clear-cut andcould be a very useful tool of analysis ifadequate data are available.

141

Chai rman 's S u m m a r y

B . A . K r a n t z *

Five papers were presented on socioeconomicfield assessment of prospective technology.The papers by Chaudhari, Faye, and Traorewere based on actual experience w i thresearch-cum-extension or extension schemesto spread specific technologies in areas of India,Senegal and Mali respectively. The other twopapers, by Doherty and Benoit-Cattin, dealt withthe approaches and methodological aspects inconducting studies for socioeconomic fieldassessment of prospective technology.

Chaudhari's paper records the extent of adop-tion of improved farming practices by farmersin the Indo-UK Project on Dry Farming at Indore(India). From 1974-75 to 1977-78 farmers havebeen able to adopt a number of new practicesthat increased cropping intensity, income, andemployment. In particular, cropping of the trad-itionally kharif (rainy season) fal low lands hasincreased from 34% of net sown area in 1974-75 to 68% in 1977-78. Some farmers reportedlack of capital, supplies, and risk as the mainconstraints.

Traore presented results using linear prog-ramming as a method of analysis. The resultsshowed that larger farmers were earlier andmore frequent adopters of new technology.Capital was found to be a major constraint facedby slow adopters of technology.

Faye presented some ideas emerging from a research-cum-action program conducted in theExperimental Units of Sine-Saloum (Senegal).He concluded that proper identification of targetgroups based on full understanding of socialstructure is essential if the prospectivetechnologies are to reach the small and poorfarmers.

Benoit-Cattin also used the Unites Ex-perimentales Project in Senegal as his referencewhile presenting a methodological frameworkfor conducting research for socioeconomic fieldassessment of prospective technology.

Doherty's main concern was human organi-

* Farming Systems Research Program, ICRISAT.

zation, to facilitate adoption of technologywhich is technically or financially indivisibleand therefore beyond the capacity of individualfarmers to adopt it. Adoption of the watershedas a unit of resource management and of ananimal-drawn Tropicultor which facilitated con-servation and efficient use of moisture were twoitems in the context of which the need for groupaction was emphasized.

Dr. Amann initiated the discussion afterbriefly summarizing the contents of paperspresented. Except for a few questions seekingclarif ication on specific aspects of theresearch-cum-action programs, most of thediscussion centered on the issues of methodol-ogy and research approach discussed by Doh-erty and Benoit-Cattin in their papers.

In particular, the participants questioned thevalidity of ideas involving specification of de-sired numbers of people in a group to ensuredurability and viability of group action for adop-tion of watershed based technology, as discus-sed by Dr. Doherty. Also questioned was therelevance of the framework of hunter andgatherer societies for the study of an integratedcommunity of Indian farmers to learn aboutconditions required for viable group actionamong farmers. The need for recognition of'profitabil i ty' or 'uti l i ty' as a key factor in induc-ing group action was also emphasized. Mr.Benoit-Cattin's suggestions about the unit ofinvestigation (i.e., residential unit or oper-ational holding) were also debated; it wasconcluded that flexibility in the approach wasrequired and it depended on the purpose of thestudy.

In terms of policy conclusions the fol lowingwas the feeling of the group:

1. Since the term prospective technologymay mean different things (e.g., innova-t ion at its early stage or the technology yetinside the fences of experimental sta-t i ons ) , t he m e t h o d o l o g y for itssocioeconomic field assessment wi l l differaccordingly. The questions relating to thearea, t ime span, and sample sizes for such

142

studies can also be decided only on thebasis of the characteristics of the technol-ogy.

2. It would be desirable that field assessmentstudies involve the local agencies respon-sible for development programs in thearea. It may facilitate execution of thestudy as well as the subsequent fol low-up.

3. If quantitative models are developed forthe purpose of socio-economic field as-sessment of prospective technology theirvalidity should be fully established on thebasis of the extent to which they approxi-mate the real-world situation.

4. The t ime lag between conduct of studiesand reporting of results should be as shortas possible to facilitate quick feedback toresearch stations so that timely amend-ments in the technology can be made.

143

Chapter 4

I s s u e s i n F o o d g r a i n M a r k e t i n g

Reforming Gra in Marke t i ng Sys temsin West A f r i ca : A Case S tudy of Ma l i

Ell iot Berg*

Abstract

This paper examines the foodgrain-marketing problems of the West African republic of

Mali and analyzes why numerous proposals for reform have proven infeasible or too

difficult to implement Among the principal findings are: (1) the Mali government is

severely limited by physical, financial, and organizational factors in what it can

implement; (2) the present mixed system of marketing (government and private sector)

cannot be easily improved as recommended in some studies; (3) uncertainty over

prices and general market disorganization divert farmer effort to cash crops and may

reduce farmer willingness to undertake greater efforts or new ventures in grain

production; (4) since existing cooperative organizations are instruments of government

used mainly for grain requisition purposes, farmers are reluctant to set up true

cooperatives that could better defend their interests; (5) external assistance including

food aid and a line of credit in the Operations Account in Paris has diluted the impact of

grain-marketing policies and allowed the Mali government to maintain policies without

having to fully absorb the consequences; (6) until very recently, the government had not

been presented with well thought through proposals. The author concludes that in any

successful reform the State grain agency will have to play a major role — even under a

"minimalist" assumption about the State's role in grain marketing — a n d that major

improvements will result from indirect measures such as improvement and extension of

feeder-road networks, better information on crops and marketing and better d i s s e m i n a -

tion of such information, closer attention to relaxation of production constraints on food

grains, and improved policy analysis within government Such indirect changes will

widen the options for reform and increase the probability of their adoption.

The West African republic of Mali may hold a

record in the world of foodgrain marketing,

price policy, and storage: in the past 5 years, no

fewer than 11 missions have studied Mali's

problems in this area and have written reports

thereon. And this is not an accident. Mali's

difficulties have been and remain deep-rooted,

and finding a way out is not simple. One happy

result of all this attention is that Mali 's grain-

marketing system and price policies are unusu-

ally well documented. Since marketing and

price issues are important in the country's rural

development, Mali provides a highly pertinent

case study. The nature of the problems — t h e

* Professor of Economics and Research Scientist,Center for Research on Economic Development,University of Mich igan, Ann Arbor , Mich igan, USA.

deficiencies of exist ing structures and

policies — are clearly observable, as are the

obstacles to reform; and in many respects, the

Mali situation has close parallels throughout

West Africa. In this paper I wil l focus on the

question: why has reform proven so difficult?

Def ic iencies in Marke t i ngPer fo rmance and Pol icy

There is little disagreement in most of the

reports, and in the understanding of concerned

Malians, about the kinds of inadequacies that

characterize grain marketing performance and

policies in Mali. These diagnoses have been

described in detail in CILSS/Club du Sahel

(1977, Vol 1, and Mali in Vol 2). Here we wil l

only summarize some of the main problems.

147

Policy Objectives Not Ma t

First, the present marketing policies are failureswhen judged by their own objectives. The mainobjectives of the present marketing policy, asthese are conceived by Malian representativesand writ ten in relevant documents, are (1) tocontrol the grain market (maitriser) so thatproducers can be guaranteed a remunerativemin imum price; (2) to guarantee grain suppliesfor deficit regions, including urban areas; (3)to stabilize prices to both consumers and pro-ducers. Related objectives are the protectionof peasants against possible exploitation bytraders and the development of "order ly"trading arrangements.

It is not uncharitable to observe that none ofthese objectives has been achieved. As can beseen in Table 1, one-half to three-fourths of thetotal volume of marketed millet/sorghum pas-ses through the " t radi t ional" sector — i.e., pr i-vate traders. In addit ion, the state grain tradingagency, the Office des Products Agricoles duMali (OPAM) has not been able to maintain a floor or ceiling price on grain purchases andsales, nor has it been able to stabilize consumeror producer prices seasonally or interannually.Table 2 is indicative.1 Moreover, OPAM doesnot even supply most of the grain to deficitregions; it was estimated in the mid-70s thatprivate traders handled 60% of the grain supplyin the remote and particularly food-scarce SixthRegion.

The reasons for these failures to attain statedobjectives are well known. OPAM's shortagesof f inancing, storage capacity, trucks, and per-sonnel played a part. Inflows of food aid and thedemands of the 1972-74 drought led OPAM toplay a major role in channeling imported grain,to the neglect of its internal marketing role. A price policy imposed on it by government (lowconsumer prices for millet and sorghum) re-duced its l iquidity, and a policy of purchasesand sales at uniform national prices led tofurther operating deficits and a reduction inliquidity. At least up to 1976, the slowness wi thwhich the grain quotas were set and delays inrelease of crop-financing funds by the banking

1. For addi t ional data on divergence between of-f icial and actual market prices for foodgra ins, seeOLSS/Club du Sahel) (1977 vol 1, p 54).

148

system meant that ORAM was never evenpresent in the market until January, after thepeak postharvest marketing period was over.

Thus, OPAM and the existence of a legalmonopoly has not protected the produceragainst presumed "exploi tat ion," as it wasdesigned to do. Nor has the hope that stategrain marketing would impose a more orderlyand efficient organization on the grain tradebeen borne out by experience. There is con-siderable uncertainty over marketing responsi-bility, not only between public and private sec-tors but within the public sector as well. Forexample, ambiguity over marketing jurisdictionhas recently arisen between the cooperatives,OPAM and Operation Mils-Mopti, which hastaken over responsibility for grain marketing inthe area of its activity.2

OPAM Operating Deficitsand Inefficiency

In recent years the Mali economy has been

2. A recent evaluation report by Berthe and Meyer-Ruhle (1977, p III-10) comments : "Af ter the be-g inn ing of the 1976-1977 campaign the govern-ment suddenly decided that in the Bankass andDjenne cercles the exist ing cooperatives wou ldtake over the grain col lect ion and del iver directlyto OPAM. It is, however, expected that this wasonly a temporary measure (the reason of which Isnot officially known) and that in the comingseason Operation Mills wi l l again be charged w i thcommercial izat ion in Its who le intervention

area. "

characterized by severe budgetary and balanceof payments disequilibrium. Budget deficitsfrom 1972 to 1976 were between one-fifth andone-third of total expenditures and close toone-third of recurrent revenues. During thesame period the overall balance of paymentsshowed large and growing deficits, amountingto between 20 and 50% of recorded imports.

The accumulation of debt by OPAM (andother public enterprises) was a major factor inthe economic deterioration of the 1970-76period. During these years OPAM accumulateda debt of 38 billion Malian francs (about U.S. $80mill ion), 12 billion of it in 1974 alone; this wasequal to 40% of total budgeted governmentexpenditures during that year.

Some of this debt arose during the 1972-74drought, when Mali imported large amounts ofgrain to stave off famine. These imports weresold at heavily subsidized prices and werefinanced by bank credits to OPAM. But much ofit arose from the meager margins betweenOPAM's buying and selling prices for grain,which make it impossible for the agency tocover its marketing costs.

A significant proportion of OPAM's deficitalso derives from management inefficiency,particularly with respect to transport and stor-age operations. Transport deficiencies com-monly delay movements of grain stacked out-doors under tarpaulins, exposing it to the firstrains. The IDET/CEGOS (1976) Report is onlyone of the many that refer to substantial lossesbecause of poor storage management. Theburden of these losses is passed on to pro-ducers and their cooperatives, removing anydirect incentive for OPAM to improve its per-

149

formance.3 In other cases, grain stored indoorssuffers heavy losses because of insufficient orimproper fumigat ion, infested sacks and failureto rotate grain properly.

Poor Market ing Services to Farmers

The private grain trading system is unde-veloped. The number of individuals workingexclusively as traders is very small. More im-portant, there is extremely little specialization infoodgrains, just as there appear to be very fewtraders who are ful l - t ime traders and nothingelse, so there are few traders who work at thegrain trade alone.4 Most buy and sell consumergoods and other crops. All this means that thedistinction between producers and traders isfrequently obscure, as is the distinction bet-ween traders and consumers. The person whoacts as millet assembler also trades in con-sumer goods, in rice, in groundnuts.

In addition to small scale and lack of special-ization, the grain marketing system is charac-terized by poorly developed rural credit ar-rangements. Private traders do provide somecredit, but it is unclear how much. Al though it iswidely believed that a substantial proportion ofproducer millet sales are made to pay offpreharvest debts, such evidence as existssuggests that this is a relatively minor factor.The IDET/CEGOS study found that in only 15%of the villages they surveyed did traders appearbefore the harvest. They report that most of thesales in the villages come f rom reduction ofstocks as the new harvest approaches. The 1973FAO (Panhuys) Report est imated — veryroughly — that perhaps 5000 tonnes of millet/

3. If not stored properly, grain del ivered to OPAMdeteriorates rapidly. Each year impor tant quan-t i t ies (thousands of tons) cannot be carried awaybefore the onset of the rainy season and are thusrendered unf i t f o r consumpt ion. The loss is ab-sorbed by the producers, i f OPAM has not alreadypaid them and by the cooperatives, i f OPAM'sfunds were distr ibuted, since OPAM demandsrepayments (IDET/CEGOS 1976.)

4. The IDET/CEGOS Report states categorically:' T h e r e are no private traders in Mali w h o deal inmi l le t alone: mi l let purchases at the farmer leveland wholesale are considered by the main agentson ly as a supplement to market ing of otherproducts (especially sheanuts and groundnuts) . "

sorghum were sold to traders in debt repaymentsat harvest t ime. This would have been less than5% of marketed output.

Similarly, very l imited amounts of grain arestored by traders. It is rare to find private traderswith a storage capacity of as much as 50 tons.Most storage, like most trading activity, ismixed — i.e., grain and other commodities, fre-quently including consumer staples such assugar, wil l be stored in the same place, gener-ally a room or two in a rather rudimentarybuilding.

The private sector marketing facilities pro-vided to sellers of foodgrains are thus veryrudimentary. In comparison with what is avail-able for export crop producers, they are mar-kedly deficient; the public grain trading agen-cies do not provide much more. Neither publicagencies nor private traders maintain an effec-tive presence in the villages. For first echelonstorage (at cercle chefs-lieux) in the state market-ing circuit, the farmer himself, or the villageauthorities acting in the name of the villagecooperative, must arrange for transport of grainfrom the farm to the cercle. But the cooperativesare paid too little by OPAM: costs of transportf rom village to the district (arrondissement) level are not fully covered. Sometimes the localcooperatives are not paid at all. Moreover, they(that is, the farmers) must bear the costs ifinadequate storage and transport delays causelosses of stored grain.5

Illustrative of the differences in public market-ing services provided for cash crops and thosefor food crops is the fact that the Operations wil lsend trucks to bring groundnuts to main stor-age points whenever a village or group ofvillages can assemble 80 sacks — and this at nocost to the producers (CRED 1976, pp 119-120).

It is much the same wi th other servicesnormally provided by a marketing system: cre-dit, trader-provided storage capacity, off-farminputs. Credit is available via the export crop

5. In its vi l lage survey, the IDET/CEGOS team notedthat in almost 40% of the vi l lages, peasants"spontaneous ly" declared that their costs oft ransport of gra in to the arrondissement chef-lieux were not paid for. The cooperatives (Federa-t ions de Groupements Ruraux) take the rebatethat OPAM gives for this purpose, to meet theiro w n expenses (IDET/CEGOS 1976, vol 3, p 41).

150

promotion agencies, as are fertilizers and otherinputs. However, with a few exceptions (e.g.,the Operation Mils-Mopti), food growers mustrely on " t radi t ional" sources of credit, little ofwhich appears to be provided by traders. Andfew input-provision schemes are available, ex-cept for export crops.

Negative Production Effects

It is not possible to know the extent to whichinadequacies of marketing arrangements andpolicies have affected grain production. Table 3 shows existing estimates of output trends formajor crops during the past 20 years. It is clearthat until 1974 foodgrain production was stag-nant or declining; the same is true forgroundnuts, though not for cotton. But it isimpossible to separate the effects of marketingpolicy, price policy, general economic policy,and weather variations.6 One must nonthelessstrongly suspect — as so many observorshave — that the uncertain, thinly structured,and relatively poorly-functioning marketingsystem has had negative effects on grain outputand on producer willingness to sell.

Negative Equity Effects

The marketing system has many inequitablefeatures. There is, first of all, the quota orrequisition system, by which decisions aremade on the amounts of grain that villagesmust deliver to OPAM at official prices. In a yearof poor harvests, this system requires the de-livery of grain at below market prices. Thosewhose grain is requisitioned may have to buyfor their own consumption later in the year, atfree market prices.

Similarly, until very recently, OPAM has hadalmost no motor pool of its own. It had to rely ontruckers, public and private, who were paidaccording to an officially fixed tariff structurethat was generally too low to make OPAMassignments profitable. As a result, OPAM (withlocal administrators) regularly "requisi t ioned"private trucks — i.e., forced them to carry grainat the unprofitable, edicted rates. One conse-quence was that many transporters took their

6. Given the data uncertainties, it is often not evenclear what has to be explained.

trucks into neighboring countries when thebuying season began.7

A third form of inequity arises from OPAM'snormally l imited capacity to meet demands inurban areas and deficit regions. A dual market iscreated, since a favored group is provided wi thOPAM's stocks at official prices while othersbuy in the free market at a much higher price.Some idea of the magnitude of the differencesis given in Table3 above; grain purchased in thefree market in early 1978 cost two or three times,as much as OPAM-provided grain. The FAO(1978) describes the consequences as fol lows:

When (OPAM) has insufficient stocks it reducesthe quota which consumer cooperatives receivefor delivery to households. These householdsthen must buy on the free market, often at twicethe prices. As for the public sector, and civilservants, the quota is generally maintained,which makes these consumers a privilegedgroup. But they are only 2 percent of the laborforce...

Obstacles to Reform

If it is true that most technicians would agreewith the main elements of the preceding diag-nosis, then the question arises: why has itproved so difficult to make appropriatechanges? After all, as we noted at the outset,Mali's grain marketing and price policy ar-rangements have not gone unstudied, nor isthere any lack of proposals for change.

Politics

One part of the explanation must be found in thepolitical environment. The political factor isindeed obvious and always mentioned. TheMalian political authorities have persistentlyhesitated to raise official foodgrain prices tourban consumers, presumably because of un-willingness to risk severe political reaction.Their perception in this matter may be right;examples of disturbances and political upheav-

7. Tementao 1977, p 54. However, tariffs have beenrevised upward since 1977, and OPAM has beenendowed w i th a considerable vehicle pool of itsown (some 40 trucks). So the problem of trans-port requisit ion may be less substantial now. Butit can be counted on to reappear as soon as priceschange and/or OPAM's motor pool shrinks in size.

151

152

als have in fact frequently followed efforts toeliminate food subsidies enjoyed by urbanpopulations — for example, in Egypt and Peruin recent years.

The political factor is clearly present andimportant. Two aspects of it are worth specialcomment. First, it is evident that reformers,particularly outsiders, tend to give less weightto the political risks of change than do thepolitical authorities concerned. This is true evenwhere the objective political realities are simi-larly perceived. But such similarity in percep-tion is rare; reformers wil l naturally tend to seefewer risks than politicians.

Secondly, in Mali in particular, the politicalauthorities appear to have been especially sen-sitive politically and hence resistant to riskypolicies. This is reflected in the economicpolicies and performance record on the pastdecade. Of all the Francophone West Africancountries, Mali has had the most persistentbalance of payments deficits and the mostpersistent budget deficits.8 These deficits havealso been the largest, relative to the size of theeconomies in question. In Niger and UpperVolta, countries similar to Mali in many ways,and also under military rule in the last decade,public expenditures have risen 5 to 6% a year;in Mali the rise has been about 15%. The majorsource of the budget deficit is generally said tolie in the Malian government's policy of hiringsecondary school and university graduates. Butsuch a policy does not exist in neighboringstates.

The employment policy is only one of severalelements suggesting that the Malian govern-ment has tended to be more anxious thansimilarly placed governments to avoid impos-ing reductions in income and economic welfareon the bulk of its urbanized population. Itsimport policies in 1973 and 1974 are anotherindication. Mali's total tonnage of food imports(aid and purchases) was higher than any of theother drought afflicted countries of the region;in 1974, the Malian government paid out of itsown resources for 55 000 tonnes of rice, at a t ime when lower-priced coarse grains wereavailable. Similarly, Mali was the last of theSahel states to raise official consumer prices for

8. Chad, in the midst of the Civil War, may have hadbudget deficits that were as large a share of totalexpenditures as in Mal i .

foodgrains closer to world prices; it held outuntil early 1975 while the other governments ofthe region let prices rise in the autumn of 1974.Similarly, Mali's wage policy response wassomewhat faster and stronger than itsneighbors (Table 4).

Mali's grain-marketing and price policies,then, are part of a larger pattern of economicpolicies that reflects not only an unwillingnessto risk urban political unrest but a more generalunwillingness to accept tough discipline ineconomic management. There are manyreasons for this (Berg 1975). One of the mostfundamental is the existence, in the FrenchTreasury, of the line of credit known as theOperations Account. Mali has been able to incurlarge budget deficits — and balance-of-payments deficits — by drawing on this Oper-ations Account. There are really two factors atwork here relevant to grain marketing reform:the Malian government has been especiallyreluctant to impose hard choices in grain pricepolicy, and it has been able to avoid thesechoices, as it has been able to avoid or softensimilar unpleasant constraints in the budgetand credit areas, by drawing on its line of creditat the Operations Account.

Different Donor/RecipientPerceptions

A second obstacle to reform or more generallyto the introduction of economically efficientpolicies derives from the different interests andperceptions of the parties to aid transactions.What lenders might regard as economicallyefficient is not seen as such by the borrowingcountry.

The clearest and most relevant example is inthe area of storage programs and policies. Onthe basis of strict cost-benefit accounting (riskquestions aside) it may be possible to de-monstrate persuasively that a fore ignexchange-based food security policy, with a relatively small " f i rst l ine of defense"emergency reserve, is the most cost-effectivestrategy for Mali. This would involve a public,centralized storage capacity of far less than thehundreds of thousands of tonnes now existingor envisaged. But from the Malian govern-ment's point of view things look different. Thereis first of all the fact that the Malian authoritieswil l see greater likelihood of famine and wil l

153

Table 4 . Wage changes by sk i l l l eve l , 1 9 6 7 - 1 9 7 5 .( 1 9 6 7 - 1 9 6 9 = 100)

MaliUnskil led rateMiddle-level manpower rateb

University graduatec

NigerUnskilled rateMiddle-level manpower rateUniversity graduate

SenegalUnskil led rateMiddle-level manpower rateUniversity graduate

Upper VoltaUnskil led rateMiddle-level manpower rated

Consumer Price Index, Capital City(1970 = 100)Mal ia

NigerSenegalUpper Volta

1970

100100100

100100100

100100100

99n.a

100100100100

1971

100100100

100100100

100100100

99n.a

121105106102

1972

100100100

100100100

100100100

100100

13011511499

1973

136100103

100100100

106100100

109100

168128135107

1974

193111109

140100100

145120109

134108

194137147116

1975°

217116115

160112104

212136121

150175

----

Source: Berg, (1975, pp 118,122).a. January

b. Typ ica l rate, midd le- leve l worker , Civi l Service (except for Upper Vo l ta , whe re It is pr ivate sector rate.)c. Star t ing rate, Univers i ty g raduate (Licence), civi l service.d. Actual rate, one emp loyer , pr ivate sector.«. Foodstuf fs on ly .

weight its impact more heavily. But aside fromthis, the Malians know that there are almost noopportunity costs to aid funds available toconstruct grain storage facilities. From the localpoint of view, only if discounted future localcosts of silo maintenance and operation exceeddiscounted benefits will it be advisable to adoptan "objectively determined" cost-effective so-lution.

In these circumstances, it is not surprisingthat large increases in storage capacity haveoccurred or are under consideration not only inMali but throughout the drought-prone Sahel.This is despite the generally weak analytic andtechnical support for the storage strategiesimplied. It is not hard to see why local per-ception and local interest tend to prevail in thisarea. Silo construction is relatively cheap; silosare attractive to donors because of their visibilityand their apparent contribution to a politi-cally attractive goal ; they can conceivably bring

154

dramatic benefits in t ime of catastrophe; andthere are many competitive donors anxious tohelp in food security matters.

Underestimation of Present Defects

Many Malian officials and other observers arefully aware of the deficiencies of the marketingsystem as outlined earlier. The public positionmost often expressed by responsible local of-ficials is to recognize these problems, and tosuggest that the reasons for them lie not in thestructure of the situation but are rather due toOPAM's lack of resources and experience. Theypoint out that OPAM is, and has always been,short of credit for financing crop purchase,short of trained staff, wi thout its own transportfacilities, and with only l imited storagecapacities. They also point out that the extra-ordinary demands of the drought years made itimpossible for OPAM to meet its marketing/

price policy objectives, and that indeed OPAM'soperational life has been too short for anymeaningful evaluation.

Because this general issue is so central to theunderstanding of marketing/price policy ques-tions and to the problem of reform, it is impor-tant to address it directly and at some length.The defects of the marketing system are in factstructural in nature, not incidental or transient.The present system contains a number of basiccontradictions, major flaws that cannot beeliminated except by transformation of its es-sential characteristics.9

Difficulties in PeacefulPublic-Private Sector Coexistence

First, the peaceful coexistence of public andprivate sectors in the grain trade requires thatprices be the same in both sectors. If not, grainsales will tend to f low to one or the other sector,depending on harvest size, as is presently thecase. As things now work, if the official pro-ducer price is above the market-determinedprice, farmers will sell to OPAM, if they can.Thus, in years of bountiful harvest, OPAM canbuy all the grain it wants, or for which it hasfinancing and storage capacity. In lean years, itcan buy little. This tendency can be clearly seenin OPAM's pattern of purchases. In recent yearsit has been able to buy insignificant quantities ina year of poor harvest and high market prices;such purchases are almost always made duringbumper crop years.

This all-or-nothing tendency can be avoidedonly in one of two ways. One is for the Govern-ment to abandon any positive price. The Gov-ernment's official producer price would be thesame as, or close to, actual free market prices.This presents an obvious inconvenience: itmeans abandonment of a primary raison d'etre of the state trading system itself. The secondsolution is to make available the financial re-

9. The present system has the fo l lowing centralfeatures: partial, f ictional state monopoly of thegrain trade; purchase of part of the crop at f ixedprices by imposi t ion of quotas; sale of publiclypurchased grain, mainly in cities and other deficitzones; use of price averages (perequation) suchthat official producer prices are uni form nation-ally and official consumer prices are similarlyun i form.

sources and storage capacity required to im-plement a true price stabilization effort. OPAMcould then buy for a buffer stock in good yearsand sell in lean years. Similarly, OPAM wouldsupport grain prices throughout the year neartheir desired level by buying during the post-harvest period and selling during the soudure ("lean" or "hungry" season). The question is, towhat extent is such stabilization feasible anddesirable?

The question of stabilization wil l be taken uplater, but here it can be noted that year-to-yearprice stabilization is certain to be very expen-sive in Malian conditions because of (1) widevariation in rainfall, output, and especially inmarketed supply; (2) a possible tendency forfarmers to subsitute public storage for somevillage-level private storage; (3) risks of heavystorage losses. Its desirability is also open toquestion on the grounds that, if successful, itcould destabilize producer incomes fromfoodgrain sales, a result that may not be in linewith public policy objectives. Most important, ifprice stability encourages production of food-grains for the market, and/or if — as is likely — the stabilized price is higher than the averagemarket-determined price,10 then additionalmarketed supply might be called forth in vol-umes burdensome to the economy.

Just as price differentials create awkwardconditions on the producer side, so do theycreate problems on the consumer side.Whenever prevailing market prices at the retaillevel are higher than official retail prices, far-mers, traders, and some consumers havestrong incentives to bypass the state structureand deal with each other directly.

The existence of a private trading sectorimposes restraints also on another basic aspectof state grain price policy: the principle ofuniform grain prices over the whole nationalterritory. This perequation principle can bepursued by OPAM without involving losses(deficits) only if losses incurred by OPAM ac-tivities in areas where marketing costs exceedthe average are balanced by profits f rom pur-chases and sales in markets of easy access andhence lower than average costs. But in a systemof coexisting public and private trading sectors,

10. Assuming that the price stabil ization scheme isnot a " p u r e " stabilization effort, but has price-raising intent.

155

the private traders occupy the profitable mar-kets and leave the unprofitable ones to the stateagency. The traders buy where and when pro-ducers offer grain for sale at relatively lowprices — i.e., in the more productive regions,and those served by better roads, at harvesttime. They sell, similarly, in the most attrativemarkets (especially Bamako and the othertowns) where unit marketing costs are relativelylow. OPAM is constrained to buy and selleverywhere, and at the uniform national prices.Under the circumstances, there is no wayOPAM could avoid deficits; if its buying andselling prices were fixed at levels reflecting trueaverage costs, the result would be to abandonthe market to the private sector, or uniformnational pricing would have to be abandoned.11

A related problem has to do wi th smuggl ing.So long as private traders are allowed to oper-ate, and a policy of uniform national pricingcontinues, there will be a strong tendency forgrain to move over frontiers. Only if govern-ment could effectively impose an exportmonopoly (or assure harmonized pricingpolicies around frontiers) could extensivesmuggling be avoided. This, incidentally, pro-vides additional incentive to seek alternativemeans of meeting the government's regionalequity objectives, to the extent that such goalsare of high and continuing priority.

Another fundamental problem in the mixedtrading system has to do wi th price and qualitydifferentiation. The state trading sector cannoteffectively compete wi th the private tradersunless it establishes a much more refined pricestructure than now exists. At present, extremelylittle differentiation exists. OPAM offers thesame price for millet, sorghum, and maize.Prices are the same for different kinds of milletsand sorghums (e.g., white millet and red millet).Buying and selling prices do not differ withrespect to the quality of millets and sorghums.

The more egregious aspects of thishomogeneity could presumably be dealt wi threlatively easily — i.e., different prices for mi l-let, sorghum, and maize. But even here, delicate

11. Some other means migh t be sought to imple-ment the Government 's objective of favor ingremote and uncongenial regions. Grain transportcosts, or all t ransport costs, could be directlysubsidized; civi l servants (and other wage ear-ners) could be given a salary supplement, etc.

decisions might be required and wrong deci-sions could lead to serious distortions. Whatshould be the relative prices of maize, sorghum,red millet and white millet? Demand conditionsvary from region to region and year to year, asdo supply conditions, due to rainfall variations.Unless it reproduces the private sector pricestructure, OPAM risks f inding itself with un-wanted surplus stocks of some grains andrecurring shortages of others. Or, if OPAMenforces a price structure on the private sector itcould give wrong signals to producers — as, forexample, it is presently doing by paying " toomuch" for sorghums as compared wi th milletand making sorghums more profitable to pro-duce even though they are less desired byconsumers.

Quality differentiation poses even more prob-lems. Under the present system, OPAM obtainsthe worst quality grains, since it pays the sameprice regardless of quality. For OPAM to try toreproduce the rich variety of quality and pricedifferentials that typify even the most isolatedrural market is almost unthinkable. It wouldrequire enormous manpower and surveillanceand would involve so much bureaucratic discre-t ion as to invite extensive abuse, particularly ifused in conjunction with the quota system forrequisitioning grain. In the long run, farmerswould only sell to OPAM if the quality gradingsand price differentials available there were asfavorable as in the private market.

Difficulties in Primary Market ing

The organization of the state sector's croppurchase is extremely cumbersome. The sys-tem works as follows. Crop size is estimated inevery village in July. These estimates are sent tothe Commandant de Cercle. who — advised bya cereals committee — transmits these esti-mates to the Governor of the region. The Gover-nor sends the estimates to the Ministry ofInterior, which then determines the quota ofgrain that wil l be required of each district(arrondissement). This arrondissement esti-mate is taken as a quota that the administrativehead of the arrondissement must deliver toOPAM at the cercle headquarters, whence it isstored or shipped to deficit regions.

At every step of the way, these arrangementsgive rise to grave problems and potentialabuses:

156

a. In July the villagers have only a very roughidea of the size of the coming harvest. Theyhave in any case every incentive to understatetheir estimate. The arrondissement head alsotends to make minimal estimates, so as toreduce his risks of nonperformance. The encad-reurs, however, have an interest in overestimat-ing production. '2

b. The primary marketing is traditionally theresponsibility of the cooperative organizations,the groupements ruraux. But these are adminis-trative units. Membership is obligatory. Theexecutive committee of the cooperative is fre-quently the village council. The Federation desGroupements Ruraux (FGR) is chaired by theregional administrative head. In general then,there is no arms-length relationship between"the administrat ion" and the "cooperatives."So the whole process of quota determinationtends to set farmers against the Government.When the quotas cannot be met, force isused — "arguments of authority," as the IDET/CEGOS Report puts it.

c. The system of previsions is often notcompleted until December. But this is too latefor ORAM to take account of the estimates in itsfinancing request to the Malian DevelopmentBank. So the credit allocation to OPAM is notonly based on uncertain crop size estimates butis very late as well. This seems to be the mainreason why, in most years, OPAM has beenunable to enter the market immediately whensales are at their peak, at or near harvest t ime(September-November).

d. The cooperatives (i.e., the farmers them-selves) are responsible for transporting theirquotas from their fields to the cercle, whereOPAM's responsibility begins.13 At best, the

12. Hans Guggenheim (1977, p 12) points out thetendency for young encadreurs " t o feel that theirpride is involved in producing high yield figuresand in commercial izing as much as possible."Guggenheim also ment ions one vil lage wherethe quota one year was 40 tonnes, and thecommandant asked for 400 tonnes the next year.The quota was f inal ly f ixed at 160 tonnes. Giventhe lack of knowledge of yields and output it iseasy to see how such problems could arise.

13. The groupement rural (village) takes It to thedistrict (arrondissement) headquarters, the Fed-eration des Groupements Ruraux takes it f romthere to the cercle.

payment made by OPAM for this service is (orhas been) below costs; in fact, very often theFGR keeps the OPAM transport rebate (frais decollecte) "w i th or without the agreement of theproducer." (IDET/CEGOS, 1976, Vol 2).

e. The FGR has to rely on private transport.Until 1978, the official tariff was too low toinduce truckers to carry grain, given the lack ofback-haul cargo, the long waiting t imes, the badroads, etc. So the truckers have been frequently"requisit ioned" — i.e., forced to carry grain. Asnoted earlier, this leads some transporters totake their trucks out of the country at harvesttime, thereby accentuating the shortage oftransport.

f. Organizing primary marketing and seeingto transport and storage involves senior ad-ministrative officials in the demanding andcomplex business of the grain trade. The radiosand trucks of the prefects and commandants decercles are mobilized in the annual effort tocoordinate truck movements. The administra-tive officials can force truckers to take assign-ments. Many mistakes inevitably occur: in onearea trucks may be sent for a delivery that is toosmall. Elsewhere, not enough transport is madeavailable, so the grain stock cannot be picked upbefore the rains begin.

g. On the consumer side there are furtheranomalies, at least as of 1977.14 The "coopera-tives," in addition to purchasing and transport-ing the crop, are responsible for sales to ruralconsumers. In theory they buy millet and sor-ghum from OPAM's warehouse at the cercle chef-lieu. They sell it to their constituents wi th a legally prescribed mark-up — an amount gener-ally insufficient to cover the costs of moving thegrain by truck from cercle to arrondissement What is especially striking is that the coopera-tives cannot keep grain themselves, for sale totheir members, because of OPAM's legalmonopoly. They must sell grain at OPAM'sretail price. The farmers thus are given noincentive in the official marketing system tostore grain through their cooperatives. Nor dothey have any incentive to utilize the officialmarketing chain. They buy and sell on the

14. Recommendations for change have been made,but it's not known whether they have beenadopted.

157

" t radi t ional" (i.e., private) market.5

Distributional Inequities

These have already been described. It is thedifficulty of avoiding them that is stressed here.So long as the state sector is responsible foronly a portion of the crop, and a dual pricestructure persists, benefits and costs of thesystem wil l be unequally distributed. In badcrop years, losses will be suffered by thoseproducers forced to sell to OPAM at the lower-than-market price. In good years, producerbenefits wi l l go to those who sell to OPAM at thehigher-than-market price. Comparable in-equities exist on the consumer side. Under thepresent arrangements, the benefits of low of-ficial retail prices go to the relatively well-offcivil servants who tend to be fully provided byOPAM at the official price; others often have topay two or three times as much as the freemarket. These perverse distributional effectscould conceivably be reduced, and distributionof low-priced grain effected by such methods as"fair price shops" for low income people, butsuch finely tuned distributional efforts are ex-tremely burdensome administratively. How thedistributional problems on the producer sidecould be handled is not obvious. Using higher-than-market purchase prices as an instrumentof income distribution policy would demandgreat administrative and organizational inputs,might have undesirable incentive effects, andmight not be feasible anyway. It would certainlyopen up very substantial opportunities for ad-ministrative discretion and corruption.

The Worst of Both Worlds

As we have seen, the present trading systempoorly provides the services producers seek in a trading system: protection against monopsony,access to credit, transport, storage, and market

15. The IDET/CEGOS Report (Vol 2) summarizes thesituat ion this way : ' T h u s producers pay for thisservice (the storage of mil let) wh ich Is performedby their o w n organizat ion, and pay at a muchhigher price than it costs, and they pay to anorganization wh i ch doesn't g ive the service(OPAM). The result is that when producers needm i l l e t they trade w i th other producers in t rad i -t ional m a r k e t s . . . "

informat ion. Competit ion among buyers,stimulated by free entry into trade, is not legallyencouraged or even allowed, but no stable andeffective state-provided alternatives exist. Oneof the main justifications for public interventionin marketing is the presumed inequality ofbargaining power of the peasant vis-a-vis thetrader, but the present arrangements do notbalance the bargaining situation. A relativelysmall share of marketed output is bought at theofficial price by the grain agencies. The bulk ofprimary marketing is still in the hands of privatetraders. Given the risks and uncertainties of thelegal situation, there are fewer traders thanthere would otherwise be, and the price de-manded by traders for their services is probablyhigher than it would otherwise be.

From a longer-term and development point ofview, the present arrangements do notencourage — perhaps do not even al low — thestrengthening of private marketing skills. Theirambiguity discourages technical progress intrading practices and techniques, the growth oftrading capital, and the emergence of morecomplex entrepreneurial skills.

Conflict Between Private TraderEfficiency and Doctrinal Factors

Because the mixed system has serious disad-vantages while the preconditions for an accept-able coexistence between public and privatetrading sectors are so demanding, drastic mod-ifications are called for if marketing deficienciesare to be effectively addressed. This is wherethe fundamental problem for reformersemerges. For there are only two ways to go:toward liberalization of the trading system — i.e., a greater role for private traders — ortoward more thorough state control. And whi lethe technical or economic advantages areoverwhelmingly on the side of liberalization,doctrinal or ideological predispositions call forstrengthening the state trading monopoly.

There can be little doubt about the substantialeconomic advantages of a private trading sys-tem compared wi th a state grain tradingmonopoly.

The private or "tradit ional" grain trade usesresources more efficiently than state trad-ing agencies. The supply of traders' services ishighly elastic at relatively low levels of remun-eration. Private trade is, for many thousands of

158

Malians, a part-time activity. Farmers, urbanworkers, school children, women, all may do a little grain trading. They tend to do so in slackperiods of the year. They also trade in connec-tion with local market activities that have a social component. In such circumstances, theopportunity costs of engaging in trade are verylow. It is the same with casual or informaltraders — chauffeurs and others with com-mand over empty cargo space. These traders'services are offered in relative abundance andat low cost.

Even full-time traders do not deal in grainalone. The volume of trading activity and therate of return from grain trading seem too low toencourage specialization. Thus, costs of tradingoperations are shared by general trade — incash crops and consumer goods, as well asfoodgrains.

Since foodgrains are heavy in relation to theirvalue, transport costs are critical; and it is in thetransport area that the private trade has particu-larly large cost advantages. First, some consid-erable part of the privately-sold grain is broughtdirectly to local periodic markets by the farmeror a member of his household. Secondly, somegrain is moved between local and regional ornational markets by the informal trader men-tioned above. Most important, the specializedstate grain trading agency normally faces fear-some cost problems because of lack of returncargo. Especial ly in the remote re-gions, there is little cargo brought by truckerscarrying grain f rom local chefs-lieux or wher-ever the primary bulking point is located. Theprivate trade is much more economical in itsuse of transport — for example, by searchingfor two-way cargo, mixture of cargo, etc. Whatis true of transport is true of other inputs. Thetraditional trade in general uses human andphysical capital more efficiently than the publictrading structure. The latter requires formallytrained manpower — managers, accountants,clerks — whereas the traditional trade relies onhuman energy and skills developed informallyin the market place. Because it tends to be oflarger scale and more complex, the state struc-ture requires physical facilities — offices,warehouses, trucks, cars, etc. — that are moremodestly provided in the traditional trade. Italso requires inputs that are scarcer than physi-cal and human capital: coordination, organi-zational capacity, and information. A de-

centralized private trading system economizeson all these.

There are many well-known, general reas-ons why sma l l o rgan izat ions or p r i va teindividuals tend to be more efficient than largerorganizations, especially state organizations:speed and flexibility in making decisions; free-dom to hire, dismiss and reward; detailed andspecialized knowledge of the activity in ques-t ion; the stronger urge for material incentives.Many aspects of the grain trade do not lendthemselves easily to large-scale operations.16

To illustrate, one need only consider the proba-ble difference between public and private re-sponses to a situation of deteriorating grainstocks in a warehouse. The private trader wouldsurely be more likely than a public servant toprevent infestation or improper rotation. On thestate side, incentives and capacities to preventinfestation are limited by personnel and budgetconstraints. The f low of information to manag-ers is slow and uncertain. The capacity torespond to unforeseen local situations is l imitedby poor communications, lengthy administra-tive procedures and diffusion of responsibility.

But it is perhaps less the advantages of thefree market solution than the disadvantages ofthe monopoly solution that provide thestrongest arguments for liberalization:

Foodgrain market structures do not lendthemselves to state monopolization. Grain isgrown over much of the country. It is traded inthousands of villages and hundreds of ruralperiodic markets. As a result of tens ofthousands of small transactions, the bulkingfunction is performed: small traders put to-gether marketed supply virtually bag by bag.The distinction between traders and farmersor consumers is generally unclear, and the in-formal or casual traders, who utilize transportcapacity for small adventures into grain trading,play an important role. Moreover, Mali hasthousands of kilometers of virtually open fron-tiers, with ready buyers on the other side. Andfor food crops there is an alternative not avail-

16. This is to say that marketing activities that benefitf rom significant economies of scale in the cir-cumstances of semi-arid West Africa are substan-tially outweighed by activities characterized bythe quick onset of scale diseconomies.

159

able for many export crops: increased storageor consumption.

Complications related to price policy arise ina monopoly situation. In a year of bad rainsand short crops, cereal prices in intravillagetransactions wil l tend to be higher than the stateagency's price, as they wil l be in any permittedgrain transactions. (Prices may also be higher inneighboring countries.) Farmers may, there-fore, prefer: (1) to store more grain or (2) to sellonly at higher than official prices (i.e., in a parallel market). It would presumably be neces-sary to extend the system of quotas, requiringdelivery of all marketed grain to the state atofficial prices, with the implications of cost andcoercion implicit in all such policies. Rigidcontrols over grain shipments would be re-quired.

Imposition of an effective monopoly wouldrequire an agent of primary marketing to re-place the traders who presently handle two-thirds of the marketed crop in most years. Theprime candidates for primary marketing wouldbe the cooperatives. But, as already stated,these are virtually nonexistent in Mali, as au-tonomous producer organizations in any event.They lack structure, money, trained staffwarehouses, and transport. To give such em-bryonic organizations the task of primary mar-keting of food grains would overwhelm them.17

Because of the weakness of the cooperatives,proposals for fuller monopolization of the graintrade in Mali frequently recommend use of theoperations de developpement, the regional de-velopment agencies that provide extension andother assistance to farmers on given crops ingiven regions of the country — the Operation Arachides et Cultures Vivrieres (OACV), forexample, or the Operation Mils-Mopti.

The regional development organizationshave the technical capacity and administrativestructure that is frequently lacking among the

17. A recent FAO paper (1976) comments as follows. . . "cooperatives as forces for the encourage-ment of rural development, whi le offering suchpromise, tend to disappoint and too much mustnot be expected of them. Given good manage-ment and a clearly def ined and l imi ted role theycan per form w e l l . . . When they are allocated toomany responsibi l i t ies they tend to sink beneaththe bu rden . "

cooperative organizations. They also have mar-keting experience, since many of them areresponsible for marketing the cash crops thatare their major focus. Frequently, they have a large number of buying points and substantialtransport capacity. They are therefore possiblecandidates for taking on the task of primarymarketing. Were they to do so, however, severeproblems could be anticipated.

The development agencies have shownthemselves capable of attracting considerableresources from aid donors, and this explainstheir relatively strong administrative and fi-nancial position. Their success in attracting as-sistance has also allowed them to operate withrelative autonomy. It has allowed them to at-tract and hold capable staff, by payment ofbetter remuneration (especially fringe benefits)than is available in the Civil Service. The accessto technical assistance in some cases has alsohelped. Also, access to nonbudgetary fundinghas allowed these agencies financial flexibilitybeyond what is usual in the public sector.

Despite their strengths, then, the operations are vulnerable. It remains unclear how they willf ind internal sources of self-financing on a long-term basis. Their work of agricultural ex-tension, combined with other activities thatnowadays make up "integrated rural develop-ment" thus has a certain urgency. The hopemust be that, whatever happens in the decadesahead, the efforts of the development agencieswil l bring about irreversible and self-sustainingchanges in agricultural practices in the directionof modernization and improved productivity.

The basic task of these agencies is the stimu-lation of production, providing the rural popu-lation wi th trained cadres, introducing newmethods, more and better education andequipment. The production task alone makesenormous demands on available resources inmoney and manpower. It might be imprudent tofurther burden these agencies with the market-ing function. This is especially so since pastexperience attests to the existence of price-making priorities which strongly tend to favorlow consumer prices. A policy of low consumerprices for foodgrains, combined with high costsof marketing, can be expected to put pressureon the primary marketing agents, who can f indthat marketing of foodgrains is costing themmore than they are being paid by the otheragents further along in the distribution chain.

160

This has been the experience in Operation Mils-Mopti. The OMM buys the grain fromfarmers and transports it to OPAM storagedepots in arrondissement-level chefs-lieux. Berthe and Meyer-Ruhie (1977) indicate thekinds of diversions and problems that havearisen:

OPAM is supposed to receive the cereals at thechef-lieux of the arrondissements, OMM beingresponsible for the transport from the collectionpoints to the chef-lieux. However, since OPAMdoes not have available agents and storagefacilities in most of those places sufficient(owned or contracted) transport capacity the ruleis that OPAM receives the merchandise only inthe capitals of the cercles, thus increasing thetransport volume of OMM.

There are major difficulties in handling thegrain after delivery by the farmers:

• insufficient transport facilities to meet theincreased transport requirements,

• no storage facilitles at the collection points,• delays in discharge and weighing at the

OPAM reception points in the peak season(at the end of the campaign).

These shortcomings are causing losses of grainbecause of open-air storage with serious dam-ages when evaluation is delayed beyond thestart of the rainy season. Last year, OMM was involved in grain transport until the month of August which considerably affected their exten-sion activities for the new season. However, since damages and losses are at the risk of OMM until the grain is handed over to OPAM, evacua-tion of the cereals constitutes a priority activity to the Operation. (Italics mine.)

OMM tries to recuperate sacks and moneyfrom farmers who have not delivered the envis-aged quantities. Sometimes one or both of themcannot be recuperated. Figures of those losseswere not available.

This suggests a final consideration: the impacton farmers. Grain marketing is full of uncertain-ties. Transfers of funds to finance crop purch-ases can be delayed or reduced. Transport,storage, advances of credit for purchase ofinputs — all can create the kinds of pressuresindicated in the OMM case. Good rains cancreate a large disposable surplus which cannotbe marketed at announced official prices ortransported and stored appropriately. Bad har-vests wil l unloose producer pressures to sell onparallel markets for higher than official prices.Unless an effective price policy is introduced,these price problems wil l be inevitable.

The involvement of the development agen-cies in primary cereals marketing thus setsloose a whole array of potential conflict — or atleast adversary relationships — between thedevelopment organization and its clients, therural producers. There are high risks that result-ing distrust could affect peasant attitudes to-ward the development agency, its personnel,and its efforts to increase production.

For all these reasons, it would seem highlyundesirable to give the Operations ful l , or evenmajor, responsibility for grain marketing. Butother alternatives are either not feasible orscarcely more desirable. The absence of a suitable substitute for the private trader inprimary marketing is a major constraint to statemonopolization.

The present mixed arrangements, then, aredisadvantageous from the point of view of botheconomic growth and social equity. They alsohave many elements of built-in instability. Atthe same time, introduction of more extensivestate monopoly is inappropriate to the structureof grain marketing and certain to bring on heavysocial, economic, and administrative dislo-cations. The conclusion would seem inescapa-ble: private trade in grain should be legalizedand private traders allowed, even encouraged,to perform marketing services. The trouble, ofcourse, is that this simple yet compelling con-clusion conflicts with deeply held ideologicalconvictions

In Mali, as in so much of the wor ld, thereexists among urban people, civil servants andintellectuals, a widely shared vision or model offarmer behavior and rural market perfor-mance. The majority of farmers are believed tohave intense demands for money income atharvest t ime to pay taxes and debts, meet thecosts of marriage, and postharvest ceremoniesand celebrations. They meet these demands byselling part of their grain crop. But they sell it inthe immediate postharvest period when pricesare at their lowest. They buy back grain later inthe year when prices are at their peak. Farmersare believed to be widely indebted to traders,who demand repayment at harvest t ime, payingthe farmer extremely low prices — i.e., exactinga very high real rate of interest.

All of this occurs, according to this view, inrural grain markets where the winds of compe-tition are notably absent. Traders conspire tokeep buying prices low. They easily and invari-

161

ably exploit the peasant, who is seen in thisview as isolated, lacking information, withoutalternatives and denuded of means to resist.

The belief that grain markets work this way isextremely widespread in Mal i , as elsewhere.18

However, there are very few empirical studiesthat confirm this belief. The number of carefulstudies of the structure and functioning of grainmarkets is extraordinarily small, even in placeslike India where these questions have been thesubject of heated controversy for decades (Lele1971, Spinks 1970). Ruttan's comment (1969,p 83) — referring to South Asia a decadeago — is still applicable generally:

Much of what passes as analysis in themarketing literature represents little morethan a repetition of the conventional wis-dom regarding middlemen behavior withlittle or no empirical con ten t . . .

This sparsity of empirical studies supportingthe model of the trader-entrapped peasant andthe monopsonized market is certainly true ofMali. To my knowledge, there are no studiesthat discuss more than casually the functioningof Malian grain markets. The CEGOS study, onunclear evidence, claimed that in 50% of thevillages surveyed, some part of marketed cere-als output was sold sur pied (on the stalk). Butthey also assert that most of the sales that tookplace in the villages came f rom stocks sold off

18. The Uni ted Nat ions, Economic Commiss ion forAsia and the Pacific (1975, p 2) reports "Thestereotype of indigenous marketing systems forthe smal l farmer is that it is exploi t ive, col lusive,economical ly inefficient and operating w i th highprofi t margins for the trader. At the bo t tom is thesmal l farmer, poor, often il l iterate and unor-ganized, whose smal l vo lume of business is ofpoor qual i ty, unstandardized, costly to handleand relatively un impor tant to the trader. Thegeneral poverty of the smal l farmers and theirchronic indebtedness to money lenders, w h o areoften the traders w h o buy their produce, weakenthe farmer's bargaining power , especially at har-vest t ime. This weakness is aggravated by thefarmers ' lack of knowledge about prices andalternative market ing p rocedu res . . . The inhe-rent weakness of the small farmer means that heis an easy target for exploitat ion — under-we igh ing or under-assessment of the produce,charging h igh interest rates, etc."

as the harvest approached.19

It is no exaggeration to say that the "monop-sonized market" model is at best unproved, atworst pure myth. Nor is it unfair to observe thatit is a priori dubious, given the structural charac-teristics of grain markets in the country.

There exists an alternative model, a differentway to see farmer behavior and market per-formance. The "average" peasant, in this view,reflects in his behavior hundreds of years ofcultural experience and social adaptation. Heplants as much grain as he wil l need to feed hishousehold on the assumption of normal rains,wi th some safety margin. He maintains, at thevillage or household level, a storage capacityequal to at least 1 year's consumption, andperhaps 2 years, in order to protect himselfagainst the bad rains he knows wil l comeperiodically. He knows very well that he wil lneed cash income at the t ime of the harvest. Heprepares for it during the dry season — bymigrating or engaging in some local income-earning activity. He prepares for it also in hisproduction decisions — by growing cash crops,for example. He is perfectly aware that grainprices wil l be lowest at harvest t ime and highestduring the soudure, and tries to arrange hispurchase, sale, and storage decisions accord-ingly.

According to this way of looking at things,similarly, the grain market is characterized byreasonably effective competit ion. Entry is easy.Anyone can become a petty trader; little isrequired in terms of capital or skill. Since in-comes available in other rural occupations arerelatively low, the elasticity of supply of traders'services is surely-very high. Even the mostisolated farmer need not sell his grain at anunsatisfactory price; all he has to do is journeyto the nearest periodic market to sell it there,either selling it himself or giving it to a smalltrader to be sold. In the rural markets, there are

19. See IDET/CEGOS (1976, Vol 3). Panhuys ment ionsa Mal ian trading circuit of what he calls a "usur i -o u s " type,.on wh ich debts are re imbursed in k indat harvest t ime. Typical ly, he notes, 1000 MF werebor rowed, and 100 kg of mil let were given asrepayment at harvest t ime. (This refers to theearly 1970s.) He estimated that perhaps 5000tonnes entered the market this way — less than5% of total market mi l let /sorghum.

162

always passers-by who act as casualtraders — civil servants, bus drivers, teachers,others — who are anxious to fill empty cargospace wi th grain to sell in the larger towns andwhose presence gives a strong presumption ofcompetition on the buying side, a presumptionthat would exist anyway because of the largenumber of traders in the market.

This second model of a calculating peasantand a competitive market is, of course, conge-nial to the preconceptions of many economists.But it is more than that. It also seems to fit wellwi th what has been discovered by a number ofrecent studies in countries wi th socioeconomicand ecological structures similar to Mali. But,in the present context, the point is not the truthor falsity of these ideas; it is rather that the"exploited peasant/collusive market" model isthe way most policy-makers in Mali see thewor ld, and it creates the unwillingness to movethe marketing system more openly and morefully into reliance on private trade.

S u g g e s t e d R e f o r m s

The policy analyses done for and/or by theMalian Government contain numerous propos-als for marketing reform. All of these, however,had serious ambiguit ies or f laws that havereduced their relevance and suitability. One setof proposals suffers from narrowness of focus;these can be called the " improvement" propos-als. A second set of proposals for fuller statemonopolization leaves unanswered many fun-damental questions, which render it inapplica-ble. The most recent reform proposal, involvinggreater liberalization, is more fully developedthan previous analyses, but also has question-able aspects.

" Improvements"

The so-called improvements aim at improvingpresent arrangements without raising funda-mental questions about the allocation of mar-keting responsibility between state and privatesectors or the scope for price policy. An un-friendly critic might call these "band-aid" pro-posals. They focus on specific and immediateproblem areas, not the system as a whole.

One example is the 1973 FAO Report (thePanhuys Report). Although this report is quite

far-reaching in its discussion of the marketingsystem and its deficiencies, and generally per-ceptive in its diagnoses, the principal proposedchange is that official grain prices be raised by a specific amount.

A more striking example is the 1975 report ofthe BDPA (Bureau de Developpement de laProduction Agricole) done under the auspicesof the French Ministry of Cooperation. Thisdocument analyzes the structure and function-ing of OPAM and concentrates on internaladministrative factors that contribute to its in-efficiency. Although there is considerable dis-cussion of OPAM's "relationship to its externalenvironment," no questions are raised aboutthe terms of coexistence of the private sector orabout the nature and limits of price policies. Thedrafters of this report explicitly reject structuralchanges relating to OPAM's role (BDPA, 1975,P 68):

It should be noted that some people wetalked to urged that OPAM's role be morelimited, more oriented toward imports,storage and the supply of towns andgrain-deficit regions, leaving buying re-sponsibility to the operations in mil let,maize, rice, groundnuts and cotton andsales to other, better-adapted organi-zations;In addition to the fact that this proposalwould lead to a considerable reduction in(OPAM's) monopoly status, it is not clearthat overall efficiency would be improvedby allocating (grain purchase transport/storage and sa le ) . . . to three different in-stitutions.

Fuller State Monopolization

The second category of reform proposals recog-nizes the elements of basic instability nowexisting. The proposals address this problemdirectly, by suggesting a strengthening of thestate monopoly. This was the principal recom-mendation of the IDET/CEGOS (1976) study,which was financed by the World Bank andwhich was intended to be a comprehensive ifnot definitive study of the problem.

The CEGOS Report provides much usefulmaterial on how the marketing system worksand is unsparing in describing its deficiencies.The report makes recommendations on a broadfront. It urges that producer prices be increased,

163

that quality grading be introduced, that market-ing margins be increased and the baremes bemade more realistic, that peasants be allowedto repurchase grain f rom OPAM at a price closerto the producer price. A series of recommen-dations are also devoted to improvement ofOPAM's administrative and financial situation:to suggestions that OPAM's debts be lightened,that it be given adequate working capital, thatthe banking system release credit for croppurchases earlier and more flexibly, thatOPAM's grain storage capacity be increased,and that a drought reserve stock of 60 000tonnes be established.

Most of these recommendations, and othersnot mentioned, were addressed to real prob-lems and are generally acceptable. It is themajor structural proposals that raise questions.

First of all, the report recommends exclusionof private traders from the primary marketingand the wholesaling of foodgrains; the traderswould be restricted to retailing. No real justifica-t ion is presented. In fact, the report gives l imitedattention to the question of market structure. Itasserts in passing, that grain markets aremonopsonized. Thus, in discussing farmer re-venues from grain production, it states:20

. . . when crops are short, (farmer) incomes arecut due to the shrinkage of marketed volume... ;the rise of prices on the parallel market does notre-establish the level of income from food cropsbecause only a small part of the increased retailprice is transmitted to farmers, because of ag-reements among traders.

The considerable problems implicit in thisproposal are very summarily treated. The roleof primary marketing is given to the FGRs (the"cooperatives"), whose nonoperational charac-ter the report elsewhere recognizes. The FGRslack staff, structure, experience, resources,even autonomy f rom the administration. Yetthe CEGOS report proposes they be giventrucks and other equipment and a full monopolyover purchase of the grain crop. The privatetransport of grain f rom surplus to deficit regionswould be prohibited by this proposal, but only a line or two is devoted to the administrative

20. It shou ld be noted that no reference to such priceagreements is f ound elsewhere in this report, norare any other studies cited as evidence.

implications and none to the economic implica-tions. Finally, even the restricted role left to theprivate sector is, it turns out, quite untenable.This extraordinary admission is worth citing inful l :

The control of the legalized retail trade wouldconsist essentially of making sure that traderssell at official prices and that they sell "OPAMbrand" grain. The system would only permit theescape outside the OPAM circuit of a little grain,and this only on local markets. The retail shopswould buy their grain from OPAM. They wouldhave no interest in selling below the official price.But they would have an interest in buying belowOPAM's wholesale price (prix de retrocession). The possibility would be limited by putting a brand name on the commodities.This brand name matter raises delicate prob-lems. It seems that the short-term solution wouldbe "micro-packaging" (micro-conditionnement),for example, packaging in 5- or 10-kg bags. Butthe costs of this operation, which would have tobe done all over the country in order to avoidraising transport costs by making grain ship-ments obligatory, is high, at the same time thatresults are uncertain because of the difficulties ofcontrol.We are forced to recognize that there's a problemhere: it will only be overcome when innovativeactions will make OPAM's services superior tothose of traditional trade...

Other basic recommendations in the CEGOSReport have major inadequacies or inconsis-tencies, though perhaps none so fundamentalas the one just described.

The report recommends a rise in producerprices for foodgrains, but the rationale for thisincrease is nowhere explained or defended. Infact the main thrust of argument is that a pricerise is unnecessary (because marketed output isnot price-responsive) and undesirable (sincemarketed output may come mainly from largerfarms).

The report proposes the purchase at a fixedproducer price of a given quota of grain, withthe rest being bought at prices determined byharvest size. The incentive and equity effects ofsuch a system are unclear, and are not muchdiscussed in the report.

The report proposes to subsidize farm-levelstorage, while making access to FGR grainstores cheaper and easier.

The CEGOS proposals were adopted by theMali Government, which created a commission

164

to make recommendations on implementation.The report of this commission supports thebasic CEGOS proposal to strengthen the OPAMgrain-marketing monopoly (Republique du Mali1976). It introduces a complex system of ar-rangements, however. In years of good har-vests, private traders could act legally at thewholesale level. When harvests are moderateor short, they would be prohibited from doingso. Thus the commission report estimates thatin 3 years out of 5, private traders would berestricted to retail sales; no private shipment ofcereals of more than 1 tonne would be allowedbeyond the arrondissement level.

The enforcement of this prohibition of grainshipment presents mind-boggling administra-tive difficulties in West African conditions. TheIPGP commission proposal required control ofall grain movements exceeding 1 tonne — i.e.,10 sacks. But thousands of vehicles carry suchquantities of grain in Mali. Every cart, everyprivate vehicle, every bus, would be a potentialviolator of grain shipment prohibitions. UnlessOPAM were to use only its own trucks, everyprivate transporter would have to be givenappropriate papers, with specification of cargoand destination. Multiple check points wouldpresumably be required. Some surveillanceand control might be required at the localperiodic markets, of which there are hundreds.Opportunities for administrative abuse and cor-ruption would be abundant.

The economic disadvantages of systems ofshipment control are well-known. To the extentthat these controls are effective, they raiseprices in deficit regions and reduce them insurplus regions. They discourage productionand marketings in the more productive zonesand encourage local self-sufficiency rather thanregional specialization. They raise prices gene-rally by substantially increasing transactioncosts for illicit grain movements, to the extentthat these continue.

Liberalization

These recommendations to enforce the state'slegal monopoly have not been adopted by theMali Government. And since 1976 a third set ofproposals has come forward that recommendmovement in the opposite direction: toward a greater role for private trade and a dilution ofOPAM's monopoly status.

One of these proposals was not specific toMali, but arose in the context of the Club duSahel (CILSS/Club du Sahel 1976). The GrainMarketing, Price Policy and Storage WorkingGroup of the Club commissioned a "diagnosticstudy" in 1976, which was done by the Centerfor Research on Economic Development (CRED)of the University of Michigan. An early draft ofthe report of that study argued that basic im-provement in the grain marketing situation inMali (as in Niger and Upper Volta) required thatthe private trade be legally recognized andprivate traders encouraged. This recommenda-tion was hotly debated at a meeting of the ClubWorking Group in early 1977 and was resound-ingly rejected by that group, which consisted ofrepresentatives of the Sahel governments(many of them officials of state grain marketingmonopolies) and several donor agencyspokesmen. The final report made no recom-mendations; it defined a set of options, includ-ing liberalization, and assessed advantages andcosts.

The most recent proposals come from a multilateral mission composed of representa-tives of five bilateral donors under the sponsor-ship of FAO (1978). This mission report firststresses the gravity of the present situation ingrain marketing and the urgency of the need forchange. The "restructuring" of the cereals sec-tor, it states, has become " imperat ive." Threereasons are given: the "excessive" annualbudget deficits to which OPAM's "exorbitantindebtedness" makes a major contribution; thepresent system's failure to encourage produc-tion, "which stag nates despite a vast potential,"and its parallel failure to supply consumers withmoderately priced foodgrains; the big differen-tial in grain prices between Mali and herneighbors, which encourages smuggling sothat even in a normal year (1977-78), the coun-try had to call on food aid and commercialimports amounting to some 60 000 tonnes (FAO1978, pp 8-9).

The report rejects at the outset the possibilityof an effective state monopoly, as suppressionof private trade. Traders are said to perform"indispensable functions they will alwaysexist because they have a useful and Importantfunction which cannot be otherwise per-formed." The monopoly option is "unrealistic"(FAO 1978, p 6).

Three "practical" options are set out. The first

165

is to give OPAM greater resources, al lowing it toraise producer prices and also raise its market-ing margins, so as to reduce its deficits. Retailprices would rise appreciably. While this optionis said to have numerous advantages (reducedOPAM deficits, reduced smuggling) it is re-jected because it would require increased pub-lic expenditure, compensating wage increasesand hence increased budget strains. It wouldalso leave intact the double market system,which OPAM would be unable to dominate.

The second option is to turn the entire domes-tic grain market over to the private traders (whoalready have three-quarters of it anyway).OPAM would retain a monopoly of imports andexports and would manage food securitystocks. The report lists many advantages forthis opt ion; it displays, in fact, a restrainedlyricism about the results.

The advantages of this option are first its simplici-ty and its economy. It costs the state l i tt le, sincesecurity stocks w o u l d continue to be externally-f inanced. OPAM's debts and structural losseswou ld vanish. The right prices (la verite des prix)wou ld prevai l , and the role of private tradersw o u l d be recognized. The double market w o u l ddisappear. Smuggl ing wou ld be discouraged inyears of bad harvests, once local prices wou ldrise. The supply ing of towns and deficit zones incase of scarcity is assured by the state th rough itssecurity stocks (FAO 1978, p 10).

But, says the report, the principal social objec-tives are not attained by this option: " . . . thereis no guarantee that the cities wil l be properlysupplied other than in periods of emergency,nor that retail prices will be affordable. Nor isthere any guarantee that producers wil l receiveremunerative prices in periods of abun-dance . . . "

For these reasons the mission proposes theadoption of a third option. Private trade wil l belegalized (traders wil l be licensed), and wil lcompete in primary marketing wi th the Opera-tions, and, to the extent possible, wi th thecooperatives. OPAM wil l control the market bybuffer stock operations around a ceiling (retail)and a f loor (producer) price. The floor andceiling prices would be freed " in principle" forseveral years, but the report is vague on thispoint; it also calls for annual reconsideration ofprices. The report also proposes that the floorand ceiling prices be regionalized — set higherin deficit regions, lower in surplus areas.

166

There is little doubt that this multilateraldonor/FAO report addresses fundamental is-sues and puts forward more operational andmore promising proposals than the earlier re-ports. The private sector is legalized and given a central role. The proposals for market controlstress indirect measures — use of the marketmechanism via purchases and sales from bufferstocks, rather than direct administrative mea-sures. The report deals wi th broad strategy, notdetail, and proposes a more deliberate, in-depthset of studies to guide implementation. Thereport comes with a stamp of approval, in somesense, of the five most concerned bilateraldonors, so its impact and audience may belarger than is normally the case wi th suchdocuments.

The proposals nonetheless present certaindifficulties that threaten their viability. Themajor one is price policy — in particular thescheme for interannual price stabilization.

Certain elements in the physical andeconomic environment in Mali put constraintson any price policy that stimulates supply above"no rma l " or "market-determined" levels. Thisis the case wi th policies that attempt to sethigher-than-market prices for foodgrains, as itis for policies that aim at stimulating productionby providing stable prices to producers. First,grain production is highly variable f rom year toyear because of wide annual variations in rain-fall and the close dependence of harvest size onthe volume and distribution of rainfall (IBRD1976, p 25). Secondly, marketed grainoutput — the "disposable surplus" — varies byeven more than total output. A good harvestincreases the surplus available for sale by a multiple of the increase in total production.21

Thirdly, the price elasticity of demand forfoodgrains is probably relatively low — i.e.,consumption does not increase by as much as

2 1 . If product ion is 1 mi l l ion tonnes, of wh ich 150 000are marketed, a rise in product ion to 1.1 mi l l iontonnes wil l increase the saleable surplus by asmuch as 67%. Of course, some of the increase inproduct ion wi l l be consumed by the producers,either because they former ly produced too l i t t leto meet their own needs or because they want toconsume more grain as part of the generalincrease in Income impl ied by the larger harvest.But It Is probable that most of the increase wi l l besold.

price falls, nor fall by as much as price rises. Thismeans that changes in marketed output tend tolead to sharp inverse changes in grain prices.Finally, in normal years, domestic productionalmost entirely satisfies domestic demand — i.e., foodgrain imports are marginal. Oppor-

tunities for import substitution are small, sincewheat is the only significant grain import and isimported in relatively small quantities.

Assuming that grain production, and espe-cially marketed supply, is responsive to price, a positive price policy — one that sets producerprices higher than market-determined levels, ormaintains a stability of prices that stimulatesoutput — wil l generate "excess supply." Thequestion of its disposal wil l have to be faced.There are four main possibilities: the grain canbe sold to consumers at a subsidized price; itcan be stored for stabilization or emergencyreserve purposes; it can be used in newways — as feed for cattle or poultry, for exam-ple; or it can be exported.

The first possible use of a price-induced"excess" cereals supply, sales on the domesticmarket, founders on the troublesome barrier ofsluggish demand. Given the low price elasticityof demand for millet and sorghum — the basicgrains — the consumer price would have to fallvery low in order for consumers to absorbsignificant quantities of additional output.However, this implies subsidization if the "posi-t ive" producer price policy is to be maintained,and subsidization raises questions aboutequity, incentive effects, and macro-economicconsequences for which there are no obviousanswers in the present state of knowledge. Thequestion of who benefits and who pays, forexample, depends on the socioeconomic posi-tion of those who buy grain and those who sell.If it is true, as sometimes claimed, that the bulkof marketed millet output comes from a rela-tively small number of larger farmers, whilepastoralists and deficit peasants in poor regionsbuy much of the marketed millet, then what isinvolved is a transfer of income from relativelypoor to relatively better off groups.22

22. There exists no study of this question in Mal i , norindeed are there many studies of it elsewhere.The IDET/CEGOS Report suggests that better-offfarmers are responsible for much of marketedoutput.

On the incentive side, the higher food grainprice may lead not to an expansion of aggregateagricultural production but to a change in thecrop mix, wi th millet and sorghum being substi-tuted for cotton and/or groundnuts. Themacro-economic effects of this change may notbe positive — e.g., national income wil l proba-bly be lower and the balance of payments lessfavorable as a result. There are also budgetaryimplications, with revenues declining and ex-penditures rising as a result of the grain priceand subsidy policy.

If the "excess" grain that is purchased oneyear can be stored and then sold the fol lowingyear when the harvest is poor, the problem wil lbe reduced but not eliminated. If producerprices are set above the long-run averagemarket-determined price, there remains theproblem of what to do with the induced increasein grain supply in the face of highly inelasticdemand. If price stability as such providesincentives to higher production, there wil l bemore output than otherwise. Moreover, it is notpossible to postulate a nicely oscillating cycle ofgood years and bad. It is highly likely that therewil l be a number of successive good yearsbefore a bad one. In the absence of exportmarkets, the accumulation of grain stocks is themost likely consequence. The costs of thisstorage, including costs of losses through infes-tation, spoilage, and quality deterioration, arelikely to be considerable.

Surplus foodgrains could conceivably f indother uses, particularly given the relatively lowprices at which they might be offered for non-food consumption. The most obvious and fre-quently mentioned possibility is in the area offeedgrain for poultry and cattle. This is certainlyan interesting and important possibility, but isnot clear that the potential feedgrain market isat present capable of absorbing more than a relatively small volume of grain.

There remains the export possibility, which isreal. The recent West African Rice Study doneby the Stanford Food Research Institute inconjunction with WARDA, indicates that Malican export rice to the Ivory Coast at competitiveand socially profitable prices, assuming worldrice prices behave as now projected. TheSenegalese rice market, dominated by importsof much cheaper Asian brokens, is moredifficult to penetrate. On the other hand, Malimay be able to export millet/sorghum to

167

Senegal and other neighboring countries.23

While export potentials certainly exist, thereremain many uncertainties, and in all casesmany obstacles to be overcome, particularly intransport and in marketing. The export pos-sibilities for millet and sorghum seem lesspromising than for rice mainly because exportof surpluses would be avoidable mainly ingood years, when external demand is low. Itwould be imprudent to base a general food-grains price policy on export potentials withoutknowing more about the nature and dimen-sions of these potentials.

We have thus far concentrated on price levels,not stabilization, which the FAO Report stres-ses. Many of the above considerations apply tothe proposal to stabilize grain prices interannu-ally. There are other reasons why interannualprice stabilization is difficult, risky, and expen-sive.

First, as noted earlier, marketed supply is a small proportion of total production and issubject to especially large changes as totalproduction changes. Thus interannual stabili-zation efforts will require considerable storagecapacity relative to the total size and value ofmarketed output. An interannual stabilizationeffort is, in this sense, likely to be expensive.

Second, interannual grain stabilization hascertain inherent contradictions. The presumedprimary purpose of attempting to stabilizeprices is to reduce the producer's uncertainty;fluctuations in price are believed to deter effortsto expand production. However, a price stabili-zation scheme that effectively reduces uncer-tainties could be sustained only at great cost, forit implies a support price that would be main-tained over a period of years, regardless ofharvest size. Such an inflexible support pricewould mean large storage costs should there bea succession of good harvests. Moreover, if thereduction of farmer uncertainty leads to in-creased grain production and marketings, thistoo would have to be bought up by the stabili-zation agency in order to maintain the supportprice. On the other hand, if the grain agencyreduces its support price as annual harvests and

23. According to Michal lof 's (1977, p 26) estimate,Malian so rghum can be del ivered in the interior ofSenegal, more cheaply than U.S. sorghum (1976prices). The reason is that t ransport costs toKaolack f r om Dakar are substantial.

market conditions change, it undermines itsprimary objective: reduction of producer uncer-tainty by reduction of price fluctuation.

Third, price stabilization schemes can bedestabilizing if the stocks held by the stabiliza-t ion agency are not large. If the harvest is bad,traders may recognize that sales from the bufferstock wil l be inadequate to maintain the officialprice ceiling. They will have nothing to lose byhoarding grain while they await the inevitablerise in price when the buffer stock is exhausted.

Finally, the presumed incentive effects ofstable prices remain unproved. One can arguethat what Malian farmers need and want is anassured market, better marketing services, re-duced exposure to arbitrary action and the rightto sell grain at market prices. These kinds ofchanges would have a greater effect in stimulat-ing production and marketed supply thanwould any price stabilization scheme.

Given the high costs, high risks of failure, andlimited benefits attached to interannual grainprice stabilization schemes, it is unclear whysuch a scheme is proposed in the FAO Report. Ifthe authors believed in the importance andefficacy of such arrangements, they do notdemonstrate it in the report. An intra-annualprice stabilization scheme would have beentactically preferable. Such schemes, while notwithout problems, are easier to implement, lessrisky and less costly. They might also havesome genuine social benefits. If distress salesare common (peasants forced by indebtednessto sell at low prices immediately after theharvest), an effective intra-annual stabilizationscheme could help. In general, intra-annualprice stabilization is more finely targeted interms of reducing possible exploitation ofpeasants by traders due to imperfect marketsand unequal bargaining power.

Despite its dubious recommendation of in-terannual price stabilization, the FAO Reportprovides a workable basis for meaningful re-form, unlike many other reform proposals thathave been made in Mali in the past few years. Ifthe proposed ceiling (retail) price can be kepthigh enough, and the floor (producer) price lowenough, the proposed fourchette idea mayprove an ingenious device for giving freer playto market prices. What remains highly prob-lematic is whether in t ime of short harvests, theMali government wil l al low official consumerprices to rise to the market-determined level,

168

and whether it will allow OPAM to pay pro-ducers the market-indicated prices.

External Support

Foreign assistance reduces the economic costsof maintaining the existing marketing arrange-ments, and in this sense impedes reform. Foodaid, though of course necessary in t ime of crisisreduces the impact of inadequate policy. Morespecifically, inflows of aid in the form of grainthat is sold by OPAM, facilitates the generationof working capital for OPAM and thereby re-duces the stresses of financial deficits causedby price policies and marketing/storage in-efficiencies. Also, the existence of the line ofcredit available to Mali in the Operations Ac-count in Paris has generally diluted the negativeimpacts of grain marketing policy — as it hasdone with other economic policies — allowingthe Mali Government to maintain ineffectivepolicies without having to fully suffer the con-sequences.

Conclusions

I have tried in this paper to outline the mainproblems of grain marketing policy in Mali andto indicate the obstacles to improvements inthat policy. Mali is the focus of discussionbecause it is well-documented and because it isanalytically simpler to discuss one country thanthe region as a whole. Much of what has beensaid is applicable elsewhere in semi-arid tropi-cal West Africa.

Of the many general conclusions that emergefrom this analysis, six seem worth special em-phasis.

1. It is clear that the West African environ-ment imposes numerous and severe con-straints on public policy options ingrain market-ing. Total grain output is highly dependent onrainfall, hence it is variable. Marketed output isonly a small share of the total and is potentiallyvolatile in volume. The country is large inphysical size; grain production is spread overvast areas; there are few specialized graintraders; traditional marketing is small in scaleand dispersed in space. Frontiers are numerousand highly porous; traditional trade has alwaysignored them. The financial environment issimilarly constraining. Budgets are in deficit.

Expenditures are mainly on salaries, whichmoreover are under continual erosion in realterms. Little public money is available for ma-terials, supplies, and maintenance. The budgetprocess is at once rigid and uncertain. Intrapub-lic sector accounts are in frequent disarray. Ontop of this, there is the limited availability oftrained people, scarcities of organizational in-puts, and particular scarcities of coordinationcapacities. These physical, financial, and or-ganizational factors put down clear limits onwhat the Mali government can effectively im-plement, either in terms of price policies ordirect state trading operations.

2. The present mixed system of marketingcannot be easily improved. Stable coexistencebetween an OPAM operating along its presentlines and the private marketing sector requireseffective price stabilization or abandonment ofsignificant price support objectives, introduc-tion of a much more refined price structure,abandonment of uniform nationwide pricing,and harmonization of prices with neighboringcountries. And even if all this were done, mostof the flaws and inequities of the present systemwould still be present.

3. Grain-marketing policies have probablyhad significant negative impact on grain pro-duction and the attainment of the priority objec-tive of food self-sufficiency. Uncertainty overprices and the availability of marketing services,and general market disorganization, have prob-ably led to diversion of farmer effort into cashcrops where this has been possible and mayhave reduced farmer willingness to undertakegreater efforts, or new ventures, in the grainproduction area. These effects have not been oflarger magnitude, or at least more apparent,because of the partial nature of OPAM'smonopoly — i.e., the fact that government hasnot been able to fully implement its policies.The substantial availability of food aid duringthe last decade has also been significant indiluting the effects of policy, as has the avail-ability of credit from the French OperationsAccount, which has indirectly financed OPAM'sdeficits by underwriting Mali's ongoing budgetand balance of payments deficits.

4. The longer-term impact of the presentmarketing policies may be of greater sig-nificance than the impact on output in the shortrun. As things now stand, the development oftrue cooperatives is extremely difficult. The

169

existing cooperative organizations are instru-ments of government, used mainly for grainrequisition purposes. So long as the requisitionsystem endures, farmers wil l hesitate to set uporganizations for defense of their interests (inthe areas of storage, credit, and crop sales forexample). So the democratic development ofthe countryside is impeded. Similarly, the de-velopment of trading competence, techniquesand capital is severely obstructed, since traderspresently operate with strong discouragementor at best, uncertainty. Finally, to the extent thatmarketing functions are transferred to the oper-ations de developpement, the production-oriented activities of those organizations wil l becompromised. These operations are highly de-pendent on foreign financing, which cannotcontinue forever. They thus have a moment inhistory, so to speak, to help Malian farmersincrease their productivity. Diversion from thiseffort, or weakening its effectiveness, couldtherefore have substantial effects on long-termdevelopment.

5. Of the many factors that have been re-sponsible for the slow reform of the marketingsystem, two seem particularly critical. The firstis the lack of f i rm knowledge, due to a lack ofbasic studies, about how the grain marketsactually function in Mali. The sparsity of know-ledge allows continued circulation of beliefsthat may have little foundation in fact, such asbelief in monopsonistic markets. In any case, itis difficult to frame suitable policies in theabsence of better information and understand-ing about farmer decision-making, the patternof grain selling and buying, the structure andfunctioning of grain markets, on-farm storageeconomics, and the behavior of grain prices.

The second factor has been poorly articulatedreform proposals. Until very recently, the Maligovernment had not been presented wi th prop-osals that were well thought through and opera-tional. Some have been highly management-oriented, indicating how OPAM could organizeits work better. But these never addressed thebasic problem of what OPAM's work could orshould be. The most extensive reform propos-als, those of the IDET/CEGOS Report, containedloose ends of a serious nature. They nevershowed how the ful l state monopoly that wasproposed could be implemented. The MaliGovernment commission that took the CEGOSrecommendations the next step put forward an

170

extraordinarily complex proposal involvingexclusion of traders from wholesale trade dur-ing " b a d " crop years, and a draconian system ofshipment control, to prevent movement ofgrain from surplus, low-price areas to deficit,high-price areas. Present proposals for reformare more viable, though there remain questionsabout the feasibility or desirability of key re-commendations.

6. Certain policy implications should be clearfrom this discussion, although they have notbeen made explicit; it is not the purpose of thispaper to propose policies but rather to makeclear the nature of the problems and the obsta-cles to change. It is, first of all, evident that nomarketing reform can hope to succeed withoutallowing the private trading sector a fuller role.It is equally evident that, as noted above, scar-cities of manpower and other critical inputsseverely constrain what the state sector caneffectively do. But it should at the same t ime bestressed that a state grain agency has a majorrole to play, even under "minimal is t" assump-tions about the State's role in grain marketing.Such an agency would manage a grain securitystock; manage grain imports, particularly crisisimports; provide market information; and per-form market inspectorate functions. It mightbuy and sell in the open market for specialpurposes — e.g., localized production crises. Itmight operate a buffer stock for seasonal pricestabilization. It could do grain storage extensionwork, especially for new grains such as maize. Itcould provide for the needs of collective con-suming units, such as the army, prisons, etc.

It is also, equally obvious that major im-provements in grain marketing can and willcome from indirect measures; improvementand extension of feeder road networks; betterinformation on crop size, prices, marketings,etc. and more effective spread of this infor-mation; closer attention to relaxing productionconstraints in foodgrains; improved policyanalysis capacity within government; morebasic studies about how grain markets actuallywork, about farmer behavior wi th respect toproduction decisions and crop disposal, aboutgrain storage, about price behavior in marketsat different echelons of the distribution chain.More and better knowledge and changes thatindirectly improve market structure and per-formance wil l widen the options for reform andincrease the probability of adoption.

References

AGRICULTURAL SERVICES DIVISION, FAO (Food and Ag-

riculture Organization of the UN). 1976. The catalyticrole of various types of marketing enterprises inst imulat ing the expansion of local production. Paperpresented at the OECD/FAO International Seminaron Critical Issues on Food Marketing Systems inDeveloping Countries. 18-22 Oct 1976, Paris.

BERG, E. 1975. The recent evolut ion of the Sahel. AnnArbor, Mich, USA: Center for Research on EconomicDevelopment, University of Michigan.

BERTHE, M., and MEYER-RUHLE, G. O. 1977. Report on

the first joint evaluation of Operation Mi ls-Mopt i .Bamako, Mal i : U.S. Agency for International De-velopment.

BDPA (Bureau De Developpement De La ProductionAgricole). 1975. Mission de restruction de I'Office deProduits Agricoles du Mal i (OPAM). Paris: BDPA.

BDPA (Bureau De Developpement De La ProductionAgricole). 1977. Seconde mission etc restructura-t ion de I 'opam.

COMITE PERMANENT INTER-ETATS DE LUTTE CONTRE LA

SECHERESSE D A N S LE SAHEL/CLUB DU SAHEL. 1977.

Market ing, price policy and storage of food grains inthe Sahel: A survey. Vol. 1. Synthesis w i th statisticalcompi lat ion and annotated bibl iography. Vol. 2.Country studies. Ann Arbor, Mich, USA: Center forResearch on Economic Development, University ofMichigan; also Washington, DC: U.S. Agency forInternational Development.

COMITE PERMANENT INTER-ETATS DE LUTTE CONTRE LA

SECHERESSE D A N S LE SAHEL/CLUB DU SAHEL. 1978.

Etude sur le stockage des Cereales dans les pays duSahel. 2 vols.

CRED (Center for Research on Economic Develop-ment). 1976. Mal i : Agricultural sector assessment.Ann Arbor, Mich, USA: CRED, University of Michi-gan.

ESCAP (United Nations Economic and Social Com-mission for Asia and the Pacific). 1975. Problems ofmarket ing of smal l farmers in the ESCAP region.Economic Bullet in for Asia and the Pacific 26(2 -3 ) : 2.

EJIGA, N. O. O. 1977. Economic analysis of storage,distr ibut ion and consumpt ion of cowpeas in north-ern Nigeria. Ph.D. thesis, Cornell University, Ithaca,NY, USA.

FAO (Food and Agriculture Organization of the UN).1973. Rapport au Gouvernement du Mali sur leprobleme de la commercial isat ion des Cereales,etabli sur la base des travaux de H. Panhuys. Rome:FAO.

FAO (Food and Agriculture Organization of the UN).1978. La pol i t ique cerealiere au Mali . Rapport pre-

pare sur la base des travaux de la mission Inter-nationale placee sous la coordination de H. de Meel.Rome: FAO.

FAO (Food and Agriculture Organization of the UN).

1978. Report on cereals policy in Mal i . Rome: FAO.

GUGGENHEIM, H. 1977. Tradit ional and moderntechniques in grain storage and transportat ion:Problems and solutions for Operation Mi ls-Mopt i .

Report to AID/Bamako.

GILLES, P. H. 1965. Storage of cereals by farmers innorthern Nigeria. Samaru research bullet in 42.Zaria, Nigeria: Institute for Agricultural Research,Ahmadu Bello University.

GILBERT, E. 1969. Marketing of staple foods in northernNigeria: A study of the staple food marketing sys-tems serving Kano city. Ph.D. dissertation, StanfordUniversity. Stanford, Calif, USA.

HAYS, H. M. 1975. The marketing and storage of foodgrains in northern Nigeria, Samaru miscellaneouspaper 50. Zaria, Nigeria: Institute for AgriculturalResearch, Ahmadu Bello University.

IBRD [International Bank for Reconstruction and De-velopment (World Bank) ]. 1976. West Africa foodgrain study. Washington, DC: IBRD.

IDET/CEGOS. 1976. Etude des structures de prix etmecanismes de la commercial isation des mils etsorgho. 3 vols.

IPGP (Institut De Productivite Et De Gestion Pre-visionnelle). 1976. Rapport final de la commissioninterministerielle sur la restructuration de I'OPAM.

LELE, U. 1971. The marketing of food grain in India.Ithaca, NY, USA: Cornell University Press.

MICHAILOF, S. 1977. Remarques generales sur lacommercialisation et la polit ique du prix des cere-ales au Mal i , Bamako, Mal i : Caisse Centrale deCooperat ion Economique, Services d'EtudesEconomiques et Financieres.

O F F I C E D E P R O D U I T S A G R I C O L E S D U M A L I /

AGROPROGRESS. 1976. Etude des problemes de lacommercialisation du mil/sorgho pendant la cam-pagne 1976/77 dans les regions de Sikasso, Segou etMopt i . W. Sachers.

OFFICE D E PRODUITS A G R I C O L E S D U M A L I /

AGROPROGRESS. 1976. Implantation opt imale desstocks. W. Sachers. 49 pp.

RICHARD, PH., and VAN DE BERG, X. 1976. Mission de

restructuration de I'OPAM. Paris: Bureau de De-veloppement des Produits Agricoles/Ministere de laCooperation.

RICHARD. PH., and HERPIN, J. 1977. Rapport de mission

relatif a la restructuration de I'OPAM et a I'organisa-t ion du marche des cereales. 2 vols. Paris: Bureau deDeveloppement de la Production Agricole (BDPA).

RUTTAN, V. 1969. Agricultural product and factormarkets in Southeast Asia. In Agricultural coopera-

171

t ives and markets in developing countries, eds. D. K.R. Anschel , R. H. Brannon, and E. D. Smi th . NewYork: Praeger.

SPINKS, G. R. 1970. Myths about agricultural market-ing. Month ly Bulletin of agricultural Economics andStatistics 19(1).

TEMENTAO, S. 1977. Analyse du systeme de commer-cialisation des cereales au Mali . Memoire , EcoleNationale d 'Administ rat ion, Bamako, Mal i .

172

Foodgrain Marke t ing and Agr icu l tura lDeve lopment in India

M. von Oppen, V. T. Raju, and S. L. Bapna*

Abstract

This paper examines foodgrain marketing in India and its impact on agricultural

development by presenting the approaches and results of four major areas of research in

agricultural marketing presently under way at ICRISAT. Agricultural market channels in

India are competitive and pricing efficiency in general is good. Consumer preferences

(both evident attributes and cryptic characteristics) for sorghum and pearl millet have a

measurable effect on market prices, and a quality preference index can be derived from

such preference measures for selecting "good quality" grain. ICRISAT's foodgrains in

India are primarily grown for home consumption. However, price elasticity of supply for

groundnut and pigeonpea in Andhra Pradesh is positive and sizable. Though pricing

efficiency of markets is generally high, it varies from market to market. Pricing efficiency

was found to be determined by a number of factors from which it could be concluded that

more markets with fewer traders per market, equipped with equally distributed

telephones and supervised by a frequently changing market secretary, would be

conducive to improved pricing efficiency. Investments in improvements of agricultural

market exchange have measurable payoffs. The removal of interregional trade restr ic-

tions was found to cause aggregate foodgrain production to increase, through relatively

minor shifts in cropping patterns.

Agricultural marketing in India is relativelywell-developed. There are at present nearly4000 regulated markets scattered across thecountry to which farmers can bring their pro-duce and where auctioning ensures competi-tive prices. Market prices are reported nation-wide on the basis of metric measures. Manytraders are equipped with telephones so thatindividual markets are well-integrated into a wider network of interregional market chan-nels; and since 1977 the policy of " food zon-ing," which used to severely inhibit privateinterregional trade of various foodgrains, hasbeen abolished.

Thus markets are operating in India, and onemight ask, how can research on foodgrainmarkets in this country contribute to develop-ment?

This paper presents in brief the approachesand results of four major areas of market re-

* Principal Economist and Economists,Program, ICRISAT.

Economics

search at ICRISAT that will demonstrate howthis can be done. The areas reported are:(1) consumer preferences, (2) supply re-sponse, (3) market channels and pricing ef-ficiency, and (4) foodgrain policies and agricul-tural productivity. We conclude that there isneed for more market research that helps toquantify the impact of market channels onallocation of agricultural activities, includingadoption of new technologies. Such quantita-tive evidence will help decision makers in Indiaas well as in other semi-arid tropical (SAT)countries of the world to make decisions oninvestments in market improvement that willovercome market-related constraints to agricul-tural development.

P r e f e r e n c e I n d e x o f F o o d g r a i n st o F a c i l i t a t e S e l e c t i o n f o rC o n s u m e r A c c e p t a n c e

ICRISAT has been assigned the task of generat-ing improved cultivars of four foodgrains —

173

sorghum, pearl millet, chickpeas, and pigeon-peas.1 In order to be adopted rapidly, newcultivars — apart f rom producing high and sta-ble yields — must also be acceptable to thetastes of consumers. The relatively low degreeof commercialization of farmers who producethese crops restricts adoption of new varietiesand therefore places a special premium ongood quality. Adoption rates will dependlargely not only upon yields but upon how wellthe new varieties are liked compared with thetraditional types.

As wil l be shown below, "good quali ty" thatensures consumer acceptance is not con-ditioned by only one or two visible character-istics; it is based upon an optimal configurationof a complex of several qualities, many of theminvisible. In order to incorporate this optimalquality mix into new cultivars, breeders mayhave to adopt a quality index for preferencesthat wil l facilitate selection for consumer accep-tance. This index would be established on thebasis of the degree to which a number ofmeasurable well-defined and relevant charac-teristics are available in a particular variety orline, each of these qualities being weighted by a parameter derived from the analysis of con-sumer preferences for known varieties.

Information on Quality PreferencesGenerated by the Market

The answer to the question of how to determineand measure what is liked seems rather simple;a grain is of good quality if it consistentlyfetches a higher price in a competitive market.

A market where innumerable consumerschoose among the different qualities of a par-ticular crop and pay certain prices for thevarious qualities can be regarded as a large,continuous consumer panel. People expresstheir preferences by paying higher prices forqualities they prefer, and vice versa (see Fig. 1).A competitive market forces traders to offerprices strictly according to the position of sup-

1. Market preferences for groundnuts, the f i f th ofthe crops included in the ICRISAT Crop Improve-ment Program, have not been s tud ied; the ap-proach to be taken for this pr imar i ly commercial

crop wou ld not be comparable to that for food-

grains.

Figure 1. Variation of prices of individual lots

of food grains transacted on a

selected market day.

ply and consumers' demand. Our observationsindicate that different lots of a particular food-grain traded in any market day are very carefullyassessed according to their appearance as wellas origin (if the information is available), andprices are formed accordingly. It is interestingto note that on a day-to-day basis the quantitiesof grains of different qualities arriving at a particular market place have no immediateinfluence on market price. This is so because theamounts in which different qualities happen toarrive at a place may be skewed towards poorquality on one day and good quality on anotherday. Also, there may be days with tall peaks inthe frequencies of quantities sold at certainprices, while on other days the distribution maybe even. Analysis of the coefficients of skew-ness and kurtosis of frequencies and pricesover 15 market days has shown that on anaverage the skewness is not significantly differ-ent f rom zero, and the kurtosis is not sig-nificantly different from 3.00; i.e., over severaldays the grain lots of different qualities arrive insuch a way that the frequency distribution oftheir prices can be considered as statisticallynormal. Thus, probably in anticipation of thenormal distribution also of the qualities arriv-ing, traders in the short run do not adjust priceto quantities of different qualities.2

2. This assumption of a normal frequency distr ibu-t ion of qualities and of a perfectly elastic demandfor quality can be tested only after all of therelevant quality characters have been identif iedso that a meaningful classification of quality and a test for consistency in pricing can be carried out.

174

On the basis of these observations, it may beassumed that in the short run, and at a singlemarketplace that represents only a small frac-tion of total quantities traded, the market de-mand for a given quality is perfectly elastic; i.e.,on a particular day virtually any quantity of a certain quality can be sold at a price reflectingquality rather than quantity.3

A competitive market also ensures integra-tion of rural areas into the commercial system.Even though the degree of commercialization islow, market prices in a competitive system aresignaled to the villages. Farmers may sell littleor even nothing at all of their produce. However,they know the price differences for differentqualities; while witholding a particular quality,they are actually aware of the opportunity pricethey are paying for their home-consumed qual-ity. Thus, in a competitive market, prices can betaken as representing the joint effect of rural aswell as urban preferences, under given condi-tions of supply.

Measurement of Market Informationon Grain Quality

There has been increasing interest recently inthe analysis of consumer demand as a functionof not only price and income (the traditionalapproach) but as a function also of consumerpreferences and qualities of the goods con-sumed (the "new approach"). The consumerwho acts in accordance with preferences and isgiven the opportunity to choose from some setZ of characteristic collections, will choose thecollection that maximizes his utility functionover Z (Lancaster 1971).

Lancaster has shown that consumer demandcan be mapped from the "good space" into the"characteristics space" with the help of a "con-sumption technology matr ix" that specifies therelationships between these two. Theil (1976)uses a similar approach: he defines a "compo-sition matr ix" for transforming individualgoods into preference independent "trans-formed goods" before analyzing consumer ex-penditure on the transformed goods.

3. This is not so in the long run and for the aggregateof all markets. In this case, prices of differentqualit ies have varying elasticities at differentsupply positions (see von Oppen 1978b).

In the case of a particular foodgrain for whichthe basic characteristics relevant to consumerpreference have yet to be identified, the de-mand analysis suggested by Lancaster/Theilcan be (and for reasons of data availability hasto be) simplif ied: by explaining price variabilityon a particular market day as a function ofvariability in quality, the analysis leaves out thequantity-consumer expenditure dimensions ofthe problem. In doing so, it is implicitly as-sumed that for the time being the price andincome elasticities are independent of quality.This implicit assumption has to be accepted inview of the fact that: (1) the primary goal of thisresearch is to identify relevant characteristicsthat will help plant breeders select for generallygood consumer acceptance, and (2) data thatwould permit a Lancaster/Theil type demandanalysis are not available.

To make use of the information on qualitypreferences contained in market prices, thefollowing approach was evolved.4 A centralwholesale market, i.e., a major secondarywholesale market that draws arrivals f rom a larger number of primary local assembly mar-kets, was selected. Centrality of the market isessential because in a central market widerranges of qualities may be observed than existin local assembly markets. At the selectedmarket, over a period of one to several days — preferably dur ing the main market ingseason — grain samples (about 50 g each) werecollected from all lots transacted. Price (as wellas other information, if available) was recordedwith the sample.

Laboratory analyses of the samples havebeen carried out as summarized.6 This list ofanalyses may not be complete, and research isgoing on to find out whether there are ad-ditional relevant characteristics. Such researchmay increasingly have to concentrate on exp-laining why certain attributes are relevant indetermination of consumer preferences.

In order to relate price information and qual-ity information, a multivariate regressionanalysis technique has been developed. Thisanalysis explains the differences in prices of

4.

5.

Nevertheless, there is need for further refine-ment, and we hope the discussion of this paperwi l l result in such improvements.See von Oppen et al. (1979, Appendix Table 1).

175

individual lots f rom the market day's averageprice as a function of quality differences inindividual lots from average qualities of all lotsof the market day. In other words, the averagequality and the average price per market day aretaken as references against which to comparethe individual qualities and prices.6

Application of the EstimationApproach and the ParametersDerived for Sorghum and Pearl Mi l let

Progressive investigations to identify relevantcharacteristics for sorghum and millets have

6. For more details on the mathematical formula-t ion of the est imation model see von Oppen 1976.

revealed the fol lowing:• Cryptic as well as evident quality charac-

teristics are of significance in explainingmarket price variation (Table 1).

• In sorghum as well as in pearl millet wefind as a preliminary result that the swel-ling capacity,7 dry volume,8 and proteincontent9 are reflected in market prices.

7.

8.

9.

Weight increase over vo lume increase after soak-ing, this variable measures the capacity of wateruptake wi thout swel l ing and should be probablymore appropriately called swel l ing " incapaci ty" ,or resistance to swel l ing.The vo lume of 10 grams of dry seed, measured bythe water-displacement method.Determined by Technicon autoanalyzer.

Tab le 1 . M a r k e t pr ice a as a func t ion o f qua l i ty character is t ics .

Quality characteristics

EvidentColor Mix- w h i t e b

- y e l l o w b

- r e d 6

- w h i t e b

- g r e y b

Glumesb

100 seed weighta

Moldiness- l i g h t b

- severe b

CrypticDry vo lume a

Swell inga

Protein contenta

R-2

Observations

Crops

Sorghum c

-.05** a

- . 1

- . 2 2 * * *. 3 * * *

- . 0 9 * * *— 2 4 * * *

.52*- . 1 8 * *

. 2 2 * * *.7765

Pearl mi l le td

- . 0 5 *

-- .02

.05*

- . 0 6 * * *- . 1 2 * * *

.43*- .12.16* * *.8325

a. Var iable was def ined as the natura l l oga r i t hm of the rat io of the observed value of a part icular gra in samp le over the average

value f o r al l samples on the market day.

b. Var iab le was def ined as the d i f ference be tween the observed percentage of seeds w i t h th is qual i ty in a part icular g ra in sample

a n d the average of a l l percentages of seeds of th is qual i ty fo r a l l samples on the market day.

c. Samples col lected in Hyderabad market over fou r days In 1977.

d. Samples col lected in Jodhpu r market over t w o days in 1978.

e * • • - Signi f icant a t 1%

** - Signi f icant at 5%

* - Signi f icant at 2 0 %

. Indicates var iable does not apply .

- Indicates var iable was not inc luded.

176

The two sets of coefficients for all character-istics that are comparable between sorghumand pearl mi l le t—i .e . , molds and the crypticqualities — have the same signs; in fact, a statistical test on an estimation model thatincludes only these comparable variablesshows no significant difference in consumersassessment of these qualities, regardless ofwhether they occur in sorghum or pearl millet.The sorghum samples for this analysis werecollected in Hyderabad in 1977 and the pearlmillet samples in Jodhpur in 1978; thus, as faras they are representative, these samples indi-cate that consumer preferences for cryptic quali-ty characters are probably of more than onlyregional relevance. Unlike the evident qualities,for which preferences vary considerably fromone region to another, cryptic qualities wouldseem to face universal preference patterns,across similar crops and across regions.10

These estimated coefficients suggest thatceteris paribus, an increase in the protein con-tent of sorghum f rom the average of 8.25% to,say, 10% would result in an increase in themarket price of 4% above the average.

The other coefficients can be interpreted inthe same way11; the ceteris paribus conditionimplies that all of the other quality character-istics remain the same while only one is beingchanged. Plant breeders' experience shows thatthese characteristics are to some extent relatedto each other, and that one cannot easily bechanged independently of all of the others. Thesimultaneous impact of several different qualitycharacteristics on consumer preference re-quires, when aiming at cultivars of higher con-sumer acceptance, the simultaneous consider-ation of all such relevant characteristics. Theestimation equation, is, in fact, designed toform the basis for such a simultaneous qualityassessment. It permits derivation of an index ofestimated consumer preferences.

10. It is interesting to note that in a study on wheatqual i ty, Bhatia (1973) reported that protein con-tent has a signif icantly positive relationship wi thmarket prices paid by private traders in twomarkets in Delhi.

11. An increase in 100-seed weight f rom an averageof 2.75 to 3.75 g would increase the market priceby 10% above the average price. For swell ingcapacity, an Increase f rom an average of 1.10 to1.25 wou ld reduce market price by 2 percentbelow the average.

As an example, we collected 15 sorghumsamples in the Hyderabad market and esti-mated the consumer preference for each. Thesesamples were taken to a village in Maharashtrafor ranking by farmers.12 They were also testedby villagers in Karnataka.13

It was found that the Maharashtra farmers'ranking and the ranking based upon our qualityindex are positively correlated with a corre-lation coefficient of .60; the correlation betweenKarnataka farmers' ranking and our qualityindex ranking was .56; both correlations weresignificant at the 5% probability level. Thisindicates that the quality index derived frommarket prices in Hyderabad is representative ofpreferences of farm families in Maharashtra, aswell as in Karnataka.

Even though these results demonstrate thefeasibility of our approach, it is felt that theestimation of the quality index can be furtherimproved. Consequently, the above resultsmust be regarded as preliminary while our workcontinues in search of additional cryptic qualitycharacteristics relevant in explaining consumerpreferences as expressed in market prices.Similar work on quality preferences for pulses isunder way. Here the major factors affectingprice seem to be related to ease of dehulling,percentage of hulls, and cooking quality of thedhal (split pea). However, there are difficultiesin producing standardized measurements ofthese qualities. Also the diffuse nature of themarket channels for pulses, which generally donot pass through central market places in largequantities as do cereals, makes reliable priceobservations more difficult. For these reasonsour work on quality preferences for pulses hasnot yet led to conclusive results.

E s t i m a t e s o f S u p p l yF u n c t i o n s f o r S A T C r o p si n A n d h r a P r a d e s h

Compared with the large number of supply-response studies done in India for irrigatedcrops, only a few studies have been done for

12.

13.

The consumer preference test in the vil lage wascarried out under the guidance of Dr. S. L. Bapnaand w i th the assistance of Mr. K. G. Kshirsagar(Report forthcoming).Personal communicat ion by Miss. T. PadmasiniAsur i , Regional Home Economist, Bangalore.

177

rainfed crops.14 In view of their importance forprice policy decisions in the SAT, we decided tostudy the supply response of the five ICRISATcrops in India. This note reports first resultsobtained for Andhra Pradesh, which is an im-portant SAT State in India. Andhra Pradeshcontributes about 1 1 % of the total area underthe five ICRISAT crops, which makes it thefour th largest producer. So rghum andgroundnut predominate in the rainfed croppingpattern of the State, whi le the other threecrops — pigeonpea and especially pearl milletand chickpea — contribute less. Sorghum andgroundnut are mainly grown as nonirrigatedcrops. Sorghum is grown in two seasons kharif (rainy) and rabi (postrainy), while the othercrops are only grown in kharif (groundnuts,pearl millet) or rabi (chickpea). In AndhraPradesh, 29% of the total cropped area isirrigated; the main sources of water are canalsand tanks.

Methodology

Past supply-response studies generally suf-fered f rom the lack of significance of coef-ficients and high level of aggregation, and theyoften provided conflicting estimates of supplyelasticities. Further, these studies concentratedon area response and not production re-sponse.15 The present study attempts to over-come these problems. Maximum use of theavailable data is being made by combiningdistrict-wide cross-sections and annual t imeseries.

wherei = 1 M independent variablesj =1 N districtst = 1 T years

= dependent variable= independent variables

The total number of observations availablewith this procedure are N x T. Estimating theabove type of equation by ordinary leastsquares (OLS) would provide inefficient esti-mates, because the use of OLS assumesEee' = This does not hold true for this case,

14.15.

For details, see Bapna (1979).For a detai led examinat ion of the methodologiesused, see Bapna (1977).

dependent variables and some of the indepen-dent ones; other variables appear untrans-formed.

Results

The results for the supply equations of sor-ghum, groundnut and pigeonpea are presentedin Table 2. The price elasticities obtained forproduction of groundnut and pigeonpea arepositive. The supply elasticity for groundnut is0.53 and is highly significant; for pigeonpea thecoefficient is positive (0.23) but only significantat the 10% level. For sorghum the supplyelasticity is negative but it is not statisticallysignificant. Sorghum in Andhra Pradesh isgrown in two seasons; therefore, our estima-tion procedure may confound annual with in-terseasonal supply response. This problemcould be avoided only if seasonal data wereavailable, which they were not.

For area, all three crops have positive supplyelasticities, but here again the sorghumcoefficient is not statistically significant. Thesupply elasticities for area are 0.30 forgroundnut, 0.29 for pigeonpea and 0.04 forsorghum. It is interesting to note that the elas-ticities for the three crops broadly correspond to

the extent of commercialization of these crops,with groundnut being the most commer-cialized.

The area supply elasticities obtained here forgroundnut are much lower than the estimate of0.69 obtained by Cummings (1975) for AndhraPradesh. For sorghum, the elasticity broadlycorresponds to the National Council of AppliedEconomics Research (1969) estimates. Forpigeonpea, no estimates are available for com-parison.

Rainfall is a very important factor in theproduction of nonirrigated crops. To capturethe effect of rainfall on productivity, the variablewas included in log-linear and log-quadraticform. In all cases, the linear term yielded posi-tive coefficients and the square term negativecoefficients, as expected. Irrigated area exhi-bited negative coefficients for all the threecrops, which indicates that with the expansionof irrigation, farmers devote a greater area totypical irrigated crops such as rice and sugar-cane. An increase in the irrigated area under thecrop concerned did not produce a measurableincrease in production, except for pigeonpea.Road density seems to have a negative impacton the production and area of sorghum, aninsignificant effect on the production and area

Tab le 2. Product ion and area supply aquat ions for sorgh um, groundnut , and p igeonpea. in AndhraPradesh.

Explanatory variable

InterceptIn (Rain)(In Rain)2

% Total irr igation% Crop irr igated% High yielding varieties

Road densityMarket densityDeflated price

Regression coefficients*

Production equation

Sorghum

2.990.64- .05

- .04- .170.15

- . 0 6 3 * * *- . 2 1 * * *- . 2 4 b

Groundnut

-13.494.26**- . 3 1 * *

- . 1 1 *.31@

-

.04b

- . 2 6 * * *0.53***

Pigeonpea

-9.813.19*- . 2 5 *

- .012.54*-

- .006— 2 4 * * *

0.23*

Area equation

Sorghum Groundnut

3.88 - .530.35** 1.00**- . 0 3 * * - . 0 9 * *

- .50* - . 1 2 * *- .26 - .01- .08 -

- . 0 3 5 * * * .045** *- . 1 3 * * * - . 2 2 * * *0.04 0 .30** *

Pigeonpea

-3.221.62***- .16

- . 0 6 b

1.80***-

.001- .07 b

0.29***

a. The equat ions w e r e f i t ted as doub le l oga r i t hm t ransformat ions of equa t ion - (2) In the text .

b. S ign i f icant at 2 0 %

* * * Signi f icant a t 1 %

** Signi f icant at 5%

* S ign i f icant at 10%

179

of pigeonpea and a positive impact ongroundnut production and area. Coefficients ofmarkets were significantly negative for all thethree crops.16

The estimation of supply elasticities — once itis extended to cover all districts of major produc-ing regions of ICRISAT crops in India — wil lhelp in assessing the impact of priceand marketpolicies. Wherever possible and not yet done,similar studies should be caried out in India andalso in West Africa.17

Efficiency of Market Channelsfor ICRISAT Crops

A study was conducted to describe the marketchannels of ICRISAT crops in different regions,to measure market costs and market efficiency,and to identify criteria determining efficiency.Results of this research are reported in this sec-tion.

Methodology

Surveys were conducted in 29 regulated mar-kets in seven SAT states of India. The marketsfor each state were selected on the basis ofprobability, proportionate to the proportion ofICRISAT crops entering the markets. The criteri-on for inclusion of markets in the sampling wasthat ICRISAT crops represent at least 25% oftotal arrivals. The selected markets and some oftheir particulars are listed in Table 3.

Primary and secondary data for 1974-75were collected from market committees of allselected markets and also f rom farmers andtraders. Farmers and traders in the marketswere selected at random and interviewed toobtain estimates of marketing margins andcosts (not reported here), on quantities pro-duced and sold, as well as directions of sales.Pricing efficiency was measured in terms of thecoefficient of correlation of weekly prices bet-ween pairs of markets. Multiple regression

16.

17.

Research reported elsewhere (von Oppen1978c), indicates that market effects on produc-t ion have a lag structure that requires carefulspecif ication. This problem is under study.To our knowledge, the only study of this kind isAbalu (1975).

analysis was used to explain the pricing effi-ciency of all pairs of markets as a function ofindependent variables such as distance bet-ween two markets, average market turnover,market age, number of traders, telephones,market secretaries, market area, density of pro-duction, density of population, etc. Hypotheseswere formulated stating in which way theseindependent variables were expected to affectpricing efficiency.

Estimates of Flows of ICRISAT Crops

On the basis of information collected fromtraders and farmers in the selected marketssurveyed in different states, estimates ofinflows into the markets and outflows from themarket were made for each crop. Estimates ofmarket arrivals based on farmers' reports indi-cate that only 22% of the sorghum producedcame to the markets and the remaining 78%was retained on farms (Table 4). In the case ofpearl millet, it was estimated that about 26% ofproduction reached markets; for pigeonpeait was 35%; for chickpea 45%; and forgroundnut 80%. These estimates of marketarrivals are somewhat higher than those pub-lished by the Government of India. This can beexplained by the fact that the 29 markets wereselected to ensure that at least 25% of theirarrivals were ICRISAT crops.

Estimates of outflows f rom selected marketsof different states (Figs. 2 to 6) indicate that a large proportion of these crops were tradedwithin the districts where the markets werelocated. A smaller proportion moved to otherdistricts within the State. Nevertheless, theremaining trade, which often involved long-distance interstate transport, was not negligi-ble. In comparison with sorghum and pearlmillet, the quantities and distances involved ininterstate exchange of pigeonpea and chickpeaare much greater. Groundnuts seem to bemostly processed locally. These estimatesshow that interregional trade in sorghum andpearl millet is concentrated within the SATareas of India where these crops are tradition-ally grown and consumed, whi le the pulses f lowto other, non-SAT areas. In fact, the majorproducing areas for sorghum (Maharashtra)and pearl millet (Gujarat) turned out also to bethe major importing regions for these crops in1974-75.

180

Table 3 . List o f se lec ted m a r k e t s and some o f their par t icu lars, 1 9 7 4 - 7 6 .

Statea / Market name (District)

AP/Warangal (Warangal)AP/Khammam (Khammam)AP/Tandur (Hyderabad)

MP/lndore (Indore)MP/Uj jain (Ujjain)MP/Khandwa (Khandwa)

MP/N. Gunj (Sehore)

MH/Poona (Poona)MH/Latur (Osmanabad)MH/Malkapur (Buldana) MH/Malegaon (Nasik)MH/Dondiacha (Dhulia)

R/Ganganagar (Ganganagar)R/Nadbai (Bharatpur)R/Hindone (S. Madhopur)R/Begun (Chittorgarh)

G/Patan (Mehsana)G/Santrampur (Panchmahal)G/Damnagar (Amreli)

K/Bagalkot (Bijapur)K/Gadag (Dharwar)

K/Chitradurga (Chitradurga)

UP/Shahabad (Hardoi)UP/Ujhani (Badaun)UP/Jafarganj (Sultanpur)UP/Orai (Jalaun)UP/Bendki (Fatehpur)UP/Jarar (Agra)UP/Panwari (Hamirpur)

Year marketregulated

193319371949

1953193019641968

19591931191719491939

1964196519651970

195119521954

194619431951

1972196719731967196819691971

Total marketturnover

('000 tons)

1163521

18462347

13684382420

7914146

527

0.1

844528

33302016138

0.07

Share ofICRISAT

crops (%)

386165

46722142

3451173839

17536337

73249

735742

33542521362974

No. ofcommission

agents

12013085

1606377

-

159127

241952

152538837

17689

10910377

22135

1021026

-

No. ofwholesale

traders

39213985

504171135

14

443326

478676

25369

11860

3379120

161293

84

825422825332

5

No. oftelephones

500250130

680210200-

650400

70100

50

4501008526

3003010

60150

15

1535-

52016—

a. AP = Andhra Pradesh; MP = Madhya Pradesh; MH = Maherashtra; R = Rajasthan; G = Gujarat; K = Karnataka andUP - Utter Pradesh.

Pricing Efficiency

Generally it is accepted that the higher thecorrelation of prices between pairs of markets

for a particular product, the better the marketsfor this crop are integrated and hence the moreefficiently they are operating.18

Weekly market prices for 1974-75 of the five

18. This logic may not apply ful ly i f one considersonly a l imi ted number of market places (seeHarriss 1978); for instance, if f lows of a particularcommodi ty between any two markets do notalways have the same direct ion but, for reasonsof vary ing supply and demand in both marketregions, somet imes cease end then reverse, then

the correlation of market prices wil l be reducedeven though the t w o markets may be perfectlyintegrated. However, for the type of analysisproposed here, wh ich involves all pairs of a relatively large number representing the majorparts of SAT India, such a consideration and itsimpact on price correlation wi l l not affect theanalysis and its conclusions.

181

Short arrows (regardless of d i rect ion) represent f lowsto other markets wi th in the d is t r i c t .Medium length arrows (regardless of d i rect ion) re-present f lows to markets in other d i s t r i c t s w i th inthe s ta te .Arrows across state borders represent f lows to marketsin other s tates.Width of arrow represents re la t ive proportion of f low.

For example, in Madhya Pradesh, about 22% of thesorghum arr iv ing in the se lec ted regulated marketsf lows to Maharashtra, 15% to Gujarat and 3% toKarnataka. Of the remainder about 35% goes to marketsin other d is t r ic ts of Madhya Pradesh and the remaining25% stays wi th in the distr icts of the markets wheref i rs t so ld .

Figure 2. Production of sorghum, market ar-

rivals as per cent of production, and

total flows (as percent of market

arrivals) from selected food grain

markets in selected states of India,

1974-1975.

ICRISAT crops for every market surveyed werecorrelated (by pairs) with those of all othermarkets. Unwe igh ted mean corre lat ioncoefficients between pairs of all selected mar-kets for all five crops are given in Table 5. Mostof the correlation coefficients were positive andsignificant at the 1% level.

Correlation coefficients were highest for

Arrows expla ined in Figure 2.

Figure 3. Production of pearl millet and total flows, as percent of market arrivals, from selected foodgrain markets in selected states of India, 1974-1975.

Table 4. Estimates of arrivals (percentage ofproduct ion) of ICRISAT crops Inselected regulated markets, 1 9 7 4 - 7 5 .

Arrivals in selectedCrops regulated marketsa

Sorghum 22Millet 26Pigeonpea 35Chickpea 45Groundnut 80

Arrivals in all regu-lated marketsb

11NRC

NR28NR

a. Source: ICRISAT Survey.b. Source: Government of India (1976).c. Not reported.

chickpea and pigeonpea, the two crops wi thmore interstate trade. Sorghum, pearl millet,and groundnut prices correlated positively andsignificantly but considerably below the levelsof correlation measured for the other crops. For

182

groundnut, the emphasis on local processingand l imited interregional trade might be re-sponsible for the low market price integration.In the case of sorghum and pearl millet, despiteinterregional trade, market prices were not wellrelated, probably due to difficulties in pricereporting, given the variability in quality ofmarket arrivals, especially of sorghum (SeeFig. 1).

Table 5. Mean correlat ion coeff ic ients of 54week ly marke t prices of ICRISATcrops, pairs o f 29 markets in S A TIndia, 1 9 7 4 - 7 5 .

Crops

SorghumPearl milletPigeonpeaChickpeaGroundnut

Mean correlation coefficient

0.240.370.620.720.31

Arrows explained in Figure 2.

Figure 4. Production of chickpea, market arri-

vals as percent of production, and

total flows las percent of market

arrivals) from selected foodgrain

markets in selected states of India,

1974-1975.

Arrows explained in Figure 2.

Figure 5. Production of pigeonpea and total flows, as percent of market arrivals, from selected foodgrain markets in selected states of India, 1974-75.

183

Fac to rs A f f e c t i n g P r i c i n g E f f i c i e n c y

Pricing efficiency can be affected by severalfactors, and in this study an attempt was madeto identify with the help of the data collected,some factors for which the fol lowing hypothe-ses were postulated:

where:CClj = correlation coefficient between

markets i and j.DT l j = distance between markets i and

j measured in kilometers. It ishypothesized that if the marketsare distant their prices will notbe highly correlated. Hence a negative sign is expected.

SM l j = average size of the two marketsmeasured in terms of total arri-vals in thousand tonnes. Largermarkets are more efficient in

pricing; a positive sign is ex-pected.

= average turnover of ICRISATcrops in percent of total tur-nover of the markets measuredin terms of arrivals in tonnes.The hypothesis for this variableis that if the share of ICRISATcrops in total size is higher bet-ween the markets then thesemarkets are efficient for IC-RISAT crops. A positive sign isexpected for this variable.

= age of markets. This is mea-sured in years by deducting theyear of regulation from the year1975. It is hypothesized thatolder markets are more efficientthan new markets. A positivesign is expected for this vari-able.

= number of wholesale traders in

Arrows expla ined in Figure 2.

Figure 6. Production of groundnut and total

flows, as percent of market arrivals,

from selected foodgrain markets in

selected states of India, 1974-1975.

184

markets . The number ofwholesale traders in a particu-lar market measures the degree(not quality) of market accessfor fa rmers . " The more tradersin one market place, ceterisparibus, the fewer are the mar-ket places and the longer is thedistance farmers have to travelto obtain market access; hence,market efficiency suffers andthe sign of this variable is ex-pected to be negative.

= number of commission agentsin markets. The hypothesis andsign for this variable is similarto NWij.

= number of telephones per unitturnover in markets. It ishypothesized that the larger thenumber of telephones per unitturnover, the higher is the mar-ket efficiency and a positive signis expected for this variable.

= number of telephones pert rader in markets. Thehypothesis and sign expectedfor this variable are similar to

= number of changes in marketsecretaries from 1965 to 1975 inthe markets. It is hypothesizedthat the larger the number ofchanges in market secretaries,the higher is market efficiency.If a market secretary stays in a market for many years he islikely to be biased and toleranttowards collusion of traders re-sulting in inefficiency. A posi-tive coefficient is expected forthis variable.

= utilization of market yards in themarkets. This is measured interms of market arrivals per unitsize of the market yard. It ishypothesized that proper utili-zation of the market yard results

19. Note that there are only t w o markets in oursample w i th fewer than 20 traders but 23 marketsw i th more than 50 traders (see Table 3).

in market efficiency and hence a positive sign is expected for thisvariable, up to a maximumbeyond which congestion leadsto inefficiencies, thus a nega-tive sign for the squared term isexpected.

= average turnover of the tradersin the markets. This variable ismeasured dividing total marketturnover by the number ofwholesale traders in the mar-ket. It is hypothesized that mar-kets wil l be efficient if their tur-nover is shared by many smalltraders rather than a few largertraders. A negative sign is ex-pected for this variable.

= population density in the dis-trict where markets are located.It is hypothesized that marketsare more efficient where morepeople live. Hence a positivesign is expected for this vari-able.

= density of crop production inthe districts where the marketsare located. It is hypothesizedthat more production of a par-ticular crop is a sign of speciali-zation which implies more mar-ket participation by farmers andbetter market efficiency. A posi-tive sign is expected for thisvariable.

In addition to the above variables, dummyvariables for states and crops were also in-cluded in the regressions.

Regression results are summarized in Table 6.This table shows that most of the variables havethe expected signs and appear to be statisticallys igni f icant .2 0 Most of the hypotheses

20. The significance levels as computed for standardregression problems should probably be usedonly as indicators of relative significance amongvariables, because of the nature of our data.Except for distance between markets, which rep-resent actual observations, all independent vari-ables def ining market characteristics were gen-erated by pairing markets, thus creating n actualobservations and n (n -1 ) /2 data points. It is notdear to what extent such a data manipulat ionreduces the statistical significance of the results.

Table 6. Summary of regression resul ts . a

Variables wi th positivesigns*

Market Size***Age of market***No. of Commission agentsNo. of telephones per turnover***

Change of market secretaries***Utilization of market y a r d * * *Density of populationDensity of crop product ion**Distance***

Share of ICRISAT crops**No. of wholesale t raders***No. of telephones per t rader***Average turnover of t raders***

Expectedsigns

YesYesNoYes

YesYesYesYesYes

NoYesNoYes

a. For the complete regression equation see Raju and vonOppen (forthcoming).

b. *** Significant at 1% level.** Significant at 5% level.

proposed were not rejected. In two cases,the proposed hypotheses cannot be acceptedas the variables have unexpected signifi-cantly negative signs. These were the share ofICRISAT crops to total market turnover, andnumber of telephones per trader. Two variableswere not statistically significant, i.e. number ofcommission agents with an unexpected posi-tive sign, and density of population wi th anexpected positive sign.

The positive sign for share of ICRISAT cropssuggests that markets are inefficient if they aredominated by ICRISAT crops. Thus marketstend to be more efficient the larger the variety ofother crops arriving in this market. The negativeeffect of number of telephones per trader onpricing efficiency seems to imply that a fewtraders monopolize the telephones, and marketefficiency is adversely affected.21

Conclusions on Market Channels

It is concluded on the basis of estimated arrivals

21. This variable should probably be included as a linear and a squared term wi th the expectationthat the linear te rm Is positive and the squaredterm negative, thus indicating an op t imumnumber of telephones per trader.

185

of ICRISAT crops into selected markets thatsorghum and millet are produced mainly forhome consumption. From the estimates ofoutf lows of ICRISAT crops from the selectedmarkets surveyed in different states, it is con-cluded that in comparison with the highervalued pulses, the lower valued sorghum andmillets are only traded over short distances.This is because consumption of the latter cropsis higher in the producing areas, whi le con-sumer demand is lacking in most other areas.Possibly trade restrictions imposed on coarsegrains and other crops during the period ofobservation have also had an impact.

On the basis of price correlation analysis, it isconcluded that for ICRISAT crops most of theselected markets are well integrated, and gen-erally price formation is efficient.

Analysis of the price correlation coefficientsbetween pairs of markets leads to the fol lowingconclusions: The fewer the traders per marketthe higher the pricing efficiency in the presentlyexisting markets; therefore, without necessarilyreducing the total number of traders, moreregulated markets wi th fewer traders in eachwould increase pricing efficiency. This wouldimply more markets in the vicinity of farmers,providing better physical access; telephonesshould be provided in these markets, but tele-phones should be distributed equally amongtraders. Market supervision is important andmarket secretaries should be exchanged fre-quently so that they can properly oversee mar-ket activities without being influenced by vested(traders') interests. Market yards should beestablished to be of a size that avoids excesscapacity and congestion.

The Ef fects o f In ter reg iona lT rade a n d M a r k e tIn f ras t ruc ture on A g g r e g a t eProduct iv i t y o f Ag r i cu l t u re

Agriculture is spread over space and its produc-tivity in any location depends upon the local andregional resource base such as soils, climate,topography, and people. To ful ly exploit thisregional and, over t ime, changing diversity inresource endowments — given the spatial ex-pansion of agricultural activity—efficient inter-farm and interregional exchange and transpor-tation is a sine qua non. The more efficiently

186

farmers and regions are able to exchange theirproduce, the stronger wil l be their incentive toallocate resources according to personal andlocal comparative advantages; by making useof the diversity in comparative advantagesthrough exchange, aggregate agricultural pro-ductivity can be increased.

The objective of this section is to evaluate theproductivity of markets and market exchangenet of other production factors. This involvesmeasurements of the effect of trade restrictionsin India on aggregate productivity by using(1) a normative activity analysis model of inter-regional trade, and (2) a positive regressionanalysis of statewide foodgrain production andfood zoning. The removal of trade restrictionsfor agricultural products has, as we shall see,positive effects on aggregate productivity; it isagreed that these may affect — at leasttemporari ly — equity and income distribution,and the policymaker may be right in askingwhether the gains in aggregate productivity areworth their political implications.22 The fol low-ing discussion is restricted to the analysis ofaggregate productivity. The approaches usedmay help to remove some of the uncertainties inassessing the gains associated with marketdevelopment so as to have a better basis forplanning physical and institutional investmentsinto this vital factor in the agricultural pro-duction process.23

Interregional Trade and AggregateProductivity: A NormativeApproach

In order to demonstrate the impact of marketf lows on productivity, an activity analysismodel has been set up. This model allows us toinvestigate commodity f lows, pricing, and cropallocation at different productivity levels onthree crops in three regions. The model is basedon data that represent the fol lowing hypotheti-cal case: In the three Indian states of Andhra

22.

23.

A study that a ims to assess the differential impactof market access on smal l vs. large farmers is nowbeing conducted.For a discussion of these issues see Raj Krishna(1965). When just i fy ing zonal food policies,Dantwala (1976) said: "The economic balancesheet of such disincentive in surplus states andincentives in deficit states is anybody 's guess."

Pradesh, Madhya Pradesh, and Maharashtra,three crops — rice, sorghum, and chickpea — are grown, and all three compete for the samelocally available resources, in this case land.24

Yield per hectare is restricting the supply, sothat the total use of land for all three cropscannot exceed its limits in each state. Supply isfurther restricted by a linear function of arearesponse to price multiplied by yield. The initialelasticities of supply are derived from availableestimates and represent — as far as they areknown — actual conditions. The model alsoincorporates demand as a linear function ofprice using elasticities available or derived fromother sources (S. L. Bapna, personal communi-cation). Transportation costs between regionscorrespond to official rail freight rates betweencentrally located places in each of the threestates.

Given this environment, a quadratic pro-gramming model allocates crop production ineach of the states so that producers' welfareplus consumers' welfare, minus total costs oftransportation is maximized. Application of thismodel produces optimal allocation of land tocrops, quantities produced, commodity f lows,and accompanying price levels.

The initial data set (i.e. case A) representsinitial conditions (1971-72); in other words,low yields and high costs of production "beforethe introduction of new technology." Supposenow that the governments of the three statesdecided to impose trade restrictions so that thequantities traded of each crop among all re-gions would not exceed 10% of the quantitiestraded without restrictions. The imposition ofthe trade restrictions under these conditionswould change production on the three food-grains in all three states as fol lows:25

• total output of rice remains unchanged;• total output of sorghum decreases by

5%;

24.

25.

Other resources such as capital and labor, whichare mobi le in the long run, wi l l tend to beallocated wherever most profitable. Under condi-t ions of restricted interregional commodi ty ex-change these resources wou ld move f romsurplus regions (because of natural comparativeadvantages) to deficit regions, where higherprices ensure higher returns.For more details of this model , see von Oppen(1978c, Table 4).

• total output of chickpea decreases by 13%;and

• total output of all foodgrains together de-creases by 2%.

Suppose now that new technologies arefound and adopted for all three crops in allregions so that yields increase by 50% in thestates that have highest yields and by propor-tionally less in the states with lower yields.26

Supply functions are allowed to shift, al lowingproductivity growth over initial levels in thesame proportion as yields growth at the sametime, an increase in demand in all regions for allcrops of 25% is assumed. In this case B, theimposition of trade restrictions causes:

• total output of rice to decrease by 1 % ;• total output of sorghum to decrease by

7%;• total output of chickpea to decrease by

15%; and• total output of all foodgrains to decrease

by 4%Under conditions of low yields and high

production costs (Case A), trade restrictionshave a measurable but comparatively smallimpact; aggregate productivity of all foodgrainsis decreased by only 2%. If, however, wi thpopulation growth, demand functions shift tothe right, and if at the same t ime newtechnologies are adopted which increaseyields, and shift supply functions to the right(Case B), differentially then regional specializa-tion and interregional exchange increase, andconsequently aggregate production rises sig-nificantly. Under such circumstances, a restric-tion on interregional trade as described above,decreases aggregate production of all food-grains by about 4%, which is about twice thepercentage decrease under the initial or low-demand, traditional supply conditions. In other

26. A comparison of wheat yields in eight majorwheat growing states in India between the years1954/55-1958/59 and 1970/71-1974/75, showsthat after the introduction of new wheattechnologies, yields had not increased by thesame percentage in every state, but rather in a proport ional fashion, i.e., percentage y ie ld in-creases were higher, the higher the original y ie ldlevel had been (with the exception of Bihar andGujarat, where yield increases were high despitelow initial yield levels). See v on Oppen (1978c,Appendix IV).

187

words, trade restrictions are more harmfulunder improved technology (Case B) becausethey considerably depress total productionbelow the possible level which could have beenreached with unrestricted trade.

Food Zoning and Aggregate Yieldof Foodgrains: A PositivisticAnalysis

Indian food policies over the past 20 to 30 yearshave varied in their degree of control of trade infoodgrains. Figure 7 presents a general pictureof the periods and states involved when foodzones were imposed. The imposition of a foodzone implies that the right to trade across theborder of a food zone becomes the monopoly ofthe Central Government through its statutoryagency, the Food Corporation of India (FCI).After 1956-57, various states were grouped intofood zones for rice and wheat. While the wheatzones were abandoned in 1961, the rice zonescontinued until 1964. In 1964 it was decided thatevery state would form a separate food zone.The state governments were thereby given theauthority to "organize" the foodgrain tradewithin the state; this often resulted in traderestrictions even at the district level, by impos-ing license systems that were intended to con-trol the movements of grain f rom one district toanother within the same state.

These statewide zones in the beginning con-trolled only the trade in wheat and rice, but after1966 most foodgrains were included in themovement restrictions. Statewide food zoningcon t inued — w i t h va ry ing degrees ofstringency — up to 1976, when the three statesof Andhra Pradesh, Karnataka, and Tamil Nadujoined again to form the southern rice zone;finally, in 1977, food zoning was abolishedaltogether.

A quantification of these changes in govern-ment intervention wi th the interregional tradein foodgrains is a complicated task. It wi l lalways remain imperfect because of two majorreasons: (1) the imposition of restrictions asrecorded in administrative documents does notprovide ful l information about the degree towhich the act was in fact implemented, espe-cially at the interdistrict level;27 (2) even though

27. Government of India, (a and b).

private trade was legally excluded from movinggrain across food zones, and the FCI had themonopoly for interzone grain movements, thisdid not in all cases imply the same degree ofrestriction of exchange among all regions. Thesometimes massive movements of foodgrainsby the FCI have to some extent fol lowed a f lowpattern similar to trade f lows that would haveoccurred in a system free of food zones.Consequently, the method adopted below toquantify "freedom of t rade" can only be re-garded as a conservative approximation to-wards a measurement of trade interference.

The degree of freedom in the trade of a particular state with other states, and in differ-ent crops for a particular year, was counted byadding the number of states in which trade wasunrestricted for each of the four crops, rice,wheat, pulses, and coarse grains. This gives a rough but ready measure of freedom of trade(FOT) for a particular state overt ime. For exam-ple in Gujarat in 1963-64, the existence of theNorthern Rice Zone restricted the number ofstates wi th which Gujarat was free to trade riceto only three (including Gujarat), while forwheat, pulses, and coarse grains, no restrictionsexisted (Fig. 7). Thus for these three grains tradewas free among 13 states; i.e. the FOT forGujarat in 1963-64 was (1 x 3) + (3 x 13) = 42. Inthe fol lowing year the establishment of Gujaratas a separate food zone for wheat and riceimplied that these two grains could be freelymoved only within the state, while for theremaining pulses and coarse grains, all 13states were open to trade. Hence the FOTin 1964-65 for Gujarat amoun ted to(2 x 1) + (2 x 13)=28. In this fashion, the FOT ofGujarat in the year 1970-71 went up to 40, whilein the next year it came down to 16.

In order to explain aggregate yield of food-grains as a function of trade restrictions andother factors, state data f rom 1960-61 to1973-74 were collected. For the estimationprocedure a generalized least squares modelwas applied which combines t ime series andcross-section data.28

It was postulated that aggregate productivity

28. For details of th is method, see Barah et a l . (1977).Discovery of a s l ight mistake in the earlier versionof COMTAC required minor corrections of resultsreported earlier. However, the conclusion re-mains the same.

188

189

of foodgrains was linearly dependent uponinputs, weather, in f rast ructura l facilities, laboravailability, and freedom of trade. The impact offreedom of trade was lagged. However, no a priori knowledge existed that would have al-lowed us to postulate a specific duration for thelag period. Consequently, lagsfrom t to t-4 werejointly entered into the equation.

The joint effect of the lags in freedom of tradeon aggregate yield was positive and the sum ofall coefficients was 1.818. None of thecoefficients appears to be significant, probablybecause of multicollinearity.29 Regarding thecoefficients of the remaining variables includedin the estimation, irrigation has a highly sig-nificant positive relationship with aggregateyield, and the same is true for rainfall, fertilizers,and density of rural population, even thoughless significantly. Surprisingly, the proportionof area under high-yielding varieties does notappear to have a statistically measurable im-pact upon aggregate yields. Truck density ispositively, but not significantly, associated withaggregate yields.

The average foodgrain yield over t ime andstates was 837 kg/ha and the average valuemeasured for the variable " freedom of t rade"was about 31 degrees. If — ceteris paribus — a complete liberalization of trade (implying a value of 52 degrees) was introduced, this wouldbe fol lowed by an increase of aggregate yield by(52 -31 ) 1.818=38.2 kg/ha, or by nearly 5% ofthe average yield. In other words, against thelevel of theoretical productivity under unre-stricted conditions of 875 kg/ha the averagerestriction reduced aggregate yields to 837 orf rom 100 to 96%, i.e. by 4 % .

Comparison of Resultsof Interregional Trade Analyses

It is interesting to note that both approaches to

29. Further research in this f ie ld Is required to bui ld a theoretical basis for explaining the lag structuresin the relationships between markets and produc-t iv i ty. One suggest ion for future analysis of th iskind is to explore weighted averages of backwardlags. Another approach of postulat ing a particularlag, e.g. F O T t _ 2 , yields a highly signif icantcoefficient of the same order of magni tude as thejo int effect repor ted, but th is approach was dis-missed as eclectic.

estimate the effect of interregional trade onaggregate productivity yield similar results.This tends to reinforce the credibility of theassumptions underlying each of the ap-proaches. The major conclusions to be drawn atthis stage are that the effects of interregionaltrade on productivity require a number of yearsto cause cropping patterns to shift. The shiftsthat occur are not dramatic. The farmer whoadjusts his cropping pattern may hardly per-ceive his own effort in view of the variability ofso many other factors, such as weather, pests,diseases, etc. to which he also has to adjustfrom one year to the next. Details of the spatialequil ibrium model reported elsehwere (vonOppen, 1978a) show that slight increases ordecreases of the order of 5 to 10% in areasunder particular crops are all that is required tobring about aggregate productivity increases ofthe order of 4%.

Summary and Conclusions

The research reported in this paper is an ac-count of our attempts to analyze informationabout agricultural marketing in India with theaim of deriving policy implications which arerelevant for agricultural development. Much ofthe work reported is still in its preliminarystages and the results may have to be modifiedas further insights are gained. Our findings areas fol lows:

• Agricultural market channels in India arecompetitive and pricing efficiency in gen-eral is good. Consumer preferences forsorghum and pearl millet have a measura-ble effect on market prices. This includespreferences not only for evident attributes(color, seed size) but also for crypticcharacteristics such as swelling capacity orprotein content. A quality preference indexcan be derived from such preference mea-sures for selecting "good qual i ty" grain.This index may be of help to plant breederswho realize that the successful adoption ofhigh yielding varieties also depends uponthe consumer acceptance of the new vari-ety.

• ICRISAT's foodgrains in India are primarilygrown for home consumption. Neverthe-less, price elasticity of supply of groundnutand pigeonpea in Andhra Pradesh is posi-

190

t ive and sizable. Even though pricing effi-ciency of markets is generally high, itvaries from market to market. Pricingefficiency — as measured in price correla-t i on coeff ic ients between marketpairs — was analyzed and found to bedetermined by a number of factors such asdistance between markets, age of marketsand crops considered; other important fac-tors were the number and size of traders,number of telephones, size of marketyards, and number of different marketsecretaries employed over the past 10years. More markets with fewer traders permarket, equipped with equally distributedtelephones and supervised by a frequentlychanging market secretary, would be con-ducive to improve pricing efficiency.

• Investments in improvements to agricul-tural market exchange have measurablepayoffs. The removal of restrictions tointerregional exchange was found to causeaggregate foodgrain production to in-crease. Positive effects to market exchangeon productivity are brought about by rela-tively small changes in cropping patterns.

Given the measurable effects of investmentsin market infrastructure on productivity at theaggregate level, the question arises as to howthese gains are distributed among small andlarge farmers. This issue is presently understudy, the general hypothesis being that bettermarket access is relatively more beneficial tothe small than to the larger farmer; we expectthat investments improving access to marketswill lead to more equitable distribution of gainsfrom it.

Acknowledgments

The authors are grateful to Messrs. P. ParthasarathyRao, R. Mahendran, and A. Pavan Kumar for theirassistance in the computat ional work. Drs. James G.Ryan, Hans P. Binswanger, and N. S. Jodha arethanked for their comments on an earlier draft of thispaper.

References

ABALU, G. O. I. 1975. Supply response to producerprices: A case study of groundnut supply to theNorthern States Marketing Board, Savanna 4 : 1 .

BAPNA, S. L. 1977. Estimation of supply funct ions forsemi-arid tropical crops: methodology and someresults, ICRISAT Economics Program, Patancheru,A.P., India (Mimeographed.)

BAPNA, S. L. 1977. Supply response studies inI n d i a - A review, ICRISAT Economics Program,Patancheru, A.P., India.

BAPNA, S. L., JHA, D., and JODHA, N. S. 1979. Agro-

economic features of Semi-Arid Tropical India.Presented at the International Workshop onSocioeconomic Constraints to Development ofSemi-Arid Tropical Agricul ture, 19-23 Feb 1979,ICRISAT.

BARAH, B. C., CHAKRABORTHY, B. K. and ESTES, J. W.

1977. Computational method fo r combin ing t ime-series and cross-section for regression analysis.ICRISAT Economics Program, Patancheru, A.P.,India.

BHATIA, G. B. 1973. A study of price differentials inwheats due to quality factors in two pr imary marketsof Delhi, M. S. thesis, Indian Agricultural ResearchInstitute, New Delhi, India.

CUMMINGS, J. T. 1975. The supply responsiveness ofIndian farmers in the post-independence per iod:Major cereal and cash crops. Indian Journal ofAgricultural Economics 39(1):25-40.

DANTWALA, M. L. 1976. Agricultural policy in Indiasince independence. Indian Journal of Agricul turalEconomics 31(4): 31-53.

GOVERNMENT OF INDIA (a) (Various Issues since 1960.)Bulletin on food statistics. New Delhi.

GOVERNMENT OF INDIA(b). (Various reports.) Reports ofthe Agricultural Prices Commission, New Delhi.

GOVERNMENT OF INDIA. 1976. Indian agriculture in brief,New Delhi, India.

HARRIS, B. 1978. Paddy and rice marketing in NorthernTamil Nadu: Studies in surplus market efficiency,technology and l ivel ihoods. Madras: Sangam Pub-lications.

LANCASTER, K. 1971. In Consumer demand: a newapproach. New York and London: Columbia Univer-sity Press.

NATIONAL COUNCIL OF APPLIED ECONOMIC RESEARCH,

1969. Structure and behavior of prices of food-

grains. New Delhi: NCAER.

RAJ KRISHNA. 1965. Rice zone policy - A note of dis-

sent. Agricultural Prices Commission Report, New

Delhi, India.

RAJU, V. T. Determinants of market efficiency for

191

ICRISAT crops in semi-arid tropical India. ICRISATEconomics Program progress report. (Forthcom-ing).

THEIL, HENRI. 1976. Theory and measurement of con-

sumers demand. 2 vols. Amste rdam: Nor th HollandPublishing Company; New York: Oxford 217 pp.

VON OPPEN, M. 1976. Consumers perferences forevident quality characters of sorghum. Presented atthe Sympos ium on Product ion, Processing andUtil ization toe Maize, Sorghum and Mil lets, De-cember 2 7 - 2 9 , 1976, CFTRI (Central FoodTechnological Research Institute), Mysore, India.

VON OPPEN, M. 1978a. Agr icul tural market ing andaggregate product iv i ty: A d imens ion to be added toagricultural market research. ICRISAT EconomicsProgram discussion paper 3, Patancheru, A.P., India.(Mimeographed.)

VON OPPEN, M. 1978b. An at tempt to explain theeffects of consumers' quality preferences on pricesof foodgrains at different levels of productivi ty.ICRISAT Economics Program discussion paper 8,Patancheru, A.P., India. (Mimeographed.)

VON OPPEN, M. 1 9 7 8 C . The effects of interregional

t rade and market infrastructure on aggregate pro-duct ivi ty of agriculture. ICRISAT Economics Prog-ram, Patancheru, A.P., India.

VON OPPEN, M. 1978d. A preference index of food-grains to facil itate selection for consumer accep-tance. ICRISAT Economics Program discussionpaper 7, Patancheru, A.P., India. (Mimeographed.)

VON OPPEN, M., RAJU, V. T., BAPNA, S. L. 1979. Food-

grain marketing and agricultural development inIndia. Presented at the International Workshop onSocioeconomic Constraints to Development ofSemi-Ar id Tropical Agricul ture, 19-23 February1979, ICRISAT.

WALLACE, T. D. 1977. Method for combin ing cross-section and t ime series data. ICRISAT, Patancheru,A.P., India.

192

Discussant 's Comments

Ba lwan th Reddy*

My remarks wil l be limited to the paper by vonOppen and his colleagues.

I comp l imen t von Oppen and others atICRISAT for their interest in important problemsof policy that are of great concern to policymak-ers in India as well as other semi-arid tropicalcountr ies. In some areas, their efforts arep ioneer ing in nature. Further theoret ica lanalysis, leading to better specification andmore empirical work, should lay a sound foun-dation for the generation of research findingsthat wil l be usable by policymakers.

In the standard elementary demand analysis,economists either derive the demand functionfrom the maximizing behavior of the house-holds or straightaway specify in Walrasian orMarshallian form the relation between price,income, and quanti ty demanded. Thus thehousehold 's impl ic i t preferences for com-modities is reflected in the characteristics of thedemand function. For plant breeders, quantita-tive estimates of these relationships per se donot provide adequate information to optimallyincorporate the relevant properties or charac-teristics in the commodity they are designing orthe grain seed they eventually want to propa-gate. However, assuming that some acceptablescaling of these properties is possible, breederswil l be interested in the gain in the form ofhigher price that the produce wil l obtain bymoving to a higher level along the scale of de-sirable properties. It is an important problem forresearchers at ICRISAT, and von Oppen and hiscolleagues seek an answer to this question.Their approach has been to specify the price ofcrops like sorghum and pearl millet as a func-t ion of properties such as color mix, seedweight, moldiness, swelling capacity, and pro-tein content to obtain tentative but meaningfulestimates. These results, together with the as-sociated costs of achieving these characteris-

* Administrat ive Staff College of India, BellaHyderabad, A.P., India.

Vista,

tics, should provide necessary data to deter-mine the optimal mix which a breeder shouldaim at in developing new strains. We must notethat the variability of some of these propertiesmay be random and this may require furtherstudy to provide more reliable guidance to thebreeders.

Plant breeders should be more interested inthe nutrient content of various crops and theirdifferent strains. The present approach of singleequation estimation has limited value in thiswider context. Lancaster's contribution, as theauthors have rightly pointed out, has a directbearing on this issue. Only after establishingmeaningful condit ions for unique mappingfrom the commodity space to the attr ibutespace wil l it be possible to estimate the preciseimpact of an attribute or quality on price. I amsure my colleagues at ICRISAT wil l explorethese avenues in the coming years.

The problem of assessing consumer prefer-ences for nutrient components can also beanalyzed within the conventional framework bypartitioning commodities into distinct catego-ries on the basis of marginal differences in anyone of their attributes. In principle if we havesufficient data, it is possible to derive a quantita-tive relationship between price differentials thatare associated with variation in any of the attri-butes or nutrient components.

The second set of results that von Oppen andhis colleagues have reported are extensions ofthe current work on supply response by addingsome 'market' characteristics to the set ofexplanatory variables. These new variables areroad density and market density. The empiricalresults do not justify the inclusion of these vari-ables in their present form. Analytically, I wouldexpect the impact of any 'market' characteristicto be reflected in prices and price differentials.Even so, in this instance as well as in othersincluded in this paper, the authors should ex-pect mul t ico l l inear i ty among some of theexplanatory variables. Some of these difficul-ties perhaps can be overcome by formulating

193

equations based on the behavioral postulates ofthe decision-making entities. If the object of thispart of the exercise is to derive some implica-tions for improvement in 'marketing' facilities,we do not have any usable results. I suspect thatthis approach may not yield such results. Itwould be preferable to measure the extra in-come that markets generate through a betterprimary wholesale price.

There can be more than one criterion to de-termine the efficiency of market channels, vonOppen and his colleagues choose pricing ef-ficiency measured in terms of coefficient of cor-relation of weekly prices between pairs of mar-kets, and at tempt to explain this in termsof distance between two markets, average mar-ket size, market age, numbers of traders andtelephones, density of production, density ofpopulation, etc. Though it is not explicit in thispaper, higher price efficiency defined in thoseterms, apart f rom being a reflection of the ef-ficiency of the information system, should helpproducers and traders in the choice of optimaldelivery or transaction locations. In qualitativeterms the empirical findings are consistent withexpectations, except in the case of number ofcommission agents and the share of ICRISATcrops. If, as suggested by these findings, fewertraders increase market efficiency, the reasoncould be economies of scale; but this requiresfurther theoretical analysis. Otherwise, mea-sures to improve pricing efficiency that fol lowfrom this work are familiar to policymakersworking in this area.

I am sure von Oppen would Iike to extend thistype of work to cover alternative indices of mar-keting efficiency. One such candidate is the dif-ference between the primary wholesale or farmgate price and the retail price. Efficiency mea-sured in terms of this differential wi l l throw con-siderable light on income distribution and po-tential incentives for higher product ion. Myown interest in this area has been to look at thestructure of submarkets that link the producerwi th the consumer. At an aggregate level, whatappears competitive may turn out to be non-competitive when we look at the structure ofsubmarkets.

I am glad that the economists at ICRISAT arerightly concerned with the implications of vari-ous types of interventions in grain trade by thegovernment. One that von Oppen and his col-leagues have analyzed in the paper is the impli-

194

cations of food zoning. Their provisional con-clusions, even though statistically not sig-nificant, are consistent with the predictions ofliberal free-trade theory. Even if they were sig-nificant, I would have hesitation in using themto shape policies until I understood the role ofother variables such as traders and their inven-tory policies. My own preference would be tostructure a more comprehensive model that in-corporates not only the major intervention vari-ables, such as zon ing , inven to r ies ,uniform prices etc., but also the various propor-tions of public and private trade. l a m confidentfuture research efforts wil l be deployed in thisdirection.

Chai rman's Summary

Jock R. Anderson

The two papers, on marketing reforms in Maliand on related developments in India, openedup a topic that is discussed again in Chapter6 — hence I can be brief knowing that the issueswil l come up again in discussions of the Harrisspaper in that Chapter.

First, some general comments were raisedduring the discussion on this subject:

1. Marketing matters are important abso-lutely, and at least as crucial as productionproblems in considering development.

2. They are essentially more complex, involv-ing challenging problems of defining andanalyzing:

• market structures (supply and demandsystems)

• information systems (and relevant data)• institutional arrangements• concepts of competit ion and efficiency

(and consequently)• economic philosophy and values of

analysts and observers.For technology design the main conclusion

was that there was a need to understand thenature of consumer preferences (especially inthe Indian work). However, it is useful to reflectthat these are not inviolate and can, and mayindeed, change over time. A good example ofthis was the change in tastes for chapatis afterthe release of HYVs of wheat in the mid-1960s inIndia.

Food grain marketing policy involves issuesof political economy and the politics of donorand government agencies and recipient gov-ernments; this is, therefore, an intrinsicallycontroversial topic. Given the profound ignor-ance of the economic realities, policy initiativesshould seemingly be taken very cautiously.

For national research programs, data prob-lems in marketing research are tremendous,and in this regard the challenges seem to bemuch greater for the African and Brazilian SATsthan in India and Mexico. To take up the sug-gestion for greater attention to micro-level sup-ply data wil l add to the challenge.

For ICRISAT socioeconomists, sadly, the im-plications are less clear to me. Although if theseeconomists can continue to do the good qualitywork reported by von Oppen and his col-leagues, we will clearly be much wiser soonerthan later.

Being realistic, ICRISAT won't want to enterinto major confrontations with Governments;perhaps the International Food Policy ResearchInstitute may have more of a role in this forum.

One thing is clear: quality studies that can befacilitated by ICRISAT in one way or another wil lenhance debate, will enhance policy making,and we hope wil l enhance the welfare of pro-ducers and consumers.

195

Chapter 5

Soc ioeconomics o f Improved An ima l -d rawn Implementsand Mechan iza t ion

Observat ions on the Economics of Trac tors ,Bul locks, and Wheeled Tool Carr iers

in the Semi-Ar id Tropics of India

H. P. B inswanger , R. D. Ghodake, and G. E. Th ie rs te in*

Abstract

This paper reviews the available survey evidence on the economics of tractor cultivation for semi-arid tropical (SAT) India. We find that tractor cultivation, as generally practiced, does not improve cropping intensity or yields and it displaces labor. This provides a justification for the emphasis on bullock power in the Farming Systems Research Program (FSRP) of ICRISAT.

The evidence now available with ICRISAT on the economics of wheeled tool carriers is

then reviewed. It is found that, even under the most favorable circumstances assumed,

such machines cannot compete on a cost basis with the traditional implements in

traditional agriculture. They must provide yield advantages of the order of 200 to 400

kglha to justify their higher costs. Experiment station evidence indicates that yield

advantage in excess of this can be achieved with the improved soil management

techniques made possible by using the tool carrier in Alfisols and deep Vertisols.

However, such evidence is not yet available at the farmer's field level.

From its inception ICRISAT has made a con-scious decision to attempt to improve existingfarming systems by technologies that, as far aspossible, are based on resources presentlyavailable in the SAT. For SAT India, in particular,this has meant an emphasis on labor-intensivetechnologies and on bullock power rather thanmechanical power.1

The justification for an emphasis on bullockpower rests on several grounds. First, Sub-rahmanyam and Ryan (1975) show that of thetotal animal + mechanical power input into In-dian agriculture, 89% came f rom animals in1966. For the predominantly SAT states ofKarnataka, Madhya Pradesh, Maharashtra, thepercentages were 89%, 96%, and 75%, re-spectively. Since1966 little mechanical powerinvestment has taken place in the SAT of India,with the exception of additions of pump sets for

* Principal Economist,Engineer, ICRISAT.

Economist, and Agricul tural

1. However, it is dear that for pumping purposes,mechanical power is replacing bullock power inthe SAT, and this shift Is considered desirable.

well-irrigation. Tractor investment has beenalmost entirely concentrated outside the nonir-rigated SAT areas (See Table B-3 in Binswanger1978). Calculations made by Stout and Downing(1975) show that tractor power is yet of littleimportance in most areas of the African SAT. Onthe basis of 1970 FAO data, they show that of thetotal arable land in Nigeria and Tanzania, morethan 98% is still cultivated by draft animals andmanual labor.2

A second justification is that — where wagerates are low and where little potential for areaexpansion exists, as in SAT Asia — tractorshave extremely low or negative private andsocial rates of return. There is evidence thattractors are used extensively for transportationand as a substitution of one source of power foranother rather than as a means of improvingmanagement or cultural practices. In addit ion,they displace labor, and their economies ofscale lead to further growth of large farms at theexpense of small farms (Binswanger 1978).

2. We are not aware of similar more recent calcula-t ions for the SAT areas of Afr ica, Southeast Asia, orBrazil.

199

These conclusions hold for almost all of SouthAsia and, by extension, to all areas with similarland: man ratios and lack of scope for areaexpansion. But they cannot be applied to SATcountries in Africa and South America. In thefirst section of this paper we wil l review in moredetail the available evidence on the economicsof tractors in SAT India, and underpin morethoroughly the justification of ICRISAT's em-phasis on bullock power.

A third consideration is that in the mechani-zation debate too much emphasis has beenplaced on the source of power rather than theimplement attached to it. As Giles (1975) pointsout, it is the implement that converts the powerinto something beneficial to the plants that isimportant. With the few exceptions wheremachines require high speeds or concentratedpower, improved implements can be pulled aswell by bullocks as by tractors. In particular, thedevelopment of various types of wheeled toolcarriers makes it technically possible to achievewith bullock-drawn implements most of theprecision in field operations that is achievablewith tractors.3

The second section of this paper reviews thetesting and modification of wheeled tool car-riers and their ancillary implements at ICRISAT.In the third section, we briefly compare the littlewe now know about the efficiency of newwheeled tool carriers wi th data from village-level studies (VLS). In the fourth section, we usesimple partial budgeting techniques to attempta rough economic evaluation of the wheeledtool carrier wi th traditional methods of cultiva-t ion. We wil l see there that, even under the mostfavorable assumptions, wheeled tool carrierswil l have to provide a distinct yield advantage tobecome competitive wi th traditional imple-ments. The final section reviews the potentialsources of such yield increases; it also brieflydiscusses the other directions of work inICRISAT's agricultural engineering program.

It should be noted that the economic assess-ment presented here is preliminary in thatexperiments now under way and further

3. An except ion to this is mechanical harvesting inthe absence of supplemental mechanical power. Inthe United States, for example, the uti l ization ofreapers in the nineteenth century was condit ionalon a shift f rom low-speed bullocks to faster horses.

analysis of the data from village studies maywell lead to revisions in the assumptions under-lying the calculations. Furthermore, these calcu-lations are useful only for the Asian SAT, sincetractor prices in other SAT countries differwidely and there is scope for area expansion inother countries.

The Economic Impactof Tractors in SAT India

In a recent review of the survey research on theimpact of tractors in South Asia, Binswanger(1978) reached the fol lowing conclusions:

The tractor surveys fail to provide evidence thattractors are responsible for substantial increasesin intensity, y ields, t imel iness, and gross returnson farms in India, Pakistan, and Nepal. At best,such benefits may exist but are so small that theycannot be detected and statistically supported,even w i th very massive survey research efforts.This is in marked contrast to new crop varietiesor i r r igat ion, where anybody wou ld be surprisedif he fai led to f ind statistically signif icant yieldeffects, even in fairly modest survey efforts.Indeed, the fairly consistent picture emergingf rom the surveys largely supports the v iew thattractors are substitutes for labor and bullockpower, and thus implies that, at existing andconstant wages and bullock costs, tractors fail tobe a strong engine of g rowth . They wou ld gainsuch a role only under rapidly rising prices ofthose factors of product ion which they have thepotential to replace.

In view of this f inding, many of the benefit-cost studies reported may have overestimatedthe benefits, both social and private, that resultfrom the agricultural uses of tractors. Except insituations where area effects are possible — orby renting or buying land from others — privatereturns to tractors from agricultural operationsmust be close to zero, or even negative atcurrent fuel prices.

That these conclusions hold especially truefor the SAT of South Asia can be seen fromTables 1 and 2. In these tables we use thefol lowing symbols to distinguish farm types:

B = Bullock farmsTO = Tractor-owning farmsTH = Tractor-hiring farms

Tables 1 and 2 are organized as fol lows. For

201

202

each study, column 2 lists the items compared.The first line has a B in column 2 and gives theabsolute value of the variables for the bullockfarms in parentheses. The following lines givethe percentage difference between bullockfarms and other farm types. For example, theline B-TO in column 3 of Table 1 indicates thatlabor use per hectare in Dharwar district was2 . 1 % higher on tractor-owning farms than onbullock farms, wi th the value of the bullock farmas the basis for the percentage difference.

All tables are on a per hectare basis. Humanlabor and bullock use is measured either inlabor/bullock days or in labor/bullock years,depending on how the authors measured it.Bullock labor is reported for single bullocks, notpairs. Intensity means gross cropped area di-vided by net cropped area expressed as a percentage. The percentage differences of in-tensity are relative to the intensity value ofbullock farms. Fertilizer use is given either in kgof plant nutrients (NPK) applied per hectare, orvalue in rupees of total fertilizer applied perhectare. Column 8 is used for various inputs,depending on information available or useful.Labor, bullocks, and production are measuredper hectare of farm size. In Tables 1 and 2 theAlfisol (red soil) area is represented by DholkaTaluk in Gujarat, while the Vertisol (black soil)areas are represented by three areas ofMaharashtra, Dharwar district in Karnataka, andNarsingpur district in Madhya Pradesh.

In the SAT areas, cropping intensity of bullockfarms is slightly in excess of 100% and theintensity of tractor farms is in no case more than10% greater. In Kundgol Taluk in Karnataka (arainy season-fallow area), tractor farms have a statistically significant lower intensity (5.7%)and in Satara district in Maharashtra the tractorfarms have an 8.4% lower intensity. Thus,evidence of higher crop intensities due to trac-tors is lacking in SAT areas. This lack of higherintensity is not so surprising since cropping isoften restricted to one season by lack of mois-ture or management techniques in these areas.New techniques being developed shouldchange this, particularly in deep Vertisol areas.

The Sapre (1969) study (Table 2, last column)attempted to measure the increase in cultivatedarea directly attributable to tractors and found itto be a modest 3.6%. Substantial area effectshave therefore not yet been documented for theSAT of South Asia.

Yield advantages of tractors (Table 1, column7) appear to exist in Kundgol Taluk of Karnatakadistrict, which belongs to the relatively highrainfall transition zone, and also in Narsingpurdistrict of Madhya Pradesh. In Kundgol taluk,however, tractor farms use 60% more fertilizerthan bullock-operated farms, and the yield ad-vantages fall far short of this figure. The re-quired fertilizer information is not available forNarsingpur. As shown in the other South Asianstudies reviewed, yield advantages on tractorfarms cannot be attributed to the tractors butare, to a large extent — possibly entirely — caused by fertilizer differences and other fac-tors.

On a per hectare basis, labor inputs ontractor-owning farms are the same or moder-ately to substantially lower than on bullockfarms, wi th the exception of Satara district inMaharashtra (Table 2). However, in Satara thetractor farms have an additional 85% of theirarea under irrigation, and the increase in laboruse is clearly an irrigation effect. As presentlyused, therefore, tractors are clearly labor-displacing in SAT India.

Evidently the use of tractors reduces bullockuse. It is clear from the two tables that bullockstocks decline by a lower proportion than bul-lock f lows (hours used) when tractors are pre-sent, and that in none of the areas have tractorsenabled farmers to completely replace bullocks.The tractorization has been selective and con-centrated on field preparation and transport,where concentrated power and high speed mayprovide the tractor with an advantage, even inthis low-wage environment.

Technical Aspects of WheeledTool Carriers

From the beginning of the FSRP at ICRISATCenter in 1973, equipment development playedan important role as a result of two relateddecisions: operational-scale systems researchwas initiated, and the ridge-and-furrow systemwas introduced as one of the major land man-agement treatments. During the first 2 years,two equipment systems were used: (1) tradi-tional bullock-drawn implements, and (2) con-ventional tractors and implements.

Tractors were not used after 1974, because itwas concluded that bullocks were going to be

203

the primary power source in SAT India. Henceonly bullock power and human labor have beenut i l ized for all cul tural operat ions inoperational-scale research since then. Thebroadbed-and-furrow system, introduced in1975 as a replacement for the narrow ridge andfurrow system, has shown superior results,particularly in the Vertisols at ICRISAT Center(FSRP Annual Reports 1975-76 and 1976-77).In the Vertisols, the practice of performingprimary til lage immediately after harvesting thepostrainy season crop was also introduced.This facilitated moisture conservation and per-mitted dry seeding before the onset of themonsoon rains. The use of double cropping inthe Vertisols, and to some extent in the Alfisols,necessitated rapid tillage and seedbed prepara-t ion after the harvest of the first crop and beforethe planting of the second crop. While it waspossible to perform all these operations wi thlocally available animal-drawn implements, theperformance generally was not good. Forexample, the t ime required for tillage opera-tions was high, whi le planting precision waspoor. It was also difficult to make uniform bedswi th this machinery.

An important objective of the equipmentdevelopment project at ICRISAT is to reduceboth the energy and t ime required to produce a given amount of crop yield. This approach issimilar to that being used at IITA in Nigeria (IITAFarming Systems subprogram report 1978). InFigure 1, two isoquants are depicted in which,for illustrative purpose only, it is assumed that(energy x time) equals a constant. The K1

isoquant represents the "convent ional" sys-tem, characterized by tractorized farming withits high energy and relatively low t ime require-ment at one extreme. At the other extreme ismanual farming, using exclusively human laborwi th relatively low energy but high t ime re-quirement. The goal is to move to a lower K2

isoquant of an "alternative technology," inwhich both t ime and energy inputs are reducedwhile maintaining the same output level. Theopt imum position on the K2 isoquant wi l l de-pend on technological, economic, and socialfactors appropriate to any particular location.

The broadbed-and-furrow system that isshowing promise consists of a bed width of100 cm and furrows 50 cm wide. The originalnarrow ridges of 75-cm width were not stableduring the monsoon rains and therefore d id not

204

Figure 1. The concept of alternative technol-ogy options.

provide the required erosion control; that wasone of the primary objectives in their use. In thebroadbed-and-furrow system, the furrows areabout 15 cm deep and laid out on a slope with a gradient of 0.4 to 0.8%, depending on soil typeand land configuration. This provides erosionprotection as well as improved drainage duringtimes of excess rainfall. The furrows have alsobeen used to provide supplemental irrigationduring times of moisture stress.

In 1975 a search was initiated for bullock-drawn machinery that was better adapted to thecultivation practices being used in the FSRPand, at the same t ime, might provide an alterna-tive technology option leading to a reduction inboth t ime and energy. The machine with mostpromise was a multipurpose wheeled tool car-rier. This is a frame supported on two wheelsand pulled by a pair of bullocks. A range ofimplements can be attached to this toolbar withsimple clamps. Depth adjustment and lateralspacing of all implements are relatively simple.Several tool carriers were subsequentlyevaluated, of which the best and most versatilewas the "Tropicultor," initially designed byMr. Jean Nolle of France.

Any good wheeled tool carrier design has twoattractive features: (1) it has one basic frame towhich a variety of implements can be attached,similar to a tractor; (2) a cart body can bemounted on the same frame, enabling use ofthe carrier for both agricultural operations andtransportation and spreading the depreciationcosts over a larger number of hours of use. The

cart body of the Tropicultor can be mounted anddismounted in a matter of minutes.

An important feature of the improved farmingsystems being developed at ICRISAT is thematching of the wheeled tool carrier and thebroadbed-and-furrow system. The wheel spac-ing on the wheeled tool carrier is adjusted to 150cm so that the bullocks and wheels alwaystravel in the furrow. The full width of a bed iscovered in each operation that is performed.Thus the distance that must be travelled tocover 1 ha is greatly reduced with the wheeledtool carrier compared with most of the tradi-tional animal-drawn implements currently usedin India. Figure 2 shows that a farmer musttravel about 66 km to cover a hectare whenusing a traditional (desi) single type plow with a working width of 15 cm. When a blade harrow of75-cm width is used, the travel distance is about13 km. However, with the wheeled tool carrier,covering 150 cm per pass, the travel distance isreduced to above 7 km. Even if the speed isreduced by 50%, because of the higher draftwhen plowing with the wheeled tool carrier ascompared wi th the traditional plow, the operat-ing efficiency is improved by a factor of five.

An essential element of the improved systemis the maintenance of the beds as a semiperma-nent feature. Thus development costs are incur-red only once. The use of min imum tillagepracticed during the dry season is an additionalattractive feature. Right and left moldboard

Figure 2. Width of bullock-drawn implement

and required travel distance to

cover one hectare.

plows are used simultaneously on the outeredge of the bed at a depth of 5 to 7 cm. Thus only40% of the bed is plowed at a very shallowdepth. In the Vertisols this is done early in thedry season immediately after the second crop isharvested. At that t ime bullocks are normallystill in good physical condition, and there issome moisture remaining in the soil. Sincethese soils are self-mulching, any small showerduring the dry season permits nature to takecare of much of the secondary tillage. Only a bed-shaping operation needs to be done beforeplanting.

Where the onset of the monsoon rains isusually quite distinct and rapid, as inHyderabad, India, it is desirable to plant dry inthe Vertisols just before the main rains begin.When soils are wet they are very difficult towork. In 1978 the planting rate on the ICRISAToperational watersheds was 21/2 to 3 ha/machine per 12-hr day. A double shift wasoperated, using a different pair of bullocks foreach shift. This planting rate was maintainedwith four wheeled tool carriers over more than50 ha. Where planting was delayed until afterthe rains had started, the planting rate wasreduced by 50%.

The planters presently in use are consideredvery good, but unfortunately they represent anadditional investment of Rs 3200 over the costof the wheeled tool carrier for a four-rowplanter. A much simpler and cheaper modelthat shows considerable promise is under in-vestigation, and subsequent cost calculationsare based on this implement. The furrowopener, which assures proper placement of theseed, and the press wheel, which assures goodseed-soil contact to enhance seedl ingemergence, remain the same. The primary dif-ference is in the seed-metering mechanism. Inthe low cost model the farmer dibbles the seedby hand. This is a traditional practice andexperienced farmers are very skilled in it.

As indicated previously, the wheeled toolcarrier is likely to prove costly for small farmers.Since the broadbed system shows promise as a means of improving yields and income, it isdesirable to develop lower cost machinery sothat these benefits can accrue to smaller far-mers also. A local wooden bullock cart found inparts of Central India has recently been modi-fied by the addition of a toolbar to make itsuitable for use on the broadbed-and-furrow

205

system. It is anticipated that the cost of thisequipment wi l l be about 25% of the cost of theTropicultor. Its life wil l probably be half as long,its versatility wi l l be lower, and it wil l not becapable of being used for transportation, exceptof people.

For those activities where there are restric-tions or l imitations on the use of animal power,the use of tractors is being developed. Thereintroduction of the tractor into the researchprogram wil l provide both technical andeconomic data to compare its performance withanimal-drawn machinery. This information iscurrently not available. The range of machinerybeing developed wil l fill different requirementsand meet a variety of needs.

Techn ica l Ef f ic iencyof Whee led Too l Carr ierand Trad i t iona l Imp lemen ts

In Table 3 we compare the bullock labor inputsof the broadbed-and-furrow system watershedtreatments, implemented on an operationalscale wi th wheeled tool carriers at ICRISAT

Center, wi th the bullock labor inputs in areaswith similar soils in the Village Level Studies ofICRISAT. The watershed treatments are thosewhich, on the basis of past experience atICRISAT Center, give the highest returns. Thesecomparisons are comparisons of systems, notindividual crops, and the fol lowing qualifi-cations have to be noted: (1) in the Alfisols anddeep Vertisols, the ICRISAT treatments implyhigher cropping intensities than the traditionalvillage practices. In the Alfisols, either a ratooncrop or pigeonpea in high densities is grown inthe early postrainy season; in the deep Ver-tisols, use is made of the rainy season whileland is mostly left fal low in the Sholapur area;and (2) the bullocks used to pull the wheeledtool carrier at ICRISAT Center are stronger thanthose used in villages.4 For these reasons, theabsolute values are not necessarily comparablebut we want to compare trends.

4. Smaller bullocks encountered in the tests in the

Vertisol villages had no t rouble pul l ing theequipment, but such a reduction in bullock sizewou ld imply a reduct ion in operating speed foroperations requir ing relatively high draft input.

Table 3. A comparison of bullock labor inputs from preferr ed broadbed-and-furrow systems usinga wheeled tool carrier with the traditional Implements in v illages in a SAT India.

Soil type

Alf isols

Medium-deep Vertisols

Deep Vertisols

Bullock pair hr per ha of net cul-t ivated area under two prefer-red t r ea tmen ts of ICRISAT

watershedsa

1976-77 1977-78

87b - (154)76c

42d (146)30e

43d (154)30e

Note : Figures in parentheses are average c ropp ing intensi t ies.

Datavi l lage average of

1975-76 & 1976-77

182 (117)(Aurepalle) f

93 (105)(Kanzara)54 (105)

(Shirapur)

Proport ion of hiredbullock labor use tototal bullock use in

vil lagesg

( % )

10

20

37

a. Data on Al f isols fo r the year 1977-78 have not been repor ted because of severe crop damage due to Insect attack. However ,

f i gu res for 1976-77 on th is so i l cou ld be used for observ ing the t rend .

b. S o r g h u m f o l l o w e d by ra toon ing .

c. Pigeonpea in te rc ropped w i t h setar ia.

d. Maize f o l l owed by ch ickpea.

e. P igeonpea In terc ropped w i t h maize.f . This is the on ly v i l l age in the studies whe re bu l lock labor used Increased sharp ly w i t h increased f a r m size.

g. In al l t he v i l lages s tud ied except Aurepa l le , there was a sharp decline in the p ropo r t i on of h i red bu l lock labor use w i t h

increased f a r m size.

206

-

First, note that the traditional systems makemuch larger use of bullock power on Alfisolsthan on Vertisols, and larger use on medium-deep Vertisols than on deep Vertisols. The lastdifference is easily understood because in thedeep Vertisol villages studied, rainy seasonfallowing prevails, whereas in medium deepVertisol villages studied the rainy season is themain cropping season. The high use of bullocklabor on the Alfisols is difficult to explain. Onepossible reason could be the comparativelysmaller sized bullocks present in Aurepallevillage than in the Vertisol villages, which mightresult in more hours of bullock labor use onAlfisols without much difference in actualenergy input. Besides this, the contribution ofowned bullocks is 90% in Alfisol villages, whilefor medium deep and deep Vertisol villages, thefigures a re 80 and 60%, respectively. The higherproportionate use of owned bullocks obviouslyencourages higher total bullock labor use.5

Further, it is now well-evidenced from theresearch work on the economic aspects of weedcontrol by Binswanger and Shetty (1977) thatweeding and interculture operations are per-formed more intensively on Alfisols than onVertisols, which further increases the per hec-tare bullock labor utilization on Alfisols.

The absolute bullock labor use figures perhectare of net cultivated area are always lowerunder watershed treatments than under exist-ing village systems, despite the higher croppingintensity on the watersheds. When thesefigures are adjusted for cropping intensities andexpressed on a gross cropped hectare areasbasis, bullock labor use on the ICRISAT water-sheds is still less than in village situations. Thewatershed treatments utilize only 28 to 47 % ofthe bullock labor used in corresponding villagesystems.

This demonstrates the much higher efficiencyof bullock labor with wheeled tool carriersunder improved systems. There thus appears tobe scope for bullock labor saving under theimproved system.

5. If we look at bullock labor use trends underwatershed treatments then a similar situation isobserved, which indicates that the above factorsmay not be the only ones to explain this observa-t ion. However, in the absence of figures for 1977-78 under Alf isol condit ions, one cannot be sure ofthe general i ty of this trend in the watersheds.

Table 4 compares the bullock labor inputs byoperation on the medium-deep and deep Ver-tisols. The comparison between the "tradi-t ional" method on ICRISAT watersheds and thebroadbed-and-furrow system of the newtechnology shows that the wheeled tool carrierappears to be able to achieve the largest in-crease in technical efficiency in tillage on themedium-deep Vertisols, while the advantageson the deep Vertisols appear to be more mod-est.

The comparison between existing villagepractices and the bed-and-furrow system wi ththe wheeled tool carrier is most instructive forthe medium deep Vertisols as well , since thetraditional system there operates at nearly thesame level of cropping intensity. It is clear thatthe wheeled tool carrier achieves substantialgains in efficiency for tillage (a reduction by a factor of nearly 2 or more) and for interrowcultivation where the reduction appears to beeven larger.6 Note, however, that when tried inthe study villages during a very wet year,farmers seemed to prefer their traditional sys-tem for interrow cultivation in the Vertisols.

It is clear from the comparison that wheeledtool carriers (with stronger bullock pairs) wouldbe able to perform the traditional operations inthe villages at a much faster rate; perhaps a factor of 2 or more may not be an unreasonableexpectation.

Wheeled Too l Carriers versusTradi t ional Imp lements :A Cost Compar ison

The simplest way to look at the economics ofthe wheeled tool carrier is to compute a contracthire rate that an enterpreneur would have tocharge his customers to make a profit on hisinvestment in a sturdy pair of bullocks, the toolcarrier, and the additional implements required.We have done this under various assumptionsabout machine costs, profit rates, and utilizationrates of the bullocks and machines.7 We can

6. This may be due in part to the reduction in weedgrowth achieved by the higher competit iveness ofthe crops against weeds in the ICRISAT systemscompared wi th the tradit ional ones.

7. The first such computat ion was done by J. G. Ryanin an internal ICRISAT memo, and we have bor-rowed his technique wi th minor modif icat ions.

207

then compare this rental cost with existingbullock hire rates in our study villages, whichcenter around Rs. 15 per day for a pair ofbullocks with driver and implement or cart, andaround Rs. 12 for a bullock pair without animplement or cart. To take into account thehigher efficiency of the tool carrier we make a comparison for the medium-deep black soils.Computations based on field capabilities indi-cate that one machine covering 15 ha of landcould be employed wi th the improvedbroadbed-and-furrow system for about 80 days.This implies a yearly bullock-with-machineinput of 5.4 days per ha. Note, therefore, that allthe calculations shown below assume a full useof the wheeled tool carrier for all operations.Such an assumption may be unrealistic if wetake as a guide the experience with the intro-duction of tractors, where only a limited number

of operations is initially carried out by tractors.Table 4 indicates an annual bullock labor inputof 93 hours per ha for Kanzara, which leads to a bullock labor input of 11.6 days per ha. Onaverage, to do what a sturdy pair of bullocks cando in 1 day with a wheeled tool carrier in theimproved system, takes nearly 2.15 bullockdays in the traditional system. Note that thecomparisons of the last section also support anefficiency factor of more than two on average.We should, therefore, compare the rental ratesfor the wheeled tool carrier and bullock pairwi th the traditional rental with implements ofRs 32. We wil l call this the equivalent rental rate.

Table 5 shows such rental rates for variousassumptions, which are described in categories1 to 7. Rental rates are computed for all of thefol lowing combinations:

• Wheeled tool carrier plus bullock pair plus

Table 4 . A compar ison of annua l bu l lock labor inputs fo r va r ious operat ions by t rad i t iona lmethods and w i t h t h e whee led too l carr ier.

Soil type/vi l lage

Medium-deepVertisols(Kanzara)

DeepVertisols(Kalman)

Operation

Preparatoryti l lageManur ing & ferti l izationSowing , trans-planting etc.Interculture

Preparatoryti l lageManur ing & fert i l izationSowing , trans-plant ing, etc.Interculture

Existing vi l lagepractice (Averageof 1975-76 and1976-77) (pair

hr/NCA).

45.7

2.4

12.5

24.8

25.6

1.5

10.3

7.6

Tradit ional methodas defined on ICRISATresearch watersheds

1977-78(pair hr/NCA)

70.3

—

6.3

30.5

-

12.2

Broadbed-and-fur row systemwi th wheeled

tool carrier 1977-78(pair hr/NCA)a

25.8*18.0c

7.Ib

5.1c

7.4b

4.5c

6.1b

5.2c

24.3b

1.84c

6.6b

3.3c

6.6b

3.2C

7.6b

7.3c

a. NCA-Net cultivated area.b. Sole maize followed by chickpea.c. Mails intercropped with pigeonpea.

208

209

driver versus wheeled tool carrier only.• A utilization of 80 agricultural days versus

180 days, where the additional 100 daysare assumed to be transport days. Notethat in nonirrigated areas of India, bullockutilization for agricultural purposes variesaround 60 days (Subrahmanyam andRyan, 1975; based on Farm ManagementStudies). Without irrigation for a secondcrop, utilization rates of above 80 days forcropping purposes wil l be difficult toachieve. Since bullock life is partly depen-dant on the utilization rate, the working lifehas been reduced to 6 years in the case ofan added utilization of 100 transport days.

• Profit rates of 10% and 20% to owners ofmachines and bullock pairs. These ratesare over and above a borrowing rate (oropportunity cost of capital) of 20% andimply attractive total rates of return of 30%and 40%, respectively.8

• Three different cost levels for the wheeledtool carriers. Other assumptions are dis-cussed in the footnote to Table 5.

It is evident that the equivalent rental rate for the traditional implements of Rs. 32 cannot be achieved even under the most favorable as-sumptions of a low-cost machine and a 10%profit rate, although the difference is only Rs. 3 in that case. Not surprisingly, increases in utili-zation rates imply large reductions in rentalrates of nearly 50%. The machine thus must bemade to work for nonagricultural purposes aswell . Reductions in the cost of machines implyequally large reductions in the rental rate for themachine only. However, note that the cost ofbullocks and the driver are the larger portion ofthe cost of the package under nearly all assump-tions. Therefore, a 50% reduction in machinecosts alone generally reduces the rental rate ofthe entire package by only around 27%.

In future research attention wil l thus have tobe focused on how to increase utilization rates.Furthermore, efficiency increases of themachine itself become of primary importancesince they would al low its operation wi th smal-ler and cheaper bullocks. Finally, It appears thatreduction in machine costs are helpful only if

8. Given the fact that these Investments ere riskybecause of the general riskiness of agriculture inthe SAT, we believe that such high rates areappropriate.

they do not have to be purchased by designchanges that reduce the efficiency of themachine, thus increasing the bullock cost com-ponent.

One of the main conclusions reached here isthat all potential rental rates exceed the equiva-lent rental rate of traditional equipment at thefactor price levels of Indian SAT farmers. Thesefarmers, especially the small ones, cannot beexpected to be interested in the package unlessit provides them wi th sufficient yield gains perhectare to compensate for the added costs.

In parentheses we give the compensatory yield increases that farmers wil l require to offsetthe increased machinery and bullock cost im-plied if they switch f rom their traditional systemto the wheeled tool carrier. These are given inrupees per ha, but since 1 kg of coarse graincosts a little less than Re. 1.00 we can think ofthe figures approximately as yield increases inkg of sorghum equivalents that must be gener-ated by the machine to offset its higher costs.

At annual utilization levels of 80 agriculturaldays, the improved machines must give a yieldadvangtage of 200 to 400 kg/ha. This is no smallrequirement, and the next section explores itsfeasibility. However, if the machines can beused for transport purposes to a considerableextent, the compensatory yield increases be-come relatively small (10 to 130 kg/ha).

We should note here, however, that the com-pensatory yield increase does not yet give thefarmer who rents the machine any benefit at all.It only covers his extra cost. To induce a smallfarmer who now owns his own bullock pair toundertake the change to renting the tool carrierrequires that the yield advantage be evenhigher than that shown in Table 5. For a switchfrom owning to renting it probably must bemuch higher because the small farmer wil l haveto be compensated for the added inconveni-ence of renting rather than owning as well.

The calculations can also be interpreted asapplicable to a farmer operating 15 ha of landwho may contemplate a switch from two orthree small bullock pairs to one large one and a wheeled tool carrier. Since he wil l capture therate of return on the investment, he may shift ifthe yield gains are equal to the compensatingyield increase.

Under such conditions the proposed technolo-gy does imply economies of scale. Unlesssmaller versions of the tool carrier suitable for

210

smaller bullock pairs can be developed, andprovide the same yield advantages as those ofthe larger models, successful introduction ofthe tool carriers favors large farmers relative tosmall ones. Furthermore, for each 15 ha farm1.5 bullock drivers will become unemployed.Such unemployment can only be acceptablesocially if the yield advantages of the machineare indeed large and consequently increaselabor requirements in other activities.

Potent ia l Y ie ld Advantages

Information on the yield impact of improvedsoil management techniques described in thesecond section comes from the Farming Sys-tems Research Program of ICRISAT. The firsttype of information is the Steps in ImprovedTechnology experiments (SIIT) described morefully in Ryan, Sarin, and Pereira (1978), and inthe Annual Reports of the Farming SystemsResearch Program. From these experiments,information is available for deep Vertisols(black soils). The operational-scale watershedsalso provide information about plots on a muchlarger scale, similar to the sizes at which far-mers would operate. Table 6 summarizes thisinformation for the 2 years 1976-77 and 1977-78. Comparable information under farmers'field conditions will only become available in a year or two.

Table 6 gives the yield increase in rupees perhectare that can be attributed to the soil andcrop management system associated wi th thewheeled tool carrier. From there we see thatyield advantages in excess of Rs. 1000 arepossible in the experimental setting when theimproved cultivation practices are imposed ontop of improved varieties and fertilizer levels.Without added fertilizer, yield advantages in theAlfisols were only around Rs. 250, whereas inthe deep Vertisols they were around Rs. 500.

In the operational-scale watershed experi-ments, the largest yield advantage reportedwas Rs. 625 for the deep Vertisols. This is a littlemore than half of the Rs. 1046 achieved withimproved varieties and fertilizers in the smallerscale SIIT experiments. On the medium-deepVertisols the soil and crop management treat-ment appears to lead to modest losses.

On the basis of Table 6, we may conclude thatyield advantages in excess of the compensating yield increases required to pay for the extra costare a distinct possibility, even if the machinesare used for only 80 days per year.

Conclusions

The available survey research evidence fromSAT India indicates that it is not possible toexpect social returns from tractors to be posi-tive in this area in the manner in which they are

Table 6 . Y ie ld increases a t t r ibu tab le to cu l t ivat ion pra ct ices w i t h wheeled too l carr ier a t ICRISATCanter. a

Variety:Fertilizer:

Alf isolsMedium-deep

VertisolsDeep Vertisols

Steps-in-lmproved Technology (SIIT) experiments,w i th improved soil managementb

Local LocalLocal Improved

234 802

546 533

Improved ImprovedLocal Improved

262 1285

479 1046

Watersheds

ImprovedImproved

Maize fo l lowedby chickpea

- 1 9 0+ 166

ImprovedImproved

Maize and chickpeaintercropped

- 2 0 6+ 625

Source: Ryan, Sarin, and Pereira (1979, Appendix Tables 1 an d 2 and Table 3}. A fuller description of the experiment i s given

there. See also Annual Report, Farming Systems Program, 1 976-77.

a. Averages are for 1976-77 and 1977-76 and are expressed in rupees per hectare of net cropped area.

The experiments reported here were carried out with seedi ng attachments that were more expensive than the ones used for

the calculations in Table 5. However, the engineers expect to achieve the same precision with the cheaper devices bud geted.

b. During the year 1976/77 the Improved management treatme nt also Included a minor level of Insect control, but the dif ferences

in 1977/78 ere fully attributable to soil management techn iques.

211

currently used. However, the introduction of

new technology could change this, so it is

undoubtedly premature to discount tractors en-

tirely. This premise wil l be investigated at

ICRISAT in coming seasons. Nevertheless, at

present, tractors seem even less appropriate in

the SAT than in other areas of South Asia. The

emphasis on bullock power is thus well justi-

fied.

On the basis of evidence f rom the Economics

Program's Village Level Studies and experi-

ments of the Farming Systems Research Prog-

ram, we find that wheeled tool carriers wil l not

be competitive w i th existing soil and crop man-

agement techniques unless they generate yield

advantages in excess of 200 to 400 kg/ha.

However, the experimental evidence suggests

that the improved system may be able to

provide substantially larger yield increases in

the Alfisols and deep Vertisols. Therefore, work

on this type of equipment and testing it in

farmers' fields is amply justif ied, even in SAT

Asia. Since the largest benefits occur in con-

junction with reasonable levels of fertilizer use

and improved varieties, testing at the village

level should be concentrated on similar situ-

ations.

Acknowledgment

The authors thank R. Sarin and Harbans Lal for theprovision of data and J. G. Ryan for the computationalmethod for the analysis of section IV.

References

ACHARYA, T. K. T., JAGANNATHA RAO, PAWAR, R., and

BHASKER RAO, B. 1974. Relative economic efficiencyof different makes of tractors on tractorized farms inKarad Taluq. Agr i . and Agro-Industr ies Journal7(11).

BINSWANGER, H. P. 1978. The economics of tractors inSouth Asia. Agr icul tural Development Counci l , NewYork, and ICRISAT, Hyderabad India.

BINSWANGER, H. P., and SHETTY, S. V. R. 1977.

Economic aspects of weed control in semi-aridtropical areas of India. ICRISAT Economics Programoccasional paper 13.

CHANDRAMOULI. 1975. The effect of tractor use in thetransit ion and dry tracts of Dharwar district, Kar-nataka State. M.Sc. thesis, Department of Agricul-tural Economics, University of Agricul tural Sci-ences, Dharwar, Karnataka, India.

DESAI, D. K., and GOPINATH, G. 1975. Impact of Farm

Tractorization on Productivi ty and Employment(Gujarat State). Center for Management in Agricul-ture, Indian Institute of Management, Ahmedabad.

FARMING SYSTEMS RESEARCH PROGRAM ANNUAL REPORT,

1975-76 and 1976-77. ICRISAT, Hyderabad, India.

FARMING SYSTEMS PROGRAM, 1978. Report of the Ag-

r icultural Engineering Sub-Program, Crop Engineer-ing Department, IITA, Ibadan, Nigeria.

GILES, G. W. 1975. The reorientation of agriculturalmechanizat ion for the developing countr ies:policies and attitudes for action programs. In Effectsof Farm Mechanization on Production and Employ-ment. Rome: FAO.

KRANTZ, B. A., KAMPEN, J., and VIRMANI , S. M. 1978.

Soil and water conservation and uti l ization forincreased food product ion in the semi-arid tropics.Journal Article 30, ICRISAT, Hyderabad, India.

MISRA, S. P. 1976. Impact of tractorizat ion: a study in a tehsi l of Madhya Pradesh. Indian Journal of Agricul-tural Economics 31(3).

PAWAR, R., JAGANNATHA RAO, and ACHARYA, T. K. T.

1976. Energy consumpt ion in different farm systemsin western Maharashtra, Department of Agricul turalEconomics, Mahatma Phule Krishi Vidyapeeth,Maharashtra.

ROY, S. E. 1970. Agricul tural Engineering. Pages489-607 in Agr icul tural Year book., new vistas incrop yields. New Delhi : Indian Council of Agricul-tural Research.

RYAN, J. G., S A R I N , R., and PEREIRA, M. 1979. Assess-

ment of prospective soi l - , water-, and crop-management technologies for the semi-arid tropicsof peninsular India. Presented at the Workshop onSocioeconomic Constraints to Development ofSemi-Ar id Tropical Agr icul ture, 19-23 February1979, ICRISAT.

SPARE, S. P. 1969. A s tudy of tractor cult ivat ion inShahada. Gokhale Inst i tute of Polit ics andEconomics, Poona. (Gokhale Institute MonographSeries No. 7)

STOUT, B. A., and D O W N I N G , C. M. 1975. Agr icul tural

mechanization policy. In Effects of Farm Mechaniza-t ion on Production and Employment. Rome: FAO.

SUBRAHMANYAM, K. V., and R Y A N , J. G. 1975. Livestock

as a source of power in Indian agriculture: A briefreview. ICRISAT Economics Program occasionalpaper No . 12, ICRISAT.

212

Anima l -d ra f t Cul t ivat ion in Francophone Af r i ca

M. Le Mo igne*

Animal-draft cultivation was introduced intoFrancophone Africa a long t ime ago. It is gener-ally agreed that it dates back to 1930. But inresearch stations and administrative sectors itstarted in 1926, and in Madagascar, as early as1900. Therefore, it has been through a very longstage of introduction.

Its promotion started more recently, gener-ally after the countries became independent(1960-1962). This developed within schemesand projects initially for production and then fordevelopment. In this respect, it differs greatlyfrom the effort in the Anglophone countries. Atfirst, the initiative was taken by the schemerather than the farmer himself. Since credit,when available, and marketing of purely cashcrops were organized, the farmer was very de-pendent on this operational environment.

Brief History

Initially, all the equipment introduced into Fran-cophone West Africa came from France: bra-bant plows (a technique directly transferredfrom European agriculture) in Madagascar;l ong -mo ldboa rd Bajac p lows or short-moldboard Bourguignon plows in West Africa.EBRA equipment, with its more helical designand Huard equipment were also used. Thesewere limited introductions, however, made byadministrative authorities or research centers.Equipment f rom Yugoslavia, Japan, and USSRwas introduced later.

After independence of the countries, the ex-tension work by the traditional agricultural ser-vices had only a l imited impact because bothfacilities and personnel proved inadequate.Moreover, supply of equipment took a longtime and restricted the work. Some models

* Cent re d 'E tudes e t d ' E p e r i m e n t a t i o n duMachinisme Agricole Tropical (CEEMAT), Antony,France.

were no longer manufactured in Europe and theothers were not necessarily suited to the condi-tions. They were generally too heavy for theavailable animal power.

Between 1960 and 1965 several factors con-tr ibuted to the large-scale development ofanimal-draft cultivation. Some manufacturersagreed to continue the production of animal-drawn equipment using more suitable patents(inventors such as Nolle and Bariani haveplayed a prominent role here) and meeting thespecifications given by the research and studycenters and then 'transferring' these by spon-soring the first factory in Africa (SISCOMAin Senegal). Much later, other factories wereestabl ished in Cameroon (Tropic) , Mal i(SMECMA), and other countries.

Agr icu l tu ra l research, — wh ich is we l l -advanced in Senegal and Madagascar, — isnow in a position after testing cropping tech-niques and equipment to recommend technicalproduction systems based on soil cultivationwith the aid of light and heavy animal-draftequipment.

Intensified promotion through the implemen-tation of "product ion" operations (rice, cotton,groundnut, etc.) w i th international aid hashelped the spread of animal-draft methods tothe entire population.

Many tests and adaptations of animal-draftequipment have been carried out over the past18 years. Such equipment now in use is prob-ably viable except for that only recently intro-duced. We wi l l now discuss the range ofequipment presently available.

Specific Equipment

• SISCOMA or SMECMA single or doubleseeders for groundnut, millet, sorghum,maize, and, to a lesser extent upland rice.These are derived from the Fabre SuperEco seeders. A special hopper has beendesigned for cotton.

• Huard or Codamm type of 10- inch plows;

213

these are generally f i t ted wi th a semi-digger body.

• SISCOMA or SMECMA hoes (ex Fabre "oc-cidentale" model) drawn by donkeys orhorses. They can also be coupled anddrawn by a pair of bullocks. Their mainfunc t ion is w e e d i n g , and the use ofduckfoot tines improves their working per-formance.

• Manga hoes are used more often for culti-vation than for weeding.

• Donkey-drawn carts wi th a 400-kg payload,horse-drawn carts with a 1000-kg payloadand bullock-drawn carts wi th a 1500-kgpayload. The latter are less popular.

Simple Tool Carriers ('multiculteurs')for Til lage Operations

• The series based on Mouzon-Nolle pa-tents. The best known are the Sine 7 and 9 hoes (with some local adaptations). Theseare manufactured in practically all the fac-tories. The use of the Ariana plow is alsospreading. Attachments include cultiva-t ion and weeding implements, cultivatortines, plows, ridger bodies, and groundnutlifters.

• The series based on Arara-Bariani patents.They are also manufactured in all the fac-tories. The equipment is the same as theearlier series and is often used for cottoncultivation; it is considered heavier thanthe Sine equipment although the weightand function of the implements is quitesimilar for the two series.

• The new 'triangles' used in Upper Volta.Their frames are different f rom the Sine buthave similar agricultural applications.

Multipurpose Tool Carriers

• The Nolle tool carrier (Polyculteur) is fittedwi th a cart platform but so far it is notwidely used.

• The Nolle Tropicultor, a wheeled tool car-rier, is also not widely used.

• The high-performance Bambey-SISCOMAtool carrier is l imited to only one region inSenegal. It is equipped wi th conventionalseeders of the Super Eco type.

• The Le Lous Tropisem which is beingtested in another part of Senegal, has three

214

seeders with plastic rollers.Other equipment has been tested but does

not need to be added to the above list.At present the technical possib i l i t ies of

animal-draft cultivation include:- plowing 12 to15 cm deep; planting of most annual crops (ex-cept uncoated furry cotton used); weeding andcultivation operations for all crops and ridgingfor cotton; and-the lifting operation in harvest-ing of groundnut (lifting operation).

In addit ion, it is always possible to transportwater and materials wi th reliable carts.

Present Research

• Equipment is being researched that wi l lenable the fol lowing to become practic-able:

• Seeders for undelinted cotton.• specially shaped of tines (Jumbo buster),

enabling work under drier conditions andthus extending the operation period.

• auxi l iary equipment for herbicides (al-though this is not exactly part of animal-draft cultivation).

In spite of this diversity, technical systemsbased on animal-draf t cu l t ivat ion face thefol lowing constraints:

• Time constraints to carrying out ti l lage op-erations under good soil conditions;

• Technical constraints to burying organicmaterial, if required.

• Labor constraints to completing mechani-cal weeding operations.

• Time and labor constraints at harvest andduring conditioning or processing of pro-duce.

Relat ive Impor tanceof An ima l -d ra f t Cu l t i va t ion

Anima l -d ra f t cu l t i va t ion f igures are of tenexaggerated to show its Importance in the Fran-cophone countries of West Africa. We shall tryto place them in the proper perspective by com-paring them wi th the area under cultivation.Censuses — especially for areas under cerealand tuber crops — are often not very accurate.Thus, the data given below must be consideredas indicators only and not as an accurate rep-resentation of the situation in 1977 (Tables 1 to4).

In 1977-78,6000 farmers in the northern Ivory

Table 1 . Equipment numbers in West A f r i ca .

Items

Bullock-drawn plowsDonkey-drawn plows

Total plowsTool carriers(mult iculteurs)

Cultivators (donkey-,horse-drawn)

HarrowsSeedersCartsRidgers (ridgers

and/or weeders)Lifters

UpperVolta

12 0504 470

16 520

2 500

21 000

Mauri-Mal i tania

106 700 24001

40 555

14 058

10 7399 707 1001

52 204

Niger

3000

4300

900

9003300

15003300

Senegal

8 000

204 000

220 00089 600

5 00088 000

Chad

58 056

1 727

14 606

(3 883)

Benin

7450

185

1. Tentat ive f igures

Source: Au tho r

Table 2 . Ave rage densi ty o f an imal -drawn equ ipment in West Af r ica (ha /un i t ) .

Items

PlowMult iculteur or hoeRidgerHarrowSeederLifterCart

Upper Volta

190160--

8800-

325

Mali Niger

12 90031 690-

116129 3020

82024 820

Senegal

27811

445-

102525

Chad

19199199---76

Table 3 . Dens i ty o f equ ipment on mon i to red dry land areas o f West A f r i ca ( h a / u n i t ) .

Items

PlowTool carrier

(mult iculteur) or hoeRidgerHarrowSeeder

Groundnut lifterCartPair of bullocksDonkeyHand-sprayer

Cotton area,

Mali

6

15.4-

386373

-

181529

4

Groundnut area

Mali

-

64--71

-

117---

Cotton area,

Senegal

13

2513.4

--

-

---3.6

Groundnut area,

Senegal

227

6.5-8.1

--

3.9-4.5

5.6-6.710.3-12.9

--

-

Maradi,

Niger

833

90111

-268

310146

--

-

215

-

Tab le 4 . Dens i ty o f equ ipment in se lec ted areas o f var ious c ountr ies ( h a / u n i t ) .

Items

PlowHarrowRidger bodyWeeding tool

SicklePedal-operated thresherCartSprayer

Rice

Segou scheme,Mali

3.614----

8.4-

Rice

Office du Niger,Mali

610----

15-

Rice

Plaine de Kou,Upper Volta

1--

1

13-1

Sorghum/cotton

Ibohamana, Niger

6.8

-

6.8

-

-

-

-

7.5Coast were reported to be using animal-drawnequipment (multiculteurs). In Togo, animal-draft cultivation is practically nonexistent; thefigure of 800 animal-drawn implements for thenorthern region is disputed.

Animal-drawn Equipment and Farms

Table 5 must be interpreted wi th caution as theconcept of a farm can be difficult to define in a general way. We shall not consider differencesbased on social structure (patriarchy, extendedfamily, households, etc.); although these arestudied, they are not really included in a census.To be more factual, we shalI examine the case ofthe cash crop producer. He is the only individualknown in the schemes, as he comes to sell hisproduce and has an account (credit) wi th them.He may be the head of a large family or of a household. Consequently, the figures may beunderestimated. Unfortunately, we do not havemore accurate agricultural statistics.

As animal-draft cultivation has not developeduniformly, some areas are better equipped thanothers. It is surprising, however, that after manyyears of sensitization and extension, the gen-eral situation has not improved.

Factors Dete rmin ing t h e Spreadof An imal -dra f t Cu l t i va t ion

The agricultural, social, and economic envi-ronment as well as the planning of the promo-tion of animal-draft cultivation present positive

216

Tab le 5 . Dens i ty o f se lected equ ipment onfarms in var ious countr ies ( f a r m s /uni t ) .

Item

PlowHoeing unitSeederCart

Upper Volta(total agri-

culturalsector)

5546

250092

Mali Niger

2.8 1507.4 42

31 5005.7 136

Senegal

381.51.43

Source : Au tho r

and negative aspects that may explain its slowspread in the deve lopment of rura l com-munities.

Expected Yield Increase

Agricultural research has obtained importantresults in areas involving farm equipment, in-cluding the fol lowing:

• Timely sowing can generate yield gains of50% on groundnut and sorghum; earlymillet is less sensitive to the planting datebut can provide a yield gain of 20%. Cottonis particularly sensitive to the date of plant-ing,

• Several mechanized crop weeding canallow yield gains of 50% for groundnut andup to 175% for millet.

• Much research has been carried out on til-lage operations. Nicou summarizes the re-sults obtained in Senegal in Table 6.

Only results that have already been publishedare mentioned here. Unfortunately, equally ac-curate data on no-tillage and minimum tillagepractices, for example, are not available.

The effects of these improved cultural prac-tices are not cumulative (i.e., the sum of thepercentages given above does not representthe overall result). But the results of tillage op-erations can be considered since timely sowing,appropriate weeding operat ions, etc. werecombined in the trials. In any case, it is worthnoting that the simultaneous use of a number ofanimal-draf t cul t ivat ion techniques s igni f i -cantly increases production.

Table 6 . Percentage y ie ld Increases f r o mseedbed preparat ion w i t h an imal -d r a w n equ ipment In Senega l .

Crop

Grain-mil letGrain-sorghumGrain-maizeRainfed riceCottonGroundnut

Effect ofp lowing

alone

192450

1031724

Effect ofplowing-burying

232473

11231

9

Source: R. N icon, Inst i tute of Tropical Agr icu l tura l Research

(IRAT), France.

Possible "Dr i f t " of Yields

"Dr i f t " denotes the inevitable difference be-tween the results obtained on research stationsand those in farmers' fields. This drift is not dueto farmers alone, as some of them obtain excel-lent results.

In the preceding section, the need for apply-ing yield-enhancing factors simultaneously wasemphasized. But this is difficult for the exten-sion services as they deal with heterogeneousgroups who differ in their technical capabilitiesand receptivity. Moreover, they lack the finan-cial means and personnel required for a wide-spread promotion of these techniques. Con-

sequently, they proceed operation by operation(e.g. sowing, then weeding, followed by til lage,and so on). The yields fall short of the target andit takes some time for the farmers to integrate allthe techniques.

Some techniques are extended imposed be-cause of agronomic considerations such as a soil-regenerating crop and the inclusion of noncash crops in a rotation, and their advantage israrely obvious to the farmer. The farmer's moti-vation is also reduced as the yield-related posi-tive factors of animal-draft cultivation are not soapparent in their actual application.

Constraints in theProduction System

These constitute the basic point in the de-velopment of any new technology. It is abso-lutely necessary to determine the constraintsfaced by a producer before proposing anytechnology package that attempts to overcomethem. If the constraints are more or less arti-ficially disguised, the problem is not solved andthe technology may meet with obstacles.

For animal-draft cultivation, let us take theexample of traditional farming communitieswho only practice manual cul t ivat ion. Thefigures in Table 7 are not comparable as theyconcern different countries using di f ferenttechniques; however, they give an idea of theworking process.

Let us compare the data used by Richard andFall for establishing the 4 S model. These are themost recent and complete data on workingtimes for animal-draft cultivation in Senegal.The accuracy of the figures in Table 8 may bequestioned but the orders of magnitude remainvaluable.

It should be noted that when using animalpower, the work days of the operator driving theanimals must also be considered (Table 9).

In manual cultivation, the size of the farms isreduced as it depends upon the working ca-pacity of the farmer. The main problem here isthe planting and maintenance of crops. The in-troduction of animal-drawn equipment enablesfarmers to increase the area cultivated withoutovercoming the p lant ing/main tenancebottlenecks which remains high on a per hec-tare basis. It seems from this that the problemhas been somewhat disguised because thefigures show that only farms with a reserve of

217

Tab le 7 . Work ing t imes fo r exclusively manua l operat ions.

Operation

Til lageRidgingFertilizationSowingResowingWeedingThinningHarvesting

IvoryCoast

Groundnut

Senegal Chad

40-10

5-

30-25

15-20--10-15-19

10--91

20-

-

Maize Mil let

IvoryCoast Senegal

days/ha

40-26 4 1

24 125 2 6 16

Benin

Cotton

IvoryCoast Cameroon

13--

7-31-29

820

582

605

50

15-20--

5 -10-30

3 - 440 -50

Source: Author

Table 8. Add i t i ona l manua l w o r k raqu i red in a fu l ly mecha n ized f a rm using ox- or horse-drawnimp lemen ts in Senega l .

Operation

Furrow markingFertilizationSowingWeedingWeeding, th inn ingHarvesting

Groundnut

_

5-80-

-

Maize

_

5--80

115

hrs/haMil let

245

18-

100170

Cotton

65

80-

100600

Source: J. F. Richard and M. Fall, Senegalese Agricultural Research Institute.

Table 9 . Number o f w o r k i n g days requi red t ocomp le te a w o r k i n g pa t te rn (p lant -ing , w e e d i n g , th inn ing) In manua land an ima l -d ra f t cu l t i va t ion .

Pure manualcult ivat ion

(required days ofCrops manual work/ha)

Groundnut 2 5 - 3 5Maize 36Mil let 18-38Cotton 3 8 - 7 5

Advancedanimal-draftcul t ivat ion

(addit ional daysof manual work/ha)

10 -1410-1418-242 4 - 3 2

Source: Author

labor can extend the area and profitably use

animal-drawn implements.

For tillage operations, work productivity has

been increased wi th the introduction of plows

and spr ing-t ime cult ivators. Yields have in-

creased requiring more labor at the t ime of har-

vest. The problem has not been solved; it has

been shifted in t ime.

Equipment Standards

In Senegal, the need to coordinate the different

means of production on a farm has become

apparent since 1977. Studies were carried out

using theoretical models in Bambey fol lowed

by on-farm models and, finally, testing in far-

218

-

mers' fields; these studies were carefully moni-tored by agricultural researchers in the Experi-mental Units in Senegal and have helped todetermine the Iimits to the use of animal-drawnimplements. These standards are good indi-cators for the Sahelo-Sudanese zone with onerainy season and for a balanced rotation of mil-let, sorghum, groundnut, cotton, maize, andeven upland rice.

In 1977, Benoit-Cattin (personal communica-tion) summarized these limits as follows:

1 seeder for 4 ha1 Sine hoe for 3.5-4 ha (depending on themodel)1 "Occidentale" hoe for 3 ha1 plow for 1.5 ha to be tilled1 groundnut lifter for 4 ha of groundnut and a pair of bullocks1 ridger body for 2 ha of maize/cotton (and a pair of bullocks)1 Arara tool carrier (multiculteur) for 4 ha anda pair of bullocks1 Ariana tool carrier (multiculteur) for 6 haand a pair of bullocks1 Ariana +1 Sine hoe for 6 + 4 ha1 high-output Polyculteur for 8 - 10 haDraft power capabilities under similar im-

proved technologies were defined as follows:1 donkey for 2.5 ha1 horse for 2.5-3.5 ha (depending on the age)1 pair of bullocks for 3-6 ha (depending on theage)1 pair of cows for 4 ha1 pair of very strong bullocks for 8 ha

In fact these standards are close to those foundin different African areas where animal-draftcultivation is practiced.

Nevertheless, to avoid the labor " t rap" men-tioned earlier, we shall add the followingrecommendation — in the definitions of animal-draft cultivation capabilities, the family struc-ture involved in these types of models must beclose to one working person/per hectare.

Inappropriate Economic Situation

Adverse economic circumstances often haddirect consequences on animal-draft cultivationand restricted its development. Some of theimportant aspects include:

• The increase in equipment prices in relationto the relative stagnation of prices to pro-ducers, except perhaps in the Ivory Coast.

In Mali, the price of tool carriers (multicul-teurs) increased by 177%, plows by 101%,and carts by 86% over a period of 3 years.

• The increase in the price of draft animalsdue to a reduction of the herds in the Northfollowing the drought and as a result ofcompetition with abbatoirs. It is not un-common for a pair of bullocks to cost up to150 000 F CFA (US$ 714.28) and donkeys up

to 18 000-20 000 F CFA (US $ 85.71-95.24).These correspond to a 3-ha and 0.5-ha grainharvest, respectively.

• Credit facilities, if any, are selective. All theimplements are not included, particularlycarts; draft animals are rarely covered.

• Official marketing is exclusively focused oncash crops such as cotton and groundnut.Rice is given a special status in certain areasbut that is insufficient. In areas whereanimal-draft cultivation is developing, it isnot really possible to rely on an incomefrom traditional cereal crops; the farmerhas to depend on income from groundnutand cotton to pay for added inputs such asanimals and equipment.

C o n c l u s i o n

This discussion of the present problems doesnot mean that we are pessimistic about thefuture of animal-draft cultivation in Africa. Butin view of the decreasing demand for equip-ment in Mali, where animal-draft cultivation hasbeen an important option, and the recent stag-nation in the Ivory Coast, it is essential toreconsider it. This is especially important asmechanized farms present a good picture.Gross margins per hectare (valued at marketprices or monetarized) ranging from 60 000 to100 000 F CFA (US $ 285.71-476.19) per hectareare common, along with food-crop productionfrom 50% of the area for self-consumption andstorage.

Among the negative factors, some are of a political nature (prices, marketing, etc.) and thetechnical person cannot influence them. Butothers are related to the system of production,which he can change. Two simple ideas shouldbe considered for this purpose:

• Although animal-draft cultivation allowsmany technical solutions, its scope remainslimited; some sectors of agricultural ac-tivity will never be involved. Therefore,

219

from the initial stages, such an alternativeshould be directly conceived in associationwith other technical solutions. Weedingoperations lead us to think of herbicides.There is a need for tractorized solutions forharvesting and processing of crops, result-ing also in an increased availability oftransport. The problems that arise f romsuch situations are not really technical.They are related to the environment whichmust first be understood and wheremechanized solutions must f ind a receptiveaudience — farmers, contractors, smallgroups, etc. Technocratic attitudes thatfarmers should systematically adopt solu-tions used in industrialized countries mustcertainly be abandoned. If the blacksmithsare establishing themselves more stronglyin Mali, it is clearly because their role andworking methods have been decided by thevillgers themselves.

• Farms constitute dynamic and infinitelydiversified systems of production. This isincompatible with the concepts of standar-dization or conformity to preestablishedmathematical models. Although these mayprove useful as working tools in envisagingseveral variants on production techniques,their role is l imited.

To help a farm develop, its problems must beexamined under existing conditions andspecific solutions must gradually be proposed.This wi l l lessen the appeal of owning a com-plete series of animal-drawn equipment. Ini-tially, it would perhaps be more useful for a producer to purchase only a ultivator and notall the equipment. But a proper solution isperhaps to temporarily continue with manualcultivation techniques supported by the use ofherbicides. Paradoxically, it is quite possiblethat the first step in the development ofanimal-draft cultivation wil l result in the intro-duction of power-driven implements for har-vesting and crop processing.

In conclusion, it seems necessary that theoptions should be directed towards two keyactivities that predetermine the future ofanimal-draft cultivation:

• Research in all areas: study of differentproblems and solutions rather than an in-depth study of a few production systems;these wil l later prove to be inappropriateand wi l l be rejected or modif ied by the

220

farmers, which is the ideal approach but itdoes involve some waste of t ime.

• Extension activities: working directly wi thindividual farms and giving advice aboutmanagement of the farm rather than serv-ing as a facility for distributing farm inputs.

References

BENOIT-CATTIN, M. 1979. La mecanisation des exploi-tations agricoles au Senega l— le cas des UnitesExperimentales du Sine Saloum. Machinisme Ag-ricole Tropical 65:44-50.

RICHARD, J. F., FALL, M., and ATTONATY, J. M. 1976. Le

Modele 4S p rogramme lineaire pour les exploita-tions agricoles du Sine Saloum sud au Senegal etcalculs de budgets automatises. Bambey, Senegal:Institut National de Recherche Agronomique/Institut Senegalais de Recherches Agricoles/ lnst i tutde Recherches Agronomiques Tropicales et desCultures Vivrieres.

Socioeconomic Aspects of ImprovedAn imal -d rawn Implements

and Mechanizat ionin Semi-Ar id East Af r ica

Bruce F. Johns ton*

Abstract

This paper examines the socioeconomic factors most relevant to decisions concerning

the role of animal-drawn implements in the semi-arid farming regions of Kenya,

Tanzania, and Uganda. Structural and demographic features that should be taken into

consideration in determining national policies for agricultural and industrial d e v e l o p -

ment also need to be considered in determining agricultural research priorities and

development strategies for these semi-arid areas.

These relationships are especially evident in Kenya, where rural-to-rural migration has

been mainly responsible for extremely rapid growth of population in the major semi-arid

farming areas, most notably in Machakos and Kitui districts. Particular attention is given

in this paper to the present situation and future prospects in Kenya, partly because they

epitomize the fact that the technical problems of devising more productive and

sustainable farming systems must be considered in relation to the broader problems of

economic transformation and social modernization.

The principal hypothesis examined in this paperis that research and related rural-developmentactivities in the semi-arid farming regions ofEast Africa should give a very high priority topromoting wider and more efficient use ofanimal-powered equipment and associated til-lage innovations. It is thus intended as a con-tr ibution to the ex ante analysis of the allocationof research resources. This is in line with theview expressed by Binswanger and Ryan (1977p 217) that economic research on the payoffsand distributional outcomes of agricultural re-search should shift its focus from ex postanalysis to ex ante research concerning theallocation of scarce research resources.

In their paper, Binswanger and Ryan suggestthat most decisions faced by research adminis-trators concern "marginal research resource

* Research Scholar, International Institute forAppl ied Systems Analysis, Laxenburg, Austr ia;and Professor, Food Research Institute, StanfordUniversity, U.S.A.

allocation problems" and that reliance cantherefore be placed on the "simpler tools ofbenefit-cost and partial equilibrium analysis"(p 218; emphasis added). Some of the evidencederived from benefit/cost analyses of cultiva-tion based on animal draft power and on trac-tors is summarized below. My approach wi l l ,however, be more in line with a broader view ofthe utilization of scarce resources for research.In the introduction to the important recent bookon Induced Innovation, Binswanger and Ruttan(1978, p 4) assert that " the most powerful forcethat man can command to alter the direction oftechnical or institutional change is the capacityfor innovation." And they go on to suggest: " I fhe is to deploy his innovative effort effectively,man must learn as much as he can about thecomplex matrix within which technical andinstitutional change interact with each otherand with the physical and cultural endowmentsof a particular society."

It is important to recognize that it wil l be a difficult and time-consuming undertaking todevise and disseminate equipment and tillage

221

innovations capable of achieving widespreadincreases in farm productivity in these areas.Moreover, in order to realize the full potential ofexpanded use of improved farm equipment inthese countries it wiH be necessary to stimulatelocal manufacture of a widening range ofanimal-drawn implements and other inex-pensive and simple items of farm equipment;and this will also be a difficult task. It seemsclear that a necessary (if not sufficient) condi-tion for mounting such an effort is the availabil-ity of evidence and arguments that are morepersuasive than a cost-benefit calculation,especially since it might be supposed that thelatter is relevant only to transitory situations.

In considering the role of equipment andtillage innovations in the third section of thispaper, considerable attention is given to theimplications of certain structural and demo-graphic characteristics of these economies aswell as to the farm-level factors that are theproximate determinants of increases in produc-tivity and output. Before turning to the generalissues that need to be considered in selectingthe technological innovations to be emphasizedin agricultural strategies for semi-arid EastAfrica, it is useful to review briefly the evidenceconcerning the relative profitability of animal-powered and tractor technologies and somerecent historical experience in the three coun-tries.

An Overv iew o f An ima lDra f t Power and TractorMechan iza t ion in East A f r i ca

Costs of Animal- and Engine-PoweredTechnologies

The evidence now available seems to bereasonably consistent in indicating that withprevailing labor-capital price relationships, re-liance on animal draft power generally repre-sents the most economical approach to over-coming the power constraint and laborbottlenecks that are major factors underlyingthe low levels of crop yield and labor productivi-ty that characterize farming in semi-arid EastAfrica. On the basis of a number of farm

222

management studies carried out in Sukuma-land in Western Tanzania in the early 1960s,Collinson (1965, p 3) asserted that "there is nocase for the introduction of tractors in areaswhere ox-plowing is established; these canperform the same function as tractors morecheaply." There has been considerable expan-sion of ox-cultivation in all three countries,beginning as early as about 1910 in Uganda. Butthat has been mainly a spontaneous processbased on the importation of either a standardmold-board or a ridging plow. Other than a modest expansion in the use of ox carts, therehas been virtually no spread in the use of a wider range of animal-drawn equipment.

A report on mechanization in Tanzania byBeeney (1975) includes a careful comparison ofmanual, oxen, and tractor cultivation. Beeneyused partial-budgeting techniques and the bestcost data available in 1974 when he carried outhis FAO/UNDP study. He concludes that both forindividual farmers and for communal cultiva-t ion in ujaama villages, animal draft power ismore economical than tractor cultivation. Theonly exception, according to his analysis, isthat in the larger and more successful ujaama villages, using tractors for min imum tillagemight be superior, provided that the equipmentwas well operated and well maintained. Evenunder those circumstances, the use of a tractorin conjunction with oxen often would offer theleast costly option.

A linear-programming analysis by Singh(1977) provides the most valuable evidenceconcerning the cost advantage of animal draftpower over tractor cultivation. His LP modelsare based on input-output coefficients esti-mated from recent farm management surveyscarried out in the Mwanza and Shinyanga reg-ions of Western Tanzania. They permit an as-sessment of how the farming system as a wholewil l be affected by different levels of mechaniza-t ion, including both a tractor-hire service anduse of tractors in a communal ujaama farm aswell as hoe cultivation with hand labor andcultivation with oxen. Singh also examines theeffects on net returns and on employment ofvarious combinations of mechanical and yield-increasing innovations, such as the use ofimproved seeds, fertilizers, and pesticides. Hisanalysis confirms the clear-cut economic ad-vantage of animal draft power over tractors.When all land in a village is cultivated as a single

operational unit so that the ubiquitous problemof underutilization of tractors is minimized, thecost disadvantage of tractor cultivation is obvi-ously reduced; nevertheless, net returns arehigher with ox cultivation than with tractors,even when it is assumed that the price of labor iswell above levels that now prevail.

Historical Experience

At various times and in varying degree thegovernments of the three countries have rec-ognized the advantages of promoting improvedox-cultivation. This makes it all the more impor-tant to consider why the results achieved to datehave been so limited.

Uganda

A promising program to test improved equip-ment and to disseminate the "Indian method"of training and controlling oxen (by the use ofreins attached to a nose ring or a nylon ropepassed through a hole punched in the nose ofthe animal), was initiated at the Serere researchstation in Uganda during the early 1960s. Butthat program was allowed to lapse because ofthe government's preoccupation with promot-ing the use of tractors imported under an aidprogram. This proclivity of policymakers toassume that farmers could and should shiftdirectly from the current predominant relianceon the hoe and human labor to tractor-poweredfarming systems has seriously weakened ef-forts to foster the use of animal draft power in allthree countries.

Tanzania

The Tanzania Agricultural Machinery and Test-ing Unit (TAMTU), set up near Arusha in 1957,has carried out considerable testing and de-velopment work on animal-drawn equipment.On a number of occasions official statementshave indicated that major stress would beplaced on promoting ox-cultivation. PresidentNyerere has been especially eloquent in hisendorsement of expanded use of animal draftpower. In one forceful statement he declares:"We are using hoes. If two mill ion farmers inTanzania could jump from the hoe to the oxenplough, it would be a revolution. It woulddouble our living standard, triple our product.

This is the kind of thing China is do ing" (quotedin Smith 1971). Similarly, in his important state-ment on 'The Arusha Declaration: Ten YearsAfter", Nyerere notes that

. . . the truth is that the agricultural resultshave been very disappoint ing. Modernmethods have not spread very quickly orvery widely; the majority of our traditionalcrops are still being grown by the samemethods as our forefathers used . . . peoplestill think in terms of getting a tractor for theirfarms — even when they are small — ratherthan learning to use ox-ploughs.

Tanzania's action — and lack of ac t ion—tofoster wider and more efficient use of animal-drawn implements is especially interesting. Itmight be assumed that in a socialist countrywith a strong commitment to an egalitarianapproach to development, the political pres-sures that have elsewhere contributed to theencouragement of inappropriately capital-intensive technologies would have been almostnonexistent. However, the gap between policypronouncements and actual implementationhas been wide. A recent monograph on " A g -ricultural Mechanization and Employment" as-serts that "Tanzania is one of the few countriesthat adopted a policy of both selective andappropriate mechanization during the last fewyears" (Bodenstedt et al. 1977, p 112). Theauthors support that statement by a referenceto the Second Five Year Plan (1969-1974),which proposed the widespread introduction ofox cultivation, especially through the estab-lishment of oxen-training centers. However, AnFAO Mission on agricultural mechanization,visiting the country in 1974, reported that they

. . . found it extremely difficult to f ind anyconcrete evidence in the regions of a con-certed effort to promote a wider and moreintensive use of animal power. The programof establishing ox-training centers, men-tioned in the Five-Year Plan, appeared to bevirtually non-operat ive. . . (FAO 1975, p 9).

My own inquiries and observations as a member of an ILO employment mission toTanzania in 1977 seemed to confirm the conclu-sion of the FAO Mission and of the subsequentFAO/UNDP study by Beeney. In addition; thereappeared to be renewed emphasis on large-scale tractor mechanization on state farms as a

223

major element in the government's program toexpand food production.1

Kenya

The government's policy to give priority toincreasing the productivity of Kenya's small-scale farmers enunciated in the Development Plan 1975-78 included an emphasis onanimal-powered equipment and other "ap-propriate technologies." Prior to about 1975,however, very little had been done in Kenya topromote ox cultivation, and many policymakersand agricultural specialists in the governmentheld the view that the country's farmers shouldshift directly f rom hoe cultivation to tractors.

There are no doubt many factors that accountfor the especially strong appeal of tractor-basedtechnologies in Kenya, including the pre-Independence example of European farming inthe Highlands and the post-Independence in-terest of many senior government officials inlarge-scale farms in the former European areas.However, a psychological factor associatedwith the "modern i ty" of tractors and a tendencyto look down on animal-drawn equipment as an"infer ior" and "backward" technology mayhave been especially strong in Kenya.

It should also be noted, however, that neglectof animal draft power in Kenya was probablyinfluenced considerably by the priority that wasvery properly given — beginning with theSwynnerton Plan launched in the 1950s—toachieving rapid increases in farm incomes bypromoting small-holder production of coffee,tea, and grade cattle for milk production inCentral Province and other "h igh potential"

1. I have suggested elsewhere that preoccupationw i th the problem of obtaining a marketablesurplus for urban centers may partly account forthis tendency to promote large capital-intensivestate farms. This Is a problem in Tanzania becauseof the diff iculties encountered by statutorymonopol ies in collecting and distr ibut ing foodproduced by several mi l l ion small-scale produc-t ion units. The lack of confidence In the vi l lagesector that th is impl ies may wel l become a self-ful f i l l ing prophecy because prior i ty al location ofscarce resources of capital and trained manpowerto the state fa rm sector is likely to mean that thegreat major i ty of fa rm households wi l l be deprivedof the resources needed to achieve increases inproduct iv i ty and ou tpu t (Johnston 1978 p 94).

areas. The steep slopes and the small size offarms in those areas alone are sufficient to l imitthe potential gains from introduction of oxcultivation. But because such profitable alterna-tives exist, the opportunity cost of land forproducing fodder for draft animals is high, andthis represents an important economic factorthat reduces the profitability of introducinganimal draft power (Adelhelm and Schmidt1975).

Obstacles to the Identificationand Diffusion of "Appropr iate"Mechanical Innovations

Tractors as the Symbol of a "Modern"Agriculture

Reference has already been made to the factthat the appeal of tractors with their speed andother technical advantages has been an im-portant factor contributing to the neglect of ef-fective research to support wider and moreefficient use of animal traction. Many govern-ment officials in East Africa and foreign advisershave tended to assume that the striking techni-cal superiority of tractors must also implysuperiority in terms of economic efficiency.

It has been said that in most cases "theimp lementa t ion of exp loyment -o r ien tedmechanization policies is essentially a politicalproblem rather than a problem of the economic,administrative or technical feasibility of suchpolicies" (Bodenstedt et al . 1977, p 108). Thedecision by the Ugandan government to shiftemphasis f rom promoting animal draft powerto tractors was related to a politically motivateddesire to obtain a large allocation of foreign aidthat was available only at that t ime for theimportation of tractors f rom the United King-dom. The essentially political decision in Tan-zania to promote tractor cultivation under theblock cultivation scheme of the 1960s alsodiverted attention from ox cultivation and otherefforts to promote a broader and more progres-sive modernization of small-holder agriculture.It is estimated that, in addition to the scheme'sheavy demands on capital and foreign ex-change, some 40 to 65% of all agriculturalextension workers in the country were assignedto it (Kline et al. 1969, p 2-113). Subsequently,heavily subsidized distribution of tractors to

224

selected ujaama villages had similar adverseeffects, including the tendency noted by Presi-dent Nyerere: "People still think in terms ofgetting a t rac tor . . . rather than learning to useox-ploughs."

Scarcity of AgriculturalResearch Resources

Research programs to identify equipment andtillage innovations adapted to the agroclimaticconditions in the semi-arid and other farmingregions of East Africa are inadequate, largelybecause of the general scarcity of financial andmanpower resources for research. These coun-tries face a formidable problem in achieving a transition from a resource-based to a science-based agriculture (Johnston 1978). Expansionof food production has not kept pace withdemand because of population growth, en-larged cultivation of export crops, and decliningyields resulting from reduced fallow periods inthe traditional bush fallow systems. As previ-ously noted, the need to evolve more produc-tive farming systems is acute in the semi-aridareas. Especially in Kenya the large number ofmigrants from densely populated, high-potential areas are using farming practices thatare poorly adapted to the marginal areas, caus-ing serious problems of soil erosion, which areexacerbated by cutting trees for f irewood andcharcoal. One serious consequence is an in-crease in the frequency and magnitude of theneed for famine relief when there is a cropfailure.

The strengthening of national agriculturalresearch capabilities in these countries is a difficult task. The general difficulties faced bydeveloping countries are compounded herebecause these countries are relatively small interms of population and GNP, but their nationalboundaries encompass a great deal of diversityin rainfall, soil, and other agroclimatic con-ditions. Those ecological differences, togetherwith considerable variation in the dominant andsecondary crops produced in different areas,complicate the problems of crop improvementand of devising appropriate equipment andtillage innovations. The impact of past researchon food production in these and other Africancountries has also been limited because bothexpatriate and national scientists have oftenlacked an intimate understanding of the con-

ditions and constraints faced by farmers invarious ecological zones.

It is noteworthy that agricultural researchexpenditures in African countries in 1974 rep-resented only 1.4% of the value of their agri-cultural product in contrast to the countriesof Western Europe and North America (andOceania), where research expenditures wereequal to 2.2 and 2.7%, respectively, of theirmuch higher level of agricultural output. Proba-bly more significant, however, is the fact thatthe 1974 figures of 1.4% in Africa represented a very modest increase from 0.8% in 1959,whereas the increase in the countries of Asiaover that period was from 0.6 to 1.9% of thevalue of their agricultural output (Boyce andEvenson 197S, p8). It seems reasonable to inferthat the threefold increase in the Asian coun-tries reflected increased awareness of the valueof agricultural research with the widespreadintroduction of high-yielding varieties of wheatand rice beginning in the mid-1960s, whereasresearch programs in Africa have not yetyielded comparable results.

Lack of an Effective Methodologyfor R & D Activities

The general problems that have limited the pastimpact of research appear to have been espe-cially serious in the case of R & D activitiesaimed at achieving more efficient use of animaldraft power. It has been noted that food cropresearch in East Africa has been focused almostentirely on plant breeding and fertilizer use, tothe neglect of research on improved farmequipment (Vail 1973). It is not surprising thatresearch stations have concentrated their ef-forts on varietal improvement, since a powerfulmethodology exists for generating biological-chemical innovations. Furthermore, theseinnovations can be used efficiently by small-holders, who account for the great bulk of thefarm population in developing countries, andthe achievements of the "green revolution" inAsia have enhanced the prestige of that re-search strategy. Indeed, the success of Kenya'shybrid maize program and the more limited butconsiderable progress in diffusing high-yielding varieties of maize in Tanzania are theprincipal examples of successful food cropresearch in East Africa.

In contrast, there are few examples of suc-

225

cessful programs to identify and disseminateequipment innovations. In fact, there is reasonto believe that the usual approach, which hasconcentrated on the development and testing ofequipment by agricultural engineers, offers lit-tle promise. There appears to be much truth inthe assertion that "Machines to fit into farmingsystems cannot be designed by farm machineryresearch engineers working on their own. Itrequires a team effort of all those concernedwith analytical research for the developmentof all aspects of appropr ia te fa rmingtechnologies" (Monnier 1975, p 224). On thebasis of their experience in Francophone WestAfrica, Monnier, Tourte, and others argue for a "systems approach" that involves several ph-ases of research and an integration of researchand extension that incorporates feedbackinformation from farmers. In an "analyticalphase" various components of a farmingsys tem—soi l and plant improvement, equip-ment, ti l lage, etc. —are studied separately butwith careful attention to the labor requirementsand other costs and to the returns associatedwith various alternatives. The goal is to designand test systems that are profitable and feasi-ble, given the constraints faced by farmers.

These systems to be extended are thendemonstrated to farmers as a factor havingrelevance to their farming operations. It is notintended that they should apply the who le newpackage of innovations in one go , but rather thatthey should themselves decide on the inter-mediate steps they should take to arrive finally ata system of product ion that is logical and ap-propriate to their condit ions. (Monnier 1975,p 226)

Equipment and T i l lageInnovat ions for IncreasingFarm Product iv i ty

In this section I face up to the problem ofdef ining "approp r ia te " innovat ions andsuggest why "appropriate" equipment inno-vations should be an important element instrategies for increasing productivity and in-comes in the semi-arid regions of East Africa. A useful point of departure is to consider brieflythe general concept of technical change, using a pair of diagrams reproduced from Binswanger(1978, pp 19, 21). The production isoquant l-l in

226

Figure 1 represents the production possibilitiesprior to technical change. The slope of the twoparallel diagonals represents the relative pricesof capital and labor. Thus point A is the economi-cally efficient point of production because,being a tangent to the price line, it representsthe min imum cost combination of capital andlabor required to produce a "un i t " of outputgiven the state of knowledge described by theinitial isoquant. The isoquants l '- l ' , l " - l " , andI ' " - l ' " represent alternative possibilities aftertechnical change has made available improvedproduction possibilities resulting from researchor learning by doing. The way that the isoquantsare drawn, the same "un i t " level of output cannow be produced at points A' or A " , usingsmaller amounts of both capital and labor or atA ' " , with a greatly reduced input of labor but a somewhat larger input of capital.

The technical change represented by theinward movement to l '- l ' is "neutral" in the(Hicksian) sense that, with no change in therelative price of capital and labor, the point thatis economically efficient, A' , results in the sameproportional saving in labor as in capital. Tech-nical change that led to shifts to A" or A ' "would, of course, be labor-saving because thereis a proportional as well as an absolute reduc-t ion in labor requirements per unit of output.Figure 2 can be interpreted as describing an

Figure 1. Rates and biases of technical change.

Figure 2. Neutrality and its reference point.

initial situation in which the capital-intensivetechnology represented by A and the labor-intensive technology at B co-exist. Even thoughpoint A is unprofitable as compared to B interms of the " t rue" prices of capital and labor,large farmers might still find it privately profit-able to operate with the capital-intensivetechnology represented by A, because theyhave access to credit f rom institutional sourcesat artifically low interest rates. (The three al-ternatives depicted in Figure 2 all representfixed-proportion technologies to sharpen thecontrast between the capital-intensive andlabor-intensive alternatives.) Technical changerepresented by A ' , i.e., an inward shift along thecapital-intensive ray OA, would now be moreprofitable even with the relative prices re-presented by the price line CD. However, in theEast African context, such an extremely labor-saving technical change would be inferior ascompared to the possibility of technical changealong a more labor-intensive ray such as OB.Given the structural and demographic featuresof these overwhelmingly agrarian economies,large quantities of labor "released" from agri-culture cannot find employment in the non-farm sectors.

Because these features are typical of develop-ing countries, it is common to partition in-creases in labor productivity into two com-ponents: mechanical innovations that mainlyincrease the area cultivated per worker andbiological-chemical innovations that mainlyincrease yields per acre. Figure 3 (reproducedfrom Johnston and Kilby 1975 p 391) presents a schematic summary of the principal factorsinfluencing labor productivity, based on thisconventional partitioning. At a particular point

in t ime, this partitioning is merely an identity,and output per worker (O/L) is equal to theproduct of acreage cultivated per worker (A/L)and output per acre (O/A). However, the rate ofincrease in O/L over time wi l l , to a close approx-imation, be equal to the sum of the rates ofincrease in A/L and in O/A.

The schematic summary (Fig. 3) also directsattention to the fact that changes in farm laborproductivity will depend on the interactinginfluence of various farm-level and sociallydetermined factors. Increases in output perworker will depend not only on factors influenc-ing agricultural capacity to produce but also onthe growth of commercial demand and agricul-tural prices as they influence the actual rate ofincrease in farm output. In addition, the rate anddirection of change in the size of the farm laborforce is obviously an important determinant ofchanges in farm labor productivity. The impli-cations of these structural and demographiccharacteristics are discussed in relation to pro-jections of future changes in Kenya's popu-lation and labor force and their distributionbetween rural and urban areas.

Ba lance B e t w e e n M e c h a n i c a la n d B io log ica l -Chemica l I nnova t i ons

In countries such as Japan and Taiwan, growthof farm productivity and incomes dependedoverwhelmingly on yield-increasing inno-vations, together with complementary invest-ments in irrigation and drainage. It has onlybeen since the mid-1950s in Japan and the late1960s in Taiwan, when the absolute size oftheir farm labor force finally began to declinesignificantly, that engine-powered equipmentbecame a major source of increases in produc-tivity. However, in earlier periods, gradualdevelopment and spread of a widening range ofsimple, well-designed implements contributedto realization of the yield potential of improvedvarieties and greatly expanded use of fertilizers(Johnston and Kilby 1975 Chs 6, 8).

Thus the hypothesis stated at the beginningof this paper can be phrased more accurately: inEast Africa, especially in the semi-arid farmingareas, research programs should be concernedsimultaneously wi th inexpensive equipmentinnovations and with yield-increasing innova-tions suited to rainfed farming. This hypothesisis based on four propositions:

227

228

1. Because of moisture limitations, thepotential for increasing crop yields throughvarietal improvement and fertilizer use is con-siderably more limited than in irrigated areaswhere high rates of fertilizer application can beexpected to have a very large and quite reliableimpact on crop yields.

2. In the long run irrigation wil l make a significant contribution to the expansion offood production in East Africa, and some atten-t ion must be given to accumulating betterhydrological information and to building upexpertise in the demanding tasks of designing,constructing, and managing irrigation systems.But in the short and medium term it wil l bemuch more economical for research and de-velopment programs to concentrate on realiz-ing the large potential that exists for expandingcultivation and raising yields under rainfedconditions.

3. In the semi-arid areas there is an excep-t ional ly high complementar i ty betweenequipment-tillage innovations and biological-chemical innovations. This is related to theeffect of the choice of implements and tillagemethods on the extent to which crops are ableto utilize the water available from limited anderratic rainfall, including such factors as infil-tration, moisture retention, timeliness of plant-ing, and weed competit ion.

4. Promoting wide use and local manufac-ture of a widening range of animal-poweredequipment can make a major contribution to thegoals of improving rural wellbeing and of trans-forming the structure of their overwhelminglyagrarian economies by making possible in-creases in farm productivity and incomes andby providing a stimulus to the expansion oflocal manufacturing capabilities.

Structural-demographic Factorsand their Implicationsfor Technical Change

The significance of the last proposition on thecontribution of animal-powered equipment torural well-being and structural transformationis especially evident in Kenya. As in all the EastAfrican countries, the manufacturing and othernonagricultural sectors in Kenya are still ex-tremely small. In 1975 less than 15% of Kenya'stotal economically active population was inurban areas. This means first of all that the

domestic commercial demand for food is smallcompared with the number of farm households.Hence the average farm unit faces a severepurchasing power constraint. Production of ex-port crops represents a highly important sourceof farm cash income; nevertheless, the purchas-ing power constraint that results from thedominance of self-sufficient farm households inthe economy severely limits reliance on "ex-ternal inputs" purchased from outside the agri-cultural sector. The small size of the nonfarmsector also means that the annual additions tothe labor force must mainly find employment inthe agricultural sector.

Rapid growth of population and labor force inthe three countries means that structural trans-formation will be a slow process. On the basis ofoptimistic assumptions about the extent towhich Kenya's growing population of workingage wil l be "absorbed" in urban areas, it hasbeen estimated that during the half centuryending in 2024, the share of the rural workforcein the total population of working age willdecline only from 86 to 65% (Table 1A).Moreover, this projection assumes that thepercentage of the urban population categorizedas unemployed or inactive would increase froman estimated 36% in 1975 to 65% in 2024 in spiteof quite rapid growth of urban wage employ-ment and a very rapid (5.5% per year) increasein employment in the urban informal sector(Table 2).

These outcomes are, of course, stronglyinfluenced by the rapid rate of population in-crease in Kenya. The Shah-Willekens (1978)projections that I am using include six scenariosof possible changes in population during the50-year period ending in 2024. According totheir "most likely estimate" there will be a fivefold increase in population from 12.8to64.3mill ion and also a fivefold increase in thepopulation of active age. (Table 1B). Even theirmin imum estimate, which assumes a 25%linear decline in rural as well as in urban fertilitybetween 1979 and 1999, involves a 3.6-foldincrease in population; and the associatedgrowth in active population would be consider-ably greater because of the time lag before a reduction in fertility is reflected fully in a re-duced growth rate of population of workingage. The assumptions used by Shah and Wille-kens in their location- and age-specific projec-t ion model yield an estimate of a 3.8-fold

229

increase in the rural population of active agecompared wi th a 12-fold increase in the urbanactive population. But, given the fact that popu-lation densities are already high in Kenya's highpotential farming areas, absorbing a 280%increase of the rural work force in productiveemployment represents a formidable chal-

Table 1. Project ions of to ta l populat ion and ofe c o n o m i c a l l y a c t i v e p o p u l a t i o n ,Kenya.

A. Projection of economically active age.

Year

197419942024

Rural---------

5.59.3

20.8

Urbanmi l l ions

0.92.6

11.1

Total-------

6.411.931.9

Rural as % of total

867865

B. Alternative projections of total population.

Year

197419942024

"Most likelyestimate"

12.828.964.3

Maximumestimate

mill ions

12.832.077.6

Min imumestimate

12.825.645.8

Source: Shah and Willekens (1978).

lenge. Unless early and substantial success isachieved in slowing population growth, therelative abundance of labor and low price oflabor relative to capital are likely to be persistentfeatures of the economy for several decades atleast.2

These same structural and demographic fac-tors also underscore the importance of thecontribution that expanded local manufactureof animal-drawn and other simple items of farmequipment can make to the growth of outputand employment outside the agricultural sec-tor. Because of the small size and l imited de-velopment of technical skills in Kenya's manu-facturing sector, local manufacture of tractorsand other sophisticated equipment would beuneconomic; however, ox-plows, cultivators,

2. From that point of v iew, it is wor th not ing thatdemographic as wel l as equity considerationsseem to warrant an emphasis on labor-using andcapital-saving innovations to enable widespreadimprovement in labor productivi ty and incomes.The available evidence strongly suggests that thechanges in attitudes and motivation that lead torapid reduction of ferti l i ty are much more likely tooccur when the great majori ty of a developingcountry's rural populat ion is actively involved inprocesses of technical and economic change.Moreover, institutional innovations capable of in-creasing the effectiveness of government popula-t ion policies wou ld also contr ibute signif icantly toan improvement in income distr ibut ion as well asto more rapid g rowth of average per capita GNP(Johnston 1978; Johnston and Meyer 1978).

Table 2. Employment project ions In t h e urban areas of Kenya.

Total active-age populationUrban wage employedInformal establishmentsHigher educationUnemployed/inactive% Unemployed/inactive

1975

876 000387 210

74 100100 000315 000

36%

1999

1 600 000896 900277 400232 000

1 193 70075%

2024

11 100 0002 151 6001 097 000

556 0007 195 400

65%

Sources: Shah and Willekens (1978 p 38). The 1975 estimates a re from Kenya, Statistical Abstracts, 1976, p 271, and Ken ya,Economic Survey, 1977, p 40. The projections for 1999 and 20 24 are based on the following assumptions:

1. The annual growth rate in wage employment in urban areas in Kenya was 3.5% for the period 1966-75. This rate of gro wth isassumed to continue to 2024.

2. Informal establishments are assumed to grow at 5.5% annu ally over the periods 1975-1999 and 1975-2024. This is eq uivalentto half the growth rate of 1 1 % over the period 1974-76.

3. The active age population receiving higher education is assumed to grow at 3.5% annually up to 2024.

230

and many other implements can be manufac-tured wi th tolerable and increasing efficiency bysmall- and medium-scale workshops usinglabor-intensive techniques that economize oncapital and foreign exchange. Thus it is im-portant to stress that the existence of that"potential internal market can be used to fosterthe growth of a local farm machinery-manufacturing industry" (Swamy Rao 1977p 100). An indication of the magnitude of thatpotential market is provided by an estimate that84% of Kenya's arable area is still cultivatedwith hand tools, compared with 12% cultivatedwith oxen and 3.5% with tractors (Muchiri andMinto, 1977 p 57). Part of the land cultivated byhand is unsuitable for cultivation with oxenbecause of tsetse infestation or other reasons.However, some of the most promising oppor-tunities lie in increasing the effective demandfor new implements such as interrow cul-tivators in order to make fuller and moreefficient use of the draft power available.

Metalworking skills are of pervasive impor-tance in modern industrial processes and, byfostering local manufacture of farm equipment,they play a unique role in raising the technicaland managerial capabilities of rural basedworkshops. In India, Pakistan, and especiallyTaiwan, such workshops have frequently pro-gressed from the manufacture of animal-drawnplows and cultivators to seed drills, grain driers,electric motors, diesel engines, stationarythreshers, or power tillers (Johnston and Kilby1975 Ch 8). A recent study of rural small-scaleindustry in China provides further evidence ofthe role of decentralized manufacture of farmequipment in fostering the development ofmetalworking skills and evolutionary growth ofan increasingly efficient domestic manufactur-ing industry (Perkins et al. 1977). By diffusingand upgrading technical skills, such rural work-shops also perform a critical role as a technologi-cal training ground in developing the capacityto build the machine tools that permit a pro-gressive upgrading of the processes and pro-ducts of the domestic manufacturing sector.Since manufacturing capacity in East Africa ispresently confined to small urban enclaves, thisimpetus to spreading the growth of suchcapabilities in the countryside is of specialsignificance in fostering widespread increasesin productivity, employment opportunities, andincomes.

C o n c l u s i o n s

In his opening address at the Rural TechnologyMeet for East, Central and Southern Africa heldin Arusha in 1977, Prime Minister Sokoine ofTanzania made the candid observation that"our own efforts to develop appropriatetechnology. . . have been isolated, patchy, un-coordinated and sometimes have lacked a sense of seriousness" (1977 p 31). This paperhas mainly examined the reasons why suchefforts should be pursued seriously. It is thusmerely an introduction to the problems thatneed to be confronted in determining researchpriorities for identifying a profitable and feasi-ble sequence of equipment and tillage inno-vations adapted to the needs of farmers in thesemi-arid areas of East Africa.

In fact, I have very little to add to an earlierattempt that Gichuki Muchiri and I made tosuggest such priorities on the basis of ideas thatwe had solicited from a number of knowledge-able persons and our review of the literature(Muchiri and Johnston 1975). We examined theprincipal requirements for equipment andtillage innovations in relation to six problemareas: (1) improved equipment and tillage sys-tems for seedbed preparation and weed con-t ro l ; (2) improved practices for seeding andplanting; (3) use of narrow-based terraces, levelterraces, bench terraces, or other land-development measures to conserve moistureand soil; (4) improved techniques for training,handling, and maintaining draft animals;(5) measures to secure the most effective utili-zation of the limited mechanical power cur-rently available and likely to become availablein the short and medium term; and (6) variouscrop production innovations that need to beconsidered concurrently with tillage andequipment innovations in order to devise moreproductive farming systems. Fortunately,Muchiri is now actively engaged in research toprovide factual information pertinent to the firstproblem area (Muchiri 1978). Reference shouldbe made to the FAO/UNDP Agricultural Equip-ment Improvement Project in Kenya, althoughwork is only now getting under way at theSiakago subcenter, which wil l be carrying outequipment trials under semi-arid conditions.But considerable progress has been made inobtaining a range of equipment for testing andin devising evaluation procedures that include

231

an assessment of an implement's "suitabilityfor local manufacture" whenever that is a rele-vant possibility. Thus there is now some basisfor hope that Kenya is initiating the seriousand sustained effort required to make more ef-ficient utilization of the indigenous resourcerepresented by animal draft power.

References

ADELHELM, R., and SCHMIDT, H. 1975. Economics As-

pects of Ox-Cult ivation. In Sidney Proceedings ofWorkshop on Farm Equipment Innovations for Ag-ricultural Development and Rural Industrialization,eds. B. Westley and B. F. Johnston. Institute forDevelopment Studies Nairobi occasional paper 16,University of Nairobi , Kenya.

BEENEY, J. M. 1975. United Nations DevelopmentProgram Report to the Uni ted Republic of Tanzaniaon Agricultural Mechanizat ion. Rome, Italy. FAO.

BINSWANGER, H. P., and R Y A N , J. G. 1977. Efficiency

and equity issues in exante allocation of researchresources. Indian Journal of Agricul tural Economics32(3).

BINSWANGER, H. P. 1978. Induced technical change:evolut ion of thought . In Induced Innovat ion, eds. H.P. Binswanger and V. W. Ruttan, Bal t imore, Md. ,USA: Johns Hopkins University Press.

BINSWANGER, H. P., and RUTTAN, V. W. 1978. In Induced

Innovation, eds. H. P. Binswanger and V. W. Ruttan.Balt imore, Md. , USA: Johns Hopkins UniversityPress.

BONDENSTEDT, A. A., et. al. 1977. Agr icul tural Mechani-zation and Employment . Heidelberg; Germany: Re-search Centre for International Agrar ian Develop-ment.

BOYCE, J. K., and EVENSON, R. E. 1975. National and

International Agr icul tural Research and ExtensionPrograms. New York: Agr icul tural DevelopmentCouncil.

COLLINSON, M. P. 1965. Economics of block cult ivat ionschemes, Uki ruguru, Tanzania: Western Region Re-search Centre (Mimeographed.)

FAO (Food and Agriculture Organization). 1975. As-sistance in Agricultural mechanization — Tanzania(Miss ion repor t : September-October , 1974).Rome: FAO.

JOHNSTON, B. F. 1977. Food, health, and populat ion indevelopment. Journal of Economic Literature.

232

JOHNSTON, B. F. 1978. Agr icul tural product ion poten-tials and smal l farmer strategies in sub-SaharanAfrica. In Two studies of development in sub-Saharan Afr ica. Wor ld Bank staff work ing paper 300,Washington, D.C.: Wor ld Bank.

JOHNSTON, B. F., and KILBY PETER. 1975. Agr icul ture

and structural t ransformat ion: economic strategiesin late-developing countries, New York:

JOHNSTON, B. F., and MEYER, A. J. 1977. Nutr i t ion,

health, and populat ion in strategies for rural de-velopment. Economic development and culturalchange, Oct. 1977.

KLINE, C. K., et al. 1969. Agr icul tural mechanization inequatorial Afr ica, East Lansing, Mich : Institute ofInternational Agr icul ture, Michigan State Univer-sity.

MONNIER, J. 1975. Farm mechanization in Senegal andits effects on product ion and employment . In Reporton the Meet ing of the FAO/OECD Expert Panel onthe Effects of Farm Mechanization on Productionand Employment. Rome: FAO.

MUCHIRI , G. 1978. Effect of fa rm equipment innova-t ions on product ion and employment in semi-aridKenya Progress Report No. 1, submit ted to theInternational Labour Office.

MUCHIRI, G., and JOHNSTON., B. F. 1975. Equipment and

t i l lage innovat ions for Kenya's med ium potential(semi-arid) farming regions. In Proceedings, Work-shop on Farm Equipment Innovations for Agr icul-tural Development and Rural Industrial ization, eds.Sidney B. Westley and B. F. Johnston. Institute forDevelopment Studies occacional paper 16, Univer-sity of Nairobi , Nairobi , Kenya.

MUCHIRI , G., and M I N T O , S. D. 1977. A n approach to

research and development in agricultural mechani-zation in developing countries. In Report of the RuralTechnology Meet for East, Central and SouthernAfr ica, Arusha, Tanzania, 29 A u g . - 9 Sept. 1977.London: Commonweal th Secretariat.

NAS (National Academy of Sciences). 1977. Re-sources for agriculture: Report of Study Team 4. InSupport ing papers: Wor ld food and nutr i t ion study,vol . 2. Washington D.C.: Commission on Interna-t ional Relations, National Research Council.

PERKINS, D. et al. 1977. Rural small-scale industry inthe People's Republic of China. Berkeley: Universityof California Press.

SHAH, M., and WILLEKENS, F. 1978. Rural-urban popula-

t ion projections for Kenya and impl icat ions fordevelopment, Laxenburg, Austr ia: International In-stitute for Appl ied Systems Analysis.

SINGH, I. J. 1977. Choice of agricultural technologies: a case study, Washington, D.C.: Wor ld Bank.

S M I T H , W. E. 1971. We must run whi le they walk. NewYork.

SOKOINE, N D U G U E. M. Prime Minister's opening ad-dress: In Report of the Rural Technology Meet forEast, Central and Southern Afr ica, Arusha, Tanzania,29 A u g - 9 Sept 1977. London: Commonweal th Sec-retariat.

S W A M Y R A O , A. A. 1977. Local manufacture of agricul-tural and rural industrial equipment in developingcountries. In Report of the Rural Technology Meetfor East, Central and Southern Africa, Arusha, Tan-zania, 29 A u g - 9 Sept 1977. London: Common-wealth Secretariat.

TOURTE, R., and M O O M A W , J. C. 1977. Tradit ional

African systems of agriculture and their improve-ment. In Food crops of lowland tropics, eds. C. L. A.Leakey and J. B. Wills. Oxford, UK: Oxford Univer-sity Press.

VAIL, D. J. 1973. Induced farm innovat ion and derivedscient i f ic research strategy: The choice oftechniques in developing smallholder agriculture inland abundant areas. Journal of Rural Development6, (1 & 2).

233

Discussant 's Commen ts

S. D. B iggs*

It is a pleasure to open the discussion onthese three papers because the authors havepresented us with many of the important issuesand problems associated with ex poste andex ante analysis of agricultural technology. I think the most useful thing I can do is to focusour attention on a few general points that haveimportant implications for the developmentand use of technology in the future.1. Determinants of Rural Technologies. It isclear, both from Johnston's paper on EastAfrica and f rom the paper by Binswanger et al.for India, that actual past mechanization prog-rams have been far from optimal in promotionof appropriate agricultural technology to meetgovernment's declared production and equitygoals.

In East Africa, governments made com-mitments to the promotion of improved ox-cultivation but, as a result of reasons such as (1)the tractor being seen as a symbol of "modern"agriculture, and (2) the terms and technologicalcontent of aid donor projects, these desirablepolicies were only feebly implemented, if at all.

In the case of India, one of the justificationsgiven by Binswanger et al. for researchemphasis on bullock power is because "wherewage rates are low and where little potential forarea expansion exists, as in SAT Asia - tractorshave extremely low or negative private andsocial rates of return. There is evidence thattractors are used extensively for transportationand as a substitute of one source of power foranother rather than as a means of improvingmanagement or cultivation practices. In ad-dition they displace labor, and finally theireconomies of scale lead to further growth oflarge farms at the expense of small farms."

Unfortunately the authors of the Indian paper

* CIMMYT, c/o ICRISAT, 23 Golf Links, New Delhi,India.The opinions expressed in this paper are those ofthe author and do not necessarily reflect those ofCIMMYT or any other inst i tut ion.

do not go into the reasons why inappropriatetechnology has been promoted in the past.Was it because the Planning Commission got its"choice of technique" analysis wrong and un-witt ingly promoted inappropriate techniques?Or was it because, although straightforwardeconomic principles were (or could have been)applied, other factors had more influence on thedecision-making process of policy makers? I suspect that the latter is true. Because of this, I would suggest if social and technical scientistsare interested in influencing the process ofagricultural technology decision-making, thatwe place more emphasis on what Johnstoncalls "obstacles to the identification and diffu-sion of 'appropriate' mechanical innovations."Such analysis wil l quickly lead us into issues ofpolitical economy, dependency relationships,the activities of multinational firms and aidagencies as well as the professional training ofthose who are supposed to analyze technologi-cal problems. It seems to me that unless thesetypes of determinants of technology are moreexplicitly taken into account we shall, I fear, in10 years again be looking back at a decade ofgovernment action and function where thepoorest group in rural areas of low incomecountries have not been the main beneficiariesof technological change.2. Resources for Research and Devel-opment (R & D). Johnston suggests that a scarcity of resources for agricultural research isone of the reasons we do not have moreappropriate technology available and beingpromoted in low income countries. While I would agree with him that there could be moreinvestment in research, I feel this type of in-vestment is like any other investment — it de-pends on the composition and t iming of the in-vestment as regards who gains and who losesas a result of the investment decision. I can thinkof many examples of agricultural universities,agricultural research institutes etc., whereteaching and research are not oriented towardsrural productivity and equity goals. In my opin-ion, the more important and difficult issue is

234

how to change the system of incentives andrewards in teaching and research institutions(in high and low income countries) so as toencourage and support those people who areworking on problems facing poor farmers, ag-ricultural laborers, and rural artisans in the lowincome countries.

3. Methodologies for Developing New Tech-nologies. I agree with all three papers thatthis is an area that needs far more attention. I would suggest that the main thrust in the futuremust be given to having social and technicalscientists work together with village peopleright f rom the start, and throughout, any re-search and extension program. The main focusof this workshop is on how social and technicalscientists can effectively work together inanalyzing technological problems and helppromote new technologies, relevant to theproblems facing the poorest groups in ruralareas of low income countries. Le Moigne, in hisconcluding remarks on research in West Africa,places emphasis on a diversified approach tosolving specific problems in order to find practi-cal solutions, rather than creating entire newsystems of production that are impractical orhave to be modified to a major degree byfarmers. He also calls for extension systemsthat provide good advice and technology tofarmers of different types, rather than extensionprograms which are mainly concerned with thedistribution of inputs. I would agree with hisemphasis.

In the paper by Binswanger et al. they de-scribe their experience in the development of a wheeled tool carrier. So far they have found thateven under the most favorable circumstances,such machines cannot compete on a cost basiswith the traditional implements in traditionalagriculture. This is a disappointing result for a group of researchers and for the farmers forwhom they are working; however, withoutbeing unfair, I suspect that the methodologythey are using is not one which is most likely toresult in the rapid f inding, development, anddissemination of useful agronomy and agri-cultural engineering technologies. Theirmachine comes out of a research processwhere: (1) animal-draft power is the focus ofattention rather than a specific cultivation prob-lem that might be solved through various typesof draft power, management methods, or otheragronomy and agr icul tural engineering

techniques; (2) the machines are first de-veloped and tested on the research station, thensent out for further testing to cooperating centers as well as to villages; and (3) suitablemanufacturers are identified at an early stage ofthe testing program in order to have the input ofproduction engineers.

As I see it, the problem with this approach isthat farmers, rural artisans, rural service enter-prises etc. are not the starting point of theresearch and extension process and are nottotally integrated into the R & D processthroughout. I would suggest an alternative ap-proach where:

(i) Engineers (and other technical scientists)as well as applied social scientists spend 2 or 3 weeks together in rural areas conductingexploratory research, i.e. talking to rural peopleabout their problems, coming up with alterna-tive hypotheses of what might or might not becritical problems and constraints on techno-logical development. This is likely to result injointly having a whole range of academic so-lutions rapidly dismissed because of such fac-tors as what villagers say about their environ-ments, what the engineer says is technicallyfeasible, what the economist says about currentand future factor and product markets, what theresearchers learn about people's motivationwith respect to production and consumptiongoals, risk, and uncertainty and attitudes to de-pendency relationships etc.

(ii) The agronomy and agricultural engineer-ing research is, as far as conceivably possible,conducted in the environment for which it isbeing designed. There is certainly a place foron-station research, but it has to be justified inthe context of the harsh multidisciplinaryproblems being faced by farmers. In my experi-ence I have not found a situation where youcannot find some farmers who wil l not givesome land for experimentation purposes. Thisis not surprising, as many farmers and ruralentrepreneurs are innovators who are alwaysconducting informal experiments in agronomyand agricultural engineering and finding waysto develop markets for goods and services. Inthe past not enough agronomy and agriculturalengineering and institutional research has beendone in villages by researchers of the formal R & D system.

Many of the problems in the "transfer oftechnology" would not arise if irrelevant

235

technologies and associated institutionalstructures that governments are trying to trans-fer to villages had been screened out by discus-sions between social scientists, technical scien-tists, and villagers, at an early stage in theformal research and extension process.

(iii) Collaboration with "cooperating centers"is seen as a starting point of a research processrather than a relationship where techniques aresent out for further testing. These "per iphera l "cooperating centers should be seen as a majorsource of information to influence the ob-jectives and programs of the "centra l" prog-ram. If it is thought that the cooperating centersdon't have the skills and capacities to take onsuch a role, then it would seem necessary toplace major emphasis on strengthening theselocal capacities.

(iv) Multidisciplinary groups of researcherswho are working on developing agronomy andagricultural engineering technologies start touse the " in fo rmal " R & D system, which is inplace in most communities and is already ef-fective in developing new viable technologies.Village people are researchers and innovators,and it is t ime we started to keep up w i th them.

I am not sure of the case in the semi-aridtropics, but certainly in the northern part ofIndia and in Bangladesh it has been farmers,village artisans, entrepreneurs of local machineshops etc., who have carried out much of theagronomy and agricultural engineering R & Don cropping patterns, irrigation technologies,management techniques, threshers, animal-drawn disc plows etc., as well as having foundmethods of repairing and maintaining equip-ment when similar public sector activities haveremained inactive because of a lack of foreginspare parts, a lack of effective demand for theservices etc. Not only is it necessary for us tofind methods to keep up wi th the results of thisdynamic research system, but also I expect wecan learn a lot about relevant methods ofresearch from people who have to worry aboutthe cost-effectiveness of identifying correctlytheir research priorities and methods.4. Analysis on the Effective Demand for Technology. The last point concerns the futureeffective demand for technologies. I wouldsuggest that a multidisciplinary macro-analysisbe conducted on the likely effective demand forproposed technologies in the future. What arethe future relative prices for fuel and forage that

236

are implicitly assumed when a researchprogram embarks upon an animal-draft powerprogram rather than a power-tiller or large-tractor program? What are the cooperativestructures or markets for services that we im-plicitly assume will work when we develop largeequipment, for technical reasons of economiesof scale, rather than smaller 'divisible' equip-ment that fits other institutional structures?What are the implicit relative prices of differentcrops that we assume when we develop onecropping system because we say it is betterthan another? Have we assumed that ruralpeople wil l have employment and be able tobuy foodgrains and thereby keep up the effec-tive demand and prices for cereals, or have weassumed high unemployment and a lack of aneffective market demand for food? What are themin imum policy conditions that wil l have toprevail in order to enable the poor to benefitfrom the new technologies?

It seems to me that a considerable amount ofcurrent research could be improved if some ofthese assumptions and questions were clearlylaid out and stated at the start of a researchprogram. It is my impression f rom the Africanpapers that little attention was given to thesefactors in the preparation of past researchprograms on ox-cultivation. Unfortunately, it isfarmers and other rural people who are left toface these harsh realities when they are ex-pected to " a d o p t and benef i t f r o m "technologies in environments that are not con-ducive to the use of such technologies.

Emphasis on multi-disciplinary ex-anteanalysis on the future effective demand fortechnology, and for the goods and services thatf low f rom that technology, brings me back tomy opening remarks about what are the de-terminants that promote technology to meetrural production and equity goals. In the pastmany far-sighted technical scientists haveworked on technologies designed to addressthe problems and needs of the poorest groupsof society. However, these scientists have oftennot been rewarded by their profession or hadthe satisfaction of seeing their technologiesused because government policies to inducethe use of those methods and techniques werenot implemented.

Chairman's Summary

R. J. H. Chambers*

Three papers were presented at this session.The paper by Binswanger et al. examined theeconomics of tractors, bullocks, and wheeledtool carriers in SAT India. The situation inFrench-speaking Africa was reviewed in M. LeMoigne's paper. Bruce Johnston dealt with theEast African scene.

The discussion centered around three mainissues. Firstly, it was felt that past research inmechanization, both in India and Africa, wasgenerally not consistant with either the statedgoals of government policies or the real needsof the farming communities. For example, whileweeding is perhaps the most important con-straint in African agriculture, preparatory tillagereceives primary attention in research. Simi-larly, while intercropping is an importantcharacteristic of SAT agriculture, it has beenlargely ignored in mechanization research. Itwas felt that noneconomic considerations oftenplay a role in promoting mechanizationtechnologies that are inappropriate.

This led to a discussion on appropriatemethodology for designing technologies. It wasargued that usually field testing under theactual farm environment comes at the laststage, often ending up with frustrating conclu-sions regarding economic viability of themachine as pointed out by Binswanger et al. Itwas suggested that it would be far moreefficient to substitute an approach based on anintensive preliminary interdisciplinary recon-naissance of the actual farming environment intarget villages, discussions with farmers re-garding immediately obvious solutions andunacceptable alternatives, and then comingback to the experiment station and working onthe basis of very specific objectives. The sub-sequent work also must involve constant con-tact wi th the target group.

This generated some discussion on past ex-periences with testing of technologies in

farmers' fields. It was pointed out that theAfrican as well as Indian experience in thisregard does not indicate that this is an easy taskor that it yields the kind of results describedabove. The difficulties were recognized to beserious and it was mentioned that such an effortwill be wasteful unless it is accompanied by a significant change in attitudes of scientists to-wards such work and also in the reward andincentive systems in the scientific profession.

The following recommendations resultedfrom the discussions:

1. Technology design. Programs for researchshould be drawn up only after a careful multi-disciplinary appraisal (by engineers, ag-ronomists, and economists) of the problems ofthe target group and area. While this wasextremely crucial for national programs, theinternational centers would need to look at theproblem in an optimum-mix context.

2. Agricultural policy. A careful evaluation ofgovernment policies is needed — particularlyin the field of mechanization — and economistsmust play a leading role in this evaluation.

3. Socioeconomic research at national in-stitutions. National systems must take a verycareful look at their research programs inmechanization and examine the relevance andcost-effectiveness of this research in relation tothe needs of the target population.

4. Socioeconomic research at ICRISAT.ICRISAT should have a diverse program ofresearch in mechanization in view of the widelydiffering resource endowment and farm pro-duction costs in different SAT countries. In allcases, however, the emphasis must be on realpower bottlenecks and cheap implementsand/or machines.

* The institute of Development Studies, UniversitySussex, Br ighton, Sussex, UK.

of

237

Chapter 6

Analy t ica l Rev iews of L i terature on the Semi-Ar idTropics of West Af r ica

Farm-Level Studies in t he Semi-Ar id Tropics ofWest Af r ica

M. N e w m a n , I . Ouedraogo, and D. N o r m a n *

Abstract

This paper presents a brief analytical review of earlier studies on socio-economic

problems of farm and village production systems of the semi-arid tropics (SAT) of West

Africa. It is to help identify high-priority areas for research by ICRISAT and help provide

crop-technology guidelines for those areas. Because new technology may increase

inequalities in income distribution at the village level, the authors argue that in designing

support systems developers of technology must recognize the heterogeneity of farmers,

so that small poor farmers are not ignored and growth is accompanied by equity. Limited

knowledge of data-collecting and analytical techniques, coupled with discipline insu la r -

ity, have ignored distributive justice in the past, but interdisciplinary work can now be

done by technical personnel, scientists, anthropologists, sociologists and agricultural

economists. Such interdisciplinary work should develop appropriate strategies for the

future; this is where the authors envisage a definite role for lCRISAT. With regard to West

Africa, they suggest a concentrated input by ICRISAT in one or two localities for several

years and going beyond the five crops currently under its mandate so that the entire

farming system (including livestock) is considered as a unified whole and developed

with firm linkages to national institutions.

The Area

The geographical area discussed in this paperconsists of Senegal, Gambia, and Upper Volta,and substantial parts of Mali, Niger, Chad,Nor thern N iger ia , Ghana, Ben in , andCameroon — the main part of the semi-aridtropics (SAT) of West Africa (Table 1). It consti-tutes about 17% of the world's total SAT area andincludes 23% of the world population living insuch areas. An unusually high proportion of thetotal area is in the SAT ecological region; 67%(average) of the West African countries are insuch a region, compared wi th an average of33% for SAT countries in the world.

The SAT is defined by Troll (Lindsberg et al.1966) as areas where the precipitation exceedspotential evapotranspiration for 2 to 7 monthsof the year. In the SAT of West Africa, theeconomies have been traditionally dependentupon agriculture, particularly rainfed agri-

* Research Assistants in Agricultural Economics,Michigan Stats University and Professor of Agricul-tural Economics, Kansas State University, USA.

culture, which tends to be seasonal in naturebecause of the short growing season. Cochemgand Franquin(1967) found that the coefficient ofvariation of rainfall was inversely related to theaverage annual rainfall. Consequently, togetherwith the decrease in the annual rainfall andlength of the growing season as one movesnorth through this area, there is a concomitantincrease in rainfall variability, particularly at thestart and end of the rains. In general terms, cashcrops (cotton and groundnut) and millet andsorghum in the south give way progressively toshort-season millet as one moves north.

Terms of Reference

This paper is a brief summary of some of themajor points in a report requested by ICRISAT(Norman, Ouedraogo, and Newman 1979).1 The

1. In the interest of brevity, few quantitative empir i -cal data are presented here; rather, emphasis islaid on giving general conclusions arising f romthe review, supported by a selected, rather than a complete, list of references.

241

Table 1. The semi-ar id t rop ica l areas of West A f r i ca . a

Country

SenegalGambiaMal iUpper VoltaNiger

NigeriaChadGhanaBeninCameroon

Total

Semi-aridtropical area('000 sq. km)

19610

721274507

69351411992

119

3245

Proport ion ofcountry

semi-aridtropical (%)

10010060

10040

7540508025

67

Semi-aridtropical

populat ion(mil l ion)

41554

423421

71

Number ofreferencesb

187237076

113

2161634

329

767 (1052)c

Source: Ryan and Associates (ICRISAT, 1975).

a. These areas are def ined as in the text .

b. Includes all references in the part ia l ly annota ted b ib l iog raphy submi t t ed to ICRISAT (Ouedraogo, N e w m a n and Norman,

1979).

c. Includes references that are not coun t ry speci f ic .

terms of reference included, among others, ananalytical review of the available researchstudies — published and unpublished — onsocioeconomic aspects of farm and village pro-duction systems of the SAT areas of WestAfrica, to help identify high-priority topics forresearch by ICRISAT and to "provide guidelinesfor the design of viable crop technology forthese areas."

The focus of ICRISAT plus the limited t imeavailable for the review imposed certain re-strictions. For example, neither irrigated ag-riculture nor livestock (pastoralism) were con-sidered, except where they directly interactedwith rainfed crop production.

Methodo logy

The methodology used in data collection de-pends on a large number of factors: objectivesof the study, degree of accuracy required, re-search resources and t ime available, know-ledge and discipline(s) of the investigator(s),etc. The methods used range from the casestudy-direct observation approach to one utiliz-ing a statistical sampling method, where a

sample is subjected to various interviewingprocedures ranging from single to frequent,and unstructured to structured questionnaireapproaches. Although generalizations may bedangerous, if appears from the literature thatanthropologists, sociologists, geographers,and most individuals working in the Franco-phone countries tend to use the case-studyapproach while agricultural economists tend touse statistical sampling procedures, particu-larly in Anglophone countries. Depending onthe data needs, both approaches have validity,and it is interesting that, increasingly, there areattempts to find common ground between thetwo. Too often the methodology used in datacollection has been inappropriate to the ob-jectives of the studies and in terms of measur-ing specific variables, thereby decreasing thevalidity of the results and making comparabilitybetween different studies virtually impossible.For example, labor f low data having a continuousnonregistered characteristic (Upton and Moore1972) require the use of a direct observation orfrequent interviewing approach, if measure-ment errors are to be reduced to a reasonablelevel. Unfortunately, however, other ap-proaches that severely l imit the value of the

242

results have on occasion been used, such asinfrequent interviewing, which is valid only forsingle-point registered types of data. Fortu-nately, increasing attention is now beingfocused on rationalizing the approach to beused in data collection (AMIRA 1978; Collinson1972; Connell and Lipton 1977; Palmer-Jones1978a).

A detailed discussion of all the specificproblems of data collection and analysis is notpossible in this paper. However, three problemareas should be mentioned because they arecrucial to the subject matter of this review.These are:

(1) In the micro-level studies it is apparentthat three biases of the research workers tend toinfluence their investigation of farmers (farm-ing families):

• "Point" bias is the tendency to consider thefarmer at one point in t ime. This bias causesresearchers to overlook the fact that the farmertoday is a product of what has happened in thepast, and that what he or his descendants will bein the future is partly a function of what happensnow. The "vert ical" dimension (which can besubdivided into the "histor ical" sub-dimensionin terms of the past, and the "prospective"sub-dimension with reference to the future) isoften missing from studies.

• "Individualistic" bias occurs when re-searchers consider the farmer as an indepen-dent decision-maker. But a farmer's decisionsare partly the function of his dependency on andrelationships to the world around him (i.e., thecommunity he lives in, linkages with the worldoutside the village, etc.). That type of relation-ship, which constitutes the "horizontal" di-mension, is often ignored by the studies.

• " H o m o g e n e o u s " bias occurs whenfarmers as a group are assumed to behomogeneous. Studies wi th that bias fail toshow the dimension of "heterogeneity."

A number of reasons, some unavoidable,cause one or more of these three dimensions tobe neglected. This produces severe ceterisparibus conditions in most studies. A reductionof such conditions, which are sometimes notexplicitly stated, brings wi th it a greater recog-nition of the dynamic environment the farmeroperates in, being a product of the "historical"and "hor izontal" dimensions, which in turn willhelp determine the future.

(2) There is the difficulty in defining a sampl-

ing unit representing a production decision-making unit (exploitation agricole) (Kleene1976; Maymard 1974; Monnier et al. 1974) that isvariably synonymous with a consumption unit(i.e., the farming family). To obtain some cor-respondence between the two is important insocieties where much of the labor on the familyfarm is from family sources and much of theproduction is consumed within the household.Complications arise in defining the family be-cause of the progressive breaking up of thetraditonal family system, the increased sig-nificance of individual decision-makers with thefamilies (Rocheteau 1975; Kleene 1976), andseasonal fluidity in composition of the unitsresulting from changes in the level and compo-sition of economic activity at different times ofthe year (Hill 1972).

(3) There is difficulty in measuring work bydifferent family members and in reducing it to a common denominator in terms of productivity.The different approaches to this problem haveled to the actif concept in the Francophonecountries (Echard 1964; Ancey 1975a) and themale adult equivalent, characteristic of theAnglophone countries (Matlon 1977; Lowe inDunsmore et al. 1976). The actif concept is oftenused in situations where only the stock of laboris measured.2 Therefore, it should include in itsweighting system both the productivity per unitof t ime and the length of t ime worked by suchindividuals.3

The male adult equivalent approach, however,is often used in situations where labor f low dataare being collected. In such situations, theweighting system needs to reflect only differ-ences in the productivity per unit of t ime oflabor. The weighting systems employed arecritical in determining the validity of the results.Unfortunately, it is not always clear what a weight-ing system was employed and why specificweights were chosen. The relevant weights tobe used are further complicated by the fact thatthe productivity of individuals of different sexand age will vary according to the task and its

2. The Experimental Units in Senegal have recog-nized the expense and difficulty of collectingaccurate information on labor f low data and havein general, therefore, decided that such datashould not be collected in their work.

3. It in fact often ignores work done by chi ldren (i.e..usually those under 15 years old.)

243

urgency (Delgado 1978; Haswell 1975).4 In viewof the seasonality of agriculture in the SAT ofWest Africa and the resulting variability of laborinput, this is a major problem and makes com-parisons of different studies difficult.

Schemat ic F ramework

A simplistic framework of some of the possibledeterminants of the farming system practicedby a farming family is given in Figure 1. Thefarming family in West Africa is both a produc-tion and consumption unit. In order to achieve a specific farming system, the farmer allocatescertain quantities and qualities of basic types ofinputs — land, labor, capital and manage-m e n t — to which he has access, to threeprocesses — crop, livestock, and off-farmenterprises — in a manner which, within theknowledge he possesses, wil l maximize theattainment of goal(s) he is striving for. Some ofthe underlying determinants of the farmingsystem, however, are outside the control of theindividual farmer.

Simplistically, the environment can be di-vided into two parts (IER 1977): the technical(natural) element and the human element. Thetechnical element determines the types of, andphysical potential of, livestock and crop enter-prises and therefore forms the necessary con-dition for what the farming system can be.Constituents of the technical element includephysical and biological factors, often modifiedto some extent by man as a result of technologydevelopment. However, the farming systemthat actually evolves is a subset of what ispotentially possible. The determinant that pro-vides the sufficient conditions for the presenceof a particular system is the human element,characterized by two types of factors: exoge-nous and endogenous.

Exogenous factors (i.e., the social environ-ment), largely outside the control of the indi-vidual farmer, wi l l influence what he wil l doand/or is able to do. They can be divided into

4. This discussion considers using the data for workunits only. Consumpt ion units wou ld Involve a different weight ing system. Unfortunately thisdifference is no t always recognized in the studiesreviewed.

three broad groups: community structures,norms, and beliefs; external institutions whichon the input side include extension, credit, andinput distribution systems and on the outputside include markets; and other influences suchas population density and location.

Unlike exogenous factors, the endogenousfactors are controlled by the farmer himself; heultimately decides on the farming system thatwill emerge, influenced and sometimes con-strained by the technical element and exoge-nous factors.

The schematic framework in Figure 1 explicitly takes into account the "hor izontal"and "heterogeneous" dimensions. The "vert i -cal" dimension, though not explicitly con-sidered in Figure 1, is the cumulative function ofwhat happens in terms of the "hor izontal" and"heterogeneous" dimensions over t ime. Therelationship is schematically represented inFigure 2.

The holistic view of the farmer and his place inthe environment is important, we feel, in spiteof the considerable problems it created whenwe tried to l imit the content of the reviewrequested by ICRISAT.

We believe that constraints faced by farmersmay be articulated in one or more of the di-mensions enumerated and that the types oftechnology that might be developed byICRISAT, in order to be relevant, must takecognizance of the "histor ical" and "hor izontal"d imens ions and must recognize theheterogeneity of farmers. It should also takecognizance of what might happen in the futureas a result of its adoption ("prospective" di-mension) such as increasing inequality in termsof benefits and increasing divergence of short-run private benefits and long-run social costs.Therefore, put rather crudely, improvedtechnology, in order to be relevant, must betime-, place-, and, to some extent, individual-specif ic— a daunting task!

In the rest of this paper we discuss some ofthe major points arising from the review, usingthe schematic framework in Figure 1 as a backdrop. In each section, along wi th a briefdescription of the characteristics, implicationsthat are considered important in terms of thedevelopmental process and the ICRISATmandate are also mentioned. Although an at-tempt is made to divide the paper into sectionsfor easier reference, it should be noted that the

244

245

1. The commonapproach

Note: The farmers are assumed to behomogeneous and ceteris paribus condit ions often ignore the " v e r t i c a l "and " h o r i z o n t a l " dimensions.

2. The idealapproach

" H i s t o r i c a l "sub-dimension

"Hor i zon ta l "dimension at time

Note: The interaction between the "hor izon ta l 'dimension and the " h i s t o r i c a l " sub-dimension is important in inf luencingthe heterogeneity that exists at present.

3. Hypothesis concerning what is happening over time

Time period

Traditional situation

Increasing influence of externalinst i tut ions and improved technology

t - 2

t-1

t

t + 1

Note: It is hypothesized that the breakdown of community structures, norms and bel iefs,together with the increased influence of external institutions,may be leading toincreased heterogeneity and inequal i t ies developing over t ime.

Figure 2. The farmer and the dimensions in his environment.

246

t - 1

t

topics are interdependent. In certain places, thismay cause some duplication in the discussion.

E x t e r n a l I n s t i t u t i o n s

The "histor ical" dimension is important in de-termining the type and organization of externalinstitutions. Even in the post-colonial era, it isapparent that the different colonial experiencesof the Francophone and Anglophone countrieshave left their mark (Kohler 1971; 1976; Amin1975).

In countries of both types, some com-monalities exist, such as commodity-based re-search programs, a bias towards research oncash crops during the colonial era (Lele 1975)and to a lesser extent in the post-colonial era,and traditionally an absence of linkages be-tween the social and technical sciences (DeWilde 1967; Kleene et al in IER 1978).

However, the differences are more striking.Until recently, much of the research in theFrancophone countries was directed fromFrance (Kassapu 1976) while much greater de-centralization characterized the research spherein the Anglophone countries. Furthermore, thelink between research and implementation, par-ticularly for cash crops (i.e. groundnuts andcotton), has been much stronger in the Franco-phone countries where support systems onboth the input and product side have beenstrongly coordinated with respect to a particularcrop (e.g. SODEVA for groundnuts in Senegal,CMDT for cotton in Mali, etc.).5 In general,commodity-based implementation programsare not widespread in the Anglophonecountries.6 Consequently, the levels of yields ofcotton and groundnuts, and the relative degreeto which oxen have been successfully intro-duced, tend to be superior in the Francophonecountries. However, the implementation of in-tegrated rural deve lopment projects — popularized by the World Bank — that embraceboth food and cash crops is bringing an in-creasing degree of congruency between the

5.

6.

These parastatal organizations have evolvedf rom French-based organizations.Tobacco and tomatoes are rare exceptions (Nor-man , in Hunter, Bunt ing, and Bottrall 1976).

Anglophone and Francophone countries.Two major factors have influenced govern-

ment policy on food production, particularly inthe Francophone countries: the importance ofencouraging export cash-crop production toprovide much needed foreign exchange and thedesire to provide cheap food for the vocal urbansector (Stryker 1978; CRED 1977).7 Thesefactors have created problems of stimulatingfood production for the market. Partly becauseof the drought in the early 1970s and the generalabsence of improved technologies for milletand sorghum, some countries have increas-ingly emphasized rice production on the low-lands (e.g. Gambia, Senegal, Mali). Efforts are to be made to reverse the consumer preferencefor rice (Government of Senegal 1977), which isnutritionally of less value than millet and sor-ghum. However, Harriss (1978) has suggestedthat for political and economic reasons, it isunlikely that governments, through pricingpolicies alone, will be able to stimulate theproduction of millet and sorghum. It wouldappear that the only practical way of reversingthe trend towards increased rice production isto generate relevant improved technologies forthe other crops. However, the success of such a program is likely to hinge to some extent on thecompetitive position of rice. Unlike the Fran-cophone countries, the Anglophone countrieshave mainly had a free market in millet andsorghum.8

The external institutions (i.e., supportsystems) play a crucial role in creating favorableconditions for the development and adoption ofimproved technologies for millet and sorghumby farmers. Without this support, it will not bepossible for research agencies, includingICRISAT, to develop technologies that will beadopted by farmers.

7. This has met wi th only l imited success (Harriss

1978).8. Nigeria has within the last 2 years taken the first

tentative steps in government involvement in themarketing of food crops. Indeed i t could beargued that even in the Francophone countriesthe market has in practice been fairly free, sincewell over 70% of the marketed cereals usually gothrough private channels (i.e., the parallel mar-

ket).

247

C o m m u n i t y N o r m s , S t r u c t u r e s ,a n d B e l i e f s

Traditionally, villages in the SAT of West Africa,whether nucleated or dispersed in a settlementpattern, had a strong sense of community, oftenwith a strong hierarchical system of control(Remy 1977; Ramond, Fall, and Diop 1976). Thiscontrol was not considered to be very exploit-ative, however, especially since individuals canonly possess usufructuary rights to the land(Hill 1972; Maymard 1974).9 Haswell (1975) haspointed out that communities have had a "shared poverty" concept, wi th poverty beingpr imar i ly de termined by the technicalelement — climate and soil. However, recentpopulation increases, concomitant with in-creasing contact with the outside wor ld, haveresulted in a breakdown of the community spirit,an increase in individualization,10 and less as-sumption of responsibility for one's fel low man.Some of the developments associated with thisbreakdown can be summarized as fol lows:

1. There is a trend in many areas for thefarmers' goal of selfsufficiency to be partiallysacrificed in order to produce cash crops. Thereare no longer serious social sanctions for failingto have enough food in the granary to feed thefamily until the next harvest.

2. Poverty is becoming individualized11 andnew power groups in the village (traders,money lenders, etc.) who may or may not bemembers of the traditional elite, are holdingeconomic power over the more disadvantagedgroups, without the feeling of responsibilityepitomized in the patron-client relationships oftraditional communities (Haswell 1975).

3. There is an increasing tendency towardsindividualization of land tenure (Kohler 1968)

9.

10.

11.

Others have, however, quest ioned the lack ofexploitat ion in tradi t ional societies (Ernst 1976;Kafando 1972). But for reasons ment ioned in thepaper, we believe that the potential for exploita-t ion is likely to be exacerbated as t ime goes on .Unfortunately, many development programshave concentrated on individual izat ion to thedetr iment of the tradit ional strengths of thecommuni ty .Haswell (1975) terms It " inst i tu t ional izat ion" ofpoverty.

248

although in most areas this is not recognized bylaw. Usufructuary rights to the use of land areincreasingly being rented, purchased, andpledged, particularly in areas where populationpressure is becoming a problem (Goddard1972; Raynaut 1976). In certain areas, wheresome preferred types of land are very l imited,cases have been reported of land being com-mandeered by the elite who have acquired thebenefits of that land for their own use (Agbonifoand Cohen 1976).

4. The characteristics of non-family labor,although not a major component of farm laborfor many areas in the SAT of West Africa, havechanged. The traditional systems of communaland reciprocal labor are being replaced by moreimpersonal labor paid in kind or increasingly incash, by the job or by the day (Raynaut 1973;Unite d'Evaluation 1976). Even the strangerfarmer, or the navetane system, strongly as-sociated with the cultivation of cash cropgroundnuts in Gambia and Senegal, is becom-ing increasingly monetized.

In spite of rapid social changes taking place,the traditional hierarchical structures are stillimportant in influencing like at the village leveland, as a result, care should be taken to draw onsuch leadership in the introduction of agricul-tural change (Dunsmore et al. 1976). However, itis also important that inequalities in livingstandards at the village level are not greatlyincreased through reinforcing traditional socialpower with newly obtainable economic power.

G o a l s

To our knowledge, little rigorous empirical workhas been done on the goal(s)12 of farmers andfarmers' attitudes to risk and uncertainty. Indirectevidence is often given that farmers are riskaverters (e.g. the practice of mixed croppingincluding different species and different va-rieties of the same species, crop diversification,etc.). Implicitly and sometimes explicitly, theauthors of the studies reviewed have assumedthat the goal of meeting food needs from farmproduction is important. They have also often

12. Ancey (1975b) has provided a useful discussionon the mult ipl ic i ty of goals that are likely to existin farming famil ies in West Afr ica.

assumed that spare resources over and abovethose required to meet the subsistence goal aredevoted to maximizing profits through produc-ing products for the market.13 Some studieshave suggested that, on occasion, as a result ofexternal pressures — such as skewed supportsystems (Lele 1975) or the compulsion and needfor money to pay taxes (Campbell 1977; Nicolas1960), changing desires (as for independenceon the part of individual members of the familyRocheteau 1975; Albenque 1974) or economicnecessity (Matlon 1977)— there has been a diminution of the subsistence goal in favor ofprofit maximization involving the production ofexport cash crops. There is no question thatmuch more work needs to be done on thegoal(s) and attitudes of famers towards risk anduncertainty.14 Such knowledge is an essentialingredient in designing relevant improvedtechnology. It is also important to assess theimpact of increasing integration of farmingfamilies into the market economy on theirperspectives, management, and goal(s) and toascertain how this will influence the way inwhich their resources will be allocated in thefuture.

Land Input

As mentioned earlier, individuals in the SAT ofWest Africa usually only possess usufructuaryrights to land. Rights can be inherited throughthe operation of the customary or Maliki law.15

13. Both of these goals or that of simple profitmaximizat ion have been assumed in the fewlinear programming studies done in the SAT ofWest Africa (Richard, Fall, and Attonaty 1976;Ogunfowora 1972; Bourl iaud, Boussard, and Leb-

lanc 1975). Hopkins (1975) also using a linearprogramming approach, has attempted to cometo grips w i th the problem of uncertainty byincorporat ing a game theoretic approach. Bar-nett, work ing at Purdue University, is at presentcomplet ing a goal p rogramming study.

14. Work of the type that is currently being underta-ken by Binswanger (n.d.) in India could prove tobe of particular value.

15. Changes towards forms of individual tenure aretaking place as a result of increased populat ionpressures, particularly In and around urbanareas, sett lement and irr igation schemes, andlarge, mechanized farms util izing tractors.

In areas of low population pressure, the amountof land farmed is in general a function of familysize (labor force) and the quality of the land. Aspopulation density increases, two and perhapsthree significant changes are taking place:

1. Farm size is decreasing (De Wilde 1967),with the result that the short-run private oppor-tunity cost of leaving land fallow is increasing(Norman and Pryor 1979). The problem offinding new ways to maintain soil fertility be-comes more and more urgent, since both thelength of fallow and amount of land fallowed isdecreasing.

2. Distribution of land among producers maybecome an increasing problem. Land, becauseof the mode of tenure and the historically lowpopulation pressure, has been fairly evenlydistributed. Unfortunately, statistics givingtrends over time are not generally available.However, there is fragmentary evidence thatinequalities are growing. Increased populationpressures have on occasion encouragd influen-tial groups in villages to increase their relativeshare of the land cultivated (Swanson 1978;Dubois 1975). Such moves have been en-couraged by the overall shortage of land and/orby the introduction of certain types of improvedtechnology, such as dry-season tomato pro-duction in northern Nigeria (Agbonifo andCohen 1976), and oxen in the Gambia (Weil inMcLoughlin 1970). Increased monetization ofthe economy and the trend towards individual-ization of land tenure mean that usufructuaryrights are more frequently being obtainedthrough pledging, purchase, leasing, and rent-ing (Goddard 1972). Such trends obviouslyhave serious implications in terms of the futureand, because of the apparent increasingheterogeneity of farmers in terms of resourcebase, for differentiation in terms of the de-velopment of relevant improved technologies.

3. In much of the area, particularly where theMaliki law applies, there is progressive frag-mentation of farms. Although there are occa-sional references to attempts at spontaneousconsolidation of land (Goddard 1972), much ofthe impetus tends to come from the outside(e.g. in the Experimental Units in Senegal asdescribed by Faye and Niang 1976; near Fana inthe CMDT area of Mali). Although a certaindegree of fragmentation has some advantagesit does cause particular problems whenmechanization is being introduced.

249

Labor Input

Family Structure

Mention was made earlier about defining a family or decision-making unit. A commontrend in the SAT of West Africa is that thetraditionally preferred extended or complexfamily unit, consisting of more than one mar-ried man plus dependents, is breaking up intonuclear simple family units — one married manplus dependents — within the same com-pound, and these then may eventually formseparate compounds (Nicolas 1960; Benoit-Cattin 1977b; Buntjer in Mort imore 1970). Theunderlying reasons, as with the changes dis-cussed in community structures, norms andbeliefs, revolve around increased contact wi ththe outside world and monetization of theeconomy. However, the rate at which thischange is taking place depends on a number ofcomplex interactions. It has been suggestedthat the introduction of cash crops, seculareducation, increased off-farm employment op-portunities, new settlements, and migrationencourage this breakup, although the speed atwhich this takes place may be tempered by thestrength of the traditional hierarchical structure,the ethnic origin of the people concerned, theownership of cattle, etc.

A number of important implications ariseconcerning such a trend. Two of them are:

1. Fields farmed by families are traditionallydivided into common and individual fields. Thecommon fields — controlled by the familyhead —tradit ional ly provided food for all mem-bers of the family. An increasing proportionof the fields are, however, coming under thecontrol of other individuals in the family. Thereis a decrease in the obligations of family mem-bers to work on the common fields, and there isno longer the assurance of food from the familyfarm to meet subsistence needs. Increasedindividualization of fields and the need forcash to pay taxes have both cont r i -buted to encouraging the growth of exportcash crops for the market. Decisions are increas-ingly made by individuals wi th in families ratherthan the family head; this creates problems inintroducing improved technology, especially ifan extension or institutional credit program isinvolved, because such programs tend to bedirected at family heads.

250

2. In many areas the breakup of families isresulting in smaller farms (although with simi-lar land-per-resident ratios), increased frag-mentation of fields, younger and relatively in-experienced family heads, and often increaseddependent-per-worker ratios, with resultingpoorer net worth and cash liquidity levels. Suchtrends raise questions about the appropriate-ness of certain types of technology (e.g. oxen)16

and processes (e.g. cattle ownership).17

Total Time Worked

Farm work in the Sahelian and Savanna areasrevolves around crops, livestock, and off-farmenterprises. In much of the area labor, not land,limit the expansion of production (United'Evaluation 1978; Ogunfowora 1972; Lewis1978). However, it is unfortunate that there arevery few detailed labor f low data by season, job,age, and sex. Nevertheless, some general-izations are possible:

1. The major input on the family farm tendsto be provided by family members (Lowe inDunsmore et al. 1976) and often by family maleadults (Matlon 1977). The significance of femaleadult labor on the farm varies according tocustom (Hill 1972), tasks to be performed (Nor-man 1972), urgency of the work, and the cropgrown (Lowe in Dunsmore et al. 1976). In termsof hired labor, age-set, reciprocal and com-munal labor have, as mentioned earlier, in-creasingly given way to labor paid by the hourand job (i.e. contract), remunerated in kind, orincreasingly in cash, with contract work receiv-ing the highest remuneration when reduced to a per-hour basis.

2. The annual total work by family maleadults often appears to be rather low. However,in evaluating it, one must consider the overrid-ing significance of the seasonality of agricul-ture. The coefficient of variation for monthlyinputs on the family farm increases as onemoves north through the region (Norman and

16.

17.

This observation concerning oxen Ignores thepossibil i ty of hir ing them or using them to docontract work for other farmers.The poorer net wo r th and cash l iquidity levels arelikely to reduce the possibil i ty of purchasingcattle whi le ownership wou ld , due to labor l imita-t ions, entail management by herders.

Pryor 1979). This creates two problems: laborbottlenecks during the rainy season, and under-employed labor during the dry season.

Seasonality

The allocation of t ime worked by family mem-bers in each of the three processes attempts toeven out the annual f low of labor. For example,off-farm work is emphasized during the dryseason but substantially reduced during therainy season. Nevertheless, for a number ofreasons, complete compensation resulting inthe same amount of total work each monthduring the year is not usually achieved (Normanand Pryor 1979) although obligatory social ac-tivities can account for a lot of so-called sparet ime during the dry season (Delgado 1978).Efforts to increase the productive use of laborduring the dry season also involve cultivatingthe l imited amounts of low land by using re-sidual moisture or irrigation (Lahuec 1970),going on short-season migration (Faulkingham1976), etc. Attempts to increase annual laborproductivity results in longer term migration tonew settlement areas (Benoit-Cattin 1977a),urban areas, or more humid areas to the south(Byerlee et al. 1976).

A number of seasonal labor bottlenecks havebeen identified in different studies. These mustbe assessed w i th care because laborbottlenecks can be a function of the aggregationperiod (e.g. week or month), length of thegrowing season (the shorter it is, the morepeaked its labor activity), the type of technologyemployed, and the power source. At the risk ofover-simplification, the fol lowing general-izations appear to be possible:

1. With only hand labor using indigenous(traditional) technology, weeding is often con-sidered to be the most demanding operation(Norman and Pryor 1979).18

2. Introducing improved land-intensivetechnology (e.g. seed and fertilizer) withoutchanging the power source shifts the bottleneckperiod f rom weeding to harvesting the in-creased yields that result from its adoption

18. Land preparation is also sometimes stated tocause a labor bottleneck. This Is particularly sowhen t im ing is important due to the shortness ofthe growing season (Unite d'Evaluation 1978).

(Norman in Kowal and Kassam 1978). However,care should be taken in interpreting this sincetiming is a particularly critical factor in theweeding operation (Haswell 1953; Matlon andNewman 1978). Analysis of the labor f low datain aggregation periods of less than a monthwou ld probably enhance the weedingbottleneck period compared with harvesting.

3. A change from hand to animal power,using indigenous technology and ridgingequ ipment , accentuates the weed ingbottleneck bacause it enables preparing largerareas of land, which then have to be weededmainly by hand (Tiffen 1971). Also, the harvest-ing bottleneck may under certain conditions besomewhat more concentrated because landpreparation, and hence planting operations, arecarried out more efficiently and quickly.

4. The combination of animal power withridging, planting, and weeding equipment — together w i t h improved land- intensivetechnology —tends to accentuate the harvest-ing bottleneck even further (Faye, personalcommunication), although this can be easedsomewhat by carting the harvest from the field.

Traditionally, farmers have used a number ofstrategies to alleviate labor bottlenecks such asweeding. Some well-established ones are:working more days and longer hours per day onfarm work at such periods (Jones in McLoughlin1970); reducing t ime spent on off-farm work(Matlon 1977); in some areas using relativelymore labor of women and children; when pos-sible, hiring labor (Kohler 1971; Lewis 1978)although for a number of reasons this haslimited potential;19 and growing crops inmixtures.

In view of the seasonal labor bottlenecksdiscussed above, the implications for develop-

19. Unlike India, West Africa at present has no classof landless laborers in rural areas. The opportun-ity cost of hired labor, which comes mainly from other farming famil ies, is therefore high. Anadditional complicating feature is that such laboris often required when the cash l iquidity levels offamilies are poor (Matlon 1977). It could thereforebe argued that In West Africa the negative effectsof labor bottlenecks are more important than thepossible posit ive effects that have beensuggested in India in providing a source ofemployment for landless laborers.

251

ing relevant improved technologies are im-portant. Too often, the technological develop-ment aim of maximizing yield per unit area, orof increasing the area cultivated, has resulted inthe development of irrelevant improvedtechnology, which, if adopted, would result ineven greater seasonal bottlenecks (Kafando1972). Use of animal power to alleviate specificbottlenecks has been most widely applied,while chemical methods (e.g. herbicides) andbiological improvements (e.g. varieties andpractices avoiding demand during peak laborperiods) are still in their infancy. But efforts toimprove labor productivity in such periods asweeding are of critical importance. This is illus-trated by the fact that the marginal productivityof labor at such a period can be as much as threeto four times higher than the wage rate (Nor-man 1970). It should also be recognized that, indeveloping improved technologies, overcom-ing old problems may result in the creation ofnew ones.

C a p i t a l a n d C a s h I n p u t

Traditionally, the capital owned by farmingfamilies, apart f rom livestock, consisted largelyof goods produced by them through a directembodiment of labor, such as hand tools, grainstores, etc. Consequently, capital levels havetended to be low (Hopkins 1975; Ernst 1976).Livestock have mult iple uses: they are a form ofsavings and investment (Eskelien 1977),sources of meat and manure and, depending onthe type, draft power and/or by-products.Ownership of livestock, apart f rom cattle, tendsto be widely dispersed, both between andwithin families. Cattle ownership, on the otherhand, tends to be unevenly distributed, beingconcentrated towards the wealthier, often moreinfluential, families (Lowe in Dunsmore et al.1976).20 Ownership and management functionare sometimes separated, the latter being in thehands of nomadic herders, usually the Fulbe,who also own cattle in their own right. Suchdifferentiation causes potential problems in theimprovement of livestock husbandry (Eriksen1978).

With the introduction of improved technol-

20. Management, and maybe also ownersh ip of cat-tle, is also a funct ion of ethnic or ig in.

ogy, there has been a change in character ofcapital to a type that involves purchasing in themarket, and which is no longer simply producedthrough utilizing labor at the farm level (e.g.inorganic fertilizer, animal equipment, etc.).

Traditionally, most explicit farm expenditureshave been for payment of nonfamily labor(Norman and Pryor 1979). Such payment wasoften in kind, although this is being replacedincreasingly by cash, particularly when cash isavailable, as after harvesting. The seasonalcash f low, which tends to be inversely related tothe level of agricultural activity, has causedparticular problems. The t ime when the level ofagricultural activity is approaching its peak — June to September — coincides with the t imewhen cash resources are at their lowest ebb.With the introduction of improved technology,this problem is likely, initially at least, to beexacerbated. Traditional sources of credit, oftenobtained for reasons of consumption during thepreharvest season, reputedly often carry highimplicit, if not explicit, interest rates (Dubois1975), although this is to some extent a functionof source (King 1976), posit ion, and collateral(Haswell 1975). This has led to attempts tointroduce institutional forms of credit, bothshort-term (e.g., fertilizer, improved seeds, etc.)and medium-term (e.g., oxen, equipment).

Two criteria are often used in evaluating suchcredit: repayment rates and, perhaps lesscommonly, equitability of access. In terms ofrepayment rates, it appears in general that highlevels are only achieved when such programsare carefully coordinated wi th other externalinstitutions and support systems, particularlyinput distribution and marketing of the product(King 1976). Certainly this has been true wi threspect to the introduction of oxen draftsystems, which are closely correlated with thepresence of a viable cash crop (e.g., groundnutsin the Sine Saloum area of Senegal and in theGambia, cotton in Mali Sud and the Gombe areaof Nigeria). Cooperatives or "precooperatives"have often been used by government to en-courage equitability of access to institutionalcredit (e.g., Senegal, Niger, Nigeria). However,because in traditional village societies the pre-conditions for true cooperatives to functioncannot be met, or have not been met, success isoften l imited (King 1976; Gentil 1971; Storm1976).

In India, oxen are an integral part of the

252

subsistence sector; in West Africa, however, itis difficult to envisage millet and sorghum,which are mainly food crops, sustaining theeconomic viability of oxen and equipment viacredit programs. This is at present being tried inareas where no export cash crop exists (e.g.,Operations Mils in Mali) but the present pricingpolicies — and their implementat ion — onsuch crops in Francophone countries, togetherwith a lack of improved technologies, wouldappear to reduce this possibility even further.

Cropping Systems

C r o p s G r o w n

The crops that can be potentially grown are, asmentioned earlier, determined by the physicaland biological factors that comprise the techni-cal element. However, the crops that wil l begrown are a subset of this and reflect both the"hor izontal" (i.e., exogenous and endogenousfactors) and "histor ical" dimensions. Theseperspectives are necessary to appreciate thedynamic and evolving nature of the cropsgrown in the SAT of West Africa. Several writershave, as noted earlier, recognized the impact ofcolonial and post-colonial policies in encourag-ing production of export cash crops. Probablyone of the best documented studies has beenthat undertaken by Haswell (1975) who studieda Gambian village at different times over a period of 25 years. She found, for example, that:

1. Early millet and hungry rice, the traditionalearly-harvested crops that help shorten thehungry season, had decreased in importancebecause of the decreased labor available fromwomen and children who had traditionally culti-vated the crop. Women had changed fromworking on upland (rainfed) crops to swamprice, made possible when the government builtcauseways.

2. Groundnuts, due to favorable governmentpolicies, had increased in significance in termsof both area cultivated and proportion of totalproduction harvested.

3. Nevertheless, until the introduction ofoxen in a government supported scheme in the1960s, there was a downward trend in the areaof rainfed land cultivated, a trend that hadcommenced with the move of women toswamp-rice work. With the introduction of

oxen, this downward trend was reversed. Al-though some of the increased cultivated areawent to groundnuts, there was a resurgence ofinterest in growing late millet, particularlymixed with groundnuts. Weil had also noted inMcLoughlin (1970) the trend back to food cropsas a result of the introduction of oxen. However,the food/cash crop ratio is still lower than earlieron upland crops, although some of the slack hasbeen taken up with the increase in swamp rice,where the yield potential is much higher.

As the preceding example shows, it is impor-tant when developing an improved technology,to understand the dynamic environment inwhich the farming family operates. It is im-portant to assess the possible relevancy of thetechnology, not only in terms of the targetgroup to whom it is being directed but also interms of the external institutional supportsystem required for its adoption. Also an at-tempt needs to be made to assess the possibleramifications of its adoption, such as, changingresponsibility and reward ratios, particularlyamong the sexes within the family (Eskelien1977), increasing inequalities in the society,etc.21 There is no question that these issueshave often been ignored in the technology-development22 and implementation phases.

C r o p p i n g P rac t i ces

It is not possible to discuss the cropping prac-tices of farmers in detail. These practices haveevolved through generations (Swift 1978;Johnson 1972) and reflect adaptation to theenvironment. The relatively recent acceleratedpopulation increases have, however, upset the

2 1 . We recognize that such an assessment ex ante islikely to be crude but we believe that such anattempt is necessary in order to provide a brea-thing space for determining possible researchpriorities, developing possible corrective de-velopment strategies, etc.

22. Also, most research has ignored the determinantfactors related to the marketable surplus of foodcrops, preferring to assume that a certain propor-t ion of total production wi l l be marketed. ForICRISAT purposes, it is important to understandthe effect on farmers' behavior of the differentialrisk attached to an assured market price forexport crops wi thout a similar assurance for

food crops.

253

traditional adaptation process, which requirestime. This highlights the necessity of develop-ing relevant improved technology. This doesnot mean that traditional practices should beignored. Indeed, it is apparent that whi le manyof these practices are f irmly entrenched withinthe farming systems and wil l be difficult tochange, there are many which can be used asbuilding blocks for developing improvements.Practices discussed in some detail in variousstudies include ridge, mound, and flat cultiva-t ion systems; use of plant indicators in asses-sing soil fertil ity; burning bush, crop residues,and sometimes manure; and explicit (year toyear) and implicit (within year)23 rotations.There are, however, two additional specificpractices that deserve special mention in thecontext of West Africa: " r i n g " cultivation, andmixed cropping systems.

" R i n g " C u l t i v a t i o n S y s t e m

Throughout many parts of the savanna areas ofWest Africa, a system has traditionally beenpracticed that involves the permanent cultiva-t ion of some fields, usually near the compound,where fertility is maintained through manuring.Fields farther away are cultivated for a fewyears, after which soil fertil ity is restoredthrough fal lowing (Marchal 1977). However,increasing population densities have upset thisbalance. In some areas, the increasing landshortages are resulting in a higher proportion ofpermanently cultivated fields, while the remain-ing fields are left fal low for progressivelyshorter periods. Traditionally, there have beensymbiotic relationships between livestock her-ders and sedentary crop farmers in which man-ure for fields is an important element. There iscertainly evidence that rates of manure appliedincrease as the proportion of land that ispermanently cultivated increases, and that itbecomes more of an economic good (Normanand Pryor 1979). However, it has also beennoted that, apart f rom a few exceptional areas(Mortimore and Wilson 1965), this has not fore-stalled the decrease in yields, a problem that

23. By this is meant the practice of g row ing crops inmixtures. Spl i t t ing ridges every year (Buntjer1971) means that the crops g rown on a particularpiece of land wi l l vary f r o m year to year a l thoughthe same mix ture may be present as a whole.

has been of particular concern in Francophonecountries where attempts have been made toalleviate the problem wi th the introduction ofoxen.

M i x e d C r o p p i n g

Results in three areas in northern Nigeria (Nor-man and Pryor 1979) indicate the overall domi-nance of mixed cropping24 over sole crops.While yields of individual crops were depressedwhen grown in mixtures, such reductions weremore than offset by other crops in the mixture,resulting in a higher return per mixture whenexpressed in value terms. Annual labor inputswere higher per hectare for crop mixtures,although the differential was reduced when thelabor put in during the labor-bottleneck periodwas considered. In spite of the higher laborinputs, the returns per annual man-hour and toan even greater extent, per man-hour put induring the labor-bottleneck period, were higherfor crop mixtures than for sole crops. The latterresults indicate that mixed cropping helps al-leviate the problem of the labor-bottleneckperiod, which in this case was weeding (Ogun-fowora 1972). Finally, the results show thatgrowing of crops in mixtures under the existingtechnological levels in the areas studied wasnot only more profitable but also more depend-able (Abalu 1976).

In general, the traditional practices in grow-ing crops have been neglected in the develop-ment and introduction of improved technology.The situation as it exists today is once again thefunction of the interaction of the "vert ical" and"hor izontal" dimensions. For example, in Fran-cophone countries, where export cash cropswere introduced and yields of such crops sub-stantially increased, much of the crop is grownin sole stands. This is probably a reflection ofthe technology having been developed for solestands and the success of the external supportsystems (i.e., external institutions), which en-courage the growing of these crops accordingto the official recommendations. In areas whereyields have not been raised so dramatically.

24. For the purpose of this paper, crop mixtures wi l lbe def ined as a combinat ion of t w o or more cropsof different species or varieties present on a g ivenpiece of land at the same t ime.

254

because the improved technology has not beenadopted, the export cash crops are still oftengrown in mixtures (e.g., Nigeria).

The practice of mixed cropping is still verydominant in food crops25 where, as mentionedpreviously, improved technology is not gener-ally available. Mixtures of millet or sorghumand cowpea — for grain but also as forage — are very common throughout the area (Repub-lique du Niger 1973) wi th millet/sorghum andmillet/groundnut or sorghum/groundnut mix-tures sometimes found. Although such combi-nations are often mentioned, empirical analysisof them is generally lacking.

ICRISAT studies in India have observed thatthe high-yielding varieties have increased thesignificance of sole cropping (Jodha 1977). It istempting to conclude that adoption of improvedtechnology wil l and should lead to the demiseof mixed crops. However, technology de-velopment and extension programs have con-centrated on working with sole crops and onlyrecently has empirical work undertaken bytechnical scientists — particularly at ICRISATand the Institute of Agricultural Research innorthern Nigeria — demonstrated the potentialfor growing crops in mixtures, using improvedtechnology. It has also been demonstrated that,although the number of crop mixtures appa-rently decreases wi th the introduction of animaltraction, it is certainly not imcompatible with itspresence (Unite d'Evaluation 1978).

In the light of these results, there appears tobe considerable justification for increased em-phasis on developing improved techno-gies for mixed crops in the SAT of West Africa,especially as the potential for sequential cropp-ing is circumscribed by the shortness of therainy season. Such emphasis should concen-trate on working wi th crops that, exhibit com-plementary relationships. Complementaritywil l be enhanced when one or more of thefol lowing characteristics offset the competitiverelationships between the species under con-sideration: different growth cycles, differentrooting habits, symbiotic relationships betweendifferent species, compatible labor demandsand practices, etc. One criterion in evaluatingthe results of some crop mixtures should be the

25. This is t rue in both the Anglophone and Fran-

cophone countr ies.

multiple use of the products, for both humanand animal consumption. As mentioned earlier,it is likely that in some areas crop residues f romcowpeas for livestock food may be a moreimportant product than the grain, which is usedfor human consumption.

Crop Livestock In terac t ion

O t h e r t h a n A n i m a l T r a c t i o n

The potential benefits of some degree of inte-gration between crops and livestock have longbeen recognized in the SAT of West Africa. Theintegration of crops and livestock can lead tomore efficient use of land unsuitable for cropproduction, provide a use for crop residues andby-products, provide manure, and be a sourceof income, savings, and investment. This re-lationship has developed in spite of the fact thatoften livestock ownership and management isin the hands of nomadic herders, while cropsare grown by farmers. Such a symbiotic re-lationship developed traditionally in areas withrelatively favorable land/labor relationships.However, large increases in population densityhave forced and are forcing changes in thetraditional relationships. The diminishingavailability of land is resulting in problemsconcerning alternative use of land, conflicts be-tween herders and farmers (Campbell 1977),conflicts concerning other resources such aslabor and capital being devoted to products forhuman consumption or to animal production(Delgado 1978) and declining soil fertility. It isone of the paradoxes of the ever-decreasingland/labor ratio that the increasing conflict be-tween devoting land to crop or animal pro-duction inhibits the beneficial effect that lives-tock can have in preventing the decline in soilfertility. It is of paramount importance, in theinterest of long-run ecological stability, that thepresent competitive relationships that are de-veloping be reversed and symbiotic relation-ships be reestablished. Apart f rom present re-search on animal traction, this is a neglectedarea in terms of technology development andimplementation programs in the SAT of WestAfrica.

Animal Traction

Animal traction in the West African context, in

255

contrast to that of SAT India, has a history of lessthan 50 years (Hasif 1978). The introduction ofanimal traction can help increase the produc-t ivi ty of labor through the use of equipmentdesigned to increase the efficiency of labor atseasonal bottleneck periods. The more tradi-tional idea of draft power to aid in increasing thearea of cultivation (i.e., extensification) is beingreplaced, particularly in the Francophonecountries, by the concept of draft power as a way to increase the productivity of soil (i.e.,intensification) through manure application,deep plowing enabling the burial of crop re-sidues, etc.

How successful has animal traction, includingnot only oxen but also donkeys and horses,been on lighter soils? Animal traction in theWest African context is, as mentioned earlier,closely l inked w i th the commercial izedeconomy. Its successful introduction has re-quired the production of an export cash crop26

to provide the revenue to pay for the equipmentand sometimes the animals, complemented bya strong support system (i.e., external institu-tions) in the form of an input distribution sys-tem, institutional credit, extension service, andmarket for the product.

Many studies have considered various as-pects of animal traction.27 However, some of theproblems mentioned earlier include:

• lack of trained animals and operators,especially for inter row cultivation (De Wilde1967, Wilcock, personal communication);

• weakness and insufficiency of draft ani-mals, caused by lack of supplemental feed (Weilin McLoughlin 1970);

• inappropr ia te equ ipmen t (Lowe inDunsmore et al. 1976, Rocheteau 1975);

26. Lowe (Dunsmore et a l . 1976) and the IER studiesindicate that i t is extremely diff icult for animaltract ion to be prof i table w h e n land Intensificationtechniques are not used to raise yields of exportcash crops. However, somewhat in contradic-t ion to th is are the results f r om the Gombe area inNigeria, where yields of cot ton, a l though higherthan those obtained tradi t ional ly, were sti l l we l lbe low the level at wh ich they potent ial ly couldhave been (Tiffen 1971). Even so, animal-tract ionintroduct ion has met some degree of success inthis area.

27. Much of the work of CNRA, Bambey, has addres-sed this issue.

• inadequate facilities for repair and servic-ing of equipment;

• nonavailability of suitable equ ipmen t -such as ridgers and plows — during peakperiods (Tiffen 1971);

• under utilization of animals during the yearas a whole (Zalla 1976);

• fragmented holdings28 that reduce workefficiency;

• damage to equipment from the large num-bers of tree stumps28 in the fields; and

• lack of finances to help farmers hire draftanimals.

There are in addition two major problems thatdeserve special mention:

1. The economics of animal draft power isbeing questioned, even for families that previ-ously had large enough farms and were locatedin areas where export cash crops could beproduced. In recent years, prices of cash cropshave increased relatively less rapidly thanprices of animals and equipment (CRED 1976;Traore and Toure 1978). This trend is slowingdown the adoption of animal draft power andreducing the beneficial interaction betweencrop production and livestock. Another dangerfurther aggravates the dual economy that ap-pears to be developing between farmers whoown oxen and equipment and those who donot.29 Relationships between families, who owndraft animals and those who do not, in whichservices provided in the form of plowing arepaid for in terms of labor, are reported invarious studies (Ernst 1976). The potential forexploitative relationships developing is obvi-ous, especially if such labor is demanded attimes when its opportunity cost is high.30

2. Incorporating residues by deep plowing(enfouissement) after harvest, which is the cor-nerstone of the intensification policy, has not

28.

29.

30.

The Experimental Units have specifically addres-sed these issues by giv ing incentives to farmersto destump their fields and by encouraging con-sol idat ion of fields (Faye and Niang 1977).Of course, there always have been farmers w h ocould not reasonably expect to have oxen. Thoseeither have farms that are too smal l or are locatedIn areas where there is no viable export cash crop.It wou ld appear that this wi l l always be apparentto a certain degree, a l though the potential for thisis l ikely to be higher when relatively few famil ieso w n oxen.

256

been successful to date (Hopkins 1974). Far-mers have tended, certainly initially, to haveseen the use of draft animals more as a meansof extensification rather than of intensification(Millevelle, personal communication). Al-though this is likely to change over t ime aspopulation density increases, it is apparent thatthere is a divergence between the short-runbenefits of deep plowing and the long-runsocial costs of doing nothing, which wil l resultin declining soil fertility.

The central problem is that deep plowing isextremely t ime-consuming, and the period av-ailable for doing it — fol lowing harvest butbefore the soil hardens — is too short for theoperation.31

T e c h n i c a l C h a n g e ,L e v e l s o f L i v i n g ,a n d I n c o m e D i s t r i b u t i o n

Comparison of levels of living in rural WestAfrica is virtually impossible on the basis ofresearch to date. Income figures are difficult todetermine because many of the farm inputs andproducts do not pass through the market placeand because incomes are rarely reported fromall sources. Hence figures derived are t ime andlocation-specific, and not often related to thecost of l iving.

Of specific interest, in terms of this review,would be information on inter- and intra-yearvariations in the level of living and on what ishappening to the distribution of resources thatraise the standard of living wi thin communit iesover t ime. Haswell (1975), in expressing farmproduction in kilograms of rice equivalent perhead per year, indicates that the average level ofliving has increased in the rural area she studiedin Gambia over the last 25 years.32 Also, evi-dence of the proliferation of metal roofs, iron

3 1 . The Experimental Units in Senegal are t ry ingother possibil i t ies. However, these wi l l Involve a move to tractor mechanizat ion, a scheme that isunlikely to have any wide applicabil i ty.

32. Whether these increases are satisfactory, espe-cially when compared to the nonagricultural sec-tor, is not discussed in this paper. However, theincreases in rural-urban migrat ion (ORSTOM1975) wou ld indicate that such increases are notkeeping pace.

beds, transistor radios, and bicycles wouldindicate that this is true, especially when draftanimals are being satisfactorily employed.However, obvious breaks in the trends arefound in a t ime of drought such as occurred inthe early 1970s. Various strategies encouragingor forcing families increasingly into the marketeconomy have made them more vulnerable todrought (Lewis 1978)33 and hence thesestrategies potentially increase the annual varia-tion in levels of l iving, unless incomes can beraised substantially above the subsistencelevel.

Mentioned frequently in the literature is theseasonal variation in levels of l iving, referred toas the hungry gap (soudure) (Raynaut 1973).Food availability is often at its lowest level whenthe demands of the agricultural cycle are high-est. Severity of this period is inversely depen-dent upon the supplies of food remaining fromthe previous harvest, the ability to purchasefood during this period, and the success of earlymaturing crops. If a hungry gap develops, thereis often a loss of body weight because of thereduced food intake in relation to the increasedworking burden (compare Grant 1950 wi th Plattin Banks 1954). At the same t ime, this has a further debilitating effect in increasing thechances of nutritionally related diseases and a reduced resistance to other illnesses (Cham-bers and Longhurst 1978). Seasonal hunger isimportant in terms of both research and im-plementation programs. Two important impl i-cations arising from a consideration of this areas fol lows:

1. It is unlikely that increased productivity oflabor, except perhaps through changing thepower base, will be possible for many farmingfamilies during peak labor demand periods,without an improvement of their nutritionallevels.

2. The hungry gap has particularly adverseeffects on the more disadvantaged members ofthe society. With the hypothesized change fromshared poverty and social power to increasedindividualization and an economic power base,such families are becoming more vulnerable toexploitation, resulting in what we earlier termedindividualized poverty. This change results

33. For example, through decreases in the amount offood grain kept in tradit ional grain stores.

257

f rom such families selling their labor to better-endowed farmers to the neglect of their ownfarms, and therefore having to accept a lowerincome f rom their own farms for short-termsurvival until the next harvest (Matlon 1977).Another strategy with equally severe long-runconsequences is for such farmers to obtainconsumption credit which is repaid at highexplicit or implicit rates of interest at harvest(Dubois 1975).

Apart f rom Matlon's (1977) research inNorthern Nigeria, little rigorous empirical workhas been done on the distribution of incomes.Both Hill (1972) and Matlon have suggested thatthere is a degree of heterogeneity among farm-ing families, but laws of inheritance, a relativelyegalitarian land tenure system, availability ofsurplus land, traditionally handpower-basedtechnologies and presumably traditionallycommunity-minded ruling elites meant thatincomes were fairly evenly distributed in tra-ditional settings. However, Matlon also pre-sents evidence supporting comments madeearlier in this paper that the degree of equality isinversely correlated with village size, popula-t ion pressure and the degree of involvement incash markets. The trend towards increasinginequalities in income distribution at the villagelevel that this implies needs to be viewed wi thsome concern. We may ask: is this unavoidable?To the extent that technology development hasnot recognized the heterogeneity amongfarmers and also that the support systems (i.e.external institutions) have been geared towardsthe better endowed or more influential farmers,the answer is yes.34 In such cases, if thesocieties of the West African SAT have a genuine concern for the goal of growth wi thequity, it is extremely important that organi-zations such as ICRISAT develop potentialstrategies that wil l avoid such potentialabuses.35

Why have distributional problems in terms of

34.

35.

Whether i t is possible to avoid the breakdown oftradit ional communi ty structures, norms, andbeliefs w i thout a basic change in ideology isbeyond the scope of th is paper and, indeed, themandate of ICRISAT.For example, a range of improved technologiesshould be developed that are sensitive not only tothe heterogeneity in the technical element butalso in the human element.

societal power, inputs, and incomes not re-ceived such attention in the West African set-ting? We believe that two reasons are: limita-tions of relevant data-collection techniques andanalytical techniques, and discipline insularity.Much of the work undertaken by agriculturaleconomists, for example, has been based on a neo-classical marginalist framework and in-volved transferring, w i th little modif ication,ideas and approaches that have been " learnt"in the so-called "capital ist" societies (Palmer-Jones 1978b). A "classical" approach has beenused, for example, in arguing that farmersallocate their resources to products in aneconomically efficient manner. As a result, it isoften argued that they have a goal of profitmaximization and that the only way to improvetheir level of well-being is through developmentof improved technology (Norman 1970). Thepreoccupation wi th allocative efficiency hasdetracted from the importance of also looking attechnical efficiency. There is increasing empir i-cal evidence that differences in technical ef-ficiency can and do exist among farmers (Mat-lon and Newman 1978). Another area that hasbeen neglected has been that of institutionalsupport systems and their efficiency in terms ofensuring equality of access.

The implications of these are very importantand return us to an important theme of thepaper — trends in distribution. The differencesin technical efficiency, which can be caused by a number of interdependent factors, go farbeyond economic variables to embrace techni-cal and non-economic (societal) variables, and,if not recognized and corrected by appropriatestrategies, may result in the development offurther inequalities. The role of interdisciplinarywork — that is, technical personnel, scientists,anthropologists/sociologists, and agriculturaleconomists — in correctly diagnosing the in-terdependences and in developing appropriatestrategies is obvious.

I m p l i c a t i o n s f o r I C R I S A T

In recommending an approach to be used bythe Economics Program in the SAT of WestAfrica, it is important to bear in mind the officialobjectives of ICRISAT (ICRISAT 1977). Althoughthese do not explicitly state a distributionalobjective, the former director has stated that the

258

job of ICRISAT is to help the poorest segment ofhumanity (ICRISAT 1976). This would imply theneed for developing relevant types of tech-nology for all classes of farmers. Space does notpermit a detailed discussion on all the factorsthat contribute to what might be termed rele-vant improved technology, but in the light of theabove and remarks made earlier in the paper, itscomponents should definitely include the fol-lowing (Chambers 1978):

1. Compatibility wi th the technical element,exogenous and endogenous factors (i.e. "hor i -zontal" dimension), so that the level of liveli-hood is improved — net l ivelihood intensity inChambers' terminology — through increasedproductivity of scarce resources.

2. The recognition of the heterogeneity offarmers ("heterogeneous" dimension) in termsof both resource base and social and/oreconomic power — that is a product of both the"histor ical" and "hor izontal" dimensions — through adapting different technologies thatfulfil l equity considerations.

3. A recognition of possible repercussionsfrom the adoption of the technology (i.e. "pros-pective" dimension), not only in terms of en-vironmental instability, but also in terms ofunequal distribution of benefits, and the designof technologies and institutions that wouldensure that such abuses would not develop.

Throughout this paper there have been refer-ences to specific items bearing on the abovecriteria, but we contend that it is necessary toapproach these in a complete and holistic man-ner. We believe that an appreciation of theinterdependency between the technical ,economic, and noneconomic variables is es-sential to understanding the farmers' environ-ment and in developing and disseminatingrelevant improved technologies. Without suchan approach, it is unlikely that practical resultswill be achieved that could fulfil l the criteriadiscussed above. This implies the need forICRISAT to use an interdisciplinary team oftechnical scientists, agricultural economists,and anthropologists, adopting an approach tofarming systems " f rom the bottom up." Obvi-ously, there is in principle nothing particularlyinnovative in this suggestion, either in terms ofICRISAT's present work or in terms of work atpresent being undertaken or proposed in theSAT of West Africa. However, we believe thereis considerable room for innovation in terms of

methodology and analytical procedures to beemployed. We believe that ICRISAT can andshould play a role in developing this area. If thisapproach is to be successful, it is of primeimportance that lCRISAT accept the fo l lowing: a concentrated input in one or two localities overa number of years; a more holistic approachthat goes beyond the five crops under themandate of ICRISAT and considers the wholefarming system, including livestock; and thedevelopment of very f i rm linkages wi th localand national institutions, both in order to helpbuild up local expertise through providing train-ing possibilities for nationals and to enablecoordinated work to be undertaken on theinstitutional aspects so important in the de-velopment and dissemination of relevant im-proved technology.

R e f e r e n c e s

ABALU, G. O. 1976. A note on crop mixtures underindigenous conditions in northern Nigeria. Journalof Development Studies 12(3) : 212-220.

AGBONIFO, P. O., and COHEN, R. 1976. The peasant

connection: A case study of the bureaucracy ofagri- industry. Human Organisations 35(4) : 3 6 7 -379.

ALBENQUE, D. 1974. Organisation du travail dans lecarre wolof. Bambey, Senegal: Centre National deRecherche Agronomique.

A M I N , S. (ed.). 1975. L'agriculture africaine et lecapitalisms. Paris: Editions Anthropes. Afr ican In-stitut for Economic Development and Planning/Institut Africain de Developpement Economique etde Planif ication.

AMIBA (Groupe D'Amelioration Des Methodes D'ln-vestigation En Milieu Rural Africain). 1978. Bilan destravaux. Paris: Association Francaise des Institutsde Recherche pour le Developpement.

ANCEY, G. 1975a. Les notions d'activites d'actifs a I'interieur d'une exploitat ion agricole. AMIRA 11.Paris: Institut National de la Statist ique et desEtudes Economiques (INSEE).

ANCEY, G. 1975b. Niveaux de decision et fonct ionsobjectifs en mi l ieu rural africain. AMIRA 3. Paris:Institut de la Statistique et des Etudes Economiques(INSEE).

BANKS, A. L. (ed.). 1954. The development of tropicaland subtropical countries. London.

259

BENOIT-CATTIN, M. 1977a. Projet Terres Neuves 1 1 ;rapport sur le suiv i agro-socio-economique de lacampagne, 1976-77. Bambey, Senegal: Centre Na-tional de Recherche Agronomique.

BENOIT-CATTIN, M. 1977b. Type d'exploitat ion etniveaux d 'equipement dans I'Unite Experimentaledans Sine-Saloum au Senegal: Mecanisation etagro-socioeconomie. Bambey, Senegal: Centre Na-t ional de Recherche Agronomique.

BOURLIAUD, J. , BOUSSARD, J. M., and LEBLANC, J. 1977.

La programmat ion l ineaire comme outi l descriptifdu comportement des paysans africains: une etudepilote au Senegal. Mondes en Developpement17:49-74.

BUNTJER, B. J. 1971. Aspects of the Hausa system ofcult ivat ion around Zaria. Samaru Agricultural News-letter 13 (2): 18-20.

BYERLEE, D., EICHER, C. K., LIEDHOLM, C, and SPENCER,

D. S. C. 1976. Rural employment in tropical Africa:Summary of f indings, Afr ican Rural Economy work-ing paper 20. East Lansing, Mich , USA: MichiganState University.

CAMPBELL, D. J. 1977. Strategies for coping wi thdrought in the Sahel: A study of recent populat ionmovements in the Department of Maradi , Niger. Ph.D. thesis, Clark University, Mass, USA.

CHAMBERS, R. 1978. Towards rural futures: An ap-proach through the planning of technologies. Insti-tute of Development Studies discussion paper 134.Br ighton, UK: University of Sussex.

CHAMBERS, R., and LONGHURST, R. 1978. Review of a

Conference on Seasonal Dimensions to Rural Pov-erty, Br ighton, July 1978. Institute of DevelopmentStudies, University of Sussex, Br ighton, UK.

COCHEME, J . , and FRANQUIN, P. 1967. Rapport

technique sur une etude d'agrocl imatologie de I'Af-r ique seche au sud du Sahara en Afr ique Occiden-tale. Rome: Food and Agr icul ture Organization(FAO) of the UN.

COLLINSON, M. P. 1972. Farm management in peasantagriculture: A handbook for rural developmentplanning in Afr ica. New York: Praeger.

COMITE PERMANENT INTER-ETATS DE LUTTE CONTRE LA

SECHERESSE DANS LE SAHEL/CLUB DU SAHEL. 1977.

Market ing, price policy and storage of food grains inthe Sahel: A survey. Vol. 1. Synthesis w i t h statisticalcompi lat ion and annotated bibl iography. Vol . 2.Country studies. Ann Arbor, Mich , USA: Center forResearch on Economic Deve lopment University ofMich igan; also Washington, DC: U.S. Agency forInternational Development.

CONNELL, J . , and LIPTON, M. 1977. Assessing vil lage

labour si tuat ions in developing countries. Delhi :Oxford University Press.

CRED (Center for Research on Economic Develop-

260

ment). 1976. Mali agricultural sector assessment.Ann Arbor, Mich, USA: University of Michigan.

DELGAOO, C. L. 1978. Livestock versus foodgrain pro-duct ion in southern Upper Volta, a resource alloca-t ion analysis. Ann Arbor, Mich, USA: Center forResearch on Economic Development, University ofMichigan.

DE W I L D E , J. C. 1967. Experiences wi th agriculturaldevelopment in tropical Afr ica. Vols. 1 and 2. Balti-more, M d , USA: Johns Hopkins University Press.

DUBOIS, J. P. 1975. Les Serer et la question des TerresNeuves au Senegal. Cahiers ORSTOM, Serie Sci-ences Humaines 12 (1): 81-120.

DUNSMORE, J. R., RAINS, A. B., LOWE, G. D. N., MOFFATT,

D. J., ANDERSON, I. P., and WILL IAMS, J. B. 1976. The

agricultural development of the Gambia: An agricul-tural , environmental and socioeconomic analysis.Land Resource study 22. London: Land ResourceDepartment, Overseas Development Ministry.

ECHARD, N. 1964. Socioeconomic study of the AderDoutchi Majya valleys. Etudes Nigeriennes 15.Niamey, Niger: Institut Fondamental d 'Afr iqueNoire (IFAN).

ERIKSEN, J. 1978. Livestock Development, Gambiariver basin. United Nations Mul t idonor Mission tothe Gambia River Basin. Abid jan, Ivory Coast: Reg-ional Economic Development Services Office, U.S.Agency for International Development.

ERNST, K. 1976. Tradit ion and progress in the Africanvi l lage: The non-capital t ransformat ion of ruralcommuni t ies in Mal i . London: Hurst.

ESKELIEN, R. K. 1977. Dogon agricultural system:Sociological aspects relating to development inter-vent ion. A report to the Research Foundation of theState University of New York. Chapel Hil l , NC, USA:University of North Carolina.

FAO (Food and Agriculture Organization of the UN).1976. Perspective study on agricultural develop-ment in the Sahelian countries, 1975-1990. 3 vols.Rome: FAO.

FAULKINGHAM, R. H. 1976. Ecological constraints andsubsistence strategies: The impact of drought in a Hausa vi l lage: A case study of Niger. Berkeley, Calif,USA: University of California. (Mimeographed.)

FAYE, J., and N I A N G , M. 1977. An experiment in

agrarian restructuration and Senegalese rural spaceplanning. Afr ican Environment 2 (4) and 3 (1):143-153.

GENTIL, D. 1971. La naissance d'un systemscooperatif: L'exemple du Niger. Options Medi-teraneennes 6 :88-95.

GOODARD, A. D. 1972. Land tenure, land hold ing, andagricultural development in the central Sokoto closesettled zone. Savanna 1 (1): 2 9 - 4 1 .

Government of Senegal. 1977. Food investmentstrategy. Dakar, Senegal: Government of Senegal.

GRANT, M. W. 1950. Nutr i t ion f ield working party.Colonial Office/20032. London: Her Majesty'sStationery Office.

HARRISS, B. 1978. Cereal surpluses in the Sudano-Sahelian States. Hyderabad, India: ICRISAT.

HASIF, E. 1978. L'emploi de la traction animale dansles exploitations agricoles. Paper given at the Com-ite Permanent Inter-Etats de Lutte contre la Sec-heresse dans le Sahel/lnstitut d'Economie RuraleMeet ing, Feb. Bamako, Mal i : Institut d'EconomieRurale.

HASWELL, M. R. 1953. Economics of agriculture in a Savannah vi l lage. Colonial research study 8. Lon-don : Her Majesty's Stationery Office.

HASWELL, M. R. 1975. The nature of poverty. London:Macmi l lan.

HILL, P. 1972. Rural Hausa. Cambridge: CambridgeUniveristy Press.

HOPKINS, E. 1974. Operation groundnuts: Lessonsf rom an agricultural extension scheme. Interna-t ional Development Service bullet in 5 (4 ) :59-66 .

HOPKINS, E. 1975. Wolo f farmers in Senegal: A studyof responses to an agricultural extension scheme.Ph.D. thesis, University of Sussex, Br ighton, UK.

HUNTER, G., BUNTING, A. H., and BOTTRALL, A. (eds.).

1976. Policy and practice in rural development.London: Croom-Helm.

IBRD [International Bank for Reconstruction and De-velopment (World Bank)]. 1976. West Africa foodgrain study. Washington, DC: IBRD.

ICRISAT. 1975. Proceedings, International Workshopon Farming Systems, November 1974, ICRISAT,Hyderabad.

ICRISAT. 1976. Annual report 1975-76. India:ICRISAT, Patancheru, A.P. 502 324. 233 pp.

ICRISAT. 1977. Annual report 1976-77. India:ICRISAT, Patancheru, A.P. 502 324. 239 pp.

IER (Institut D'Economie Rurale). 1977. Pour un prog-ramme de recherche sur les systemes de product ionagricole. Bamako, Mal i : IER.

IER (Institut D'Economie Rurale). 1978. Note de pre-sentation du programme de recherche socio-economique appl iquee dans la zone de product ioncotonniere, Region Sikasso. Presented at the 18thsession of the Comite National de la RechercheAgronomique, 25 -28 Apr 1978. Bamako, Mal i : IER.

JODHA, N. S. 1977. Resource base as a determinant ofcropping patterns. ICRISAT Economics Programoccasional paper 14, Patancheru, A.P., India.

JOHNSON, A. W. 1972. Individuali ty and experimenta-t ion in tradit ional agriculture. Human Ecology1 : 149-159.

KAFANDO, T. W. 1972. Contr ibution a I'etude du de-veloppement integre du Liptako-Gourma: Introduc-tion a I'etude des systemes agro-pastoraux. Dakar,Senegal: Institut Africain de DeveloppementEconomique et de Planification,

KASSAPU, S. 1976. Les depenses de recherche agricoledans 34 pays d'Afr ique tropicale. Paris: Organiza-t ion for Economic Cooperation and Development(OECD).

KING, R. 1976. Farmers cooperatives in northernNigeria. Zaria, Nigeria: Ahmadu Bello University;also Reading, UK: University of Reading.

KLEENE, P. 1976. Notion d'exploitation agricole etmodernisat ion en mi l ieu Wolof Saloum. AgronomicTropicale 31 (1) : 63-82 .

KOHLER, J. M. 1968. Activites agricoles et transforma-tions socio-economiques dans une region de I'ouestdu Mossi . Ouagadougou, Upper Volta: ORSTOM(Office de la Recherche Scientif ique et TechniqueOutre-Mer) Mission.

KOHLER, J. M. 1971. Activites agricoles et change-ments sociaux dans I'Ouest-Mossi. Memoires deI'ORSTOM 46, Paris: Office de la Recherche Scien-t i f ique et Technique Outre-Mer (ORSTOM).

KOWAL, J. M., and KASSAM, A. H. 1978. Agricultural

ecology of savanna: A study of West Africa. Oxford:Clarendon Press.

LAHUEC, J. P. 1970. Les jardins de saison seche a Zaongho (Region de Koupela). Cahiers deI'ORSTOM,Serie Sciences Humaines 5 (2) : 67-87.

LELE, U. 1975. The design of rural development:Lessons f rom Africa. Balt imore, Md , USA: JohnsHopkins University Press.

LEWIS, J. V. 1978. Descendants and crops: Two polesof production in a Malian peasant vi l lage. Ph.D.thesis, Yale University, New Haven, Conn, USA.

LINDSBERG, M. E. et al. (eds.). 1966. Wor ld maps ofcl imatology. New York: Springer-Verlag.

LIPTON, M. A., and MOORE, M. 1972. The methodology

of vil lage studies in less developed countries. Insti-tute of Development Studies discussion paper 10.Br ighton, UK: University of Sussex.

MARCHAL, J. Y. 1977. The evolut ion of agrarian sys-tems in Yatenga. African Environment 2 (4) and 3 (1) : 73-85.

MATLON, P. J. 1977. The size distr ibut ion, structureand determinants of personal income among far-mers in the north of Nigeria. Ph.D. thesis, CornellUniversity, Ithaca, NY, U S A

MATLON, P., and N E W M A N , M. 1978. Production ef-

ficiency and income distr ibut ion among farmers inthe north of Nigeria. East Lansing, Mich , USA:Michigan State University. (Mimeographed.)

261

MAYMARD, J. 1974. Structures africaines de produc-t ion et concept d'exploi tat ion agricole. Premierepartie. Un exemple de terroir afr icain: les confinsDiola-Manding aux bords du Sonngrongron(Senegal). Cahiers ORSTOM, Serie Bio logie24 : 27-64.

MCLOUGHLIN , P. F. M. (ed.). 1970. Afr ican food produc-t ion system: Cases and theory. Balt imore, M d , USA:Johns Hopkins University Press.

MONNIER, J., DIAGNE, A., Sow, D., and Sow, T. 1974. Le

travail dans l'exploitation agricole senegalaise.Bambey, Senegal: Centre National de RechercheAgronomique.

MORTIMORE, M. J. (ed.). 1970. Zaria and its region: A West Afr ican savannah city and its environs. De-partment of Geography occasional paper 4. Zaria,Nigeria: Ahmadu Bello University.

MORTIMORE, M. J., and W I L S O N , J. 1965. Land and

people in the Kano close-settled zone. Departmentof Geography occasional paper 1. Zaria, Nigeria:Ahmadu Bello University.

NICOLAS, G. 1960. Un vil lage haoussa de la Republiquedu Niger: Tussao Haoussa. Cahier d'Outre-Mer13 : 421-450.

N O R M A N . D. W. 1970. Init iating change in tradit ional

agriculture. Proceedings of the Agricul tural Societyof Nigeria 7 : 6 -14.

N O R M A N , D.W. 1972. An economic study of three ZariaProvince vil lages. Part 2. Input output study. Vol. 1.Text. Samaru miscellaneous paper 37. Zaria,Nigeria: Ahmadu Bello University.

N O R M A N , D. W., OUEDRAOGO, I., and N E W M A N , M. 1979.

The farmer in the semi-arid tropics of West Afr ica.Vol. 1. Interpretive review of the literature.Hyderabad, India: ICRISAT. (Internal Report.)

N O R M A N , D. W., and PRYOR, D. H. 1979. Technical

change and the smal l farmer in Hausaland of north-ern Nigeria. East Lansing, Mich , USA: MichiganState University; also Zaria, Nigeria: Ahmadu BelloUniversity; also Manhat tan, Kans, USA: KansasState University.

OGUNFOWORA, O. 1972. Derived resource demand,product supply and farm policy in the North CentralState of Nigeria. Ph.D. thesis, Iowa State University,Ames, Iowa, USA.

ORSTOM (Office De La Recherche Scientifique EtTechnique Outre-Mer). 1975. Enquete sur lesmouvements de populat ion a partir du pays Mossi . 3 vols. Paris: ORSTOM.

OUEDRAOGO, I., N E W M A N , M., and NORMAN, D. W., 1979.

The farmer in the semi-arid tropics of West Africa.Vol. 2. Partially annotated bibl iography. Hyderabad,India: ICRISAT. (Internal Report.)

PALMER-JONES, R. W. 1978a. Field research methodsused in fa rm management studies. Zaria, Nigeria:

262

Institute of Agricul tural Research, Ahmadu BelloUniversity. (Mimeographed.)

PALMER-JONES, R.W. 1978b. Peasant differentiation inrural Hausaland. Zaria, Nigeria: Institute of Agricul-tural Research, A h m a d u Bello University. (Mimeog-raphed.)

RAMOND, C, FALL, M., and DIOP, T. M. 1976. Prog-

ramme moyen te rme Sahel (Cellule de Liaison).Taux de penetration des techniques et incidencessur les rendements des cultures de mil et d'arachide.Bambey, Senegal: Centre National de RechercheAgronomique.

RAYNAUT, C. 1973. La circulation marchande descerea les e t les mecan i smes d ' i nega l i t eeconomique: le cas d'une communaute vi l lageoiseHaussa. Cahiers des Centres d'Etudes et de Re-cherches Ethnologiques 2.

RAYNAUT, C. 1976. Transformat ion du systems deproduct ion et inegalite economique: le cas d 'unvil lage haoussa. Canadian Journal of AfricanStudies 10 (2) : 279-306.

REMY, G. 1977. Enquete sur les mouvements depopulat ion a partir du pays Mossi. Rapport deSynthese, Fascicules I et II. Ouagadougou, UpperVolta: Office de la Recherche Scientif ique etTechnique Outre-Mer (ORSTOM).

REPUBLIQUE DU NIGER. 1973. Enquete agricole par

sondage, 1972-73. Niamey, Niger: Gouvernementdu Niger.

RICHARD, J. F., FALL, M., and ATTONATY, J. M. 1976. Le

modele 4S p rogramme lineare pour les exploita-t ions agricoles du Sine Saloum sud au Senegal etcalculs de budgets automatises. Bambey, Senegal:Institut National de Recherche Agronomique/Institut Senegalais de Recherches Agricoles/ lnst i tutde Recherches Agronomiques Tropicales et de Cul-tures Vivrieres.

ROCHETEAU, G. 1975. Pionniers mour ides au Senegal:Colonisation des Terres Neuves et t ransformationsd'une economie paysanne. Cahiers ORSTOM, SerieSciences Humaines 12 (1) : 19-53.

STOREM, R. 1976. Government cooperative groundnutmarketing in Senegal and Gambia. Journal of RuralCooperation 5 (1) : 29-42 .

STRYKER, J. D. 1978. Food security, self-sufficiencyand economic g rowth in the Sahelian countries ofWest Afr ica. Palo Al to , Calif, USA: Food ResearchInstitute, Stanford University.

S W A N S O N , R. A. 1978. Gourmantche agriculture.Ouagadougou, Upper Volta: U. S. Agency for Inter-national Development.

SWIFT, J. 1978. Notes on tradit ional knowledge, mod-ern knowledge and rural development. Presented atthe Institute of Development Studies Workshop onthe Use of Indigenous Technical Knowledge in Rural

Development. Br ighton, UK: University of Sussex.

TIFFEN, M. 1971. Changing patterns of farming inGombe Emirate, North Eastern State, Nigeria. Sam-aru miscellaneous paper 32. Zaria, Nigeria:Ahmadu Bello University.

TRAORE, N., and TOURE, M. 1978. Le machinisme ag-

ricole au Mali. Presented at the Journees d'EtudesTechnico-economiques sur le Tracteur SwaziTiukabi : Promot ion de la Construct ion grace a laCooperation entre les Pays en Developpement d'Af-r ique. UNIDO, 9 - 1 4 Oct 1978, Swaziland. Bamako,Mal i : CEPI and DMA.

UNITE D'EVALUATION. 1976. Evaluation d'Operat ion

Arachide et Cultures Vivrieres: Resultats d'une en-quete descriptive de la zone d' intervent ion deI'OACV en 1976. Bamako, Mal i : Institut d 'EconomieRurale.

U N I T E D EVALUATION. 1978. Evaluation de I'Operation

Arachide et Cultures Vivr ieres: Etude agro-economique de 32 exploitat ions agricoles en zoneOACV. Bamako, Mal i : Institut d 'Economie Rurale.

ZALLA, T. 1976. A proposed structure for the med iumterm credit program in the Eastern ORD of UpperVolta. Afr ican Rural Economy work ing paper 10.East Lansing, Mich , USA: Michigan State Univer-sity.

263

Discussant 's C o m m e n t s

W . H . M . M o r r i s *

I f ind the Newman, Ouedraogo and Normanreview excellent and have no differences to f indin their observations, although sometimes I donot agree wi th their conclusions.

I have only few comments:Basically, I see farmers throughout the region

attempting to meet their subsistence needswith a margin of safety; the subsistence cropmarket wi th its price cycles discourages farmersfrom using it as a vent for surplus; this place istaken by cash export crops.

There is a real question as to whether inten-sification or extensification is the best path toincrease farm income.

The farmers exploit all the ecological oppor-tunities that are available to them; employmentof labor in production opportunities outside therainy season crop production period has a veryimportant potential as a means of increasingincome. These opportunities include f lood re-cession, bas foods, and small-scale irrigation.

There is an important area which might be ofspecial interest to ICRISAT, the question oflong-term trends in fertil ity and yields in over-populated areas.

One purpose of this session is to suggestwhat the intervention of ICRISAT's EconomicsProgram should be in the Sahelian countries.Jim Ryan has outlined a rather comprehensiveprogram. However, a contract proposal wasoffered to U.S. Universities to launch a farmingsystems unit in Upper Volta in association wi thICRISAT/SAFGRAD research unit at Kamboinse,Upper Volta. My University, Purdue, has wonthis contract and wil l be starting research inApril 1979. Obviously this is a collaborativeresearch between Purdue and ICRISAT scien-tists.

The development of institutional linkages is,obviously, essential to the long run benefit ofthe research program. Al though we can train

* Department of Agr icul tural Economics,University, West Lafayette, Indiana, USA.

Purude

Voltaic and other Sahelian state students, ourability to train professionals in our work inUpper Volta is l imited by the inability to obtaincounterparts for our research workers.

The major problem that all of us face in doingresearch in the region is how to check thecontinuous decline in the real per capita netincome of the farmers.

264

Going Aga ins t t he Gra in

B a r b a r a H a r r i s s *

Abstract

This paper examines, through an interpretive review of the literature, interrelationships

between interventionist policy, private grain marketing systems, and grain production in

the semi-arid tropics of West Africa. It analyzes how the domestic marketing system may

contribute to rural inequality and may depress grain production; how the interventions

of the State exacerbate this process, fail to achieve their ostensible objectives and

contribute in other ways to the stagnation of grain production; and how the indigenous

production and marketing system, in turn, affects the form of State intervention and

contributes to "an intractable structural crisis in commercialization." The author

concludes that the latitude for either reformist or radical change within the present

socioeconomic structure of the West African SAT appears to be extraordinarily limited.

This essay "goes against the g ra in" metaphori-cally in four ways. First, it analyzes how the"domest ic" marketing system may contributeto rural inequality and may depress grain pro-duction. Second, it examines how the interven-tions of the State exacerbate this process, fail toachieve their ostensible objectives, and con-tribute in other ways to the stagnation of grainproduction. Third, it assesses how the indigen-ous production and marketing system, in turn,affect the form of State intervention and con-tribute to an intractable structural crisis incommercialization. And fourth, it goes againstthe grain personally for it is not written f rom theexperience of direct field work; its sources areconfined to isolated, and regionally and topi-cally unsystematic, published and semipub-lished material, so that it wil l appear to someblithely to disregard important regional varia-tions and to other to be dangerously ahistorical(for some remedies, see Meillassoux, 1971;Hopkins, 1973); it is tentative, far f rom defini-t ive; it suggests hypotheses for further researchrather than presenting the substantive contribu-t ion based on the original f ieldwork that thesetopics really require.

Agricultural marketing systems play a dual

* Research Associate, Overseas Development Insti-tute, London; and Research Fellow, School ofDevelopment Studies, University of East Angl ia,Norwich, U.K.

role in economic development in states whoseresources are primarily agricultural. On the onehand, they are channels along which moneycommercializes a not necessarily static butnoncapitalist peasant society. Increasing de-mands for money wi th which to purchase othergoods may lead to a increasing sensitivity torelative prices on the part of producers, to theirspecialization on those crops where returns aregreatest subject to particular cultural, ecologi-cal and economic constraints, and thus to anincrease in total production. Clearly it is themarketing system that transmits the crucialprice signals. On the other hand, and in order tosustain nonagricultural development, re-sources have to be extracted from the agricul-tural sector— physical resources to guaranteesupplies of food and raw materials for agro-industry and financial resources for investmentin all aspects of the nonagricultural economy aswell as for reinvestment in agriculture. Taxationis one way of transferring resources. So is theaction and relationships of the internal terms oftrade as mediated by the exchange systems.Even with ceteris paribus assumptions aboutproduction, the marketing system has to putconstraints on any production increases pow-ered by a response to comparative advantageby denying production its full resources toexpand. Naturally the form and direction, thesize and the pace of this transfer of resourcescan be highly varied.

Most exchange systems in the semi-arid

265

tropics today are simultaneously acting as re-source allocators and resource extractors.

The net effect of an agricultural marketing orexchange system upon society depends verymuch on the relationships of people to thesystem of production. These relationships de-termine who consumes how much, where, andwhy and therefore also determine distributionor consumption. The three spheres of theeconomy interact, as outl ined simply inDiagram 1.

DIAGRAM 1. Interact ions b e t w e e n economic

spheres

The theoretical problem we are to f lush outwi th facts and are to analyze in this interpreta-t ion of literature is the simultaneous occurrenceof potentially contradictory processes (the pro-cesses of resource allocation and resource ex-traction by the market, the simultaneous oc-currence of petty subsistence trade allowing noaccumulation of capital and monopoly trade byvarious institutions al lowing large-scale ac-cumulation of capital), the effects of theseprocesses on the cultivation of food crops of thesemi-arid tropics of West Africa — most nota-bly millet and sorghum — a n d the implicationof these effects for technological change.

It fol lows that it is necessary to examineinteractions between the economic spheres ofDiagram 1. The nature of each sphere and allcombinations of interactions have been re-viewed elsewhere (Harriss 1978). Here we wil lfocus on the interactions between productionand exchange in our effort to understand therole of the agricultural marketing system. Thisis, of course, not to deny the importance of theforces and relations of production in tran-sitional economies (see Post 1970 for ananalysis). Because the state has intervened incomplex ways in the mobil ization and distri-

bution of resources in the economies of theSahel, it is clear that this intervention must haveprofound implications for the marketing ofagricultural products. Accordingly, it is neces-sary for the purpose of an analysis orientedtowards marketing to subdivide the sphere ofexchange into that of the private sector and thatof the state, and to examine the interrelation-ships between these two subspheres and bet-ween each of them and production. We are atthe mercy of the literature, as befits a review, inthe details accorded to each interrelationship.

In the Sahel, agriculture absorbs 76 to 90% ofthe population but, significantly, provides 40 to55% of the GDP. Rural incomes are stagnant ordeclining (CILSS 1977, vol 1, p 11; Wilhelm1976a, p 91). Production of grains — millets andsorghum — stood in 1977 at 1.2 mil l ion tonnesper year in Upper Volta, 850,000 in Mal i , 1.3mil l ion in Niger, and 620,000 in Senegal (Nacro1977, vol 1, p 23). Where cereals are monocul-tivated, there are few regions that are not deficitareas. Elsewhere they are cultivated withgroundnuts and cotton. Cereal yields arelow — millet at 250 to 600 kg/ha in Upper Volta;sorghum at 300 to 900 kg/ha (Minister© duDeveloppement Rural 1976; Broekhuyse 1974);millet at 450 to 800 kg/ha in Mal i , sorghum at 300to 1000 kg/ha (Operation Mils Mopti 1973;Outtara 1977). Al l these ranges mask modalyields towards the low end.

Since 1960, whi le population has grown at2.2% per year, agricultural production in-creased at 1 to 1.7% (Wilhelm 1976a, pp91-92).Cash crops fared better than cereals, productionof which is either stagnant or has declined as inUpper Volta and Niger (CILSS 1977, vol 1, p 27).Because of their historical lack of urbanization,the West African states of the Sahel have thehighest rates of urbanization in the world (Wil-cock 1978, p. 258), so pressure for crop expan-sion comes from the demand side and cities arein such chronic undersupply that some Sahe-lian states resort to imports of rice and wheat.However, there is no production deficit attestedin the region as a whole in a "no rma l " year forrainfall (Nacro 1977, vol 1, p 13). In an averageyear, Mal i , Upper Volta and Niger all can and doexport millet and sorghum. Marketed surplusvaries f rom 5 to 20% of production, and thestates of the Sahel market between 20 and 50%of this surplus.

The fol lowing sections look in detail at the

266

operation of the market in this paradoxicalsituation.

The marketing systems we study first areneither traditional (they are not sunk in a histor-ical stasis) nor natural in the sense that theyhave not been interfered w i th ; nor are theyindigenous, since this carries the implication ofhaving avoided the involvement either of thestate or of outside trading communities. Weshall call them "pr ivate" for want of a betteradjective. We shall try to consider systemati-cally the interactions between this marketingsystem, that of the State, and grain production.

The Interactive Roles of theDomestic Marketing Systemand Agricultural Production

The interrelation between grain marketing sys-tems and grain production in Sudano-SahelianWest Africa has been explicitly investigated by a number of economic anthropologists. Theirdata must be considered valuable because ofthe careful and direct manner in which it iscollected, a manner that maximizes accuracythough this is at the inevitable expense ofgeographical coverage (a problem that surveyresearch does not automatically escape).

Our first sets of evidence relate to Hausaland, a region embracing northern Nigeria andsouthern Niger wi th "a high degree of cultural,linguistic and religious uniformity" (Hill 1972, p xiv) w i th the unusual characteristic for semi-arid tropical West Africa that " the population isso dense that farmers are obliged to farm all thecultivable land every year. This agronomic sys-tem, which has been practiced for centuries,does not necessarily lead to a progressivedeterioration in yields. Not even in the denselypopulated area of the "groundnut basin" ofSenegal is there anything comparable (Hill1972, p 21). As Hill herself stresses, and as weshall see, this agronomic and cultural dis-tinctiveness does not lead in a deterministicway to an economic uniformity, and in factpopulation density is not uniformly high.

Probably the clearest statement encounteredof the rural grain marketing system's role inagricultural production is f rom Clough's casestudy work in a millet and cotton-growinghausa village (1977a). Here 30%of farmerswhose production does not meet their food

needs have to sell grains required for sub-sistence directly after harvest at low prices inorder to meet the cost of taxes, debt repayment,household repair, and ceremonies involving thepurchase of manufactured goods, againstwhich the agricultural terms of trade are se-cularly deteriorating. Much of this grain isexported to deficit regions or towns creating anaggregate nutritional deficit in a producingsettlement. In the hungry or lean preharvestrainy season (the "soudure") this group offarmers is forced to work as seasonal, paid laboron the farms of others in order to obtain food atdouble the postharvest price, work that reduoestheir capacity to farm their own lands. Or inClough's case they are forced to borrow moneyat seasonal interest rates of between 50 and140%, to be repaid in kind at postharvest prices.Their body weights may decline and morbidityincrease during this period (Hunter 1966,Chambers 1978) further reducing thesefarmer's capacity for productive physical work.

Successful middle farmers have bimodal sel-ling patterns, also contributing to the post-harvest glut because of heavy social pressuresfor ceremonial or marriage expenditure. How-ever, they tend to sell relatively less and, be-cause they are better risks, borrow more thanpoor farmers. During the cultivation season,grain will be sold to raise money to pay wagelabor. Muslim farmers, not using female labor,will produce less than pagan ones, but, notdrinking millet beer, they will also consumeless.

Big farmers are able to withhold grain againstmidseason price rises when they sell in order topay wage labor and to invest in cattle, trade inwhich yields large speculative profits. However,their accumulated wealth is qualified by corre-spondingly larger social obligations that aresocially redistributive.

The same large farmers buy and sell grain.Prices offered vary with the quantity of theindividual lot sold and therefore with the wealthof the seller. The market is highly volatile at themicro level. In Clough's case studies, rates ofreturn on capital over a 2-day transaction byfarmer-traders vary f rom 8 to 27%. Farmer-traders may also store grain in villages againstoff-season price rises, and lend out moneyborrowed in turn from intervillage wholesalers.The latter specialize in trade (though not in thegrain trade); they frequent periodic markets as

267

well as marketless villages, have higher aver-age rates of return than the farmer-trader towhom they lend money, and operate in turn onmoney borrowed from a few urban wholesalerswi th whom they share profits and f rom whomthey clandestinely siphon capital. The urbanwholesalers receive between 73 and 114% oncapital loaned out seasonally to finance therural storage of crops destined for rural finaldestinations (Clough 1977a).

Thus the social relations of trade have forcedmany farmers to sell grain when prices werelow or to go into debt to be repaid at harvest, athigh interest rates. Through this market struc-ture a hierarchical trading system, financed byurban merchants and hereditary notables andtied by patron-client dependency relations, ac-cumulates large f inancial and physicalsurpluses through its control of production(Clough 1977a, 1977b). Yusuf (1975) adds cor-roborative evidence for Kano; Watts (1978;cited with permission) gives a very similaranalysis for a village near Katsina.

Long distance rural-urban or interregionaltrade enables the rural-urban f low of money totake place. These create a system of debt thatperpetuates and reinforces itself, enabling theextraction of a distress surplus on unfavorableterms f rom disadvantaged farmers. Farmersare forced to work as wage labor to obtain foodwhose earlier sale was equally forced on them;their labor input on their own farms is reduced1;and, since agricultural production in northernNigeria is overwhelmingly a function of land,labor, and rainfall, production of both cash cropand subsistence crops is constrained by theaction of the market for "subsistence" crops.Thus the market mechanism that progressivelycommercializes the rural economy simultane-ously reduces returns to farmers and thusconstrains its own progress.

While Clough identifies the system of produc-t ion and distribution over long distances asarticulating inequality, Raynaut's work, which isfocused on a deficit millet-producing butsurplus groundnuts-producing Hausa villagenear Maradi in Niger (1973), shows the samemechanism at work wi thin the intimacy of thevillage. Here, however, it is activated by tax-

1. A reduction in labor input is a it is not the only one.

crucial constraint but

ation, which has been at significantly higherlevels than in Nigeria; although even in Nigeria,the need for cash for paying taxes has beenidentified as a major cause of distress sale(Shenton and Freund 1978). Raynaut in Nigeremphasizes that the cash requirement for taxesalone amounts on average to 65% of the valueof the groundnut crop, forcing the poorestdeciles into the sale of their subsistence crop.Fifty per cent of family heads sold 35% of theirmillet. Raynaut estimates that two-thirds of thegoods traded are for this small-scale rural redis-tr ibution. Little even appears at periodic marketplaces from where it may be exported (4% ofgrain purchases and 17% of sales). Most ishouse or honeycomb trade, not necessarily justbetween houses but sometimes involving non-local foodgrains. This trade is carried out byitinerant traders and by women in the seclusionof their compounds wi th children as go-betweens for price information (see also Hill1972, p. 138). In Hausa, Niger, there are very fewindividuals who do not derive a supplementaryincome from trade (Raynaut 1975, vol 2, p 31).

Large numbers of women are involved incenturies-old marketplace trade in millet andsorghum and their cooked dishes (Nicolas 1962,1965, 1968). There is an intense circulation ofgoods, which pass from hand to hand withoutenriching anyone (Mainet and Nicolas 1964).Raynaut also describes the role of women asaccumulators in the sexual division of labor inexchange. Women produce 37% of total cere-als, sell 10% of all sales, but buy 24% of allpurchases, notably f rom men at harvest, resel-ling prepared food to them before harvest.

Raynaut concludes: "Cereals commerce doesnot indicate the existence of a surplus, nor doesit assure better distribution between membersof the community. Everything points to itsorigin in the economic vulnerability of certainfamily heads and it operates to accentuate theirweakness and dependence, fueled by their in-tense need for money" (1973, pp 34-35, mytranslation).

Money circulates with extreme rapiditythrough the community, but because it origi-nates in the sale of cash crops or subsistencecrops for taxes, most of it is immediatelyrestored to the "coffers of the State". Money istherefore highly scarce (Raynaut 1977, p. 171).Because food has to be repurchased later, theterms of trade for farmers selling distress

288

surpluses are automatically worse than forthose whose surpluses are not for distress(Gore 1978; Post 1970, p 10).

We see once again f rom Raynauts analysishow the action of the marketing system (heretied in with the fiscal system) may constraingrain production. Raynaut feels domestic grainproduction is "seriously distorted." Obligatorytaxes, payment for labor, socially essential giftsand essential clothes, given the declining termsof trade for groundnut and the lack of demand,or alternative infrastructure, for other crops,require a perverse supply response to price,with both production and supply of groundnutsincreasing as producer prices fall. This behavioris corroborated in Niger by Nicolas (1977), innorthern Nigeria by Forrest (draft, 1978), and inMali by Ballan et al. (1977, vol 2, pp 47-48).

For the larger number of farmers thesemechanisms leave no surplus money for in-vestment in new technology and may lead tothe physical overexploitation of the land. In-creasing monetization leads to an expansion ofwage labor (though not of total landlessness).The longer the period of wage labor, the less thepossibility of working one's own land, the lowerthe yield on self-cultivated fields, the greater thedeficit, the greater the need to sell at harvestand buy preharvest, the greater the need towork as wage labor, on seasonal migrations ifnecessary. Greater also is the need to borrowmoney. Loans with interest (often 100% andtaken because the debtor needs seed) are morecommon than are friendly loans without in-terest. The civil administration exists to sup-press such activity but often participates in it(Mainet and Nicolas 1964, p 113; BCEOM 1978,p 37). Since production is intractably con-strained, profits f rom grain marketing comefrom retrogressive activity, such as hoardingover t ime, which reinforces the constrainingmechanism.

Clough, Raynaut, and Hill report 100%seasonal swings in price; Gore 100 to 400% inGhana (1978). As Polly Hill (1972, p 137) says:"Seasonal price swings are not more severe inHausaland than to the south but the economicconsequences are much more far reaching"because of the inability of consumers to substi-tute and their consequent, predictable vulnera-bility.

Watts (1978) for Hausaland, Gore (1978) forGhana, Lallemand (1978, p 49) in Upper Volta,

and Meillassoux (1974) in West Africa generallyassert that interseasonal price fluctuations haveincreased through t ime, as might be expectedfrom a combination of increased seasonal oscil-lation of demand and constrained supply.Comparison of the percentage seasonal vari-ations for 1960-61 in various Nigerian grainmarkets — normally in the range of 25 to40%—wi th those occurring now (bimodaly dis-tributed around 60 and 140%) would supportthis contention (SEDES 1963, p 17-24; CILSS1977, vol 2, Niger p 50), though more research isneeded. There is some evidence that storagestrategies for several years' consumption re-quirements, ensuring the integration of thehousehold through t ime — strategies well-adapted to a risky environment — have de-cayed and that smaller farmers stock less. Onceagain this fits logically with an argument involv-ing the deterioration of the land, the reductionof land down to cereals with an increase in cashcropping, the stagnation of aggregate pro-duction and its decline on smaller farms and theincrease in the relative importance of cash fortaxes and marketed consumption goods(Wilhelm 1976a).

Watts (1978) remarks "there is no inevitabilityabout seasonal food shortages, just as there isno logical, automatic or predetermined relationbetween drought and famine." However, ifthese analyses of the effects of the marketingsystem on agricultural production are correct,then there is an inevitability about the seasonalbehavior of the market; and if evidence isassimilated on adaptive land-use strategiesmade during the drought that lead to a furtherdecline in lands sown to subsistence crops(Faulkingham 1977), then there is now an au-tomatic relationship between drought andfamine. And this understanding of the socialcauses and effects of increasing seasonality hasto be set in the context of possible secularclimatic and environmental deterioration,partly caused by the action of man himself. Inaddition, extreme events such as the Saheliandrought exert an irreversible "ratchet effect"(Chambers 1978), in this case towards the indi-vidual appropriation of land and herds (LeMoigne and Memni 1973; Wilhelm 1976a; Spitz1977; Swift 1978) and the marginalization ofsignificant numbers of people for whom there isno alternative employment as remunerative astheir marginal agriculture had been.

269

But is this logic correct? Polly Hill in her twobooks on rural inequality, which consider trad-ing in detail stresses in her explanation of ruralpoverty (1) environmental factors (short farm-ing seasons and uncertain rainfall), (2) lack ofproduction factors (such as capital and cattlemanure), (3) lack of alternative opportunitiesoutside agriculture, even including long-distance trade by donkey caravan, which hasbeen replaced by the urban-based trade of lorryowners, (4) socially redistributive practices(though Watts argues that these are on thedecline). Hill also argues that labor is not fullyutilized, which is not supported by Norman'sevidence from Zaria and which appears tocontradict the assertions of Clough, Raynaut,and Watts that wage labor is diverted f romself-employment rather than acting as a sup-plement to it. However, it is seasonal underutil i-zation that most concerns Hill, while it is thepeaking of labor demands during the cultivationseason when untimeliness has high oppor-tunity costs that concern the other writers (seeHill 1972, p 190; 1977, p 100). Hill, nevertheless,does expose the "balance of payments"problem of her village which "exports" 2500pounds of groundnuts and would need to im-port foodgrain to the tune of 4000 pounds tonourish everyone adequately. Invisible remit-tances do not redress the balance, and onesection of the population — the poor who sellafter harvest, are forced to work for a l iving, andare most dependent on the market economyeven for grain seed — is likely to be under-nourished.

Hill states that the yields of the poor are lowerthan those of the rich per unit area because theirpoverty prevents them f rom applying manure,let alone modern inputs; she even identifies a category " too poor to f a r m " in a land surpluseconomy, a statement that she regards ascontroversial (Hill 1977, p 162-64). She iden-tifies large farmers, mostly men but somewomen, who deliberately store for speculativeprofits over time — speculative because there issome risk of loss when the prices fall beforeharvest as farmers offload old stock onto themarket, a practice attested by other writers. Butshe does not see the process as self-reinforcing,stressing instead the dissipation of accumu-lated wealth at death. However, that "death is a great leveler" may be an exaggeration. Shesays: "Rich men may have rich fathers yet most

sons of rich fathers wil l not be rich men " (1977,p 100). Hill is not distinguishing convincinglyenough between the extent of inequality andthe extent of transmission of that inequality. I conclude that the differences between Hill'sexposition and those quoted earlier are ones ofqualification rather than being fundamental.

Matlon in his thesis on Hausaland (1977, pp419, 432) argues somewhat differently: thatincome differentials were small and due mainlyto differences in the physical productivities ofland and labor which "reflected at least approxi-mately interpersonal differences in aptitudeand work motivat ion." Palmer-Jones, in anexceedingly thorough critique (draft, 1978),shows conclusively that the analysis of the dataon which these conclusions were made is sta-tistically naive and that, similarly to my owncritique of standard marketing methodologyand conclusions (Harris 1978, Annexe 3 andearlier in the main text), Matlon's conclusionsdo not fol low from either his own or anypossible reinterpretation of the data presentedby him.

Palmer-Jones (1978, p 4) argues that "Non-market relations may mediate access to land,labor, manure, chemical fertilizers etc. in such-a way that poorer farmers cannot farmeff ic ient ly . . . a considerable portion of theirsurplus product and surplus labor t ime is ap-propriated by these non-capitalist relations,making it difficult for them to break out of theirimpoverished posit ions."

Matlon (p 425) categorically denies the ex-istence of the mechanism of usury: "No interestpayments were reported on cash loans repaidby the poorest 40% . . . and . . . credit chargeswere n o t . . . biased against poor households."So does Hays (1975, p 90); "There was noevidence that marketings tied to credit ex-tension were significant for these two grains"(millet and sorghum). So does Ejiga (1977, p 255): "None of the traders in all the samplessaid they advanced any money to farmers. Norhad any taken any loans themselves. Nofarmers had borrowed money or given cow-peas in exchange for debt." Ejiga (p 41), how-ever, deliberately identified himself with theGovernment " in order to improve his rapport."Polly Hill, in her critique of Hays, reminds usinteralia of the exceedingly secretive nature ofstorage and of allied subjects such as debt inHausaland (1976, p. 86). Al l three authors did

270

fieldwork using paid assistance unlikely to beable to retrieve "statistically respectable" dataon debt. Finally, Palmer-Jones agrees, as I do,with Wood (.1978, p 42 on Bangladesh) thatsubjects such as moneylending "are insti-tutionally disguised at the level of social rela-tions in the village, both consciously throughverbal agreement and informal arrangementsbacked by sanctions, and more pervasivelythrough the class based management of thehegemonic egalitarian Muslim kinship ideol-ogy which denounces interest rates and thelike."

Kohler's study (1977) of the traditionalgrain-marketing system for Niger statescategorically that there is little distress salesince peasants can store at little cost on thefarm; he notes almost parenthetically that"farmers get cash f rom agricultural laboring notfrom the sale of grain." He also states thatwhereas seasonal price fluctuations are rarely100%, his data show that 40% of interseasonalprice variations were between 100 and 150%.He also does not mention the role of agriculturalrates, (CILSS 1977, vol 2 Niger, p 50).

The observation of the earlier 1960-61 studyof grain commercialization in Niger by SEDES,that there was even a region where the (thenmuch lower) seasonal grain price swings wereevened out or reversed, is an isolated obser-vation still not inconsistent with a marketingsystem reinforcing inequality in the productionsystem, (1) if the postharvest grain price risewas due to the clogged and glutted groundnutmarket where prices were low, and (2) if theoff-season grain price drops involved the shed-ding of grain in anticipation of a big harvest,which would reduce prices further (see SEDES1963, p 17-18).

Hays' work in northern Nigeria on the struc-ture and performance of the grain trade in a rural-urban distribution system (1975) showsthat larger farmers sell more grain and sell itlater at higher prices than do small farmers. Hiscategories (large and small) are crude ones oflandholding, unramified by household size (Hill1976, p 85-86). As we have seen, Haysconcludes that resource allocation in marketingis "approximately opt imal" — the value judg-ment so characterist ic of convent ional"structure-conduct-performance" analyses;but he does not question the role performed bythe marketing systems in changing the income

distribution. Dough's analysis of the relationbetween overhead costs and net trading profitsshowed that the latter were two to four timesthe former, and he also queries the valuejudgments of Hays and Kohler (1977a, p 22-25).Clough tells us that profits f rom the grain tradeare invested in highly profitable cattle trading,enabling the wholesaler to employ wage laboron his farm and to lend out money at highinterest rates to this labor. The pseudo-scientificanalysis of structure, conduct, and performanceto investigate market efficiency (see Harriss1978 annexe 3, for a major critique) usually failsto incorporate ramifying social and economiclinkages, nor a systematic pursuit of the circuitof money in commerce, which is actually asimportant as the circulation of the commodity.The quality of the data in West Africa (aboutwhose use much bickering has occured) doesjustify skepticism about the meaning of marketperformance results, and the aggregates maskprecisely those micro-level variations that givetraders their profits. A low average rate of returnto trade can still be perfectly consistent withlarge-scale accumulation by a few, in thecommonly occurring combination of subsis-tence and monopoly trade. Finally, the action ofthe market in constraining production and rein-forcing inequality generally takes place at levelsof disaggregation in t ime and space belowthose for which there is statistical information.The analysis of trading profitability over space,without incorporation of t ime lags, ignores thehigher levels of profit achieved through thegreater control over time, form, and place "uti l i-ty" . (See Harriss 1978 Appendix 3 for a detailedelaboration of these points). The cost effective-ness of long-distance trade (rural, urban, orinterregional) if proven, would not imply thatlocal redistribution is similarly efficient, nordoes the the efficiency of periodic marketplacelocal exchange necessarily imply that eitherhouse trade or long distance trade is just asefficient. All these channels have to beexamined, both separately and in their inter-relationships with each other, and their com-plexity cannot be overemphasized.

We have contended that the marketedsurplus of food entering trade is much larger involume that is the net marketable surplus be-cause of postharvest distress sale and pre-harvest buy-back. We have also contended thatat the micro level, the observed variations in the

271

interpretations of the role of the marketingsystem in Hausaland are more likely to beideological (affecting the interests of a group)rather than grounded in substantive differencesin the relation between agrarian and com-mercial systems. What variations occur in therest of semi-arid West Africa?

Polly Hill refers to Hausaland as "socio-economica l l y the g rea t underexp lo redregion of West Afr ica" (1972, p xiv). My particu-lar search in the rest of semi-arid, formerlyFrench West Africa suggests precisely the re-verse.

For Upper Volta, Berg (CILSS 1977, vol 2, U-V,p 24) quotes the fol lowing ministerial note ongrain price stabilization: "The peasant is cur-rently in debt to the local trader who gives him a loan during the preharvest t ime. At harvest t imehe reimburses this trader by selling his crop at a very low price. Sometimes he sells his wholecrop, although he is later obliged to buy part of itf rom the trader at extremely high prices. Thenhe gets into debt again and wil l never be freefrom this vicious cycle." Berg then says: "Theleast that one can say about these assertions isthat they are not based on any systematic studyof rural markets and peasant behavior inUpper Volta. To our knowledge there are nosuch studies in existence." But this is some-thing of an exaggeration, or a confession ofdisciplinary blindness. Most research is on theMossi, the most numerous of the many tribalgroups of Upper Volta.

Wilhelm (1976b) describes heavily-populatedMossi country south of Ouagadougou wheremaize, red sorghum, millet, and beans aregrown for subsistence and groundnut, tobacco,and cotton for cash. Here the harvests no longerassure self-sufficiency in staples. Lallemand(1975, p 44-50) observes precisely the samephenomenon in Yatenga. The smaller theenterprise the larger the importance of purch-ased cereals, the greater the likelihood thatmarket purchases have to be put off until thesoudure and the greater the likelihood of dis-tress sale after harvest.

Broekhuyse's survey (1974) of the Mossiplateau shows that about half of farm revenue isspent on taxes and ceremonies (a too easilyaggregated category comprising nonfoodpurchases) and that approximately the sameproportion is spent on repurchasing millet andsorghum (1974); similar conclusions are

reached in a general study by Bollinger (1974).The "surp lus" in Wilhelm's study area is boughtby merchants f rom Ouagadougou, whose sup-plying dominates the reallocation of surplusand which supplies the countryside f rom urbanstocks during the soudure. This is perfectlyconsistent with Berg's findings that seasonalprice swings are greater in rural areas than inOuagadougou, which he could not explainother than by blaming inaccurate data (CILSS1977, vol 2, U-V, p 55). Ouedraogo corroboratesWilhelm's analysis in her study of the cerealsmerchants of Ouaga (1974): "Groupes au seindu syndicat des marchands de cereales ils necraignent aucune interpellation des autoritesvoltaiques aux yeux desquels ils font miroiter lepoids politique qu'ils representent."

To recover the surpluses yielded up afterharvest, farmers may migrate for work, whichreduces the strength of the work force engagedin agriculture. Much of this migration isseasonal. In Kombassiri 34% of the men bet-ween 15 and 59 are absent in the dry season(Wilhelm 1976b). Based largely on data from theearly sixties, Amin (1974, p 73) estimates anannual f low of seasonal migrants of 120,000from Upper Volta into the cocoa-coffee belt.While this is prompted by cash needs, it mayprevent the adoption of production-enhancing" innovat ions" in the source areas. Gugler(1975, p 197) cites a source which reports thatan effort to introduce cotton production amongthe Mossi failed because it interfered with theexisting pattern of seasonal labor circulation.

This response has become permanent, andUpper Volta, Mali, and Niger are performing therole supplying cheap labor for the farms andindustries of the Ivory Coast and Ghana (Amin1973; Wilcock 1978). As with seasonal circu-lation, these migrants support with remittancesthe cash demands of an increasingly unself-sufficient agriculture whose production may bedominated by women, apparently unable tomanage innovations in tillage because of ag-ricultural extension oriented to males ex-clusively (Gissou 1977). On the migrants' returnto marry, these workers resume their subordi-nate position in the family. They tend to begin inpetty trade because, as Berg notes, there arefew entry restrictions (CILSS 1977, vol 2, U-V,p 25-27). Furthermore, they have no outlet inagricultural production for their accumulatedcapital. In the studies of Mossi migrations this is

272

attested to be because decision-making on landuse is based on seniority, and because thosewith social power do not need money alone toenhance it, and because, most importantly,geographical expansion of territory now facessharply diminishing returns to labor and capital.This reinforces the domination of the marketover production and micro-level social in-equality (Ancey 1975, p 212-13; Ancey 1977;Capron and Kohler 1975, p 39-41). The lattercomment: "Tout se passe actuellement commes'il etait etabli (entre les detenteurs du pouvoireconomique et ceux du pouvoir social) unesorte de consensus tacite destine a geler (pro-visoirement) ov a mettre entre parentheses lamasse monetaire circulante." Ancey concludesthat there is an asymmetrical liaison betweenthe external capitalist system and the internalprecapitalist one that is self-perpetuating. How-ever, Capron and Kohler doubt this equil ibriumand Kohler argues that insofar as this incom-plete spread of the market economy connectedwith demographic expansion leads to the land'sacquiring value and becoming actually negotia-ble (rather than being accumulated indirectlythrough "borrowing") we can expect a reinforc-ing of inequality (Kohler 1968); but the processis slow.

The large number of intermediaries season-ally evident in rural petty commodity marketsappearing to satisfy structural conditions forperfect competition gives us little indication ofthe process underlying the pattern; and it de-tracts attention f rom the accumulation of physi-cal and financial surplus by interregionaltraders where even Berg finds price differencessubstantially in excess of transport costs (CILSS1977, vol 2, U-V, p 25-27). Explaining this withreference to lack of information, accountingskills, and transport detracts f rom the fact of theprofits and the social and economic process ofwhich these profits are a part.

In Mal i , there really are no studies of the roleof the evolving grain-marketing system and ofits effects on production. We agree withPanhuys: "Des etudes specifiques d'an-thropologie economique de terrain seraientsusceptibles d'apporter des resultats d'une por-tee non negligeable pour la politique commer-ciale a mettre enoeuvre" (FAO 1973, p 10). Thisbeing so we have to read between the lines in a rather speculative fashion.

Two studies of grain production point to

inequalities in farm size. In the region of Mopt i ,farms vary from 3 to 20 ha, bunching around 7 ha (Operation Mils-Mopti, 1973), while inSikasso to the South, a more "advanced"region, they vary from 1 to 40 ha, bunchingaround 5 ha (Institut d'Economie Rurale 1978).Of course we cannot conclude a process of ruraldifferentiation from this static evidence, and tosome extent, size of farm reflects size of family.Two further major surveys of agriculturalpractice in the south of Mali show a reduction ofland down to millet and sorghum. In 464 vil-lages covered by the Compagnie Malienne deDeveloppement des Textiles (Ouattara 1977)between 1973-76, land under cereals had de-clined from nearly 80% to 65% along with a "rural exodus," a reduction of labor on subsis-tence crops and a transfer of land to cash crops,notably cotton, the price of which is moreremunerative and for the marketing of whichbetter infrastructural facilities exist. Individual-ization of land is on the increase, as is the hiringof migrant Bembara workers (Lemoigne andMemni 1973, pp 58-67; Operation Mils Mopt i1973, p 12). Ballan et al., surveying 28 villages inthe Bamako, Segou, Sikasso, and Mopt iregions, show that the increase in yield of milletand sorghum from fertilizer residuals in anexperimental millet-cotton rotation is quite in-sufficient to maintain the marketed surplus ofmillet or to counteract the adverse effect onproduction of a reduction in land by an increasein the acreage under cash crops (1977, vol 3, pp32-34). The correlation coefficient between themarketed surplus of millet and the percentageof land under cotton is -0 .97. The relationshipcould hardly be more explicit. And the relativeofficial price of millet against cash crops isdeteriorating against millet (Ballan et al. 1977,vol 1, p 47). Again this does not prove that themechanism at work in Nigeria, Niger, andUpper Volta works in Mali; but given the risingdemand for wheat and rice from urban areasand deficit regions, effective demand for milletand sorghum must be relatively declining.

Obligatory postharvest sales to meet cash re-quirements for taxes are frequently mention-ed as an important source of supply (e.g. Ballanet al. 1977, vol 1, p 45 -46 ; vo l3 , p 34). This teamactually relegates all the redistributive activitywithin villages and via local periodic markets(where according to Bah, women have a do-minant role) to the cadre of "subsistence", not

273

to be considered as trade (Bah, in CILSS 1977,vol 2, Mali p 67). Millet is described as a stabilizer for food security.

Interregional trade, dominantly rural-urban,is fueled from 'supplies at periodic marketspicked up by the agents of urban nonspecialistwholesalers owning transport. Whether or notthey trade in mil let or sorghum does notjeopardize their livelihoods (Ballan et al . 1977,vol 1, p 134). Their trade is characterized bydiversity, precisely in order to minimize risk.Very considerable profits are made by thesetraditional urban wholesalers f rom interna-tional smuggling caravans; for Mali, in spite ofits declining acreage to cereals and low relativeand absolute prices, is a surplus country in a normal year. It is estimated that 50 000 to100 000 tonnes may cross borders in this way inyears when Mali receives food aid, indicatingnot only the likelihood of large profits throughlong-distance trade but the collaboration be-tween these traders and a partisan bureaucracy(CILSS 1977, vol 2, Mali, p 30; FAO 1973, p 14)for: "Lies aux 'aristocraties religieuses' qu'ellessoient animistes ou musulmanes, les commer-cants n'hesiterent pas, non plus, a financer plustard le jeune mouvement nationaliste, semenageant ainsi des alliances utiles dans lafuture equipe dirigeante du pays" (Diop 1971, p 135).

The distinguishing economic features ofthese mercantile groups are their speculativebehavior, collective organization, interregionaland international tradecrossing ecologicalzones, their versatality in dealing wi th cash,credit, barter, etc. The outstanding tradinggroup is that of the Dioualas whose crucialimportance to the Malian economy has had tobe recognized by all Governments. But in theabsence of a modernized nonagriculturalsector, their accumulated reserves and entre-preneurial dynamism are said to be "unproduc-t ively" reinvested in commerce (Diop 1971, pp141-51; Amselle 1969).

We may conclude tentatively that the distin-guishing feature of the cereals marketingsystem in Mali is prices that give a low rate ofreturn to production, al lowing trading groups toprofit at the expense of cultivators f rom whomthese traders are socially defferentiated, unlikeelsewhere. With greater profits to be made f romcash crops, the acreage to millet and sorghumdeclines; this renders not only the rural popu-

lation but also the urban population vulnerableto the power of trader-bureaucrats.

In Senegal, where both towns and coun-tryside are increasingly provisioned by im-ported wheat and some rice, paid for f rom theforeign exchange earnings of a stagnant exporttrade in groundnuts, sorghum and millet arerather little marketed (Yaciuk 1977, p 56). Therole of the goundnut marketing system in ex-tracting financial and physical surplus f rompeasant producers at highly deteriorating,double-factorial terms of trade is well knownand wil l not be recapitulated here (see Amin1973). Suffice it to recognize that to a certainextent the two commodity markets are interre-lated; production and sales of millet dependeither directly on the relative prices of the twocrops (CILSS 1977, vol 2, Senegal, pp 43-50) oron the relative prices of groundnut and rice, thelatter being preferred over millet in urban areasand some rural areas. Also the t iming of cashdemands (not only in tax but also for therepayment of groundnut seed) determines thepattern of postharvest sales.

There are three kinds of millet and sorghumsurplus; first, the sporadic residuals f rom the"disaster plantings" of larger farmers, releasedwhen the fol lowing season's crop can safely bepredicted; second, the continual small suppliesof millet cultivated by women and bartered forpetty comestibles and consumer goods in vil-lages or at local periodic markets (Minvieille1976, p 45; Sar 1973), where the price receivedby women for their surplus may be 75% that ofmen, resulting in and from differentiationwithin the household (SONED 1977, p 8 1 ;Kleene 1974). The Yaciuks (1977, p 43), studyingeight villages, found that 85% of all sales wereof this type, with occasional sales by men ofgrains exchanged for prepared food cooked bywomen. Most sales in rural markets were under10 kg in consignment size, and 40% of transac-tions were by women (Richard 1974). This typeof very small-scale (barter) exchange may resultf rom micro-level regional specialization. In theregion of Matam studied by Minvieille (1976),within a radius of 10 km there is specialization inthe production of (1) fish and irrigated rice, (2)souna millet, (3) sorghum on dry land, and (4)cattle, based on comparative environmentaladvantage. The third type of surplus is thefamiliar distress surplus of smaller cultivators,triply penalized in price by having to sell post-

274

harvest, to sell in small quantities, and to rebuywhen prices are high (Richard 1974), thoughthis activity seems much less important than inthe other countries we have studied (Yaciuks1977, p 43).

As Rocheteau (1970, p 74) says in his study ofthe Musl im mouride sect in West Saloum, thesurvival of families during the off-season fromthe moment when the stores are empty until theharvest and the cash begins to f l o w . . . dependson the sales, loaning, and sales on credit ofprivate traders and big local landlords. "Lecaractere encore tres personnalise de la relationentre paysan et commercant n'est pas, on levoit, exclusif de pratiques commerciales sansconcessions." Rocheteau, Dione (1975) andSONED (1977, p 70-75) in their micro-levelstudies of the cereals economy warn that usuryis a dominant element in such transactions.Seasonal interest rates of 50 to 75% are increas-ing because of the canalization of agriculturalsurplus away from private traders and throughthe cooperatives dominated by marabouts (Rocheteau 1970). The kind of hierarchical creditarrangements attested by Clough in northernNigeria are hinted at in the SONED study ofgrain marketing, as by Rocheteau, though thereare remarkably few large-scale African traders(Diop 1972, p 158). A further unique feature ofSenegal is the diminished importance of inter-regional and long-distance trade. This has beenattributed to tighter state controls on the geog-raphical direction of trade though the state iscertainly compromised by its independence onthe approval of traders in its commercialpolicies (Cruise O'Brien 1975). Wide regionalprice discrepancies and fluctuations occurnonetheless (Dione 1975, p 27; SONED 1977, p 78) giving rise also to short-distance sporadiccross-border smuggling to Gambia, Mauritania,and Guinee Bissau. Haswell (1975, pp 207-217)documents the similar internal economic pro-cess within the Gambia.

It is irresistible to conclude that even with thesparse and patchy information available to us,the Sahelian grain-marketing systems are gen-erally quite closely linked with that for money,and control of both is concentrated. Financialand physical resources are being transferredfrom country to town via rural markets. To-gether the commodity and money marketssimultaneously work to expand commodityproduction and to constrain agricultural pro-

duction by inhibiting savings by the peasantrywho might reinvest it in agriculture. Thestrength of this domination wil l vary regionallydepending on factors such as the productivity ofthe land and the population resource relation-ship, migration patterns, the tenure system, thepolitical strength of trading lineages, householdstorage capacity and practice, and the organiza-tion of markets for competing agricultural pro-duce.

It has been quite impossible to consider theeffects of the marketing system on productionwithout considering the effects of the produc-tion systems on marketing. It has been arguedhere that given declining land productivity overmuch of the Sahel, given increasing populationpressure and a slowly increasing compulsiveinvolvement in the cash economy, rural povertyis increasing, both in an absolute and relativesense. The fiscal demands of the State reinforcethe action of the market economy.

But the State makes other demands.

S t a t e I n t e r v e n t i o n i n M a r k e t i n g

Although it is most likely that the sociallysensible form of State intervention would be touse public resources to regulate the use of,rather than replace, private resources in anotherwise free grain trade, most of the ex-French Sahelian states have intervened on paper on a massive scale. The stated objectiveshave been to supply urban areas and ruralregions most regularly in deficit, to stabilizeproducer prices, to keep consumer prices low,to organize infrastructure and emergencystocks (Wilcock 1978, p 253) and to organize theexport of grain. They have also been justified interms of the uses to which the resources trans-ferred are put in comparison with the pattern ofsavings and investment of private trade.

Similar measures have been used in everyFrancophone state. To summarize they arecharacterized by:

1. Institutional fission and proliferation. Ag-ricultural marketing parastatals are only part ofthis explosion. In Senegal, for instance, 97parastatals have been established since 1973. InNiger, the parastatals administer one-third ofthe formal sector economy with a salariat ofonly 2,000 in a country of 5 mill ion people(BIT/PNUD 1977). Many of these parastatals

275

form parallel roles to, and duplicate, the civiladministration.

2. A tendency to expand regardless of finan-cial viability. The fundamental internal contra-diction is that financial, technological, andmanpower resource requirements for State in-tervention are high whereas surpluses of milletand sorghum are small in quantity and sporadicin t ime and space, as is also a significantcomponent of demand. Parastatals are requiredto pay producers more than the free marketprices, to sell to consumers at less than the freemarket prices, and to trade where private tradeeither cannot or wi l l not go, all this wi th thesocial aim of reducing the cost of marketing(CILSS 1977, vol 2, Niger). The cooperatives arethe only parastatal institutions vulnerable tobankruptcy; the others are maintained on sub-sidies.

3. Chronic instability in structure and infunctioning (CILSS 1977, vol 2, U-V, pp 7, 11). Inmultifunctional parastatal institutions, mil letand sorghum account for insignificant per-centages of total turnover (see Harriss 1978,Historical Annexe 1) and thus are affected bydecision-making on other commodit ies oftenimported and subject to external influences. InNiger there are no policies for the parastatalsector on the fol lowing areas: its role as a promoter or regulator; the economic and ad-ministrative limits of monopolies; price policy;investment policy, especially the relative rolesof local and international capital; taxes anddividends and their distr ibution; financing ofagricultural parastatals wi th the trading profitsof uranium; organizational restructuring; theattitude to employment creation; the autonomyof the administration; training in relation toeducation; and finally the legal status of para-statals (BIT 1978). In Upper Volta there is a notorious series of inconsistencies in the allo-cation of responsibility between public andprivate sectors and no policy that coordinatesthe four sets of parastatal bodies connectedwith grain marketing (CILSS 1977, vol 2, UpperVolta, p 7). In Senegal there is no policy onconsumer prices, on crop subsidies (either theirsize or their duration), or on the taxation ofagricultural crop surpluses (Min. de Devt. duRurale 1978).

4. The key grain marketing parastatal is in-variably out of direct contact wi th both pro-ducers and consumers; its control over re-

sources is weak and uncommensurate wi th itsgreat responsibility; its finances are precariousand dependent on foreign subsidies.

5. Parastatals are in any case an unstablecoalition of multiple unstable financial inter-ests, both private and public, indigenous andforeign capital.

6. Administrat ion is overcentralized andcharacterized by acute discoordination. Pricepolicy and fixed-trading margins are de-termined in various government committees inspecial Caisses de Stabilization des Prix. Thedates delimiting putative monopoly trading bythe State are established politically. Tradingquotas, sacks, and sometimes power may bearranged by the general administration, or theparastatal. Purchases and sales are inter-mittently in the hands of cooperatives, licensedprivate traders, and (sometimes internationallyfinanced) regional or crop-specific productionparastatals. The State marketing boards orga-nize transport. This may involve several means,all organized differently (donkey, camel, boat,railway, truck) and the cooption of the privatesector as we l l as re levant para-statals. Restrictions may be placed on themovement of grain by private individuals onanything but a local scale. The boards alsoorganize storage. At least until recently this wascharacterized by relatively high centralizationand high losses, in comparison with the peasantsector, as wil l be explained in detail later. Whatemergency reserves there were, were not inappropriate locations for deficit regions otherthan the capital cities. The parastatals organizefinance, in one case simultaneously negotiatingeach cash receipt with a multiplicity of banks.Storage and transport have been the targets ofexpansion everywhere. This expansion is usu-ally externally financed by bilateral foreignloans or gifts and thus subject to externalpressures (see CILSS 1977 for the whole region;Ballan et al. 1977, for Mal i ; CRED 1977 for Mal i ;Wilcock 1978 for Upper Volta; and SONED 1977for Senegal). As commented in Mali: "L ' execu-tion de ce travail n'est pas toujours faite avecjo ie " (Operation Mils-Mopti 1973, p 5) and inSenegal: " I I " faut souligner cette confusioninstitut ionnelle" (SONED 1977, p 9).

In v iew of this "confus ion" and given that thedismantl ing of State intervention is extraordi-narily unusual, there really is little point indebating, as is still done (see OECD/FAO 1977),

276

whether private trade, cooperatives, or para-statals are separately or in combination "opt i-ma l " or "most suitable." Nor, given the patent"gap between formulation and implementa-t i on " which "may be not entirely the result of aninability on the part of the State to act" (Collins1974, p 5), is there point in evaluating theseinstitutional phenomena in terms either ofstated objectives or of textbook objectives formarketing intervention "which exist only inGovernment speeches or in the hearts andminds of outside observers" (Collins 1974, p7),or in terms of testing unproven assumptions onwhich current policy rests. This is especiallytrue if there is no coherent policy and if thetesting of assumption gives ambiguous results(e.g. CILSS 1977, vol 2, sections on Upper Voltaand Niger). Here, we explicitly reject theKeynesian interpretation of State interventionas being above and outside civil society and asexisting to abolish crises by economic planning.It is clear from the condensed generalizationabout interventionism presented already thatthe State at worst may cause crises and at besthas multiple, inconsistent, and changing ob-jectives.

The relevant issues for review are the histori-cal evolution of, reasons for, and distributionaleffects of this complex situation. To study thisproperly, it must be recognized that, althoughmost of the best literature confines its assess-ment of intervention in terms of unidirectionalcausality, more is involved than the effects of"Government" (regarded as a monolithic unity)on the " tradi t ional" distribution system. In fact,both sectors may have coexisted for a fair whilein a dialectical relationship, tied together inunity and in contradiction. State interventionitself is a set of relationships. In the case ofagricultural marketing they short-circuit the pri-vate spheres of production, exchange, and dis-tribution with effects now to be analyzed.

T h e E f f e c t s o f S t a t eI n t e r v e n t i o n o n D o m e s t i cG r a i n M a r k e t i n g S y s t e m s

It is difficult to tease out the disaggregatedeffects of the various types of intervention froman inadequate (because differently oriented)literature. We shall assess in the fol lowing orderthe effects of the "monopo ly" and infra-structural interventions, trading restrictions,

and the effects of consumer price subsidies.The first salient feature is that the domestic

grain marketing system has been legislated outof existence. The efficacy of the state monopolyvaries from country to country but, as has beenseen from the preceding sections and as isgenerally acknowledged, it is weak. The un-stable state monopoly dates generally f romabout the time of the Sahelian drought. Untilafter the drought in Niger the parastatal, OPVN,operated in a regulated free market. Since thenit has operated either wi th the cooperatives orby using large-scale licensed private traders, orvillage headmen for collection and retailing(BCEOM 1978, p 11). In Upper Volta the privatesector had responsibility for supplying deficitregions and urban centers until 1970, operatedunder license to the parastatal OFNACER until1974, and was "replaced" by the RegionalDevelopment Organizations from then onwards(CILSS 1977, vol 2, U-V, pp 11-16). In Maliprivate trade still supplies urban centers inproduction zones and undercuts the parastatalOPAM (Ballan et al. 1977, vol 1). In Senegalprivate purchasing on license to the parastatalONCAD was abolished in 1976, but licensedtraders are still allowed to sell ONCAD's milletand sorghum (SONED 1977, p 77). In view of therecentness of the suppression of private trade, itis too early to judge effects.

The purchase "monopoly" is merely a post-harvest campaign, limited in duration and ceas-ing everywhere when State funds dry up forwhatever reason, or when sacks are not availa-ble, and inappropriately lagged so that farmerswith pressing needs for postharvest cash areforced by lack of alternatives onto the parallelmarket. This fact has led traders with money tobuy at low postharvest prices and stock for thesoudure, the hungry season. This is thought tohave exacerbated, rather than stabilized theprocesses resulting from the interlinkage ofcommodity and money markets and increasedthe amplitude of rural price fluctuations(BCEOM 1978 for Niger; CILSS, 1977, vol 2 U-V,p 16 for Upper Volta; Ballan et al., vol 2, pp 40-45for Mali; SONED 1977, pp 82-86 for Senegal).The prices at which small-scale redistribution inthe soudure is effected may be inflated becausecoercion by the parastatals during the cam-paign has creamed off subsistence crops (there-fore increasing off-season demand), and be-cause these commodities have been removed

277

f rom the locality (therefore reducing supplies)(see Ballan et al 1977, vol 2, p 20; Dione 1975, p 27). Traders may prospect for standing crops tomortgage-in (Ouedraogo 1974, p. 25), which isthe first stage in the private acquisition of land.The fact that the State agents of the parastatalswil l only buy round sackloads swells the con-tr ibution of the most vulnerable petty sellers tothe parallel market in producer areas and forcesthe vast majority of consumers, unable to affordwhole sacks, to buy grain privately. It is knownthat bureaucrats wi th access to the parastatalsometimes legally buy lots of whole sacks,break bulk and either distribute small lots on a breakeven basis to subordinates, or at a profit,which is illegal (Ouedraogo 1974, p 25; Wilhelm1976b for Upper Volta; Ballan et al. 1977, vol 1, p 40 for Mali). Direct sales f rom individual pro-ducers to mobile urban consumers in a markedly imperfect market may also occur — flows were estimated at 2200 tonnes in Senegalin 1976 (SONED 1977, p 72). The suppression ofprivate trade during the monopoly may result inthe temporary drying up of petty grain sales inperiodic markets, and as a result, put specialpressure on the incomes of the women forwhom trading is an independent means ofimproving economic status. The whole processof State intervention may replace employmentfor women in private trade by that for men inparastatals (although we would concede thatthe involvement of women in traditional long-distance trade is unusual but, as in the case ofTogo, not unknown). Finally, it is probablyincorrect to conceptualize State and privatetrade as separate activities. Rocheteau (1970)makes the important point that the marketingcooperatives in Senegal are dominated byMouride marabouts and used as a battle groundfor economic power between them and privatebush traders. Similar processes are at work inconsumer cooperatives (SONED 1977, p 74).State intervention thus enlarges the scope forexcess profit-making in private trade.

The intervention of the State in marketinginfrastructure has probably exacerbated thecosts of interegional trading in traditionalgoods relative to export sector goods. Colonialroad and rail systems neglected interterritorialcommunications. Railways have been builtusing four different gauges, and they do notinterconnect. The fact that there are three majorcurrency zones w i th only the franc easily con-

vertible also hinders interregional trade, as documbersome international customs dues andnontariff commercial policies within the macroregion (llori, 1973).

Interregional cereals trade is restricted notonly by the marketing parastatals but also bythe separate civil administration. In Senegal,prefets control trade in lots under 100 kg andGouverneurs consignments under 10 tonnes,and the rest is controlled by the EconomicPolice (Controle Economique). In Nigeria, localemirs historically held the reins of the graintrade whi le the native authorities (the ColonialState) were unable to enforce trade restrictions(Tiffen 1976, pp 40-45, 60; Gilbert 1969, p 211).In Mali and in Niger, permission to trade inter-regionally is in the hands of the prefets and theeconomic police (CILSS 1977, vol 2, p 13; Ballanet al. 1977, vol 2, p 9). Chronic problems ofdiscoordination and discretionary decision-making may play into the hands of large privatetraders (who may also be bureaucrats).

We have the best indications of the effects ofthe parastatals' restrictions on the pattern ofinterregional trade in Mal i , where it has beenhistorically of greatest importance. The majoreffect is to induce spatial compartmentalization.Here the State concentrates its purchasing re-sources in the southern regions with 26% of thepopulation and 4 1 % of the millet and sorghumproduction; and over one-third of its employeesare actually located in Bamako (CRED 1977, p 140; Richard et al. 1975, p 56). Even in suchsurplus-producing areas the private marketingsystem grows in importance with increasingdistances from towns (Ballan et al. 1977, vol. 3,pp 32-38) because of the increase in the costs ofState control and " the price decline to cul-tivators really accentuates the economic dis-advantages of isolated populations" (USAID1976, p 24). Meanwhile, the severely deficitdistrict of Gao (the sixth region on the desertborder proper) may be starved of suppliesbecause provincial governors may refuseOPAM's legitimate trade — refusing to partwi th their local buffers. Or red sorghum (surpluscattle feed from the U.S. imported under food-for-cash agreements) is distr ibuted, for whichconsumers have very low preferences. In anycase, given the fluctuations in production,OPAM cannot predict in advance its regionalrequirements; free interregional trade variedbetween 19 000 and 738 000 tonnes of millet and

278

sorghum between 1967-73 (Richard et al. 1975,p 36). Hardly surprisingly, long-distance andinterregional trade thrives clandestinely, andprofits f rom speculation over space and t imeincrease in security. Given that official prices forgrain in Mali are 32 Fr Maliens/kg whereas 50 FrM equivalent may be obtained in Upper Volta,80 in Senegal and up to 100 in Mauritania, thereis much smuggling across Mali's vast andunpoliced frontiers. CRED gives a very conser-vative estimate of 15 000 to 30 000 tonnes(1977, p 143 footnote), some of which may bebartered for cattle and salt in short supplydomestically (Richard et al. 1975, p 24).

Less is known about the effects of inter-national or long-distance trade in the othercountries. In Senegal, where trading lineageshave never attained the power of those in Mali,interregional grain trade appears to be sup-pressed (SONED 1977, p. 78). In both UpperVolta and Mali, private transporters andtruckers (who may also be general traders) havehad to be enlisted, sometimes coerced to oper-ate at below cost, w i th losses to be made up byhoists on prices in private trade (CILSS 1977, vol2, U-V, p 19; Ballan et al. 1977, vol 1, pp 25-28;Richard et al. 1975, pp38-45) . The accumulatedprofits f rom tolerated illegal grain-trading tendto be invested in transport facilities, urbanproperty, and in the import-export parastatalscreated from the old European merchantcapitalist f irms and sometimes still financed inmajor part by international capital.

In theory, the State subsidies on official con-sumer prices should l imit private trading profitsjust as they increase urban demand and work totransfer the financial surplus potentially ac-cumulable by the parastatal to private consum-ers instead. As Stevens (1978, p 14) explains inhis study of food aid to Upper Volta: "If thequantity of food aid is too small to influencegeneral price levels, this may be the only ef-fect." The same argument applies generally toconsumer subsidies; if not, and if the loweringof the general price level is transferred along theparallel market to producers, then farmers'incomes will also be deleteriously affected.Steven's modest conclusion for Upper Volta isthat "whi le subsidies are Government policy,price levels are not unaffected by marketforces" (1978, p 56). We have already seen thatit is petty consumers who frequent the parallelmarket. And it is not irrelevant to note that the

rich urban bourgeoisie were remarkably pro-tected over the period of scarcity of the drought.In Upper Volta, while the rural/African con-sumer price index, in which grain is weightedheavily, increased from 135 (base 100 = 1964)in 1970 to 221 in 1974; that for urban/Europeanswent up from 116 to 126 (Bollinger 1974, p 89).In Senegal, traders' profits in millet and sor-ghum are affected by State-pricing policies onrice for which millet and sorghum are substi-tutes (Dione, 1975, p 9-12 & 33). Here a third ofmarketed surplus goes to ONCAD (Yaciuks1977, p 43) which appears to exert more effec-tive regulatory controls than elsewhere in theSahel: "La double menace du monopole et duprix officiel garanti semble avoir un effectserieux sur le commerce pr ive" (SONED 1977, p 86), though this literature may refer to years ofgood harvests.

Elsewhere, it is important not to overem-phasize the regulatory effect of the inefficientoperation of the parastatals. Throughout muchof the Sahel they may be of small importance incontrolling the domestic market. Even inSenegal: "Le circuit commercial traditionnel neva pas disparaltre meme s'il devient clandestin"(SONED 1977, p 86). However, there is not muchevidence that State intervention does anythingbut accentuate the power of private traders overproducers.

T h e E f f e c t s o f S t a t eI n t e r v e n t i o n o n G r a i nP r o d u c t i o n

A negative has to be established at the outset.We are not considering the effects of inter-ventionist projects deliberately designed to in-crease production; we are confining ourselvesto marketing intervention. We shall concentrateon the effects of price and infrastructure onproduction, for it is widely contended that the(low) producer price policy is the most im-portant single reason for stagnation in the pro-duction of millet and sorghum (see the Col-loquium, Nacro 1977, vol 2).

It is difficult to determinethe relation betweenofficial prices and production. The Colloq-uium's case study of Upper Volta argued that atthe official price for millet and sorghum, theaverage return on production was between 39and 56% of the price which would pay agricul-tural wage labor at the State-decreed Salaire

279

Minimum Agricole Garanti. In Senegal, therelationship is 55 to 70% depending on theregion (Sivilia/CGPA 1978), and in Niger it is 44to 7 1 % . In Mali, the current official purchaseprice is 90% of the price based on covering thelabor input at the planned min imum dailywage — not necessarily the same level as inUpper Volta, of course (FAO 1973; Ballan et al.1977, vol 1, p 21). The subsidies on consumerprices represent an invisible tax on producers of66% in relation to world market prices minustransport costs f rom the West African coast toBamako (CILSS 1977, vol 2, Mali, p 19). Theproducer gets a third of the world market price.Peasant labor is obviously valued at a social ratefar beneath that of the wage labor market or thatof the States' decrees. Berg interprets theinput-output cost ratios as favorable (CILSS1977, vol 2, U-V, pp. 17, 38) as he does therelation between marketed surplus and relevantcash crops — groundnut and cotton. Sivilia'swork shows that in Senegal millet and sorghumprices should be raised by 25 to 30% to returnthesame net margin per unit area as groundnut(CGPA 1978). Panhuys (FAO 1973) and Ballan etal. observe that the Malian terms of tradebetween official prices of cash crops and mil-lets turned against the latter during the sixtiesand towards millets after the drought. Theyhave not yet returned to the parity of the early1960s, however; and whereas in Upper Voltathe official price for grain was relatively high atthe inception of monopoly, its attractiveness istarnished by the reduction of subsidies on cashinputs (CILSS 1977, vol 2, U-V, pp 16-38),admittedly not of crucial importance to mostpeasants. The contention that official prices arenot low, which appears in the general summaryof the CILSS/Club du Sahel (1977, vol 1, p 13),though qualified as "highly tentative," is alsohighly controversial.

There is some evidence of the depressingeffects of low relative prices on grain produc-tion throughout the region (FAO 1976a, p 41)though production statistics are notoriouslyunreliable (production sometimes being calcu-lated simply as a function of rainfall for whichthere is better datal) and though the treatmentof price by FAO is crude. In Upper Volta, it isfeared that there wil l be a further shift f romgrain to cash crops (CILSS 1977, vol 2, U-V, p 58). In Mal i , the marketings of traditionallysurplus regions are declining; the lack of dif-

280

ferentiation in price by the parastatal betweenmillet and sorghum — a feature of pricing pol-icy throughout the Sahel in contrast to thedetailed schedules for paddy and rice — mayhave led to an increase in sorghum acreage,which has lower labor requirements and thushigher returns (USAID 1976, p 8). Recent shiftsto cash crops are reported (CILSS 1977, vol . 2,Mali, p 34 ; FAO 1973). Further intervention tocommercialize crops other than millet andsorghum in southern Mali has reduced the areaunder millets — and therefore the production ofmi l lets—by more than the residual fertilizerfrom cash crops has raised yields (Ballan et al.1977, vol 3, p 30-36).

However, to hold that official price policy iscausing this stagnation assumes a marketedsurplus significantly responsive to price — anas yet unverified article of faith of the Saheliantechnocracy — and we must surely questionthis. We do not actually know whether themarketed surplus of Sahelian food grains issensitive to price at all. The one case study ofthe price elasticity of grain production forsorghum in the Sudan (Medani 1972, 1975)yielded median short-term elasticities of 0.19and long-term elasticities of 0.3, with distribu-tions skewed to low values — hardly very excit-ing and suggesting a strong insensitivity. Itwould be very surprising to find generally a netsurplus highly responsiveto price, given (1) thepriority assigned by peasants to domestic stor-age (which will mean that this year's physicalsurplus is a function of this year's productionand previous year's stocks); (2) the (declining)practice of overplanting to ensure a guaranteedminimum production in a risky cropping envi-ronment; (3) distress sales before the cam-paign and repurchases after it; (4) possiblydifferent responses to price between the stocksfrom fields cultivated by individuals and f romthose cultivated communally within the familyunit; and (5) the crop rotations necessary in theenvironment. There is no reason to suppose a linear aggregate response either, in view of thecomplex production system. There is no re-search to show whether "present stagnation" isa result of a high response to low official prices(often announced too late in the year toinfluence plantings) or a low response to highparallel market prices, or a high response toparallel market prices that are lower than officialprices. Alternatively, it might be an amalgam of

responses to a parallel market that differ-entiates In price between varieties and qualities,between types of seller (women, smallfarmer/male and large farmer/male), a marketthat changes seasonally in its relations to thestatic official price. Little is known about therelations between official and parallel marketprices, except what has been already reportedon the higher volatil ity of the latter as a result ofthe former, more so in rural areas than in urbanones (see CILSS 1977, vol 2, U-V, p 58) thoughthis will depend on the amount of marketablesurplus entering official channels. In UpperVolta, this is low and market prices are littleaffected (Wilhelm 1976b). In Mali, one sourcefinds OPAM's prices consistently exceedingthose of private trade in surplus regions (Ballanet al. 1977, vol 1, p 20, 34) and thereforesucceeding in lifting producer prices during thepostharvest campaign. Panhuys states the re-verse: that parallel market prices exceed thoseof OPAM by 10 to 100% (FAO 1973, pp 14-15),and Richard and Vandenberg (1975, p 24) statethat private purchase prices can be higher andyet consumer prices be lower than the respec-tive prices of the parastatal, because privatetrade monopolizes the least-cost f lows, leavingOPAM with problems of distributing to thecostly residual areas in addition to high-overhead costs. In Senegal, too, opinion differsas to whether free market producer pricesexceed the official ones (Ross in CILSS 1977, vol2, Senegal, pp 43-50) or the reverse (SONED1977, pp 86, 102). To some extent these rela-tionships must depend on the size of the har-vest. But to sort all this out is an importantresearch priority with considerable policy im-plications.

It is equally reasonable to suggest that non-price factors are important conditioners of ag-ricultural stagnation, that it is the poverty ofphysical and financial infrastructure for grainmarketing — especially the dislocation of trad-ing f rom the limited state production credit — that is crucial. Sincethe marketed surplus in anyone year is difficult to predict, it is hard toorganize finance for trade and repayments forfarmers. Funds may be untimely in all thesecountries. Purchase of cultivators' subsistencerequirements may be coerced. Peasants'transport costs to centralized depots may not bereimbursed. The physical removal of sub-sistence stocks f rom zones of production

exacerbates the vulnerability of farmers to thesoudure (see Wilcock 1978, Wilhelm 1976a andCILSS 1977, vol 2, U-V, pp 43 -44 for UpperVolta; Ballan et al. vol 1, USAID 1976, CRED1977 p 138, and Guggenheim 1978 for Mal i ;SONED 1977 and Financial Times, February1978 for Senegal).

Thus producers may also be starved of capi-tal, not only by the action of private trade butalso by the State. The existence of efficientmarketing infrastructure for cash crops mayinfluence cropping pattern decisions away frommillet and sorghum, irrespective of the relativeprice relationships. Also, the cultivators wil lhave a reduced and untimely quantum ofcapital unlikely to be channelled into innova-tions, of which few are available for millet andsorghum anyway. In Upper Volta, encadrement of cultivators has been hindered by their dis-trust of the Regional Development Boards,which were not adequately equipped to handlea monopoly purchasing system (Wilcock 1978,pp. 197-198; CILSS 1977, vol 2, U-V, p 19), andthese production parastatals are rejecting themercantile role. In Senegal, it is alleged to belack of capital that prevented peasants f romresponding to a 30% rise in the official price ofgrain after the drought (FAO 1976a, p 15). Bothprice and non-price factors clearly work to slowthe formation of a rural market.

T h e E f f e c t s o f G r a i n P r o d u c t i o na n d M a r k e t i n g S y s t e m so n I n t e r v e n t i o n

We cannot find any research at all that hasconsidered this subject or its implications in anexplicit way.

The main effect of production on the nature ofState intervention is that the agricultural systemdoes not generate sufficient resources — eitherfinancial or physical—for intervention to beorganized so as to have bite in relation to itsstated objectives. The impoverished nature ofthe increasing number of producers who are(net) consumers wil l reinforce the very samecheap food policy that is causing their povertyin the first place, since it is not in their intereststo pay out higher prices for food. The firstresponse of the Sahelian peasant is, if he can, to"hoard" (a most value-laden verb) —unto storeso as to enable his dependents to survive a longbut unpredictable production scarcity. This is a

281

strategy highly adapted to a risky and uncertainproduction environment, but it is unsuited tothe consumption needs of a rapidly urbanizingmarket economy. Attempts by the State totransform storage practices (mistaking the dis-tress sales and rural redistribution for a " t rue"surplus) extract excessive resources from thecontrol of the rural sector and increase itsvulnerability. Small wonder that some suchprojects are meeting wi th stubborn resistancefrom peasants. Nonetheless, policies aiming atthe reduction of household-level stocks ap-parently continue to be pursued (USAID 1976).Meanwhile, urban demand increases andsupplies are diverted from the parastatals,which are thus caught in a double bind.

Local trade is proving impossible for the Stateto replicate. It consists of small quanta of"surplus," sporadic in space and t ime, soldwhenever cash is required by both poor andrich, and sold by the rich when prices are high.These supplies are diverted from parastatalswhose min imum capital investments are high,whose technology is imported and capital-intensive, and whose manpower requirementsare large. In turn, to ensure its min imum currentobligation, let alone create a buffer, the para---atal is sometimes forced to order its agents tocoerce peasants into parting wi th grain, as inMali and Upper Volta. Alternatively, it indulges,as in Senegal, in sanctioning increasing importsof rice and wheat, which are changing thestructure of demand in rural as well as urbanareas towards economically superior, thoughnutritionally inferior, food (Yaciuk and Yaciuk1977, p 56). "Les consequences de l'evolutionde la demande interieure tendancielle sontcatastrophiques pour la nat ion" (Ministere duDeVeloppement Rural 1977). But the researchesrequired to reverse this trend: a stable andhigher yielding production technology formillet and sorghum, a product transformationtechnology, but most importantly control overwhat rich urban classes choose to eat, show nosigns of being available.

In Mal i , where the deficit regions are territo-rially vast and sparsely populated, the parallelmarketing system competes effectively wi ththat of the State in surplus districts, leaving thelatter to cope wi th reduced supplies and highercost distribution to deficit regions.

According to FAO (1976a, p 109), "15 to 20%of the marketed surplus is sufficient to control

282

the market," but this depends on the structureand performance of the market. We have seenthat the domestic marketing system, certainly inMali and Upper Volta, is too powerful to becontrolled in this way by a competing paras-tatal. Only in Senegal, where marketed surplusis small and until recently ONCAD marketed a supposedly higher percentage of this surplus,was there any indication of successful control.

Furthermore, the lack of resources wi th whichto monitor borders is an open invitation tosmuggle. It is impossible to implement themonopoly, and the lack of financial resourcesfeeds back continually into an inability to ac-cumulate adequate physical resources.

There is thus a structural crisis in the para-statals. It is expressed most obviously in lack offinancial viability in grain trading, necessitatingbailing out by international aid. In Mali, forinstance, this involves increasing dependenceon external supplies of millet and sorghum.During the drought, some 350 000 tonnes weresupplied, two-thirds as gifts sold at subsidizedprices to consumers but one-third at 191 Frs.M/kg, when the producer pricestood at 32 Frs/kg(Ballan et al. 1977, vol 2, p 2). Gifts as well asloans are continually being given to the paras-tatals by external agencies, which usually de-mand some internal organizational rationaliza-t ion and reform in return. If there is any re-sponse it takes the form of reform by a processof institutional fissions (see IBRD 1977), spawn-ing further parastatals and sharpening the in-ternal contradictions. It would seem that theexpansionist forces exceed the internal capacityto operate effectively.

In Mali and Upper Volta, members of tradingethnies have been closely involved in the opera-t ion of state trading (Diop 1971). They have aninterest in sustaining its inefficiency; fargreaterprivate resources accrue from the clandestineinternational trade that is inevitably provokedby present policies, and bureaucrats may investin trade and transport (Bollinger 1974, p 44).

The production and marketing system thuscreates conditions that help expand a type ofintervention that exacerbates pressures exploit-ing the agricultural sector. With external fundsinjected periodically, the administrators haveno interest in phasing out what they control.Thus this analysis shows that constraints oninstitutional reform are very much narrowerthan has been concluded, for example, f rom the

large study carried out for the Club du Sahel(CILSS 1977), where the dismantling of mostState commercial institutions is set out amongpolicy options.

Conclusions

We have examined the effects upon oneanother of grain production, private grainmarketing systems, and State intervention. Wecannot quantify the interacting role of themarketing system on production, nor do wepretend to deny the importance of relations ofproduction in explaining the stagnation of thesubsistence sector.

It is due to external metropolitan economicpower that the agricultural marketing systems'allocative role has favored export crops, notfood crops. Change in agricultural productiontechnology is still subject to this power and, inthe medium-term future, an appropriate ecolo-gically sound, widely replicable modern grainproduction technology, capable of employingand feeding the growing population, will almostcertainly be absent. It is an unstable coalition ofexternal metropolitan and national economicpower that uses the marketing systems forexport crops to extract resources, and for whichthe general failure of State marketing systemsfor foodgrains to generate similar resources isproblematical. External economic power is thusboth the key to and the barrier to productive andprogressive change in the agrarian economy.

Commerce is not just a constraint on agri-cultural development on a par with absence ofwater or presence of river blindness (as in FAO1976 and many other documents using a com-modity approach to rural development); com-mercial activities are part of a system of particu-lar social relations from which they cannot bedivorced. And the agricultural productionsystems of the Sahel are being impoverished,as the general orders of magnitude of the re-markable statistics assembled by Szczepanikand laccoacci (1975) show (Table 1)

State intervention appears to alter the struc-ture of private trade in favor of monopolies,thereby exacerbating what it intervenes to re-strain. Intervention has apparently failed toquash anything but information on privatetrade, though the evidence reviewed here isin major part supportive of the notion that thesecomposite marketing systems inhibit grain pro-

Table 1. Income per active worker inagriculture as a percentage ofincome per act ive worker innonagriculture.

1 9 6 0 - 6 2

U p p e r Vo l t a

M a l i

N ige r

Senega l

A l l U n d e r D e v e l o p e d

Coun t r i es

France

UK

U S A

Al l A d v a n c e d Coun t r i es

11

8

11

11

28

37

103

61

/ 54

1 9 6 9 - 7 1

10

8

911

27

39

108

7 3

55

duction. It is quite impossible to identify singleconstraints on production, or single interestgroups or classes who benefit from the operationof the composite marketing system, though it isobvious that the poorer peasantry and poor urbanconsumers are the losers.

The latitude for change within the presentsocioeconomic structure appears to be extra-ordinarily limited. It is not at all clear how furtherreformist measures can improve market ef-ficiency. It is clear that any local political man-date to liberalize trade is insufficiently powerful.It is not altogether clear that the liberalization ofprivate trade would necessarily improve cur-rently stagnant grain production since the paral-lel market still functions, albeit under peculiarcircumstances. And evidence suggests that theprivate commodity marketing systems, linkedwith the marketing of money, suppress andexploit as much as or more than they encourageand expand. The liberalization of internationaltrade is politically inconceivable as well assubject to the same caveats.

It is possible that the literature reviewed isexcessively critical (in the tradition of appliedsocial science) and/or written too early on in thecurrent multiplicative phase of State inter-vention. Since 1977 technical delegates of Africancountries have expressed concern to increase"true surpluses," necessitating actions to alterwhat they perceive as the priorities of relativeneglect but what we have suggested is deliber-ate action (see Nacro 1977, vol 2). However, theresources of their economies are very l imited.Fragile domestic budgets (even of the most

283

robust economy) do not balance, so any reallo-cation of resources from internal funds has tobe diverted from an alternative use. Externalfunds are rarely allocated for purposes actuallyarticulated by nationals (Meillassoux 1974).Furthermore, we are not dealing wi th a tabula rasa. There are precious few reformist prece-dents for the dismantling of any institutionalstructures (apart f rom village level cooperativesbuilt up not so much for the staples consideredhere as for the cash crops grown for export).Intervention expands because it createsmonopolistic trading sources of income bothfor bureaucrats and for private traders. In sodoing it creates privileged employment as well.

Finally, the literature treating price policy formillet and sorghum is curiously silent on themost important point: that low producer pricesfor grain must be designed not only to keepurban prices low (for they do not always do thatfor many of the urban poor) but also to ensurestability in, or an increase in, the production ofcash crops. In the absence of a uniquely andwidely appropriate production technology (seeIBRD 1978, pp 45-51), increases in grain pro-duction would reduce acreages to cash cropsthat earn the foreign exchange (largely con-trolled by State institutions) and, in turn, sustainan evolved structure of powerful nonagri-cu l tura l demand for nonagr i cu l tu ra lgoods, mostly imports. The import of wheatand rice, preferred over millet and sorghum byurban populations, in reducing effective de-mand for local crops also works to lowerthe general price level for millet and sorghum,and the food crop/export crop price ratio. Takentogether, these explanations for current pricepolicy represent formidable obstacles for re-formist change.

We have to face the question of whether any"development" involving substantial reversalsof present trends, as would be necessary toaffect the changes in demand in Senegal forinstance (Ministere du Developpement. Rural1977; Min. du Plan 1978) is possible, given whatwe know about the Pandora's box of inter-ventionist measures) even though it might wellbe "necessary". Until peasant producersdecideof their own accord to take initiative in the arenaof political power and economic decision-making (an event that seems unlikely in thetime-span of conventional economic planning)then this puzzle, under the status quo of the

284

Sahelian political economy, appears to be in-tractable.

Acknowledgment

This paper is based on work done by the author whi lea consultant in agricultural marketing to the Inter-national Crops Research Institute for the Semi-AridTropics, Hyderabad, India, which facil itated her ac-cess to source material but whose views this paperdoes not necessarily represent. She is grateful toElliot Berg, John Cameron, Bruce Johnston, Matthiasvon Oppen, J im Ryan, Stephen Sandford, and GavinWil l iams for detailed comments on an earlier version.

References

A M I N , S. 1973. Neo-colonial ism in West Africa.Harmondswor th , UK: Penguin.

A M I N , S. 1974. Introduction. Pages 65-124 in Modernmigrat ions in West Africa. London: Oxford Univer-sity Press.

AMSELLE, J. L. 1969. Rapport de mission surI 'economie marchande du Mal i . Cahier d'EtudesAfricaines 9 (34): 313-317.

ANCEY, G. 1975. Enquete sur les mouvements depopulat ion a partir du pays Mossi. Vol. 3. Mi l ieuxruraux Mossi : aspects economiques. Ouaga-dougou, Upper Volta: Office de la Recherche Scien-t i f ique et Technique Outre-Mer (ORSTOM).

ANCEY, G. 1977. Variat ion Mossi sur le theme: Repro-duction des mil ieux ruraux mis en contact avec lesysteme capitaliste. Pages 1-15 in ORSTOM, Essaissur la reproduct ion de format ions socialesdominees. Paris: Office de la Recherche Scientif iqueet Technique Outre-Mer (ORSTOM).

BALLON, L, LELIEVRE, D., and MANGENOT, PH. 1977.

Etude des structures des prix et de mecanismes dela commercial isat ion des mils et sorghos. Vol . 3.Bamako, Mal i : Institut d 'Economie Rurale (IER).

BCEOM (Bureau Central D'Etudes Pour Les Equipe-ments Outre-Mer). 1978. Projet de routes rurales enRepubl iqueduNiger : Etude de la commercial isat ionet du stockage des produits agricoles. Niamey,Niger: Ministere des Travaux Publiques et Trans-ports et de l 'Urbanisme.

BIT (Bureau International Du Travail). 1978. Statis-t iques financieres des societes d 'economie mixte.Niamey, Niger. (Mimeographed.)

BOLLINGER, D. 1974. Etude des circuits de distr ibut ionde la Haute-Volta. Situation actuelle et projets dereformes. Suresnes, France: IDET/CEGOS; alsoAbid jan, Ivory Coast: U.S. Agency for InternationalDevelopment.

BROEKHUYSE. J. Th. 1974. Developpement du nord desplateaux Mossi. Vol. 4. Resume, situation, besoins,proposi t ions. Amste rdam: Inst i tut Royal desTropiques.

BUREAU INTERNATIONAL DU TRAVAIL/PROGRAMME DES

NATIONS UNIES POUR LE DEVELOPPEMENT. 1977. Projet

de format ion des cadres a la gestion par assistanceaux societes d 'economie mixte et aux Offices duNiger. Resultats du projet et recommandations.Geneva: ILO/UNDP.

CAPRON, J., and KOHLER, J. M. 1975. Enquete sur les

mouvements de populat ions a partir du pays Mossi.Vol. 2. Migrat ions de travail Mossi. Ouagadougou,Upper Volta: Office de la Recherche Scientif ique etTechnique Outre-Mer (ORSTOM).

CENTER FOR RESEARCH ON ECONOMIC DEVELOPMENT.

1977. Le secteur agricole de la Republique du Mal i .Rapport f inal, Dec 1976. 2 vols. Ann Arbor, Mich,USA: University of Michigan.

CHAMBERS, R. 1978. Seasonal dimensions to ruralpoverty. Presented at the Conference on SeasonalDimensions to Rural Poverty. Institute of Develop-ment Studies, University of Sussex, July 1978,Br ighton, UK, (Mimeographed.)

CLOUGH, P. 1977a. The relationship between ruralpoverty and the structure of t rade. Oxford: OxfordUniversity Press. (Mimeographed.)

CLOUGH, P. 1977b. Rural exchange relationships andthe structure of the grain trade. Oxford: OxfordUniversity Press. (Mimeographed.)

COLLINS, J. D. 1974. Government and groundnutmarketing in rural Hausa Niger: The 1930s to the1970s in Magaria. Ph.D. thesis, Johns Hopkins Uni-versity, Balt imore, Md , USA.

COMITE PERMANENT INTER-ETATS DE LUTTE CONTRE LA

SECHERESSE DANS LE SAHEL/CLUB DU S A H E L 1977.

Market ing, price policy and storage of food grains inthe Sahel: a survey. Vol. 1. Synthesis w i th statisticalcompi lat ion and annotated bibl iography. Vol. 2.Country studies. Ann Arbor, Mich, USA: Center forResearch on Economic Development, University ofMichigan; also Washington, DC: U.S. Agency forInternational Development.

COPANS, J. 1975. Secheresses et famines du Sahel. 2 vols. Paris: Maspero.

CRUISE O 'BRIEN, D. B. 1970. The Mourides of Senegal.

Oxford: Oxford University Press.

CRUISE O ' B R I E N , R. 1975. Lebanese entrepreneurs in

Senegal: Economic integrat ion and the polit ics ofprotect ion. Cahier d'Etudes Africaines. 57 (XV-1) :95-115.

DELGADO, C. L. 1977. Economic interaction betweenpeasants and herders in the West Afr ican savannah:A case study f rom Tenkodogo, Upper Volta. USAIDcontract report REDSO/WA 77-104. Ann Arbor,Mich , USA: Center for Research on Economic De-velopment, University of Michigan.

DIONE, J. 1975. Le deficit cerealier du Senegal: situa-t ion et perspectives. Bambey, Senegal: Centre Na-t iona l de Recherche A g r o n o m i q u e , Inst i tutSenagalais de Recherches Agricoles.

DIOP, M. 1971. Histoire des classes sociales dansI'Afrique de I'Ouest: le Mal i . Paris: Maspero.

DIOP, M. 1972. Histoire des classes sociales dansI'Afrique de I'Ouest: le Senegal. Paris: Maspero.

EJIGA, N. O. O. 1977. Economic analyses of storage,distr ibut ion and consumpt ion of cowpeas in north-ern Nigeria. Ph.D. thesis, Cornell University, Ithaca,NY, USA.

FAO (Food and Agriculture Organization of the UN).1973. Rapport au Gouvernement du Mal i , sur leprobleme de la commercial isation des cereales,etabli sur la base des travaux de H. Panhuys. Rome:FAO.

FAO (Food and Agriculture Organization of the UN).1976. Perspectives du developpement agricole a long terme de la Haute-Volta. Rome: FAO.

FAULKINGHAM, R. H. 1977. Ecologic constraints andsubsistence strategies: The impact of drought in a Hausa vil lage: A case study f rom Niger. Pages148-160 in Drought in Africa, eds. D. Dalby, R. J.Harrison Church, and F. Bezzan. Afr ican Environ-ment special report 6. London: International Afr icanInstitute.

FORREST, T. C. 1978. Agricultural policies in Nigeria1910-1978. Zaria, Nigeria: Ahmadu Bello Univer-sity. (Mimeographed.)

GILBERT, E. H., 1969. Marketing of staple food innorthern Nigeria: A study of staple food marketingsystems serving Kano City. Ph.D. thesis, StanfordUniversity, Stanford, Calif, USA. DissertationAbstracts-A 30(8): 3163.

Gissou, J. 1977. Etude sur les besoins des femmesdans les villages de I'A.V.V. et proposit ions d 'unprogramme d' intervention. Ouagadougou, UpperVolta: SAED/Minist6re du Developpement Rural.

GORE, C. 1978. The terms of trade of food producers asa mechanism of rural differentiat ion. Bulletin of theInstitute of Development Studies 9(3): 20-24 .

GUGGENHEIN, H. 1978. Of mil let, mice and men :Traditional and invisible technology solut ions topostharvest losses in Mal i . Pages 109-162 in Wor ldfood pest losses and the environment, ed. D. Pimen-tel. West View, Pa, USA: American Association forAdvancement of Science.

GUGLER, J. 1975. Migrat ing to urban centres of un-employment in tropical Africa. Pages 184-204 inInternational Migrat ion: The New Wor ld and theThird Wor ld , eds. A. A. Richmond and D. Kubat.London: Sage.

HARRISS, B. 1977. Piecemeal planning in rice markets:The effects of partial government intervention on

285

marketing efficiency in a South Indian District.Monographs in development studies 1. Norwich,UK: School of Development Studies, Universi ty ofEast Angl ia.

HARRISS, B. 1978. Cereals surpluses in the Sudano-Sahelian states. 2 vols. ICRISAT Economics Prog-ram report , Patancheru, A. P., India. 330 pp.

HAYS, H. M., Jr. 1975. The marketing and storage offood grains in northern Nigeria. Samaru miscel-laneous paper 50. Zaria, Nigeria: Institute for Ag-ricultural Research, Ahmadu Bello University.

HAYS, H. M., Jr. 1976. Agr icul tural marketing in north-ern Nigeria. Savanna (2):139-148.

HILL, P. 1966. Landlords and brokers. A West Afr icant rad ing system. Cahier d'Etudes Afr icaines 6: 3 4 9 -366.

HILL, P. 1971. Two types of West Afr ican house trade.In Development of indigenous trade and markets inWest Afr ica, ed. C. Meil lassoux. Oxford: OxfordUniversity Press.

HILL, P. 1972. Rural Hausa: A vil lage and sett ing.Cambridge: Cambridge Universi ty Press. 368 pp.

HILL P. 1976a. A plea for the development of indigen-ous methods of gra in storage in the West Afr icansavannah. In Policy and practice in rural develop-ment, eds. G. Hunt ing, H. Bunt ing, and A. Bottral l .London: Croom-Helm. 391 pp.

HILL, P. 1976b. Review of H. M. Hays' The market ingand storage of food grains in northern Nigeria.Savanna 5 (1): 8 5 - 8 6 .

HILL, P. 1977. Rural Hausa: Populat ion, prosperity andpoverty. Cambridge: Cambridge University Press.

HOPKINS, A. G. 1973. An economic history of WestAfrica. London : Longman.

HOSWELL, M. 1975. The nature of poverty: A casehistory of the f irst quarter century after World War II(Gambia). London: Macmi l lan.

HUNTER, J. M. 1966. Seasonal hunger in a part of theWest Afr ican savanna: A survey of body weights inNangodi , N. E. Ghana. In Markets and Market ing inWest Afr ica. Seminar proceedings. Centre of Afr icanStudies, University of Edinburgh, 29-30 Apr 1966,Edinburgh.

IBRD [International Bank for Reconstruction and De-velopment (World Bank)]. 1970. Si tuat ion etperspectives du Senegal, Vol. 1. Washington, DC:IBRD.

IBRD [International Bank for Reconstruction and De-velopment (World Bank)]. 1977. Senegal — T h eParapublic Sector Report, Washington, DC: IBRD,West Afr ican Program Department.

IBRD [International Bank for Reconstruction and De-velopment (World Bank)]. 1978. Wor ld developmentreport. Washington, DC: IBRD.

ILORI, C. O. 1973. Trade Constraints: A barrier toeconomic cooperat ion in West Africa. NigerianJournal of Economic and Social Studies 15(3):405-422 .

INSTITUT D'ECONOMIE RURALE/INSTITUT ROYAL DES REG-

IONS TROPICALES. 1978. Programme de recherche

socio-economique appliquee dans la zone de pro-duct ion, region Sikasso. Bamako, Mal i : Ministers duDeveloppement.

KLEENE, P. 1976. Not ion d'exploi tat ion agricole etmodernisat ion en mi l ieu Wo lo fSa loum. AgronomieTropicale 31(1): 6 3 - 8 2 .

KOHLER, J. M. 1968. Activites agricoles et t ransforma-t ions socio-economiques dans une region de I'ouestdu Mossi . Ouagadougou, Upper Volta: ORSTOM(Office de la Recherche Scient i f ique et TechniqueOutre-Mer) Mission.

LALLEMANO, S. 1975. La secheresse dans un vi l lageMossi de Haute-Volta. Pages 4 4 - 6 1 in Secheresseset famines du Sahel, ed. J. Copans. Paris: Maspero.

LE M O I G N E , L., and M E M N I , P. 1973. Etude de revo lu -

t ion de I 'emploi et des effets des facteurs de produc-t ion mis en place pendant les dix dernieres anneesdans I 'Operation Coton, sud Mal i . Paris: Secretariatd'Etat aux Affaires Etrangeres.

M A I N E T , G., and NICOLAS, G. 1964. La vallee du Gulbi

de Maradi : enquete socio-economique. Documentsdes Etudes Nigeriennes 16. Niamey, Niger: InstitutFondamental d 'Afr ique Noire/Centre National de laRecherche Scienti f ique.

M A T L O N , P. 1977. Income size and rural differentia-t ion : Some case examples f r om N. Nigeria. Ph.D.thesis, Department of Agr icul tural Economics, Cor-nell Universi ty, Ithaca, NY, USA.

M E D A N I , A. I. 1972. The supply responses to price ofAfr ican farmers at various stages of development.Oxford Agrar ian Studies 1(1):57-66.

M E D A N I , A. I. 1975. Elasticity of the marketable surplusof subsistence crop at various stages of develop-ment. Economic Development and Cultural Change23(3):421-429.

MEILLASSOUX, C. 1971. The development of indigen-ous trade and markets in West Afr ica. Oxford:Oxford University Press.

MEILLASSOUX, C. 1974. Development or exploi tat ion:Is the Sahel famine good business? Review ofAfr ican Political Economy 1 : 27-33.

MINISTERE DE LA COOPERATION/INSTITUT NATIONAL DE LA

STATISTIQUE ET DES ETUDES ECONOMIQUES/SOCIETE

D'ETUDES ET DE DEVELOPPEMENT ECONOMIQUE ET

SOCIAL/FOOD AND AGRICULTURE ORGANIZATION OF THE

UN. 1963. Etude de la consommat ion al imentaire duNiger. Paris.

MINISTERE DU DEVELOPPEMENT RURAL, DIRECTION DES

ETUDES DES METHODES ET DU PLAN. 1977. Act ions

286

planifiees de product ion cerealiere au Senegal1977-1985. Dakar, Senegal.

MINISTERE DU DEVELOPPEMENT RURAL, SECTION DE

STATISTIQUES AGRICOLES, CELLULE DE PLANIFICATION

RURALE 1976. Annuaire des statistiques agricoles1970. Ouagadougou, Upper Volta.

MINISTERE DU PLAN. 1977. Cinquieme plan quadriennalde developpement economique et social. Dakar,Senegal: Nouvelles Editions Africaines.

MINISTERE DU PLAN. 1978. Senegal: Politiques cere-alieres nationales. Dakar, Senegal. (Mimeog-raphed.)

MINVIEILLE, J. P. 1976. Migrat ions et economies vi l-lageoises dans la vallee du Senegal. Dakar, Senegal:Office de la Recherche Scientif ique et TechniqueOutre-Mer (ORSTOM).

NACRO, A. M. (ed.). 1977. Colloque sur le Developpe-ment de la Production et de la Commercial isationdes Cereales dans la Communaute. Vol. 1. Docu-ment introductif. Vol. 2. Rapport f inal. Bamako,Mal i : Bureau Communautaire de DeveloppementAgricole.

NICOLAS, G. 1962. Etude de marches en pays Hausa(Republ ique du Niger) . Documents ethnog-raphiques. Niamey, Niger: Centre de Documenta-t ion , Ministere du Plan.

NICOLAS, G. 1965. Budgets collectifs de groupementsdomestiques Hausa en Republique du Niger. (Can-ton de Kantche, Region de Maradi). Enquete.(Mimeographed.)

NICOLAS, G. 1968. Un systeme numerique sym-bol ique: le 4, Ie 3, et le 7 dans la cosmologie d'unesociete Hausa (ville de Maradi). Cahiers d'EtudesAfricaines 32(8) : 566-616.

NICOLAS, G. 1977. Remarques sur divers facteurssocio-economiques de la famine au sein d'unesociete subsaharienne. Pages 159-168 in Drought inAfr ica, eds. S. D. Dalby, R.J. Harrison Church, and F.Bezzan. Afr ican Environment special report 6. Lon-don : International African Institute.

O R G A N I Z A T I O N FOR E C O N O M I C COOPERATION A N D

DEVELOPMENT/FOOD AND AGRICULTURE ORGANIZATION

OF THE UN. 1977. Critical issues on food grainsmarketing systems for small farmers. Paris: OECD;also Rome: FAO.

OPERATION, MILS-MOPTI. 1973. Reunion mensuelle de

I 'Operation Mi ls-Mopt i tenue a Kore le 10 aoot.Bamako, Mal i : Ministere de la Product ion, Directionde I 'Agriculture.

OUATTARA, K. 1977. Rapport sur I'enquete agricole1973-1974, 1974-1975, 1975-1976. Bamako, Mal i :Compagnie Malienne de Developpement des Tex-tiles (CMDT).

OUEDRAOGO, M. M. 1974. Origine des cereales con-sumees a Ouagadougou et problemes de commer-

c ialisation. Pages 16-27 in Notes et documentsvoltaiques. Oagadougou, Upper Volta: Centre Vol-taique de Recherches Scientifiques (CVRS).

PALMER-JONES, R. W. 1978. Peasant differentiation inrural Hausaland, Zaria, Nigeria: Institute for Agricul-tural Research, Ahmadu Bello University. (Mimeo-graphed.)

POST, K. 1970. On "peasant isat ion" and rural classdifferentiation in western Africa. Institute of SocialStudies occasional paper. The Hague, Netherlands.

RAYNAUT, C. 1973. La circulation marchande descereales et les mecan i smes d ' i nega l i t eeconomique: le cas d'une communaute vil lageoiseHausa. Cahiers du Centre d'Etudes et de RecherchesEthnologiques 2. Bordeaux, France: Universite deBordeaux.

RAYNAUT, C. 1975. Le cas de la region de Maradi(Niger). Pages 5 -43 in Secheresses et famines duSahel, ed ; J. Copans, Vol. 2. Paris: Maspero.

RAYNAUT, C. 1977. Circulation monetaire et evolut iondes structures socio-economiques chez les Haous-sas du Niger. Africa 47(2) : 160-171.

RICHARD, J. F. 1974. Unites Experimentales du SineSaloum: commercialisation des cereales 1973-74.Bambey, Senegal: Institut de Recherches Ag-ronomiques Tropicales et des Cultures Vivrieres/Centre National de Recherche Agronomique.

RICHARD, P. L., and VAN DER BERG, X. 1975. Mission de

restructuration de I'Office de Produits Agricoles duMali (OPAM). Paris: Bureau de Developpement desProduits Agricoles/Ministere de la Cooperation.

ROCHETEAU, G. 1970. Pionniers mourides du Senegal:changement technique et transformation d'uneeconomie paysanne. Dakar, Senegal: Office de laRecherche Scientifique et Technique Outre-Mer(ORSTOM).

SAR, M. Louga et sa region (Senegal). Dakar, Senegal:Institut Fondamental d'Afr ique Noire (IFAN).

SEDES (Societe D'etudes et de DeveloppementEconomique et Social). 1963. Les produits vivriersdu Niger: production et commercial isation. Paris:SEDES.

SHENTON, R. W., and FREUND, W. M. 1978. The incorpo-

ration of northern Nigeria into the wor ld capitalisteconomy: A study in the development and trans-format ion of an African social format ion. Presentedat the 23rd Congress of the Historical Society ofNigeria. Ahmadu Bello University, 4 - 8 Apr 1978,Zaria, Nigeria.

SIVILIA, M. 1978. Etude des prix agricoles. Dakar,Senegal: CGPA, Minstere du Plan.

SONED, 1977. Etude sur la commercial isation et lestockage des cereales du Senegal. Dakar, Senegal:Ministere du Developpement Rural et de I'Hyd-raulique.

287

SPITZ, P. 1977. Silent violence: Famine and inequality.United Nations Research Institute for Social De-velopment 78/C/7 GE 78-4387, for the UN Educa-t ional , Scientific and Cultural Organization (UN-ESCO). (Mimeographed.)

STEVENS, C. 1978. Food aid in Upper Volta. OverseasDevelopment Institute working paper 5. London.

SWIFT, J. 1978. Seasonal aspects of pastoral ism in theSahel. Presented at the Conference on SeasonalDimensions to Rural Poverty. Institute of Develop-ment Studies, University of Sussex, Br ighton, UK.(Mimeographed.)

SZCZEPANIK, G . F., and LACCOACCI, P. L. 1975. Inegalites

des revenus entre l 'agriculture et les autres activiteseconomiques a differents niveaux de developpe-ment. Rome: Food and Agr icul ture Organization(FAO) of the UN.

TIFFEN, M. 1976. The enterprising peasant: Economicdevelopment in Gombe Emirate, NortheasternState, Nigeria, 1900-1968. London: Her Majesty'sStationery Office.

USAID (U.S. Agency for International Development).1976. Draft of grain storage P I D . SenegaI. (Mimeog-raphed.)

W A T T S , M. 1978. The sociology and polit ical economyof seasonal food shortage: Some thoughts onHausaland. Presented at the Conference on Sea-sonal Dimensions to Rural Poverty. Institute ofDevelopment Studies, Universi ty of Sussex,Br ighton, UK.

WILCOCK, D. A. 1978. The polit icel economy of grainmarketing and storage in the Sahel. Afr ican RuralEconomy Program work ing paper 24. East Lansing,Mich, USA: Michigan State University.

W I L H E L M , L. 1976a. Food crisis in the Sahel: A sum-mary. Pages 85-100 in Famine risk and famineprevention in the modern wor ld . UNRISD/76/C. 19,GE 77 -1621 . Geneva: United Nations ResearchInstitute for Social Development.

WILHELM, L. 1976b. Rapport socio-economique de laregion de Kombassir i : Nam Yui et Doulougou.Enquete nutrit ionnelle INSERM-GRAINS. Ouaga-dougou , Upper Volta: Centre Voltaique de laRecherche Scientif ique.

W O O D , G. 1978. Class format ion and "an ted i luv ian"capital in Bangladesh. Bullet in of the Institute forDevelopment Studies 9(3) :39-43.

YACIUK, G., and YACIUK, A. D. 1977. Discussion des

resultats de I'enquete sur la technologie post-recorte en mi l ieu paysan au Senegal. Bambey,Senegal: Centre National de Recherche Ag-ronomique.

YUSUF, B. 1975. Capital format ion and managementamong the Mus l im Hausa traders of Kano, Nigeria.Africa 15:167-182.

288

Discussant 's Comments

Ell iot Berg*

Dr. Harriss has been given a difficult task — tosurvey a diffuse and disparate literature, muchof it consisting of fugitive materials (reports ofagencies, government documents, etc.) cover-ing many different countries and many complexissues. She has clearly done an extraordinaryjob of synthesis and analysis. This is evident inher paper, and even more evident in the reportsshe has written for ICRISAT.

It is a difficult paper to critique. It is full ofcritical observations, fresh ideas, assertions, setout in rapid order and in great variety. Thismakes for stimulating reading and new in-sights. But it also results in frequent obscuritiesand occasional nonsequiturs and irrelevancies.We are told, for example, that in Senegal: "Therole of the groundnut marketing system inextracting financial and physical surplus frompeasant producers at highly deterioratingdouble factoral terms of trade is well known andwil l not be recapitulated here." The State can,by paying producers less than world prices,"extract a financial surplus from groundnutgrowers; but whether the double factoral termsof trade are rising or falling has nothing to dowith i t ."

A few pages earlier we are told that inMali " . . . very considerable profits are madeb y . . . traditional urban wholesalers f rom inter-national smuggling caravans, for Mali, in spiteof its declining acreage to cereals and lowrelative and absolute prices, is a surplus countryin a normal year." But whether Mali is a surpluscountry or not is irrelevant; smuggling occurswhenever Mali's prices are lower than those ofits neighbors, and this would be true whether ornot the country is in "surplus" or "deficit".

The paper has other statements of this kind,deriving perhaps from its broad scope, thesweep wi th which it is wri t ten, and the author'sattempt to compress a great body of material

* Center for Research on Economic Development,The University of Mich igan, Ann Arbor, Michigan,USA.

into a relatively short paper. But rather thanpursue this further, it is more interesting toconsider the general issues, or at least some ofthem, raised by Dr. Harriss.

First, a word about general methodology andthe matter of values and biases in social scienceresearch. Dr. Harriss has elsewhere aimedsome sharp and cogent criticisms at "conven-t ional" economic analysis in marketing. Shehas also quite properly observed that none of usis value free — in particular in what we chooseto study and how we study it. It is perhaps notimproper to look at her paper from this per-spective.

With regard to choice of problems to bestudied, one point stands out clearly: the impactof the marketing system on income distribution.Of the two functions of marketing systemswhich she mentions — the "allocative" and the"extractive" — she is mainly concerned withthe "extractive." If any central theme runsthrough the paper it is that marketing systemsin the West African SAT reinforce inequality inthe production system and in the distribution ofincome.

Now, the interaction between marketing andincome distribution, and the "extracting" as-pect of marketing arrangements, are certainlyinteresting and important issues. But amongthe range of problems and issues in the market-ing area, does it really deserve so central a place? The impact of the functional statemonopolies on equity and production presentsissues of perhaps more importance to peasantsand to policy makers. The disorganization ofgrain markets, the existence of coercion; thediscouragement of traders' skill development;the use of productionist development agencies(ORDs in Upper Volta, Operations in Mali, forexample) to do primary grain marketing withpotentially devastating effects on their mainobjectives; problems of price policy; issuesrelated to on-farm storage — all are clearly highpriority matters in West African marketing.

Dr. Harriss discusses many of these issues,

289

and in her research priorities paper as wel l . * Buther emphasis is on distributional questions.Midstream in the paper she rejects considera-t ion of options for public policy and insteadaffirms " the relevant issues for review are thehistorical evolution of, reasons for, and distri-bution effects of this complex situation."

if West Africa's SAT had a merchant class ofsizable proportions, or if there was f irm evi-dence of widespread monopoly profits or ofsignificant shares of income to grain traders,this priority would be understandable. Butdespite Dr. Harriss' valiant efforts to show thepossibility of accumulation of surplus on a bigscale, most of the evidence runs the other way.In highly monetized regions of Senegal(Kaolack and Diourbel), a recent study1 indi-cates that the six biggest traders in Kaolack buyon an average 500 to 1000 tonnes of grainannually; it is believed that in Diourbel only fivetraders buy morethan 500 tonnes a year. Casualempiricism, visits to grain traders in theSahelian cities, as well as virtually all studies ofgrain trading, indicate a consistent pattern ofsmall inventory holdings and relatively small-scale operations. It is certainly hard to see a compradot class i n the making, at least for somet ime. Given the relat ively burdensomeproblems that exist in the allocative side, the"extract ive" focus in Dr. Harriss' paper seemsdisproportionate.

Let's turn briefly to Dr. Harriss' principalconclusion about the nature of the privatemarketing system and how she reaches thatconclusion. It's often said that to believe is tosee. Dr. Harriss believes that immediate post-harvest (distress) sales are common, and thatrural indebtedness, incurred at usurious rates,is widespread. Shesees confirmation of this in a number of studies. She stresses particularly thefindings in an unpublished work by Clough,though she marshals other evidence as well .She concludes (P. 5) that " the social relations oftrade thus force most farmers (emphasis mine)to sell grain when prices are low or to go into

* Editors' note: This paper, entit led "Relevant andFeasible Research", is in the document by Harriss(1978) cited in her paper in th is Proceedings.

1. SONED, Etude Sur la Commercial isat ion et leStockage des Cereales au Senegal, 1977.

290

debt to be repaid at harvest at high interestr a t e s . . . "

In reaching this conclusion, she rejects orignores contrary findings in the literature. Thusthe three most intensive studies in this field (byHays, Ejiga, and Matlon — all done in Hausaareas of Northern Nigeria) found little in the wayof distress sales and relatively little indebted-ness, especially to traders. These findings areconfirmed by the SONED study in Senegal andare suggested in the CEGOS (Ballan) study inMali. But Dr. Harriss rejects the conclusions ofthe village-level survey type studies on thegrounds that farmers are secretive about in-debtedness, so the findings on debt incidence aresuspect. This is to impute uncommon careless-ness or thoughtlessness to the researchers inquestion, each of whom spent many months indose contact with the villagers. Incidentally, itis wor th noting that while Polly Hill, in herreview of Hay's book, does note the secretivenature of debt information, she also records herenthusiastic agreement wi th the generalfindings about the competitiveness of the graintrade and presumably about the apparentlysmall amount of distress selling. In any case,why should findings that show high levels ofindebtedness be accepted while those indicat-ing the contrary are rejected?

From a policy point of view, the conclusionsof Dr. Harriss' paper are gr im. The privatetrading system is exploitative. The State tradingsystem is ineffective, inequitable, and dis-ruptive; intervention makes things worse, notbetter. Moreover, there is not much hope forameliorative change, for reform. The bureauc-rats benefit f rom it. So do the internationalagencies that support the grain parastatals.

Dr. Harriss' pessimism is not without somefoundation, but it is excessive. Among otherelements, the international donor community isseeking reform. Dr. Harriss implies that theseagencies created the parastatals in pursuit of a rational goal: development of local institutionsuseful to donor purposes. But this is only partlyso. External policy was conditioned partly bythe drought and partly by bureaucratic inexora-bility: the acceptance by local governments ofpoor programs.

Dr. Harriss also underestimates, I think, theideological factor — the fact that intellectualsand civil servants believe that traders are es-sentially unproductive and exploitative and

have therefore eagerly pursued interventionisttrading policies.

It is worth recalling also that the presentconfused state of grain marketing in West Africadates only from the middle or late 1960s in mostof the region. Private traders did the job before,and in more developed countries further south(Ghana and Nigeria) there existed until veryrecently a highly complex and apparentlyefficient food trading network. The unsatis-factory performance of the present system maylead to a will ingness to consider new options.

A final word on research priorities. What weneed is to know more about the structure andfunctioning of grain markets. More specifically,we need a large set of in-depth village-levelstudies of crop disposal. This would involveclosely linking marketing wi th production andlabor use studies and would basically start atthe harvest. The questions to be asked are: howmuch grain is stored, sold, given away, bought?Who buys and who sells, when, whereto whomand at what price? The link to the firstmarket — whether house trade, local periodicmarket, or other — should be explored in depth.There is also need for more standard types ofmarketing studies, fol lowing the f lows of grainthrough the distribution channels f rom produc-ers or local markets to major consuming cen-ters. The important point is that the basicstructure of these markets is so poorly knownthat studies of this general type would seem tohave first priority. Whether this is a priority inaccord with ICRISAT's comparative advantageis another question.

( E d i t o r s ' note: C o n s i d e r a b l e d i s c u s -s i o n f o l l o w e d B e r g ' s c o m m e n t s , a n dm u c h o f i t w a s c r i t i c a l o f H a r r i s s 'p a p e r . T h e s u b s t a n t i v e i s s u e s a r es u m m a r i z e d h e r e , a l o n g w i t h H a r r i s s 'r e s p o n s e s . )

B. F. Johnston expressed his uneasinessabout Harriss' criticism of standard "structure,conduct, performance" marketing research ofthe type done in the many West African studiesshe reviewed. He and others were especiallyconcerned about her assertions that many re-searchers had drawn wrong inferences fromtheir data on marketing in West Africa.Johnston also doubted whether acting on mar-keting systems in West Africa is an effective way

to reduce inequalities. He sees emphasis ontechnologies aimed at progressive moderniza-tion of the great majority of small-scale farmunits as the most important priority.

Harriss responded that a more detailedcritique of the "structure, conduct, perfor-mance" marketing research in West Africa isbeing published in a separate monograph ("Themarketing of Food Grains in the Sudano-Sahelian states: an Interpretative Review ofLiterature"). She doubts if studies concludingthat there are no debt or moneylending rela-tionships can be based on reliable data. Harrissbelieves that, historically, changes in produc-tion systems tend to change marketing systemsalso, rather than the reverse. However, at cer-tain stages of development, commodity market-ing systems linked with money markets maydominate production systems, often for longperiods. She agreed with Johnston in welcom-ing emphasis by ICRISAT on devisingtechnologies that assume the perpetuation ofthe peasantry rather than its elimination. At thesame time she believes marketing research tobe equally important.

Regarding Berg's comment that her paperoveremphasized the extractive role of marketsin West Africa and distributional issues, Harrissresponded that the history of market policyinterventions in West Africa is such that majorregressive effects have resulted. Furthermore,contrary to Berg's view, there is considerableevidence (cited in the Harriss monographmentioned above) of sizable merchant or trad-ing "ethnies" or groups, and trading is almostalways an income supplement for ruralfamilies. There is also evidence of considerablemonopoly profits from grain trading. These arefurther reasons why emphasis in Harriss' paperwas on distributional issues in grain marketingrather than on allocative issues, which havealready been studied (although, according toHarriss, with major flaws in both data andanalysis).

Harriss concurred with Berg that the multi-plicity of government agencies involved in cerealmarketing has led to confusion and inefficiency.Berg concludes that in 1965, before the spate ofnational and international intervention, grainflowed satisfactorily; he said that it would do soonce again if these interventions were dis-mantled. Harriss stressed that her research wasaimed at understanding why interventionism

291

had grown and to identify the forces constrain-ing liberalization. Berg believes rational organi-zation of trade is possible. Harriss believes thatbecause of the histories of the Saheliancountries, it cannot be assumed that rationalorganization of trade can occur there.

292

Chai rman's Summary

J. C. Davies*

It was a pleasure to chair this session when twothought-provoking and interesting papers wereread on "Farm Level Studies in SAT WestAfr ica" by Norman and "Going Against theGrain" by Harriss. It would be invidious for meto attempt, in the few minutes available, tosummarize these since they are already sum-maries of detailed and comprehensive consul-tancies carried out by the authors. All I can do,as a Daniel in a den of economists, is tohopefully pick out certain areas where it ap-peared there was some agreement or disag-reement in the stimulating and lively discussionwhich frequently — owning to the t ime factor — had to be curtailed, but which I am sure willcontinue over the next few months whileICRISAT's plans for economics research in WestAfrica develop. I must add a disclaimer in theapproved style that the statements and conclu-sions I draw here are not necessarily thosewhich you as a body of economists drew — theyare those of an interested observer with possi-bly an axe to grind.

The fol lowing points to my mind emerged:

1. There was some misunderstanding aboutthe ICRISAT proposals for production and mar-keting research in West Africa, in that severalcommentators stressed that they were verybroad-based, rather ambitious, and that theywould require focus. I believe that the intentionwas to use the Workshop as a mechanism to dojust this and, to an extent, the discussionssucceeded. That is not to say that the discus-sions to date do not need further considerationand digestion.

2. There was unanimity that in the produc-tion and marketing research field ICRISATshould be concerned wi th development of suit-able methodologies for conducting economicresearch and obtaining base data for SATAfrica. There was clearly a need to modify andbuild on the experience gained in the Indiancontext. In particular, some of the methodologyused in the vil lagestudies in India would requiremodification for use in Africa. Care will have to

* ICRISAT, Patancheru PO 502 324.

be taken to integrate fully with national prog-rams and institutions at the outset and toinclude a training element in projects, in view ofthe acknowledged shortage of personnel inmany of the countries of interest in West Africa.

There was some disagreement on theamount of base information already available inSAT West Africa, and a feeling that in severalcountries a great deal of information would beaccessible to ICRISAT, as an international or-ganization, if ICRISAT was interested. In gen-eral, it was felt by ICRISAT staff that this infor-mation was best obtained and summarized bynational staff. The subject of price levels overt ime for cereal crops was specified as oneexample where data were apparently availablein such countries as Nigeria. It was agreed thatsuch data were crucial, that price cycles wereimportant but difficult to follow, but that theydid have an effect on production economics.ICRISAT could contribute in collection of basedata and play a valuable role in assessment ofavailable information since independence fromlocal systems could be an advantage.

3. On the production technology side, sev-eral areas were mentioned as likely target areasfor research, including research on the effects ofintroduction of animal power and theeconomics of doing this. It was particularlynoted that issues other than straight-forwardeconomics had to be taken into account inassessment projects. The impact of improvedtechnology, introduction of new cultivars, andpossibly hybrids, on production of cerealswould have to be quantified and closely moni-tored since the effects on the rural poor mightconceivably be double-edged, particularly inview of the lacunae in our knowledge aboutactual disposal and movement of cerealsurpluses. The question of the impact of im-proved technology on soil fertility, and as-sociated with this, the benefits of deep plowing,rotational systems, and mixed cropping, werefruitful areas for project development ineconomics. Associated with rotational systemsis the perplexing problem of integration ofanimals into the pattern of ICRISAT agricultural

293

research in the SAT. Animals are an integralpart of the system in the soil ferti l ity andeconomic sense, the importance of whichICRISAT has acknowledged, but for a host ofreasons chosen to ignore.

4. The dangers of ignoring the dose interre-lationships between the West African SAT areasand the humid tropics to the south of them werenoted. The economies of these countries werelinked, sometimes subtly but almost always.Possibly studies of production could be linkedto studies of intercountry movements of cere-als. There appears to be litt le definite informa-tion on the effects of fiscal and policy decisionson movement and production of cereals bet-ween West African nation states, and in thecourse of discussion different speakers placeddifferent interpretations on the importance ofGovernment marketing policies on production.Government statistics, for instance, often con-veniently ignored the flourishing illegal trade.There was a clear indication that ICRISATshould study the nature and scope of privatemarketing systems since again there were areasof disagreement on how exploitive they were,particularly in the historical context. It wasacknowledged that recently there had be'enconsiderable changes in private trade. A studyof grain markets and f lows was a high priority,particularly the relationship between the pri-vate and parastatal marketing systems. This,irrespective of one's stance on the use, abuse,efficiency or inefficiency of the latter was a greyarea, particularly in view of the disruptions insupplies of food grains of recent years in WestAfrica. It was felt by many that at present there isno detailed, unbiased, study of grain f lows fromfarm to final point of sale in West Africa.Clarification of differing views on the extent ofenforced sales is vital in view of their impor-tance in seasonal f lows. Such studies wil l benecessary in several countries and possibly inseveral areas within a country in view of thegreat diversity found in West Africa.

5. A definite need for economic research inthe area of on-farm food storage was expres-sed. While the purely technical aspects hadreceived considerable attention, there was littleinformation on the relative economic benefitsof small farmer versus trader storage. Therewere differing views on the extent and effects ofenforced post-harvest sales on indebtednessand on the benefits of on-farm versus bulk

294

(which could often be equated wi th state) stor-age. This study would link with the study ofgrain movement f rom homestead to point offinal sale and the vexed question of food sec-urity.

6. Several participants reinforced the pointmade in the Harriss paper that the sum total ofpolicies currently in operation was to reduce theincome of the "man who produced the goods."In these circumstances, the overall incentivesfor increasing production of cereals were prob-ably nil. It is a moot point whether ICRISAT can,other than in a very oblique way, instigateresearch that wi l l highlight this fact. Certainlyrelated to this f ield, was the request thatICRISAT study labor constraints to productionand assess the importance of off-farm employ-ment, which appeared to be playing an increas-ingly important role in the economics of therural areas and in peasant farming households.Also, what is the effect on farm and ruralcommunities of the vast input of aid into thesecountries?

7. Related to these are studies that are un-doubtedly required on emigration from ruralareas—both local and long distance. Emigra-tion is having profound effects in various wayson productivity; and by repatriation of cash toareas of origin, it greatly affects the budgets,incentives for increased productivity, and thelife styles of the remaining population. To datelittle research in these topics has been done inWest Africa.

8. There was an appeal for the socio-economic/anthropological side of research not tobe swamped by the sheer weight of feasibleeconomic projects identified in the workshop.

9. The implications of the Harriss' paper onrelevant and feasible research,* and particu-larly the criteria to be adopted in assessingprojects, needs further careful and in-depthstudy by the program research committee atICRISAT before final decisions are taken onproject formulat ion.

* Editors' note: A copy of this paper ent i t led, "Relev-ant and Feasible Research," was made available toWorkshop participants but is not Included in thisvo lume. It was wr i t ten by Dr. Barbara Harriss atICRISAT's request as an addendum to the completereport she prepared on research policy in foodgrain market ing in West Afr ica, which her paper inthis Chapter summarizes.

Chapter 7

Nature and Significance of Risk

Nature and Signi f icance of Riskin t he Explo i tat ion of N e w Technology

Jock R. Anderson*

Abstract

This paper is essentially contemplative rather than attempting an empirical review of the issues canvassed. However, questions are posed — sometimes in the form of hypoth-eses intended to be susceptible to empirical examination. While the subject of overt attention is farming in the semi-arid tropics, the author's hope is that the opinions mar-shalled here may have general relevanceto matters related to risk in farming everywhere.

This essay is essentially contemplative and isnot geared as closely as the author would wishto the realities of the lives of farm families.Without doubt, had it been possible, it wouldhave been much more rewarding to present a cogent empirical review of the issues that arecanvassed. However, in the second-best ap-proach adopted several questions are posed,sometimes in the form of hypotheses intendedto be susceptible to empirical examination.

While the subject of overt attention is farmingin the semi-arid tropics, often by people who arecategorized as peasants, the hope is that theopinions marshalled here may have generalrelevance to matters related to risk in farmingeverywhere.

The word 'exploitation' has replaced 'de-velopment' in the title for two reasons. First, theauthor's ideas on the interplay between risk andthe development of new technologies for smallfarmers have changed little f rom those elabo-rated in Anderson (1974) and Anderson andHardaker (1978). Second, especially in less de-veloped agricultures, there is a difficulty thatmany research and extension workers havelittle empathy for farmers and little understand-

* Department of Agricultural Economics and Busi-ness Management, University of New England,Armidale, Austral ia.The views expressed are those of the author andare not the responsibility of the Australian Bureauof Agricultural Economics, to which he is presentlyseconded.

ing of what farmers do or why they do it — withconsequent impact on the validity of their com-parisons of possible new technologies withprevailing practices. While there have beensome appropriate studies (e.g., Misra 1964,Miracle 1968, Bell 1972, Herdt and Wickham1975, Barlett 1977, Cleave 1977, Gladwin 1977)there still seems to be a need to comprehendthe decision environment in which new tech-nologies may be exploited, before the questionof development of technologies can beadequately tackled.

S o m e Thoughts on Farmers'Pursuit of Sat is fact ion

As in most human communities, the family isthe basic unit of farming societies. The familyenters farming typically in a pssive way thatreflects previous family history and traditionand the confined openings of rural upbringings.Its initial endowment of rural resources islargely predetermined and open to restrictedenhancement through whatever credit marketsare available.

The associated human resources are alsosubject to various constraints. Social status islargely inherited. Limited educational facilitiesrestrict direct development of human capitalcapable of easy redeployment in other sectors.

The picture sketched is one of tightly con-strained physical financial and human re-sources. Decision makers, during their custodyof the physical resources, have to chart a course

297

that in some way generates the best possibleaccumulation of satisfaction for the respectivefamilies. Their task is demanding of adaptabil-ity, patience, and resilience.

The operating environment is intrinsicallyunstable and, for planning, uncertain. The deci-sion makers have to plan to exploit favorableevents and to survive the unfavorable. Thisthey do from a background of cultural andpersonal attitudes to all aspects of Iife. Concern-ing risk in particular, their attitudes are mostunlikely to be neutral (Anderson, Dillon andHardaker 1977).

Risk aversion is now seldom challenged asthe 'norm' of attitudes towards risk generallyand especially in peasant farming communities(Johnson 1971; O'Mara 1971; Moscardi and deJanvry 1977; Wiens 1977; Binswanger 1978a, b,c; Dillon and Scandizzo 1978). Accordingly,there is no point here in dwelling on or attempt-ing to review the growing body of conventionalwisdom on the implications of risk aversion inthese communities for resource allocation, in-vestment, efficiency, etc.

In spite of a paucity of reliable empiricalinvestigations of risk a version, the only featuresthat in anyway remain controversial are (a) thedegree of aversion held on average in variouscommunit ies and (b) the proportions of at-titudes that differ significantly from the averagelevels of aversion (e.g., risk preference or ex-treme risk aversion). It might be observed inpassing that the procedure of drawing infer-ences about 'average' levels of risk aversion onthe basis of matching optimal enterprisecombinations from risk-averse aggregativeprogramming models against observed com-binations, seems fraught wi th potential errorsthat are sensitive to the specification of themodels (see, e.g. Simmons and Pomareda1975).

Over the lifetime of any farm family there wil lbe a succession of environmental crises andcatastrophes, challenges and changes (see, e.g.Jodha 1978). Especially in semi-arid tropicalareas, the unpredictability and severity of clima-tic events is rivalled by the dispersion of relev-ant probability distributions and the humandifficulty of comprehending and managingprobabilities of relatively rare events.

Positive opportunities, such as technologicalinnovations, wi l l appear and be available foruptake. Each wil l be reviewed by decision mak-

ers and, on the basis of whatever information isat hand, personal judgment made about thedesirability of the innovation and about theextent and t iming of possible exploitation. Far-mers may feel that some innovations add moreto the riskiness of their operations than they addto anticipated gains on average. There issketchy investigative support for some suchfarmer appraisals (O'Mara 1971; Anderson andHardaker 1973, Perrin et al. 1976, Gladwin 1978).

The criterion presumed here to underly thejudgment is the lifetime satisfaction of the farmfamily. Unfortunately, this topic is shrouded inclouds of ignorance that are symptomatic of thesparseness of pertinent investigations. Moreresearch on decision criteria of farmers is re-quired if analysts hope to advance from merepresumptions to well-tested hypotheses. Butthis is demanding work. The general reasoningbeing advanced here is that the implementationof new technologies by farm families wil l onlybe satisfactorily comprehended if there is a good understanding of the human characteris-tics behind the pursuit of satisfaction (see alsoCleave 1977).

As a possible lead to future research on thistheme, some speculations are assembled onthe nature of the preferences of the farm family,in these speculations, the biological (Hirshleifer1977) and social (Barlett 1977) underpinnings ofeconomic behavior are emphasized.

Hypothesis 1

Preference functions of farm families areevolutionary and:

a. depend on the composition of the family,and relatedly, the stages in life cycles ofthe members of the family, and

b. feature arguments of f lows of consump-t ion and stocks of wealth, and

c. depend on social status and family self-esteem.

Pride in the present social position of thefamily, and low utility associated wi th the(perhaps remote) possibility that status andthus self-esteem might slump, can plausiblycontribute to ultra-conservative attitudes to ex-perimentation, innovation, change and risk.Perhaps the factors in Hypothesis 1 c should beextended to include educational level, off-farmemployment, religion and group member-ships? (See, e.g. Moscardi and de Janvry 1977.)

298

Hypothesis 1 b is tantamount to a call for anintegration of the New Household Economics(e.g. Sant'Ana de Camargo Barros 1976,Barnum and Squire 1977) with MultiattributeUtility Theory (Keeney and Raiffa 1976). Withthe recent advancements in these fields, thetime seems right for an insightful union of theideas and methods of both, before thespecialists in each spawn the specialized deaf-ness that fol lows more esoteric developments(Weinberg 1975).

In Hypothesis 1 a, there are many sub-hypotheses that could be tested to glean deeperinsights. For example, what are the relativecontributions of marriage partners (when thereare at least two)? One suspects that the rote ofwives is much more significant than has beenrecognized in previous surveys and models ofdecision making on farms. Case studies offarm-family preference functions before andafter separations (by death or divorce) might bea discriminating source of data on this question.If the role of wives proves to have the specu-lated high level of significance, there wil l berelated implications for education in generaland farm management training in particular forrural girls and women.

A further area of ignorance is the potentiallyconflicting preferences that impinge on in-tergenerational transfers of ownership and con-trol. Longitudinal family studies of attitudestowards risk would help to elucidate someaspects of intergenerational problems. Cross-sectional studies may, in the meantime, help toextend understanding. One could usefully con-front a more detailed disection of Hypothesis 1 a such as:

Hypothesis 2

Attitudes towards risk:a. are highly heritable,b. vary systematically with age,c. fol low a general pattern of increasing

aversion to risk with age, andd. tend to be matched similarly in marriage

contracts.Testing of Hypothesis 2 wil l be especially

challenging because of the need to control forother important hypothesized effects, e.g. theeffect of arrival of children of different sexes onattitudes to risk, and the much-discussed effectof changing wealth levels on attitudes to risk.

Hypothesis 1 a might thus be partitioned furtheras follows.

Hypothesis 3

Attitudes to risk:a. change as children enter or leave the

family unit,b. fol low a general pattern of increasing

aversion to risk with:i) increasing numbers of dependent children

and eitherii) increasing feminity of dependent children

when a dowry system prevails, oriii) increasing masculinity of dependent chil-

dren otherwise.It is not intended to imply that testing the

above hypotheses, or even gathering the re-quired data wil l be easy. Indeed, as Tversky(1975) has argued and Binswanger (1978c) hasdemonstrated, this is a challenging field ofinquiry and the difficulties should be well-appreciated before empirical work is embarkedupon.

Perceptions of risk are as important in deci-sions about new technology as are attitudes torisk (Misra 1964). They are just as difficult tostudy as attitudes because of the requirementsfor eliciting the inner feelings of people unac-customed to explication of such vague andunfamiliar concepts (see Tversky and Kahne-man 1975). Ignorance of the way in which riskperceptions change in response to new infor-mation from either on-farm experience orsources beyond the farm is profound. TheBayesian learning model seems reasonable andappropriate on a conjectural basis. However,there is need for empirical work on validatingthe model in terms of the component distri-but ions— prior, likelihood, and posterior—,and on integrating the learning model with theon-going experimentation on and managementof the farm (O'Mara 1971, 1979). Some re-searchable possibilities are embodied in thefollowing compound hypothesis.

Hypothesis 4

Perceptions of the risk associated with newtechnologies:

a. can be modelled adequately as Bayesianrevision of subjective probability distribu-tions,

b. feature 'pre-experience' prior distribu-

299

tions that differ systematically w i th :i) extent of formal education,

ii) length of farming experience in therespective agroclimatic region, and

iii) number of previous severe droughtsand other extremes encountered:

c. differ between farm families and develop-ers of new technology in that the Iikelihooddistributions of the latter group are lessdisperse for any given pieces of experi-mental evidence.

Opt ima l i t y , Cogn i t i on ,and Cogn i t i ve Dissonance

The framework thus far has been cast in thestyle of optimising agents who populate themodels of modern economics. With due regardto transaction costs associated wi th change,and to the perceptual l imitations of human be-ings, the appropriateness of implicit assump-tions about the unqualified pursuit of optimallevels of satisfaction can be challenged. As a disciple of Bernoullian decision theory, this wri-ter sees no problems with the notion that deci-sion makers should, in their own interests, seekto maximize expected util ity (Anderson 1978).

The power of the maximizing assumption isin the enrichment of the analysis of the con-sequences of an appropriately specified model.However, this is not to suggest that decisionmakers really do actively seek to maximize any-thing (Lipton 1968). The crucial presumption isthat analysis of maximizing models leads to re-levant conclusions about actual behavior. Thisp resumpt ion must also be sub jec ted tocritical test. Perhaps one day the non-maxi-mizing Simonian models may be exploitable inanalysis of farm decision making. But that day isseemingly still far off.

As decision makers proceed through theirsequences of confronted decisions, their cogni-tive limitations constrain their perceptions ofopportunities, and their perceptions of the dif-ferences in satisfaction associated wi th alterna-tive decisions. A characteristic of optimality isthat near-optima are indeed very nearly optimalin terms of the opportunity cost of the forgonevalue of the objective function (Anderson 1975).For many decisions in the dynamic existence ofa farm family, the perceived costs of change(which may include evaluations of personal dis-

comfort wi th unknown or foreign ideas) wil l ex-ceed the perceived gains in satisfaction. Persis-tence wi th famil iar and possibly tradit ionaltechnologies wil l be rational for the family insuch instances (Gladwin 1978).

Finally, it must be stressed that farm familiesdo not evolve their individual attitudes, in isola-t ion. Group pressures f rom local communitiesserve to reduce cognitive dissonance so that, inspite of individual conservatism, inertia and cir-cumstance, attitudes and perceptions wil l tendto move in the same general direction.

Second-Par ty Appra isa lo f Innovat ions for Turbu len tFarming Sys tems

People who are so bold (impertinent?) as toattempt to judge what changes are 'good ' forfarmers must run a daunting gauntlet beset byneeds:

a. for economy and thus simplicity in theirwork,

b. for a certain elegance in approach for theirown self-esteem,

c. for relevance in grappl ing (and being seento grapple) wi th the fundamental issues atstake, and

d. for responsibility in avoiding the role ofdisaster agent for families who are possi-bly in much more fragile circumstancesthan their appraisers and advisors.

Tricky t radeoffs among these needs arereadily apparent, which is perhaps why suchpeople tend to take a rather s ingleminded'strong-line' personal approach to their work.This writer holds an evolving and still ambiva-lent position. On the one hand, the dangers ofpotential self-delusion and irrelevance in nottaking formal account of the rich preferences ofaffected people should be emphasized (Ander-son, Dillon and Hardaker 1977). On the otherhand, the prospective difficulty of adequateformal but economical modell ing of complexsystems should also be recognized (Weinberg1975, Anderson and Hardaker 1978).

A working compromise that may be reason-able, especially in view of the above remarksabout optimality, is as fol lows. Highlight thepotentialities for human intuition to deal withthe complexities informally. However, let thisintuition be played out against a backdrop of

300

formal case studies of the importance of the

mul t i fa r ious components of the decis ion-

making environments of farm families.

Much appraisal work wil l be directed prima-

rily at assessing what is 'good' for a lending

agency, perhaps a government. In this case, the

foregoing remarks are still apposite, although

there is probably greater scope for more formal

modell ing than when farmers themselves are

the clients. Such modelling should probably in-

vo lve cons idera t ion of r isk avers ion and

stochastic aspects of the environment rou-

t ine ly . To the present t i m e , however ,

techniques of investment appraisal under risk

have seldom found appl icat ion in practical

farm-economic analysis (Wong 1978).

C o n c l u d i n g R e m a r k s

As with only science, from astronomy to atomic

physics, and w i th any nonsc ience, f r o m

psychoanalysis to Pandora's Box opening,

deeper investigation does not necessarily make

life more comfortable or knowledge more cer-

tain. Yet Science, including science bearing on

families and their farms, cannot turn back in its

methods — even when 'refinement in model-

ling eventuates a requirement for stochasticity'

(Mihram 1972, p. 15). So it is that further careful

study of the decision processes and environ-

ment of farm families wil l yield many revela-

tions and many new uncertainties. The revela-

tions should help to render better futures for

farm families, particularly if potentially new

technologies are subjected to scrutiny more at-

tuned to the realities of farm life. The uncertain-

ties wil l be the new challenges for future inves-

tigators.

References

ANDERSON, J. R. 1974. Risk efficiency in the interpreta-t ion of agricultural product ion research. Review ofMarketing and Agricul tural Economics 42 (3): 1 3 1 -84.

ANDERSON, J. R. 1975. One more or less cheer foropt imal i ty. Journal of the Austral ian Institute ofAgricultural Science 41 (3) :195-7.

ANDERSON, J. R. 1978. Perspective on models ofuncertain decisions in Risk, Uncertainty and Agricul-tural Development, eds. J. A. Roumasset, J. M.Boussard, and I. J. Singh. Berkeley, Calif, USA:University of California Press.

ANDERSON, J. R., DILLON, J. L. and HARDAKER, J. B. 1977.

Agricultural decision analysis. Ames, Iowa, USA:Iowa State University Press.

ANDERSON, J. R., and HARDAKER, J. B. 1973. Manage-

ment decisions and drought. Pages 220-244 in Theenvironmental, economic and social significance ofdrought, ed. J. V. Lovett, Sydney: Angus andRobertson.

ANDERSON, J. R. and HARDAKER, J. B. 1978. Economic

analysis in design of new technologies for smallfarmers. In Economics and the design of small-farmer technology, eds. A. Valdes, G. M. Scobie, andJ. L. Dil lon. Ames, Iowa, USA: Iowa State UniversityPress.

BARLETT, P. F. 1977. The structure of decision makingin Paso. American Ethnologist 4 (2):285-307.

BARNUM, H. N. and SOUIRE, L. 1977. An econometric

model of an agricultural household. Wor ld BankEconomics Depa r tmen t w o r k i n g paper ,Washington, DC, USA.

BELL, C. 1972. The acquisition of agricultural technol-ogy: its determinants and effects. Journal of De-velopment Studies 9 (1):123-59.

BINSWANGER, H. P. 1978a. Att i tudes towards risk:experimental measurement in rural India. YaleUniversity Economic Growth Center discussionpaper 285, New Haven, Conn, USA.

BINSWANGER, H. P. 1978b. Att i tudes towards risk:implications for economic and psychologicaltheories of an experiment in rural India. YaleUniversity Economic Growth Center discussionpaper 286, New Haven, Conn, USA.

BINSWANGER, H. P. 1978c. Risk attitudes of ruralhouseholds in semi-arid tropical India. Economicand Political Weekly 13 (25):49-62.

CLEAVE, J. H. 1977. Decision-making on the Africanfarm. Pages 157-177 in Contributed papers read atthe 16th International Conference of AgriculturalEconomics, Oxford, UK.

DILLON, J. L., and SCANDIZZO, P. L. 1978. Risk attitudes

of subsistence farmers in Northeast Brazil: a sampl-ing approach. American Journal of AgriculturalEconomics 60 (3):425-435.

GLADWIN, C. H. 1978. Allocative efficiency and deci-sion models. International Fertilizer DevelopmentCenter working paper, Florence, Alabama, USA.

HERDT, R. W. and WICKHAM, T. H. 1975. Exploring the

gap between potential and actual rice y ield in thePhilippines. Food Research Institute Studies 14(2) :164-81.

301

HIRSHLEIFER, J. 1977. Economics f rom a biologicalv iewpoint . Journal of Law and Economics 20(1) :1-52.

JODHA, N. S. 1978. Effectiveness of farmers ' adjust-ment to risk. Economic and Political Weekly 13(25):38-48.

JOHNSON, A. 1971. Sharecroppers of the Sertao,

Stanford University Press, Palo A l to , Calif, USA.

KEENEY, R. L. and RAIFFA, H. 1976. Decisions w i thm u l t i p l e ob ject ives: preferences and va lue

tradeoffs. New York: Wiley.

LIPTON, M. 1968. The theory of the opt imizing peasant.Journal of Development Studies 4 (3 ) :326-351.

M I H R A M , G. A. 1972. S imula t ion : statistical founda-t ions and methodology. New York: Academic Press.

MIRACLE, M. P. 1968. Subsistence agriculture: analyt i -cal problems and alternative concepts. Amer icanJournal of Agricul tural Economics 50 (2) :292-310.

MISRA, B. 1964. Uncertainties and adopt ion of newfarming practices in India. Indian Journal of Agricul-tural Economics 19 (1) :73-75.

MOSCARDl, E., and DE JANVRY, A. 1977. Att i tudestoward risk among peasants: an econometr icapproach. Amer ican Journa l o f Agr icu l tura lEconomics 59 (4) :710-716.

O ' M A R A , G. 1971. A decision-theoretic v i ew of themicroeconomics of technique: di f fusion in a de-veloping country. Ph.D. thesis, Stanford University,Palo Al to , Calif, USA.

O ' M A R A , G. 1979. The microeconomics of techniqueadopt ion by smal lhold ing Mexican farmers. In Prog-ramming studies for agricultural sector pol icy, eds.L. M. Bassoco, R. D. Nor ton, J. S. Si los, and L. Solis.Oxford : Oxford University Press. In press.

PERRIN, R. K., MOSCAROI , E. R., W INKELMANN, D. L., and

ANDERSON, J. R. 1976. From agronomic data tofarmer recommendat ion: an economics trainingmanual . CIMMYT Informat ion Bul let in 27, Apdo .Postal 6-641, Mexico 6.

S A N T A N A DE CAMARGO BARROS, G. 1976. Asking wages,

market wages, and the of f- farm labor supply by farmoperators. Ph. D. thesis, Nor th Carolina State Uni-versity, Raleigh, N.C., USA.

S I M M O N S , R. L., and POMAREDA, C. 1975. Equi l ibr ium

quanti ty and t im ing of Mexican vegetable exports.Amer ican Journal of Agr icul tural Economics 57(3) :472-479.

TVERSKY, A. 1975. On the el (citation of preferences:descript ive and prescriptive considerat ions. Pre-sented at the Workshop on Decision Making w i t hMul t ip le Confl ict ing Objectives'. International Insti-tute for Appl ied Systems Analysis, Schloss Laxen-burg , Austr ia.

TVERSKY, A. and KAHNEMAN, D. 1975. Judgment

under uncertainty: heuristics and biases. Pages141-162 in Uti l i ty, probabi l i ty, and human decisionmaking, eds. D. Wendt , and C. Vlek. Dordrecht:Reidel.

WEINBERG, G. M. 1975. An introduct ion to general

systems th inking. New York: Wiley.

W I E N S , T. B. 1977. Uncertainty and factor al location ina peasant economy. Oxford Economic Papers 29(1):

4 8 - 6 0 .

W O N G , C. Y. 1978. Risk analysis in the appraisal oflarge irr igation projects in developing countr ies: a Malaysian case study. M.Ec. thesis, Universi ty ofNew England, Armidale.

302

The Nature and Signi f icance of Riskin the Semi-Ar id Tropics

Hans P. Binswanger, N. S. Jodha, and B. C. Barah*

Abstract

This paper reports on a series of research projects on the role of risk in the semi-arid

tropical (SAT) agriculture of India. The projects are based on (1) macro data of yield, price

and income variability, (2) microlevel data of income and expenditure levels of rural

households over drought cycles, and (3) on a set of psychological experiments on risk

attitudes. So far this research has yielded the following conclusions:

1. Levels of income risk in the Indian SAT are high and come mostly from production

rather than price risk.

2. Virtually all farmers in the SAT Indian sample are risk-averse although not to an

extreme or severe extent.

3. Farmers do not have access to cheap self-insurance and risk-diffusion devices

enabling them to even out their consumption streams in the face of risky production.

We are therefore forced to conclude that risk and risk-aversion lead to u n d e r i n v e s t -

ment in SAT agriculture.

Agriculture in the SAT is almost universallycharacterized by low farmer investment in ag-ricultural inputs such as fertilizers, pesticides,and seeds. Even labor inputs and investmentsin land improvements and irrigation are usuallylow compared wi th other regions of the de-veloping world. Over the past 3 years theauthors have carried out research aimed attesting whether these low investment levels arepartly or fully caused by the risky nature ofagriculture in these areas, by the aversion to riskof farmers, and by the lack of mechanisms andpolicies aimed at stabilizing the income offarmers. In this paper we report what we havefound so far in a research project that is still farfrom completion.

Risk and risk aversion of farmers is not theonly potential source of low investment or"underinvestment"; low investment could alsoresult f rom generally low profitability (mea-sured as expected returns) or f rom credit con-straints. To establish that it is risk and riskaversion that lead to underinvestment, empiri-cal knowledge on these questions is required.

* Principal Economist, Senior Economist,Economist, ICRISAT.

and

First, of course, investment in SAT agricultureis risky. This riskiness is a well-known fact andneeds no further research, except for moreprecise quantification. However, to deal w i thpolicy alternatives later, we must be able toquantify the relative importance of yield risksand price risks in total production risks; theeffect of crop diversification on those risks; therelative importance of drought risk in total yieldrisks; and the effect of modern inputs such ash igh-y ie ld ing var iet ies, fer t i l izers , andsupplementary irrigation on yield risks. Someevidence on the sources of risk will be pre-sented in the next section.

Second, riskiness or increases in riskinesswith input use wil l lead to underinvestment onlyif farmers are risk-averse rather than risk-neutral. Risk-neutral individuals wi l l try tomaximize average or expected net returns re-gardless of the extent of variability in thesereturns. This is also a strategy that leads to thehighest returns on investment in the long run.On the other hand, a risk-averse individual wi l lforego some expected returns if this also re-duces the extent of variability of his incomestream. Thus he wil l underinvest relative to therisk-neutral or socially optimal level. The ques-tion is whether the overwhelming majority of

303

farmers in the SAT are risk-averse. Also neces-sary is the quantification of the extent of purerisk-aversion of farmers and landless laborersand the relation between risk-aversion and theriskiness of agriculture of the region, the size offarm, and the size of the investment to beundertaken.1 In the third section we presentexperimental evidence on risk aversion.

Third, it is important to realize that underin-vestment need not necessarily occur if agricul-ture is risky and if farmers are risk-averse. If theyhave effective mechanisms at their disposal forself-insurance or risk diffusion, they may stillinvest up to the risk-neutral opt imum. Forexample, an effective crop insurance systemwould al low farmers to shift the risks to theinsurance system as a whole, i.e., risk would bediffused over wider areas and across years.Such systems generally do not exist at presentand farmers have to rely largely on their ownmeans for self-insurance and risk diffusion. Wedistinguish two types of adjustment measures:(1) Risk-reducing measures that are used be-fore damage may occur and that include cropdiversification, intraseasonal adjustment ofsowing times and cropping patterns, soil- andwater-management techniques, etc.; (2) Lossmanagement, or risk diffusion measures thatare designed to deal wi th the consequences oflosses and include storage, salvage operations,accumulation of financial assets, reduction offinancial commitments in drought years, bor-rowings, and many more. They are aimed attemporal and spatial, risk diffusion, as well asdiffusion among individuals. Efficient risk-reduction and loss-management devices wouldallow a farmer to take substantial levels of riskwithout being exposed to severe reductions inhis customary consumption or wi thout loss ofproductive assets even in drought years. Thefourth section is devoted to this issue while thefinal section wil l indicate what conclusions we

1. Even if all farmers were risk-neutral, investment inSAT areas w o u l d have to be lower than in assuredzones. In good years, crops in bo th zones wou ldyield the same under given management and soilcondit ions, but in the SAT the frequency and seve-ri ty of negative deviat ions f r om the g o o d yearyields is much larger. Thus, for any given invest-ment level, the rate of return w o u l d be lower,reducing the risk-neutral op t imum investmentlevel.

304

can already draw f rom our findings and indicatefuture directions of the research program.

Riskiness o f Semi -Ar idTrop ica l Agr i cu l tu re

One of the major problems of working on therole of risk in agriculture is the near total lack ofthe farm-record type data over extendedperiods that would enable us to measure prob-ability distributions of yields for individualplots, farms, or techniques of production. Tomeasure riskiness of crop production acrossagroclimatic zones we therefore have to rely onaggregate data for fairly large administrativeunits. In India, the smallest such unit for whichstatistics are available is the district, and theyare often small enough to be entirely allocatedto specific zones.2

For farmers who participate in the market,what is important is the variability of grossreturns to owned factors of production, i.e.,gross return minus variable costs such as fer-tilizer and hired labor. We do not know whatvariable costs are or how they vary with outputlevels and therefore use the variability in grossvalue of total agricultural production per hec-tare as a yardstick to measure the level andsource of risk to which farmers are subject.

The easiest measure of riskiness of farmingas a whole is the coefficient of variation of valueof output per ha. Barah (1977) has developed a technique that permits the decomposition ofthe (square of the) coefficient of variation (cv)into four components:

1. variability in yields,2. variability in prices,3. the covariance between yields and prices,4. a residual measurement error due to vari-

ous interaction terms.This technique has been used in a pilot study

of 20 districts of Andhra Pradesh, which con-tains irrigated rice-growing districts in its coas-tal zone and mostly SAT districts wi th predo-minantly irrigated agriculture on red soils(Alfisols) in the interior. The study wil l be

2. Whi le data are reported on a distr ict basis, thesampl ing f rameworks on wh ich they are based aredesigned to achieve a certain level of accuracy forstate level estimates. Thus the distr ict level esti-mates have a fair ly high sampl ing error.

extended to several other states in the nearfuture. The major conclusions emerging fromthe pilot study are as fol lows:

1. Coefficients of variation of gross value ofcrop output per ha for the SAT districts rangefrom 18 to 3 1 % , with a mean value of 22%.Those for the coastal (irrigated) zones vary from15 to 20% with a considerably lower mean of17%. Not surprisingly then SAT agriculture issubstantially more risky, and policies aimed atrisk reduction should first be directed towardsthese zones.

2. The decomposition of the total variabilityof gross crop returns into a yield component, a price component, and a yield-price interactioncomponent, indicates that in most SAT districtsthe contribution of yield variability to totalvariability exceeds that of price variability, andoften by a factor of close to 2 or more. Con-versely, in the irrigated coastal districts, theprice component of variability usually exceedsthe yield component, often by the same factor.(However, in both zones exceptions occur.)

For both types of zones the yield-price in-teraction components are usually small relativeto the sum of the individual yield and priceeffects. The largest absolute value of the yield-price component is only 22%, and on averagethe absolute value is only 8%. Furthermore, theyieldprice interaction component is positive in10 cases and negative in 10, which means thatindividual districts are sufficiently small relativeto the national market and their weather condi-tions sufficiently independent of nationalweather conditions that yield and price compo-nents contribute to total variability in an almostindependent way. This result is somewhat sur-prising but of importance to those interested inprice stabilization schemes.3 Alternatively, the

3. The recent literature has been concerned mainlywi th the fo l lowing problem of price stabil ization. Inthe absence of price stabil ization, yields and pricesshould usually be expected to move In oppositedirect ions, i.e., there should be a negative correla-t ion between them. Any such negative correlationtends to stabilize revenue in the presence of yieldinstabil i ty, and a price-stabilization scheme couldtherefore destabilize revenue rather than stabilizeit if it breaks that negative correlation. Since far-mers are ult imately interested in revenue stabiliza-t ion , a price scheme could therefore do more harmthan good. See for example Hazell and Scandizzo,

1977.

low interaction components could have beencaused by the persistent price intervention ofthe Indian government over the study years1953-1973.

3. Barah has also computed the amount ofrevenue stabilization that arises from the diver-sified cropping patterns farmers prefer to solecropping. He computes an "insurance value" ofcrop diversification as the increase in coefficientof variation of total revenue that would arise ifthe yields and prices of all crops actuallyplanted were perfectly correlated, i.e., werebehaving like a single crop. This insurancevalue ranges from 31 % to as high as 154%, wi than average of 78%. In sharp contrast to thefailure of a negative price-yield correlation as a variance reducer, crop diversification, on aver-age, reduces coefficients of variation dramati-cally. Interestingly, there appears to be no clearcorrelation between the extent of irrigation in a district and the amount of " insurance" farmersreceive from crop diversification.

These results suggest that if at all necessary,income stabilization policies should first con-centrate on the SAT zones where variability ishigher rather than on the irrigated zones. Furth-ermore, since yield risk is the major componentof overall variability in those zones, policyshould first be aimed at yield stabilization ordirect income stabilization rather than pricestabilization.

The Extent of Risk Avers ion

Over the past 3 years Binswanger (1978a,1978c) developed an experimental technique tomeasure pure attitudes towards risk. The exper-iment has practically no theoretical restrictions;individuals choose among alternatives whereincreasing expected returns can only be purch-ased by increasing risk or dispersion of out-comes. In contrast to most work in experimentalpsychology and in economics, the experimentused large real payoffs to induce participants toreveal their preferences. The highest expectedpayoffs for a single decision exceeded monthlyincomes of unskilled workers. Furthermore, theparticipants consisted of a random sample ofthe rural population of some regions of south-ern India that contained a large variation in

305

wealth, education and other personal charac-teristics.4

The experiment, carried out wi th 330 indi-viduals selected at random f rom six villages ofthe SAT regions of Maharashtra and AndhraPradesh, consisted of a sequence of games.People were offered a set of eight alternativechoices in which higher expected returns couldonly be "purchased" for larger standard devia-tions. The alternatives A to F are described inthe top panel of Table 1. Each consists of a "good luck" and a "bad luck" outcome wi thprobability of 50%, which is decided on the tossof a coin. Alternative zero is a certain outcome inwhich the individual is simply paid Rs 50,whereas alternative F pays nothing or Rs 200wi th equal probability. The alternatives D* andD and F are stochastically dominated by alterna-tives B, C, and E respectively. Each alternative isgiven a name classifying the extent of riskaversion of the person who chooses it. Thesenames are arbitrary, and more precise mea-surements of risk aversion are discussed inBinswanger (1978b).

The game was played and payments actuallymade seven or eight t imes over a period of 6 weeks or more, wi th f rom 1 day to 2 weeks t imeleft between games for reflection. In the first fivegames all amounts shown in panel 1 of Table 1 were divided by 100, i.e., the alternative F paidRs 2 on good luck while alternative zero paidRs 0.50 for sure. This game level is thereforecalled the Rs 0.50 game level. At least 2 weekslater, two games at the 5-Rs level fo l lowed (allamounts in Table 1 divided by 10). After 2 moreweeks a subsample of 118 household headsplayed the 50-Rs game of Table 1 and only theresults for this subsample are shown in panel 2 of Table 1. Chi-square tests showed that at the5-Rs level the risk-aversion distribution of thesubsample whose results are discussed herecould not be distinguished statistically f rom therisk-aversion distribution of the other house-

4. For a more detai led descr ipt ion of the exper imen-tal methods and results see Binswanger (1978c).The sample is random and taken f r om the ruralpopulat ion that has agr icul ture or agricul turallabor as pr imary or secondary occupat ion. Themean physical weal th of the sample was Rs. 22 370w i th a coefficient of var iat ion of 137%. Mean yearsof school ing was 2.6, w i th a CV of 130%. Mean ageof the sample was 42 years, w i t h a CV of 3 0 % .

hold heads or their wives who did not play the50-Rs game.

The results in panel 2 show that, whenpayoffs are small (0.50 Rs), we find nearly 50%of individuals in the intermediate and moderaterisk-aversion categories (B and C). Over a thirdof the individuals show a nearly neutral orrisk-preferring behavior pattern (E and F), andfewer than 10% are extremely or severely riskaverse (O and A). When the stake levels rise, theproport ion of individuals in the intermediateand moderate categories rises til l it reaches80% of individuals in these two classes. Near-neutral and risk-preferring behavior virtuallydisappears; only one of 118 individuals chooseF. On the other hand, the fraction of extremeand severely risk-averse choices stays below10% up to the 50-Rs level and rises to 16% at the500-Rs level (a hypothetical question). The ex-treme risk-averse fraction never exceeds 2.5%.At higher stake levels, the risk-aversion distribu-tion is thus single-peaked, wi th most of itsweight in the two intermediate and moderaterisk-aversion classes.

The main conclusion arising from this exper-iment is that in the SAT regions studied, virtu-ally all farmers have intermediate or moderatedegrees of risk aversion, once the stakes of a game come into the neighborhood of small- tomedium-size agricultural investments. We ob-serve almost none in the risk-neutral or severelyrisk-averse classes of behavior; i.e. attitudes arestrikingly similar, despite the fact that the indi-viduals involved have widely different incomeand wealth levels. In particular, at the higherstake levels we are unable to establish clearstatistical relationships between risk aversionand the personal characteristics of wealth, farmsize, sex, age and tenancy status. On the otherhand the fol lowing variables are generally as-sociated wi th very modest decreases in riskaversion: schooling, salaried employment, re-ceipt of transfer income, and the number ofwinning draws in the sequence of experiments.

These results are in many ways surprising.Earlier work in Brazil by Dillon and Scandizzo(1978), based on interviews rather than experi-ments, had indicated a much wider spread ofrisk aversion; in particular, they found fairlyhigh proportions of extreme risk-avoiders andrisk-takers. Second, economists are very sur-prised at the weak relationship between riskaversion as measured in the experiments and

306

307

measures of wealth such as asset holdings andfarm size. The theoretical implications of thesefindings have been explored in Binswanger(1978b). The implications of these results foragricultural research and economic policy in theSAT wi l l be explored in the last part of thispaper.

S e l f - I n s u r a n c ea n d R i s k D i f f u s i o n

Risky production and risk aversion need notnecessarily lead to lower investment levels thanthe ones that maximize expected or averagereturns. We perceive the goal of a risk-averseindividual as one of achieving low fluctuationsin his consumption stream over t ime, regard-less of the fluctuations in his production andincome levels. It has been long established thatif an individual has costless means available tohim for evening out his consumption stream inthe face of a variable income stream he wi l l actas if he were maximizing expected income; hewi l l invest up to the level that is opt imal f romsociety's point of view and no "underinvest-ment" occurs.5 We know, of course, that noeconomic system ever al lows individuals toeven out their consumption streams wi thoutcost; we are therefore interested here in deter-mining whether the farmers in the SAT havelow-cost mechanisms available to even outconsumption streams. If this were the case wewould then argue that there is little need toworry about risk aversion as a source of unde-rinvestment and no justification for any policyintervention.

How should we judge the effectiveness of thefarmer's self-insurance and risk-diffusionmechanisms? The first criterion is whether theyachieve the goal: Do they stabilize real con-sumption levels, i.e., are farmers in a posit ion toconsume essentially the same quantities of

5. A r row (1970) shows that a system of cont ingentfuture markets in all commodi t ies wi l l be a sufficient condition to achieve Pareto optimality inan exchange economy, w i t h the future contractsact ing as " I n s u r a n c e " fo r t h e ind iv idua l .Economists usual ly assume that a system whichoperates w i th only a few insurance and/or futuresmarkets is all tha t is necessary to achieve Pareto

opt imal i ty .

goods during drought years as during normalyears? This, of course, is a fairly strong condi-t ion and would indicate perfect self-insuranceand risk diffusion. We may, therefore, weakenthe criterion successively and ask whether farmers are at least able to maintain the quantity offood consumption and — as the weakestcriterion — whether they are able to maintaintheir foodgrain intake (as an indicator of calorieconsumption).

Farmers wil l always attempt to even outconsumption levels by some mechanisms. In-dependent of the degree of their success inevening out their consumption streams, we canask what is the cost of the various mechanismsthey use? This wil l be our second criterion,considering high-cost mechanisms as expen-sive insurance and low-cost mechanisms ascheap insurance. Economics also tells us whatmechanisms we should look for. Price risks areusually " insured" (or rather diffused) throughthe economy via future markets in risky com-modities, whereas production risks are diffusedvia crop insurance or other disaster insurance.Both of these mechanisms are absent in theSAT. The next best alternative is " insurance"via the capital market. Individuals or smallregions should accumulate financial assets dur-ing good years and be able to borrow at essen-tially the same rates of interest for consumptionduring drought years. Indeed if savings institu-tions offer sufficiently high interest rates andofficial credit agencies lend out at only slightlyhigher rates during drought years, the capitalmarket could act as a very effective substitutefor crop insurance and future markets.

If, however, capital markets are imperfect,individuals have other options to even out theirconsumption streams. First, they can storefood. However, such storage is not a low-costinsurance even if storage costs and storagelosses are low. The cost of holding these stocksis equal to the rate of return on investment ofthe value of the stocks in some other fo rm, suchas additional fertilizer applications or financialassets.

Second, individuals wi l l have to even outconsumption streams via purchases and salesof assets. Gold and jewelry as wel l as consumerdurables can fulfi l l that funct ion. Note, how-ever, that the cost of storing gold is again therate of return that could be had on the capital.Furthermore, transactions costs wi th such capi-

308

tal items — especially consumer durables — are usually high; they can be sold only at lowerprices than they can be bought.

Third, producers can also sell producer dura-bles, such as cattle, during drought years andrepurchase them in the good years. As long astransaction costs — the difference between thesale price and the repurchase price, adjusted forthe cost of carrying the animals over thedrought — are low, nothing is wrong with thisadjustment mechanism. For example, if a drought hits one small area, farmers fromneighboring districts may come in and buy upproducer durables that can be repurchasedimmediately after a drought for only a smallcost differential. This points to the importanceof effective interregional markets for agricul-tural factors of production as a risk-diffusiondevice.

Finally, the government operates relief pro-jects in drought areas. Since agricultural labordemand is low during drought years, the possi-bility of earning income on a public worksscheme affords insurance not only to farmersbut also to landless laborers, who may often beharder hit by drought than farmers, since theyusually cannot borrow easily for lack of collat-eral and have few assets to allow them toabsorb the income fluctuations.

One adjustment mechanism also worth con-sidering is the postponement of "nonessential"consumption, especially marriages and othersocioreligious expenditures. Of course, no onelikes to be subject to such rescheduling andtherefore it is not an insurance mechanism inthe strict sense. However, it is usually assumedthat such postponement is associated wi th littlediscomfort or disutility, and we consider italong with the other mechanisms.

The evidence presented below has been as-sembled over the years by N. S. Jodha in a number of studies covering parts of or entiredrought cycles in five districts of the SAT andarid tracts of India (Jodha 1978a, 1978b).

Table 2 shows indices of drought year con-sumption expenditures (not quantities, exceptfor foodgrain) compared wi th postdroughtyears. Unfortunately, we do not have the quan-tity or price information to correctly deflate theconsumer expenditure to real expenditures.However, Table 3 contains such information forone drought cycle f rom 1962-63 to 1964-65 inRajasthan. Note that the price index constructed

on the basis of these numbers for the food itemsstood at 100 in the predrought year 1962-63, at177 during the drought year 1963-64 and fellback to 128 in the postdrought year.6 Constancyor near constancy of expenditures in Table 2 thus implies sharp reductions in quantities con-sumed during the drought years. In Table 2,note that although we are looking at nominalexpenditures, the totals drop in all areas by 8 to29% compared with the normal postdroughtyear. Furthermore, the quantity of food grainsconsumed, which is probably the most conser-vative index of fluctuation in consumptionlevels, drops by from 11 to 19%. Total foodexpenditures remained nearly constant. In theface of the price rises in the drought years, tomaintain calorie consumption at tolerablelevels, individuals had to reduce the expendi-tures on protective foods (milk, fats, sugar,fruits, etc.) by up to 57% in the case of smallfarmers in the Barmer area. In the face of theseprice rises, the drop in quantities consumedmust have been even larger. Larger adjust-ments were also made in ceremonial expendi-tures. This is the category with the largestexpenditure reduction, which is not unex-pected. Overall this table indicates that farmersare not able to maintain foodgrain consump-t ion, much less any other component of con-sumption, at a customary level during a droughtcycle.

Table 3 examines indices of real asset posi-tions, relating to predrought years, for droughtand postdrought years. It indicates substantialreductions in both productive and nonproduc-tive assets over the drought cycle, with thesmallest reductions in productive assets occur-ring in the Banaskantha and Sholapur areas.

A detailed examination of Jodha's evidenceshows that most reductions in productive as-sets are sales or outright losses of livestock.Large losses occur when households temporar-ily migrate to other areas in an attempt to keeptheir livestock alive or to find work. Further-

6. If a small region is hit by drought , an effectivemarketing system should result in prompt ship-ment of commodit ies to this region f rom otherregions and prevent substantial price rises. If a whole country or a subcontinent is hit by drought ,price rises wi l l occur even in the presence of a goodmarketing system.

309

310

more, f rom Table 4 we see that transactioncosts associated with asset sales are extraordi-narily high. They are Rs.1060 per bullock pair at1972-73 prices, which is around 50% of theprice of a bullock pair in normal years.

Also note that in most areas householdsappear to be unable to substantially recoup thestock of productive assets during the post-drought years, which indicates a real weaken-ing of their production potential as a result ofthe drought year.

Table 5 indicates the sources of sustenanceincome of households during drought years.Credit in all forms provides rarely more than 10per cent of sustenance income, which is indica-tive of a very severe capital market imperfec-t ion, since credit would be the most efficientsource of " insurance", next to futures marketsor crop insurance. The detailed evidence (Jodha1978a, 1978b) also indicates that official creditinstitutions sharply reduced their lending dur-ing these droughts, except in the Banaskanthaand Sholapur areas. In fact, many offi-cial credit agencies are prevented by their con-stitution from lending for consumption pur-poses, which in drought-prone areas is a totallycounter-productive policy. Table 4 also showsthat for the Rajasthan area, where price infor-mation is available, the average interest rate offormal and informal credit sources rose from15% in the predrought year to 23% in thedrought year, a further indication of substantialcapital market imperfections.7 In addition smallfarmers seem to have more interregionaldifficulties obtaining credit in drought yearsthan large farmers.

Wi th the exception of the Jodhpur area,wages from relief works were the single mostimportant source of sustenance income for allfarmers combined. (In Jodhpur sales of assetswere slightly more important). Such wages arepredictably more important for small ratherthan large farmers. Contrary to impressionisticevidence and evidence focusing on the poten-tial waste and corruption associated wi th public

7. Rates of interest on savings, deposits, and otherfinancial assets potential ly available to ruralpeople are usually be low 7%. The large differencebetween these rates and the borrowing rates makefinancial savings and bor rowings poor insurancemechanisms for consumpt ion streams.

works, such programs seem to be remarkablyefficient as an " insurance" mechanism.

We conclude that the combination of farmers'own efforts, official credit agencies, and publicworks, are insufficient to provide effectivesmoothing of essential consumption streams,much less of overall consumption streamswhere quantity variations are very large. Theofficial and unofficial credit systems appear tobe particularly ineffective. Furthermore transac-tion costs in markets for physical capital are solarge as to make the disposal and acquisition ofassets a particularly unattractive tool for financ-ing drought-year consumption. The only brightspot appears to be public relief programs.

Implications

Our joint research so far has yielded the fol low-ing conclusions:

• Levels of income risk in the Indian SAT arehigh and come mostly from productionrather than price risk;

• virtually all farmers in our SAT Indiansample are moderately risk-averse;

• they do not have access to cheap self-insurance and risk-diffusion devices enabl-ing them to even out their consumptionstreams in the face of risky production.

We are therefore forced to conclude that riskand risk aversion lead to underinvestment inSAT agriculture.

However, two important caveats are in order.First, we do not yet know the quantitativeimportance of the underinvestment; furtherresearch is needed to assess this. Second, wenow know that underinvestment exists in situa-tions where individuals either are risk-neutral oract as if they were (because they have access tocheap loss-management and risk-diffusion de-vices). However, unless new economic andsocial policies can be designed to improveself-insurance or risk-diffusion devices so thatthe "insurance cost" goes down, our findingsare not of much operational significance. Ifpublic risk diffusion mechanisms simply re-place the private ones without reducing theircosts, not much is gained. Social science re-search must therefore be directed into findingthese new or improved mechanisms. Dan-dekar's (1976) work is a significant step in thatdirection.

311

312

313

314

After this important qualification of the re-sults, some policy conclusions still emerge.First, it clearly makes no sense to advocate thedevelopment of risk-graded technologies inSAT India so that small farmers may adopt thelow-yield, low-risk ones and large farmersadopt the high-yield, high-risk ones. There isnot enough difference in risk attitudes amongsmall and large farmers in our SAT Indiansample to warrant such an approach. Smallfarmers should accept the same technologiesas large farmers but are prevented f rom doingso by differentially higher costs of credit andinputs, or by other constraints. Furthermore,techniques to measure differences in riskinessamong technologies are still very complicatedand data-intensive, which makes such a diffe-rential research strategy even less appealing.One may legitimately search for a " low-inputop t imum" with a given technology. For exam-ple, in a wheat variety one might explore thebest combination of fertilizers if, instead of anoptimal dose of all fertilizers, one could onlyafford one-third of the money. Since the lowranges of production functions are usually theones with highest marginal returns, pushingsmall farmers towards varieties that do notrequire fertilizer on account of their higher riskaversion is not doing them a service. This isanother case in which removal of the disadvan-tages of small farmers requires institutionalpolicies aimed at equalizing access to factor andproduct markets or to land rather than tech-nology policy. (For other such cases seeBinswanger and Ryan 1977).

Second, if it can be shown in other regionsthat risk attitudes vary as little across farm sizegroups as they do in SAT India, we have toreconsider in a new light the debate about riskaversion versus credit constraints as explana-tions for low fertilizer use. If all farmers in a given area underinvest in fertilizers (relative tothe risk-neutral opt imum), this underinvest-ment is more Iikely to be caused by risk aversionthan by credit constraints. The reason is that it ishard to imagine how a uniform credit constraintacross farm size groups could emerge, whereasrather uniform risk aversion has now beenshown in at least the SAT Indian case. On theother hand, if in a given area fertilizer adoptionincreases systematically and rapidly with farmsize (or other indices of asset holdings) theunderinvestment of small farmers must be

caused either by adoption cycle phenomenasuch as slower access to information and in-puts, or by credit constraints. If differentialfertilizer use persists long after completion of anadoption cycle, it is most likely caused bydifferential credit or input costs (or constraints)across farm sizes, since such differentials arenot hard to document.

Third, if it can be further demonstrated thatprogressive farmers are not much less risk-averse than other farmers, then rewards forinnovation are not rewards for superior risk-bearing ability, as Schumpeter hypothesized,but are instead returns to superior humancapital and superior ability to recognize andadjust to new opportunities and constraints. It isunlikely that some intrinsic attitude holds theless progressive ones back; it is more likely tobe the lack of human and physical capital.

The future directions of this research pro-gram are as fol lows: First, we need to quantifythe apparent underinvestment in SAT agricul-ture, at least as a rough order of magnitude.Second, strengthening of self-insurance andrisk-diffusion mechanisms through publicpolicy is required. Credit policies must bechanged to allow credit markets to play a moreimportant intertemporal and interregionalrisk-diffusion role. Further work on the effec-tiveness of public relief as insurance is alsorequired. Such schemes have usually beenadvocated on equity grounds; that they may beable to provide important efficiency benefits byinsuring farmers is probably a new insight. Cropinsurance deserves much closer study, espe-cially schemes such as area-based insurancewhich can overcome such problems as highadministrative costs. We wil l not, in the nearfuture, look at price stabilization or buffer stockschemes since such policies would primarilyaffect the lower cost irrigated zones which, inour view, deserve less priority on account oftheir lower income variability.

References

ARROW, K. J. 1970. Essays in the theory of risk-bearing.Amsterdam: Nor th Holland.

315

BARAH, B. C. 1977. The decomposi t ion of variabi l i ty ofagricultural income in the Indian semi-arid tropics.Presented at the Seventh conference of the IndianEconometric Society, 1 9 - 2 1 December 1977,Tr ivandrum, India.

BINSWANGER, H. P. 1978a. At t i tudes towards risk:experimental measurement in rural India. EconomicGrowth Center discussion paper 285, Yale Univer-sity, New Haven, Connect icut

BINSWANGER, H. P. 1978b. Att i tudes towards risk:impl icat ion for economic and psychologicaltheories of an experiment in rural India. EconomicGrowth Center discussion paper 286, Yale Univer-sity, New Haven, Connecticut.

BINSWANGER, H. P. 1978c. Risk attitudes of ruralhouseholds in semi-arid India. Economic and Politi-cal Weekly 13(25):A-49-A-62.

BINSWANGER, H. P., and RYAN, J. G. 1977. Efficiency

and equity issues in ex-ante al location of researchresources. Indian Journal of Agricul tural Economics32(3).

BORKAR, V. V., and NADKARNI, M. V. 1975. Impact of

drought on rural life. Bombay: Popular Prakashan.

CHAUDHARI, K. M, and BAPAT, M. T. 1975. A study of

impact of famine and relief measures in Gujarat andRajasthan. Agro-Economic Research Centre, SardarPatel University, Val labh Vidyanagar, Gujarat, India.(Mimeographed.)

DANDEKAR, V. M. 1976. Crop insurance in India.Economic and Political Weekly 11(26):A-61-A-80.

DILLON, J. L., SCANDIZZO, P. L. 1978. Risk att itudes of

subsistence farmers in northeastern Brazil: a sampl-ing approach. Amer ican Journal of AgriculturalEconomics 60(3):425-44.

HAZELL, P. B. R. and SCANDIZZO, P. L. 1977. Farmers'

expectations, risk aversion, and market equi l ibr iumunder risk. Amer ican Journal of Agricul turalEconomics 59(1): 2 0 4 - 9 .

JODHA, N. S. 1978a. Effectiveness of farmers' adjust-ment to risk. Economic and Political Weekly 13(25).

JODHA, N. S. 1978b. The role of credit in farmers'adjustment against risk in semi-arid tropical areas ofIndia. ICRISAT Economics Program occasionalpaper 20, Patancheru, A. P., India.

JODHA, N. S. 1975. Famine and famine policies: someempirical evidence. Economic and Political Weekly10(41).

NEWBERRY, D. M. G. 1970. Price stabil ization w i th riskyproduct ion, institute for Mathematical Studies in theSocial Sciences work ing paper 69, Stanford Univer-sity, Palo Al to, Calif, USA.

316

Risk and Uncer ta in ty in Peasant"Dec is ion - M a k i n g "

Frank Cancian*

Abstract

This paper provides a theoretical framework for studying the role of risk and uncertainty

in the farmer's "decision-making" process. The author urges that the distinction

between normative and descriptive economics, sometimes blurred, be maintained. The

patterns he traces in this paper suggest that poorer farmers would take a greater role in

technological change than they have often been accorded and that past hesitancy on the

part of farmers who are well off in local terms may be due more to rank protection than to

intransigence. The author believes that what a person does is importantly influenced by

his or her position in the community.

In most parts of the world risk and uncertaintyplay important roles in the management ofproduction by small farmers. Farmers can sel-dom be sure how much their investments ofeffort and resources wil l yield. This is particu-larly for those engaged in rainfed agriculture inareas where rainfall is not consistentlyadequate, and also for those who employ newand unfamiliar technology.

The research literature related to these situ-ations conceptualizes decision-making underrisk and uncertainty in various ways. Becausethe present paper does not fit comfortably intothe research tradit ion, I want to lay out a deliberate plan for its development. The paperwi l l :

1. distinguish between risk and uncertaintyfor the farmer making production deci-sions that involve new techniques orphysical inputs,

2. briefly characterize and discuss the use of"r isk" and "uncertainty" in economicanalysis,

3. elaborate a theory consistent with therisk/uncertainty distinction made in thepresent paper,

4. report on extensive empirical tests of thattheory,

5. draw out the policy implications of thefindings.

* University of California, Irvine, California.

As the title suggests, the paper's argumentturns on the distinction between risk and uncer-tainty. The conclusions are that it would beuseful to distinguish between "uncertainty-averse" and "risk-averse" farmers, and that thedistinction leads to different policies, depend-ing on the degree of adoption of newtechnologies and practices and the relativewealth or poverty of the farmers who areprincipal targets of policies.

Risk and Uncertainty

My consideration of risk and uncertainty beginswith an imaginary farmer about to decidewhether to adopt a new seed variety or a verycostly recommended level of fertilizer. I seeboth knowledge and ignorance as componentsof his decision-making process. On the onehand, he has knowledge, even if it is onlyprobabilistic, about factors like rainfall and theIikely yields of the trad itional seed variety undervarious rainfall levels. He may be less thanperfectly informed about yield and rainfall pat-terns, but let us assume an old, intelligentfarmer who observes widely, remembers well ,and processes rationally. On the other hand,even this farmer may be profoundly ignorant ofthe relevant characteristics of the new seedvariety or of the effects of fertilizer in therecommended amounts. Thus, if he is choosingbetween, say, traditional seed and new seed,

317

his level of knowledge/ignorance about hisalternatives varies greatly. He knows fairly wellwhat the traditional seed wil l do and he gener-ally knows nothing about what the new seedwill do.

Of course, new seed varieties and other inno-vations are usually preceded by informationcampaigns, but even so there are apt to bemany moments of the kind just described. Infact, most decision making by most people canbe characterized as involving substantial com-ponents of both knowledge and ignorance(Arrow 1971, p. 1). In ordinary discussion andanalysis we usually emphasize the knowledge,because we know what to do with it. Theignorance is harder to incorporate into ourthinking.

Given this vision of the wor ld, Frank Knight'sclassic distinction between risk and uncertaintyis useful. In 1921 Knight distinguished measur-able uncertainty or "r isk" from true, unmeasur-able, uncertainty (1971, p. 20 and Chapter VII). Insimple terms, it is "r isk" insofar as you know theprobabilities of various outcomes, "uncer-tainty" insofar as you cannot specify the prob-abilities.1 With these definitions, risk and uncer-tainty may vary independently of one another.In fact, the research described below looks at a situation where risk remains constant whileuncertainty varies. Its interpretation will beclearer if we first look at other uses of "r isk" and"uncertainty."

A Select ive V i e w of OtherUses of t he Terms

Broadly speaking, other uses of the terms risk and uncertainty are of two kinds: (1) those thatdistinguish the concepts in ways distinct f rom

1. Whi le I use Knight's dist inct ion, our purposes areradically different. Knight sought to explicate a quasi-philosophical f ramework for understandingthe or igin of profit . I wan t to predict behavior f r omsocial posit ion. Though they, appropriately g iventheir goals, l imi t themselves to individual andderived institutional characteristics, Balch and Wu(1974, pp. 4 - 5 ) give a ful ly Knight ian def ini t ion ofuncertainty. In this they depart f rom c o m m o npractice. I am not competent to evaluate theimplications of the more technical essays in thevolume their essay introduces.

318

Knight's, and (2) those that see the concepts asindistinguishable in practice.

An example of the first kind is provided byRoumasset (1977, p. 1) who says:

In modern decision theory, uncertainty is a state of mind in which the individual per-ceives alternative outcomes to a particularaction. Risk, on the other hand, has to dowith the degree of uncertainty in a givensituation.His distinction, stated in a report of the semi-

nar on "Risk and Uncertainty in AgriculturalDevelopment" held in Mexico in 1976, makesuncertainty a useful cover term to define thespace in which the discussion of the concep-tualization of risk may take place (see alsoRoumasset 1976, Chapter 2, and Roumasset, etal in press). In a definition that is consistent withRoumasset's, Berry (1976, p. 2) uses "uncer-tainty" to indicate incomplete knowledge on thepart of the actor and defines "r isk" as thepossibility of incurring a loss in the course ofproductive activity.

It seems to be more common to use the terms"risk and uncertainty" to indicate a single con-cept. This awkward usage and the failure todistinguish the concepts is largely explained, I think, by a position stated by Hirshleifer andShapiro (1977, p. 81). They note that there is a tradition that "attempts to formulate a distinc-tion between risk and uncertainty based onability to express the possible variability ofoutcomes in terms of probability distr ibution,"but reject it because: "This distinction hasproved to be sterile. Indeed we cannot in prac-tice act rationally without summarizing ourinformation (or its converse, our uncertainty) inthe form of a probability distribution (Savage1965)."

Though it is presumptuous for an outsider (ananthropologist) to tell insiders how they havelived their lives, this merging of the risk anduncertainty concepts in modern decisionanalysis is best understood, I think, through an"origin my th " about the development of nor-mative analysis in economics. As the storysuggests, I believe that a major barrier tosuccess in the study of behavior results f romneglect of the distinction between positive (de-scriptive) and normative analysis, and an unfor-tunate attempt to use the immense power ofadvances in normative analysis to model whatpeople actually do. In this story the simple

normative approach starts off wi th perfectknowledge about alternatives. From this andthe decision-maker's goals or values, the ap-propriate choice can be calculated.

Where risk is involved, that is, where thedecision-maker lacks perfect knowledge andcertainty, but does know probabilities of vari-ous alternatives, normative analysis producesan expected value solution. It uses the odds tospecify the decision that wi l l maximize goalattainment over the long run. Even thoughhe/she may lose at first, the decision-maker isadvised to toss a fair coin w i th a person who wi l lgive 51 cents for heads and take 49 cents fortails. Risk is involved, but by assuming thatoutcomes can be treated as aggregatable inde-pendent events, it is possible to make prob-abilities equivalent to the perfect informationemployed in the simple decision-makingmodel. This transformation is the first stepdown the road that leads away from simpleutility maximization under certainty.

The next step is the big one. The real problemfor normative analysis comes when thedecision-maker does not know the prob-abilities, for the apparatus of calculation de-pends on being able to fix the odds (or theirfunctional equivalents). What is to be donewhen the odds are not known? That is, what is tobe done when there is uncertainty?

L. J. Savage's (1954) work on subjectiveprobability solved the problem for normativeeconomics. Simply stated, Savage showed thatunder uncertainty the best way to achieve themaximum is to use whatever is known to makebest guesses about the probabil ities. That is, heshowed that uncertainty mitigated by even a little bit of " in format ion" in the unspecifiableintuition of an expert (or client) is distinctlybetter than no information at all in reaching thegoal of maximization (given the aggregationassumption). The subjective probability esti-mate provided by the decision-maker is basedon an act of wi l l that transforms uncertainreality into a calculable probability. With thisstep Savage effectively expelled the noncalcul-able part of uncertainty from normativeeconomic analysis.2 This made it sensible foreconomists to talk about "risk and uncertainty"as a single concept in economic analysis, for,from the subjective probability point of view,they cannot be distinguished.

In sum, the uses of the terms risk and uncer-

tainty mentioned above, and the developmentsin analysis associated with them, emphasize theprobabilistic definitions of risk. They do not callattention to the ignorance Knight labeled withthe term uncertainty. It seems to me that theprobabilistic definition of risk and uncertaintydoes not adequately capture (model) the lack ofknowledge under which most actors often op-erate. I hope that what follows will show thatactive separate conceptualization of that ignor-ance in terms of (in this case) the uncertaintyconcept wil l lead to useful empirical results thatare not routinely produced by relaxing theperfect information assumption in conventionaleconomic analysis.

Decision-makingunder Uncertainty

Theory

The adoption of innovations by small farmersis, for many reasons, a good place to startexploring descriptive analysis. Such adoption isinteresting because it offers great potential forimproving the human condition. It has thereforebeen widely studied. More specifically, it hasbeen the focus of studies of risk aversion byagricultural and development economists. And,finally, adoption typically occurs across a sub-stantial knowledge (ignorance) imbalance bet-ween what is known about the alternatives inthe decision situation, i.e., about the traditionalpractices (what was done last year) and the newpractices (what might be done next year).

Below I will:(1) elaborate the definitions ofrisk and uncertainty in terms of a specificresearch context, (2) state a theory that makes

2. The expulsion of uncertainty may betraced in textslike Kassouf's Normative Decision Making (1970).Kassouf discusses decision-making " in the ab-sence of probabil i t ies" but introduces the subjectby saying that, "There is no completely satisfa-ctory way to handle these si tuat ions" (p. 66).Though I am not technically competent to evaluatethe approaches to decision-making in the absenceof probabilities that he covers and the moresophisticated ones he does not cover, none seemsto reverse the reduction of uncertainty inherent inSavage. Like Savage, they require the decision-maker to crystallize complexit ies by an act of wi l l .

319

distinctive predictions about behavior underrisk and behavior under uncertainty, and (3)specify a test of the theory that provides anillustration of the utility of the original distinc-tion for descriptive analysis.

The classic-diffusion of innovation study doneby rural sociologists (Rogers and Shoemaker1971; Rogers 1976) provides a good opportun-ity to specify the distinction between risk anduncertainty. Suppose a new, high-yielding seedis being introduced (it could be hybrid corn inIowa in 1935 or dwarf wheat in Pakistan in 1968).Individual farmers must decide between buyingthe seed and sticking wi th local varieties foranother year at least. In this situation it is likelythat they face both risk and uncertainty.

The risk involved as the farmer faces hisdecision is in part the same risk that would beinvolved with traditional practices. Uncontroll-able factors such as future temperature andrainfall wil l affect the yield. Presumably anexperienced farmer has substantial knowledgeabout the Iikely variance in outcome and can, ineffect, calculate probabilities of various out-comes or estimate the likelihood, given hisreserves, that he wil l be pushed into a situationof severe hardship. This risk involves the pre-dictable variation in yield f rom year to year.

The decision to adopt the new seed wouldinvolve uncertainty as well as risk, for nobodyreally knows what the seed wil l yield under localconditions. There is no local experience withwhich to estimate its response to locally avail-able inputs. The probable outcomes cannot becalculated very well compared to the probableoutcomes for the traditional seed. Thus, at theoutset at least, uncertainty adds a new elementto the considerations of the decision-makersconsidering the new seed, and it is an elementthat is likely to be substantially less calculablethan the known variance of yields f rom thetraditional seed.

This distinction between risk and uncertaintysets the context for a theory that relates theadoption behavior of richer and poorer farmersto stages in the spread of an innovation. Sincethe full theory has been stated elsewhere (Can-cian 1972), here I wil l give only enoughbackground here to permit elaboration of theimplications for the understanding of risk anduncer ta in ty in descr ip t ive analys is ofdecision-making.

Stated in terms that are compatible with the

decision-making framework, the theory relieson the notion that poorer farmers are relativelyrisk-averse and uncertainty-preferring whilericher farmers are relatively uncertainty-averseand risk-preferring. The basic prediction, whichwil l be empirically specified below, comes fromthe idea that, under uncertainty, where thedecision-maker cannot predict the outcome, therich farmer has more to lose and less to gainfrom adoption (because it offers random re-sults). In the same situation the poor farmer hasmore to gain and less to lose from a randomchange. The farmers are understood to beinterested in maximizing their rank in a com-munity of relevant alters. It is assumed, realisti-cally I think, that all farmers operate in a socialsituation that is likely to save them from starv-ation resulting from crop loss. Thus, effort tomaintain rank by avoiding uncertainty has moreto do wi th the acceptability of falling back to thesocial level of support. Such a fall is likely to bemore unacceptable to higher ranking farmers.3

Under risk, as it is understood here, financialability to withstand known potential setbackswould dominate adoption behavior muchmore. Other factors, such as indivisibility ofinnovations, that favor richer farmers, contri-butetodetermining who wil l adopt when uncer-tainty is low (see Cancian 1967, 1972).

It is possible to test these prediction becausethere is a natural situation where uncertainty

3. It may be useful to think of the si tuat ion covered bythe theory using the fo l lowing image. Thedecision-makers are located in total darkness onhigh and low platforms. Al l w ish to be on thehighest possible plat form to see the sunrise; andall know that there are other platforms that may bereachable f rom theirs — though groping for themcould lead to a fal l .The conservative behavior of the higher rankpeople may be understood as " leave well enoughalone behavior." This has some parallels toSimon's satisf icing, but is made different because"leave wel l enough a lone" acknowledges thatpeople start f rom somewhere, i.e., that they haveaposit ion that is not total ly recreated w i th everydecision (action). The satisficing not ion is betteradapted to the microeconomic vision of the wor ld ,for that paradigm is not designed to take accountof the initial (relative) posi t ion of actors. It is moresuited to isolated individuals and marginalanalysis.

320

varies while risk stays fairly constant. When aninnovation is introduced to a community offarmers from outside, some farmers adopt itimmediately, and some adopt it in later years.Later adopters usually use the experience ofearly adopters to inform their decision. Thus,uncertainty is greater for the earlier adoptersthan it is for the later adopters. Risk remainsfairly constant.

This gives us a critical test: if uncertainty ismeaningfully distinguished from risk, poorfarmers should adopt more than to rich farmersin the early stages of the spread of an innova-tion. In later stages, the rich should be relativelyfaster adopters. All this theory is illustrated inFigure 1.

In the real world various special consid-erations apply to the very rich and the very poor(Cancian 1967, 1972, 1979). Thus, I haveconfined my predictions to the behavior of themiddle of the wealth continuum in agriculturalcommunities. These predictions are illustratedby the solid lines in Figure 2. In the language ofconcrete research: the Low Middle Rank ispredicted to have a higher adoption rate thanthe High Middle Rank in Stage 1 of the adoptionprocess, and it is predicted that that relationshipwil l reverse in Stage 2 of the process.4

Data and Tests

The two predictions stated above and illus-trated in Figure 2 were tested with data fromsurveys of sixteen communities done by ruralsociologists, anthropologists and agriculturaleconomists. The 16 cases include more than3000 farmers in eight countries.5 Despite the

diversity of data sources, it was possible toemploy fairly uniform variables across all thecommunities. The independent variable waseconomic rank within the community measuredby farm size or farm income; and the dependentvariable was percent (of a given rank) adoptingbefore the end of the specified stage. Details ofthe data gathering and analysis are described inCancian (1979).

Manipulation of the data was straightforwardand can be best understood by looking at Figure3, where 173 Wisconsin (USA) dairy farmersstudied by F. Fliegel are divided into approxi-mate quartiles by income. The Stage 1 side ofthe figure shows the percentage adopting inStage 1 by rank, where Stage 1 includes the firstapproximate 25% of the overall population toadopt. Stage 2 is defined as the next approxi-mate 25% to adopt, and the Stage 2 side of thefigure shows the cumulative figures after halfthe population has adopted.

Some of the data sets collected permittedusing a 20% band of percentile rankto calculatea moving average rate of adoption. The resultsof this calculation for 105 Pakistani wheat far-mers studied by R. Rochin are displayed inFigure 4.

Both predictions are clearly confirmed by theselected data sets displayed in Figures 3 and 4.In the larger study, 14 of the 16 communitiesconfirmed the first prediction, that is, theyshowed a higher Stage 1 adoption rate for theLow Middle Rank than for the High MiddleRank; and 12 of the 16 studies confirmed thesecond prediction. Despite the complicationsand doubts that are inevitable in a comparativestudy of this scope, the principal predictions are

4. Note that whi le this theory sees poor farmers ashaving a comparative advantage in uncertaintybearing, it does not necessarily predict any empir i -cal situation in which poor farmers adopt morethan rich farmers. Rather it predicts only that,compared w i th rich farmers, poor farmers wi l ladopt relatively more the higher the uncertainty.This complicat ion is discussed further in Cancian1979.

5. Of the 16 cases, one each came f rom India, Japan,

and Kenya, two each f rom Mexico, the Philippines,Pakistan, and Taiwan, and five f rom the UnitedStates. Full documentation on the data sources isgiven in Cancian (1979). Here I want to thank the

original investigators who gave permission to usetheir data and others who helped me collect andanalyze the data sets, especially Joseph Alao, SusanAlmy, Randolph Barker, Frederick Fliegel, JohnGartrell, Peter Gore, Barbara Grandin, Herbert Lion-berger, Max Lowdermilk, G. Parthasarathy, RobertPoison, Refugio Rochin, Everett Rogers, AliceSaltzman, and Eugene Wilkening. Each of the casesincludes a reasonable approximation to a randomsample or a complete census of a community offarmers. Sample size ranges from 91 to 540. In thecommunities outside the United States, the mediumsize of holding for the ful l t ime farmers included inthe analysis was in no case more than 2.5 ha.

321

KEY

WEALTH = Rank on size of farm orsize of farm income

ADOPTION = Percent adopting before endof specif ied stage

Early Stages

UNCERTAINTY HIGHrelative to risk

Later Stages

UNCERTAINTY LOWrelative to risk

F i g u r e 1 . T h e o r y

Stage 1

UNCERTAINTY HIGHrelative to risk

Stage 2

UNCERTAINTY LOWrelative to risk

F i g u r e 2 . P r e d i c t i o n s

322

ADOPTIONRATE

high

low middleWEALTH

high

Stage 1 Unce r ta in t y Higher Stage 2 Uncer ta in ty Lower

F i g u r e 3 . H i s t o g r a m t e s t

1 7 3 W i s c o n s i n ( U S A ) d a i r y f a r m e r s

F i g u r e 4 . M o v i n g a v e r a g e t e s t

1 0 5 P a k i s t a n i w h e a t f a r m e r s

N o t e s : S t a g e s 1 a n d 2 a r e b e s t a p p r o x i m a t i o n s t o t h e f i r s t 25%a n d t h e n e x t 25% t o a d o p t . I n F i g u r e 3 t h e a c t u a l s t a g e sa r e 2 3 a n d 2 7 % , i n F i g u r e 4 t h e y a r e 3 0 a n d 40%.

I n F i g u r e 3 t h e r a n k s a r e t h e b e s t a p p r o x i m a t i o n s o fq u a r t i l e s . F i g u r e 4 u s e s a 20% b a n d o f p e r c e n t i l e r a n kt o c a l c u l a t e t h e m o v i n g a v e r a g e .

323

strongly confirmed. It is clear that, early in theprocess of spread of an innovation, farmers oflower middle rank in a community are more aptto adopt than farmers of upper middle rank; andthat the process usually reverses itself in thelater stages of the process.

Policy Implications and Discussion

The findings reported above have two majorpolicy implications. One is straightforward; theother is complex and intertwined with the ques-tions about risk and uncertainty raised at thebeginning of this paper.

The straight forward one stems directly fromthe finding that, in the early stages of the spreadof an innovation, the " low middle" rank farmeris more apt to adopt than the "high middle"rank farmer. This finding contradicts the re-ceived wisdom that "Wealth and innovative-ness appear to go hand-in-hand" (Rogers andShoemaker 1971, p. 187), and suggests thatprogram designers must take into account theapparently strong innovative inclinations ofpoor farmers. This is especially important whenplanners and technicians are matching innova-tions to the financial resources of the targetpopulation. Under the received wisdom, aninnovation that is adopted by the rich butrejected by the "upper middles" — even whenit is made economic for them — would properlybe abandoned as inappropriate for the poormajority on the grounds that they are lessinnovative than the "upper middles." Thefindings reported above show that the " lowermiddles" are not less innovative than the "up -per middles." They imply that redesign to thescale of the lower middle rank farmer mightlead to a substantial further surge of adoption.In general, programs intended to help the smallfarmer should not be tested for viability on theupper middle ranks of the community. This is animportant caution, because, in the real wor ld ,these upper middle rank farmers frequentlyappear to be the poorest among local people forwhom there is much hope within the existingsocioeconomic framework.

The second area of policy implications in-volves what I have labeled "uncertainty aver-s ion" in approximate parallel to current usageof "risk aversion." This usage of uncertaintyaversion is meant to call attention to the ignor-ance that is crucial to decision situations, like

the spread of innovations, where even know-ledge of probabilities is in relatively short sup-ply. The parallel to risk aversion cannot beexact, of course, because risk aversion is com-monly defined in terms of elicited or imputedchoice among bets with known odds andpayoffs. The whole point with uncertainty aver-sion is that it refers to situations in which verylittle secure information about outcomes isavailable.

While uncertainty aversion is fundamentallymeaningless — or at least totally unanaly-zable — with regard to individual decision-makers, it becomes fully interpretable withregard to decision-makers operating in a com-munity where they seek to maximize their ownstanding. In this situation where relative posi-tion is the goal, uncertainty has clear implica-tions for decision-makers who are lower andhigher in the existing hierarchy. These implic-ations help explain the observed patterns oflower middle rank innovativeness and uppermiddle rank conservatism under conditions ofhigh uncertainty that are reported in this paper.

The implications for policy derived from thispicture of the world are seemingly perverse, forthey might be stated as fol lows: planners whowish to give an advantage to the small farmerover the large farmer should promote maximalignorance in the countryside. Such a promotionis inconsistent with the nature of ordered plan-ning and is unlikely to occur, but the lessperverse implications remain and are practicalwithin established institutional arrangements.Knowledge is less advantageous to the smallerfarmer, for he is less concerned with securelymaintaining the status quo and more concernedwith changing the current situation. Largerfarmers are more interested in trading off pro-duction for security and therefore less in-terested in innovation, i.e., they are uncertaintyaverse.

G iven the choice between allocating researchresources to develop better small-scaletechnology and educating farmers to use whatis available, the planner who aims to help thesmall farmer should probably consider theformer alternative a little more favorably thanhe/she did in the past.

324

References

ARROW, K.J. 1971. Essays in the theory of risk-bearing.Chicago: Markham.

BALCH, M. and Wu, S. 1974. Some introductory re-marks on behavior under uncertainty. In Essays oneconomic behavior under uncertainty, eds. M. S.Balch, D. L. McFadden, and S. Y. Wu . Amsterdam:North Holland.

BERRY, S. 1976. Risk and the poor farmer. Prepared forEconomic and Sector Planning, Technical Assis-tance Bureau, USAID, Washington, D. C.

CANCIAN, F. 1967. Stratif ication and risk-taking: A theory tested on agricultural innovation. AmericanSociological Review 32:912-927.

CANCIAN, F. 1972. Change and uncertainty in a peasanteconomy: The Maya corn farmers of Zinacantan.Palo Al to, Calif, USA: Stanford University Press.

CANCIAN, F. 1979. The innovator 's s i tuat ion: Uppermiddle class conservatism in agricultural com-munit ies. Palo Al to , Calif, USA: Stanford UniversityPress (expected about May).

HIRSHLEIFER, J. and SHAPIRO, D. L. 1977. The treatment

of risk and uncertainty. In Public expenditure andpolicy analysis, 2nd ed, eds. Robert H. Haveman andJul ius Margolis. Chicago: Rand McNally.

KASSOUF, SHEEN. 1970. Normative decision making.

Englewood Cliffs, N. J . , USA: Prentice-Hall.

KNIGHT, FRANK. 1971. Risk, uncertainty, and profit .

Chicago: University of Chicago Press.

ROGERS, E. M. 1976. New product adopt ion anddif fusion. Journal of Consumer Research 2 : 2 9 0 -300.

ROGERS, E. M. (with F. FLOYD SHOEMAKER). 1971. Com-

municat ion of innovations: A cross-cultural ap-proach. 2nd ed. Glencoe, NY, USA: Free Press ofGlencoe.

ROUMASSET, J. A. 1976. Rice and risk: decision makingamong low- income farmers. Amsterdam: North-Holland.

ROUMASSET, J. A. 1977. Riskand uncertainty in agricul-tural development. New York: Agricultural De-velopment Council Seminar Report 15.

ROUMASSET, J. A., BOUSSARD, J. M., and S INGH, I. J.

(eds). Risk, uncertainty and agricultural develop-ment. Berkely, Calif, USA: University of CaliforniaPress. In press.

SAVAGE, L. J. 1954. The foundations of statistics. NewYork: Wiley.

325

Discussant 's Commen ts

J. L. D i l l on*

Over the last decade the nature and signifi-cance of risk to farmers' decision-making hascome to be much better appreciated. Many is-sues, however, remain unresolved. Both howfar our understanding has been increased andthe ignorance we still face is well illustrated bythe three papers presented in this workshopsession. In opening discussion on these threethought-provoking papers, I would emphasizethe two primary aspects of decisions under risk(i.e., decisions where the outcome is not sure).These two aspects are, first, the attitude to riskof the decision maker and, second, his percep-tion of the risk that he faces. I should also notethat I am presently persuaded that both theseaspects are quantifiable — risk attitude throughelicitation of a utility function for the decisionmaker and risk perception through specification(either explicitly or implicitly) of the decisionmaker's subjective probability distributions forthe risky alternatives which he faces. Overall, I would say Anderson's paper adequately recog-nizes both risk attitude and perception but wi thsome bias to the former while Binswanger et al.pay virtually no attention to risk perception. Thethird paper, that of Cancian, stresses risk per-ception in both a provocative and — to mymind — confused way.

Let me now comment on the papers individu-ally. With Anderson I have no disagreement.The hypotheses he proposes about risk at-titudes and risk perception are, I think, reason-able and certainly deserve some investigation.However, in so far as they represent a refine-ment rather than a transformation of existingknowledge, I would be loathe to see too muchemphasis on them in our research portfolio. Atthis stage what we need in the risky decisionarea are more Binswanger-type estimates onthe distribution of risk attitudes in a variety ofphysical, social and cultural environments to-gether with systematized information (we havevirtually none at present) on farmers' risk per-

* Un i ve rs i t y o f N e w E n g l a n d , A r m i d a l e ,Austral ia.

N S W ,

ceptions in their current environment and underpossible changes to that environment. It mightalso be noted that Anderson poses his hypoth-eses in terms of the farm family. As I know hewell knows, group utility considerations—evenfor such an integrated group as the family — present a qualitative jump in difficulty of elicita-tion as compared to research on individual de-cision makers. Nonetheless, I like his recogni-tion of the family and the roles of spouses.

As with Anderson's paper, I have little to dis-agree with in the paper by Binswanger et al.They well substantiate their conclusions for theIndian SAT of high income risk derived mainlyfrom production risk, of virtually all their samplefarmers being moderately risk averse, and ofrisk and risk aversion leading to underinvest-ment. To these conclusions I would add anotherwhich is of methodological importance. This isthat from the risk attitude elicitation work usingreal payoffs which they report, it is apparentthat mind experiments without some means ofstandardization appear to be a most unreliablemeans of eliciting risk attitudes. I have only onecriticism of the Binswanger et al. paper, In theird iscussion of riskiness of agriculture in the SAT,they emphasize only frequency-based mea-sures. From a policy orientation, this has somelogic. However, in the farmer's decision mak-ing, it is his personal perception of the risk hefaces here and now that is important, not thefrequencies given by the historical trace. Be-cause of this, farmers having the same risk at-t i tude, through differences in their risk percep-tions, may choose differently f rom among thesame alternatives — as is implicitly recognizedby Binswanger et al. in their penultimate parag-raph. Thus I would urge that they also investi-gate risk perception among their village sam-ples. A g o o d s ta r t ing po in t w o u l d beAnderson's Hypothesis 4. Such work would befacilitated by the knowledge already gained ofthe risk attitude of their sample members.

I turn now to Cancian's paper. To an avowedsubjective expected utility decision theorist likeme, his paper is highly provocative. One doesn't

326

like to see one's favorite theory dismissed. So I disagree with the thrust of his paper, but I hopeon logical ground. The essence of our differenceis my belief, and that of other subjective ex-pected utility theorists, that the preferences ofdecis ion makers who make non- randomchoices (i.e. most people) can be expressed in a utility functi.on and that for any choice theymake from an array of risky alternatives, thischoice corresponds to an implicit subjectiveprobability distribution for the possible out-comes. Further, as proved by Savage, prefer-ence and probability are sufficient to modelchoice. Cancian doesn't accept this. He arguesthat as well as preference and probability, weshould add a further consideration of degree ofignorance or information corresponding to ourdegree of belief in the probabilities used. Statedthis way — i.e., in terms of a probability of a probabi l i ty— his position is untenable since a probability of a probability is just a probability.In short, I believe this proposed theory — whichwe might note is somewhat akin to Shackle'stheory of focus-loss and potential surprise andFellner's slanted probabi l i ty approach — iswrong because it assumes that the decision-maker's state of information about a risky alter-native cannot be fully described by his (if needbe implicit) subjective probability distributionfor the unsure outcomes associated with thatalternative. To this, Cancian might respond thatI'm talking normative while his concern is de-scriptive. I don't agree. To me, nonrandomchoice must imp ly ut i l i ty max im i -zation. Subjective expected utility theory withits possibil i t ies of mult iat t r ibute (includinglexicographic and status) considerations is richenough to encompass, I believe, all modes ofnonrandom choice, including safety-first typeprocedures. Whether expressed explicitly orimplicitly via intuitive mechanisms, degrees ofpreference (i.e. utility) and degrees of belief (i.e.subjective probability) are sufficient to describebehavior.

What then of the adoption patterns shown byCancian's empirical analysis? Could subjectiveexpected utility theory explain his result that"early in the process of spread of an innovation,farmers of lower middle rank are more apt toadopt than farmers of upper middle rank; andthat the process usually reverses itself in thelater stages of the process?" The answer is yes.Such a result would occur if lower middle rank

farmers initially saw greater expected utility inadoption than did the higher middle rank far-mers in the same community. This could occureither if lower rank farmers had more optimisticrisk perceptions or were less risk averse — bothof which are reasonable hypotheses though, sofar as I know, are not yet adequately tested.Since it is a potentially far simpler explanation, I prefer this hypothesis to Cancian's more com-pl icated approach — not least because it avoidssuch perverse policy implications as suggestingthat small farmers would be advantaged by thepromotion of maximal ignorance.

All this may sound like a rejection of an-thropological and sociological considerations.It is not. Appraisal of prospective technologymust involve both financial and social consid-erations, i.e. be socioeconomic. I trust that theexample of ICRISAT in this regard wil l be fol-lowed elsewhere.

Editors' Note:

Due to Dr. D i l l on ' s s u b s t a n t i v e c r i t i c i s m s o f

Dr. Canc ian 's paper , a n d in in te res ts o f f u r t h e r

in te rd i sc ip l i na ry d i a l o g u e o n th i s i m p o r t a n t

t op i c , Dr. Canc ian 's r e s p o n s e to Dr. D i l l o n is

a lso p resen ted here .

Response to Discussion— F. Cancian

I appreciate Dr. Dillon's attention and his clearstatement. I want to thank him for giving me anadvance copy of his remarks. His kindness hasgiven me a chance to collect my thoughtsaround two points.

First, please note that the two policy implica-tions of my paper sta nd in very d ifferent relationto the theoretical disagreement.

The idea that the lower middle rank is a usu-ally untapped source of innovative inclinationsis not affected by the disagreement. It stemsquite directly from the finding that the lowermiddle rank is more innovative than the uppermiddle rank even when no special attention ispaid to its economic limitations. The implica-tions of this finding stand, despite the con-troversy about my theory. I do not want theseimplications to be lost in the theoretical shuffle.

Things are different for the second implica-

327

t ion. The idea that small farmers are betteroff in comparison wi th large farmers underconditions of uncertainty, and the negative im-plications of this for traditional extention andeducation programs, is probably too mired inthe theoretical controversy to save. So I leave itto a later t ime.

My second topic is a comparison of expectedutility theory and my theory about behaviorunder uncertainty.

Expected util ity theory is immensely power-ful as a normative theory. It provides a powerfulway to use information to calculate future deci-sions given stated goals. However, as a positivetheory, i.e. as a theory that helps us understandbehavior, it has mostly this success as norma-tive theory going for it. Dr. Dillon himself haswritten about the limited success of expectedutility theory as an aid to prediction in the realwor ld.

Expected utility theory interpreted as a posi-tive theory has two major problems. First, it isso flexible, so all-encompassing, that it is fun-damentally not testable. Second, it is, of course,individual-oriented (microeconomic) and assuch it provides litt le guidance in assessing therole of social relations in influencing behavior.

I want to suggest that, for purposes of under-standing behavior I may have a viable alterna-tive to expected utility theory. But I do not want,in this setting, to be contentious. My theory iscompatible wi th Dr. Dillon's. We could say that I am concerned wi th specifying the risk prefer-ences of various ranks of farmers at the begin-ning of the adoption process. Thus stated, mytheory is complementary to his. I am happy tostate it in this manner if it makes it easier for himand the many who share his viewpoint to acceptit and its claims for the role of social relations(specifically economic rank) in determiningadoption behavior.

I feel compelled to say, however, that as I seeit, it is possibleto make my small addit ion to hislarge theory because he has an analyt icalframework, not a theory. It is an analyticalframework because it states that people wil l actaccording to risk preferences and subjectiveprobability estimates and leaves the content ofthese categories unspecified. Anything, includ-ing my substantive propositions, wi l l fit.

As a social anthropologist, I f ind my originalstatement simpler and better but recognize thateconomists may see the world differently.

Finally, I want to call your attention to thesubstantive propositions about social relations.They provide an example of the importance ofsocial structure to change in production sys-tems. Social inequality defined in terms of thelocal community is an important determ inant ofadoption behavior. Technology developmentshould be responsive to those facts about thereal wor ld.

328

Chairman's Summary

J . G. Ryan*

As the papers presented in this session havealready been summarized by Dr. Dillon thissummary will cover only the major issues andconclusions which arose out of whole discus-sion. These are discussed under the headingssuggested to the chairpeople by the organizers.

Policy and Technology Design

At least for the areas studied in SAT peninsularIndia there seem to be clear policy implicationsarising from the work done on risk attitudes offarmers. There would appear to be little reasonto design technologies wi th basically differentrisk characteristics for small and large farmers.Contrarily from extensive cross-country empiri-cal evidence of adoption of innovations by far-mers, it appears that in the early adoptionperiod when information is lacking on new tech-nologies— the so-called 'uncertain' phase — small farmers adopt more readily than uppermiddle- level farmers. The pol icy impl ica-tion of these cross-country results, though con-tentious, suggest that ignorance is bliss forsmall farmers and that hence a more bewilder-ing array and larger supply of prospectivetechnologies would be a preferable course tomore widespread education and extension inorder to enhance adoption by small farmers andhence to achieve equity goals. It would be aninventor's dream if indeed this were true, andcertainly make life much easier in agriculturalresearch. In fairness it must be said this viewwas challenged and again counterchallenged.

In the SAT Indian regions, it appears thatwhere irrigation is much less dominant, produc-t ion variabi l i ty dominates price variabi l i ty,whereas the opposite occurs in the highly irri-gated regions. Technological innovations in thenonirrigated areas would thus appear to offersignificant potential for reductipn of overall risk,as measured by variability, a concept that wasquestioned in the discussion.

The role of public relief in alleviating adverseeffects in drought-prone areas of India wasfound to be critically important, both in terms ofefficiency and equity considerations.

Future Research

The session seemed to conclude that we requirea far better methodological and empirical basison which to make judgments related to the im-portance of risk and uncertainty in the semi-aridt ropics. Bernoul l ian decis ion theory wassuggested as the foundation upon which tobuild the empirical framework of subjective riskperceptions. Explicit recognition of the role thatsubjective risk perceptions and probabilitiesplay in the decision-making process was some-thing that seemed to be lacking in many studiesof risk attitudes and responses of decision-makers.

There was no clear consensus on the role ofinstitutions such as universities and ICRISAT infurther refining methodological approaches torisk attitudes and perceptions. More research isclearly indicated on the role that social structureand the farm family play in the process of for-mation of risk attitudes and perceptions, andtheir effects on the adopt ion of newtechnologies. There would seem to be somecontradictions at present between approachesto technology design and adoption that em-phasize the role that social structural charac-teristics play, and those that examine and mea-sure risk attitudes across socioeconomic strata.I believe both approaches have a lot to offereach other, as was the tenor of the session. Thedialogue has been initiated on this critically re-levant factor in the technology design process,and it is my fervent hope that it fruitfully con-tinues.

There was general agreement that the basicmethodology of utilizing real monetary choicesin eliciting attitudes of farmers to risk is muchpreferred to the tradit ional interview tech-niques. The need now seems to be to employ

329

* Economics Program, ICRISAT.

such techniques in other SAT regions to see towhat extent the conclusions arising from thestudies of farmers' risk attitudes in SAT penin-sular India hold true elsewhere. Further studiesshould also embrace questions of the role of thefamily and its components in the formation ofrisk attitudes. The policy implications that f lowfrom such work both for design of technologyand infrastructural support systems are obvi-ous. It would seem that here is a prime exampleof where useful collaboration between socialscientists f rom national programs and ICRISATcould ensue.

Further research on the way in which publicrelief programs can be made more effective asrisk diffusing devices in the SAT was clearlyindicated.

330

Chapter 8

Rural Labor Markets

Rural Labor Use and Deve lopment Strategiesin East Af r ica and India

Janet Benson*

Abstract

This paper discusses the existing division of labor in East Africa and India and its

implications for the design of new agricultural technology. Primary emphasis has been

placed on smallholder cultivation rather than large-scale production, and the author has

attempted to focus on labor use in dryland cropping systems whenever possible.

However, since useful data on such subjects as seasonality and time allocation

(Particularly by sex and age) are not often available, information from other geographic

regions is introduced where relevant. Methodological problems are discussed.

This paper discusses the existing division oflabor in East Africa and India and its implica-tions for the design of new agricultural technol-ogy. "Division of labor" refers to the fact thattasks are differentially allocated within anysociety so that specialization is possible; that is,no single member need be responsible forcarrying out all activities necessary to social Iife.The most fundamental bases of allocation aresex and age. While some of the simplestsocieties (e.g. gatherers-hunters) are charac-terized by a division of labor affected mainly bythese two criteria, complex social orders suchas preindustrial agrarian states or industrialsocieties exhibit much more elaborate occupa-tional structures. Even in these societies, how-ever, task-allocation is related to sex and age.Some anthropologists now believe that divisionof labor within the family, which includes food-sharing, was fundamental to the developmentof human society and remains an importantdistinction between man and other primates. I wi l l first examine the sex variable and then age.

Divis ion of Labor: Sex

All known human cultures maintain conven-tions concerning sex-specific activities, thoughconsiderable variation occurs in the rigidity of

* Department of Sociology, Anthropology and SocialWork, Kansas State University, Manhattan, Kansas.

such task-allocation. Certain near-universal pat-terns seem clearly related to reproduct ion;since women in preindustrial societies spendmuch of their adult lives in a pregnant or lactat-ing condition, their physical mobility is morerestricted than that of men, and child care isdefined as a female responsibility.1 As a matterof convenience, work in and around the homesuch as cooking, firewood and water collection,and kitchen gardening is commonly assigned towomen, while men are allocated tasks that re-quire more physical mobility, risk, and sheerstrength. Beyond this, much task-specificationis based more on arbitrary convention than onany biological base; that is, there is often nocross-cultural consistency concerning whichsex performs a particular task. Given a certainnumber of activities that must be carried out forsurvival and reproduction, sexual specificationestablishes clear responsibilities for males andfemales in a given culture or ethnic group andfacil i tates household product ion, an issuefurther discussed below. The economic roles ofmen and women are typically complementaryrather than competitive, and childhood sociali-zation prepares the sexes for eventual partner-ship in the formation of a new household andfamily (Benson 1978).

I t shou ld be stressed that convent ionsconcerning sex-typed tasks are by no meansimmutable; even in societies that clearly distin-

1. See Brown 1970 for a fuller discussion.

333

guish "ma le " f rom " female" activities, loss orabsence of a spouse occasionally results incompromises with ideal behavior. Where thehousehold labor pool is small as in nuclearfamil ies, husband and wi fe may frequentlycooperate in the same tasks or undertake thoseof the opposite sex. Given sufficient incentives,the society-wide model of sex-typed activitieswil l also change.

Within agrarian societies of specified types,this complementarity results in different pat-terns of labor allocation. Boserup (1970), forexample, differentiates between two patterns ofcultivation which she refers to as "ma le " and" female" farming systems. The former charac-terize Asian societies, while the latter are more typical of African ones. "Ma le " systems arethose in which male labor is predominant, while" female" systems rely primarily on women'slabor.

As a number of writers have observed (seeMartin and Voorhies 1975), female farming sys-tems are normally associated with hoe cultiva-tion (horticulture) rather than " t r ue " agricul-ture. The latter, which is characterized by use ofthe plow and draft animals as well as otherintensive cultivation techniques, relies moreheavily on male labor since plowing is almostuniversally defined as a male task. Horticulturalproductivity varies greatly (Wolf 1966, pp. 2 1 -25), but under conditions of moderate popula-t ion pressure can provide an adequate livingwith relatively little male input. In the Africancontext, this has meant that men have been ableto migrate in search of wage labor, leavingwives on the land to produce subsistence crops.Watson (1958), for instance, discusses this situ-ation among the Mambwe of central Africa dur-ing the 1950s; their cultivation system requiredonly a few men for brush-clearing, wi th otheragricultural tasks handled by women. AmongEast Afr ican cul t ivators, women of ten hadprimary responsibility for subsistence crops(millet, maize, plantains), whi le men tended cat-tle and spent their t ime on politics and warfare;wi th the introduction of cash crops (tea, coffee,cotton, pyrethrum) during the twentieth cen-tury, however, men have become interested incultivating the new and profitable items. Cleave(1974), in a study of smallholder agriculture inAfrica notes that: " I t is a consistent feature ofsurvey reports and other sources that the intro-duction of a new cash crop or a new technique

tends increasingly to involve men in agricul-tural operations, frequently leading to reduc-tions in the traditional division of labor on sexlines".

Des ignat ing Asian f a rm ing sys tems as"male , " by contrast to most African farmingsystems, tends to obscure the contribution offemale labor to intensive agriculture. Accurateevaluation of female labor input is hampered bythe tendency of household heads to under-repor t w o m e n ' s con t r i bu t i on , and by thedifficulties male researchers face in interview-ing women. Sinha (1977), for example, notesthat 54% of all men and 23% of women wereenumerated in the work force in the 1951 Cen-sus of India; by 1971, due to def in i t ionalchanges, this proportion had dropped to 52.5%for men, only 12% for women. He suggests

that:A close look at the concepts and definitionsof workers suggests the possib i l i ty ofunder-enumeration in 1951 as well as 1971.In 1951, the application of the income crite-rion led to the omission of "unpaid familyworkers." The 1971 census left the deter-mination of working status to the subjectivejudgment of the respondents and, sincenon-working status of women is a mark ofsocial superiority among the lower stratawho constitute the bulk of the population, itled to a serious under-enumeration biasand a landslide in the number of femaleworkers f rom 59 mill ion in 1961 to 31 mil l-

ion in 1971.

In India, the participation of males between 15and 55 appears to be very high, over 90%; thedirect economic contribution of women, how-ever, is considerably lower on the whole andvaries widely f rom region to region (Dasgupta1977, p. 23; Gulati, 1975). The variation is prob-ably genuine although, as indicated above,under-remuneration presents a problem. Re-cent research by Ryan et al. (1979) documentsthe importance of female labor in semi-aridregions of India.

Female labor is vitally important in India, as inmany parts of Asia, since women (together wi thchildren and the aged) provide a seasonal laborpool to handle bottlenecks in cultivation. Ag-ricultural operations are often sex-specific. Inthe Telangana region of Andhra Pradesh, forinstance, men plow and prepare fields for bothwet and dryland crops; women help sow, do the

334

transplanting and weeding of paddy seedlings,cut paddy and sugarcane, and assist w i ththreshing. Often a series of agricultural ac-tivities fol low each other in rapid succession,and delay means lowered productivity or a lateharvest. Consequently, women's input, either inthe form of unpaid family labor or hired labor, isof crucial signif icance at key periods eventhough participation rates in terms of woman-days per acre per year may be low.

Women are also largely responsible for pro-cessing of grain, a lengthy undertaking that in-volves winnowing, cleaning, drying, and finallyhusk ing and g r i nd ing (either by hand ormachine). Women of merchant householdsclean and bulk the grain traded to them for re-sale in wholesale markets elsewhere. Even inhouseholds where women do not actively par-ticipate in field labor, then, they are involved inimportant agriculture-related activities. In gen-eral, it can be said that the greater the laborrequirements of the farming system, the greaterthe input by both men and women (Martin andVoorhies 1975) though this may be in the formof wage rather than family labor, distinguishingthose whose women perform manual laborfrom those whose women do not.

Division of Labor: Age

Routine and seasonal labor requirements inpaleotechnic agrarian economies also requireinputs f rom chi ldren and the elder ly, twosources often overlooked by members of in-dustrial societies where mass education andearly retirement are norms. Although statisticaldata on work inputs by children are usually lack-ing, numerous anthropological studies notetheir direct or indirect contribution to produc-t ion. While some students, considering only di-rect ly p roduc t i ve ac t iv i t ies , deva lue theeconomic importance of children in peasantsocieties (Mueller 1977), others (Nag et al. 1978)stress their contr ibut ion; this perspective isbased on the recognition that peasant societiesare characteriszed by occupational multiplicity,and the majority of working t ime is spent bymost people in nonagricultural occupations. Incomparing work inputs by age-sex groups in a Javanese and a Nepalese village, Nag et al.found that "Javanese boys and girls of 15-19years spend as much as 7.9 and 10.2 hours per

day in all work activities, while the correspond-ing figures for the Nepalese village are 9.5 and11.3 hours" (1978, pp. 294-295). In both vi l-lages, the average input of the girls aged 12-14years in all work is nearly the same as that formales 15 years and over, while the averageinput of 15- to 19-year-old girls exceeds that ofmales of the same age. In directly productivework, however, inputs of all children are lowerthan those of males 15 years and older. The totalrange of activities included animal care, foodpreparat ion , ch i ld care, other househo ldmaintenance work, reciprocal labor exchange,handicrafts, irrigated rice and garden cultiva-tion, wage labor, and trading.

Children in nonwestern societies are trainedin sex-specific tasks as soon as they have theability to manage them; a child as young as 5 or6 is often actively contributing to householdmaintenance or animal care (cf Epstein, 1962, p.70). Responsibility training begins earliest forgirls, who are most commonly taught to tendyounger siblings and perform domestic tasks.Both boys and girls may be assigned animal-tending duties. Nag et al. found that children inboth the Nepalese and Javanese villages spentmore time herding than in any other activity,and more than adults; between 4 and 5 hoursper day for boys and gir ls of 9 -11 in theNepalese village where cattle, chickens, goats,and sheep are the important domestic animals.Epstein (1962, p. 70) reports from South Indiathat boys begin work by tending bullocks, girlsby caring for buffaloes, milking, and watchingchickens. Even very young children may thusindirectly contribute to production by freeingthe t ime of adult women and men for cultiva-tion. By the age of 10 to 15, Indian girls and boysmay be earning wages as domestic servants orcasual laborers, and can provide a useful inputinto family agriculture.

Although children's labor is economicallymost important to medium- and low-incomehouseholds, where education is not a feasiblealternative to this use of t ime, school atten-dance need not result in the complete loss ofchildrens' work if chores can be done afterschool or if vacations are t imed to meet sea-sonal peaks in agriculture. One of the consequ-ences of increased education is a tendency tosend males to school and retain female childrento cope with the extra work. Many peasant cul-t ivators, however, are sparing w i th educa-

335

t ion — even if they can afford it it — becausethey fear losing the labor potential of sons whomay feel that manual labor is beneath them.

In addition to children, the elderly may alsomake signif icant economic contr ibut ions tohousehold economies. Data from 130 villagescollected by Indian Agro-Economic ResearchCentres (AERC indicate that 46.5% of those over59 belong to the workforce; there is little varia-tion by region (Dasgupta 1977, pp 46-47) . Inaddition to cultivation activities, the active el-derly attend to many of the same duties asyounger adults: tending children, food prepara-t ion , grain processing, handicrafts such ascarpentry or rope-making, and crop-watching.Unless an individual is seriously disabled orsenile he or she can almost always contribute insome way to household maintenance or pro-duction; this is particularly true for women,whose sphere of activity does not noticeablynarrow. In fact, as women reach menopausethey frequently have fewer domestic respon-sibilities and more freedom to travel, while cul-tural restrictions on heterosexual interactionalso lessen at this point. Occupations such astrading, then, are taken up by mature women inmany peasant societies (Chinas, 1973; Mintz,1967) and offer an alternative source of incometo agricultural (family or wage) labor. In EastAfrica and South Asia such marketing normallyconsists of petty vegetable-vending due tow o m e n ' s lack of access to capi ta l andspecialized skills.

Div is ion of Labor:Class and Caste

In addition to sex and age, other characteristicsof social organization affect labor use. As men-tioned earlier, many horticultural societies haverelatively simple occupational structures wi thmost tasks assigned on the basis of age-sex categories. As far as cultivation is con-cerned, not unti l the twentieth century — andeven then, most strikingly in cases of extensivecommercial izat ion (cotton plantat ions inUganda, cocoa in Ghana) — has dependenceon extra-familial labor emerged. Among hor-ticultural groups in East Africa, the most impor-tant inputs are still f rom family and reciprocallabor exchange rather than hired labor. This isconsistent wi th a rather egalitarian social struc-

ture among early stateless societies and theabsence — unt i l recently at least — of trueeconomic classes, even in pr imi t ive states(Falters 1964, p 163).

In contrast to the egalitarian social organiza-t ion of many East African societies, South Asiais characterized by marked social inequalitybased on unequal access to basic resources,namely land. Prior to the British colonization,land might be seized by a conquering warriorlineage, given as a gift to Brahmins, or settledby a kin-group in a frontier area; members ofartisan and service castes would be attracted tothe new village to serve the dominant group(see Bailey, 1957). Members of the lowestcastes formed a permanent labor pool forpaternalistic upper-caste farming householdsto which they were attached. Although therelationships between households of differentcastes had ritual aspects, the essence of thejajmani or baluta system was the formation ofproduction teams in which grain was ex-changed for goods and services.2 Barth (1960)discusses a similar system in Swat, Pakistan,which also operated until quite recently in a nonmonetary subsistence economy. In areas where artisans had land, their craft worksupplemented agricultural income and mit i-gated the seasonal ity of agriculture. With grow-ing commercialization of the rural economy, thejajmani system has been weakened or elimi-nated in many areas and contractual relationssubstituted for paternalistic ones (A. N.Michie 1976, Patrons or brokers: clientelismand the commercialization of argiculture, un-published paper.)

Another consequence of social stratificationfor labor is related to the economic contributionof women. Among both Indian Hindus andMuslims in rural areas, high status is indicatedby the ability of a household head to excuse hiswomen f rom labor outside the home, in particu-lar wage labor for other households (seeBenson 1978). An increase in household wealthis thus likely to be fol lowed by withdrawal ofwomen f rom the family labor pool and thehiring of servants and casual laborers to replacethem. The same process can be seen in Africansocieties where migrant labor is extensivelyused for cash-cropping.

2. See Benson (1976).

336

Peasant Product ionand the Household :Impl icat ionsfor Deve lopmen t

As suggested above in discussions of taskallocation and labor input by sex and age, thehousehold constitutes the basic productive unitin peasant societies; its members play com-plementary roles in household maintenance,cultivation, and other work. Smallholders relyheavily on household members, supplementedby relatives and perhaps labor exchange, toform the necessary production team. One im-portant difference between social organizationin Africa and South Asia is the prevalence ofpolygyny in the former region. Because womenhave traditionally been given the major respon-sibility for cultivation throughout much of Af-rica, mult iple wives increased a man's wealth aswell as status. Women themselves often ap-preciate additions to the household labor forceand may even ask a husband to bring in anotherwife to ease the burden of work.3 In India, whilenuclear-family households are most common,joint families consisting of parents and marriedsons may emerge where landholding or otherlabor requirements favor large households.Though the household is a basic productionunit in both East Africa and India, then, itsstructure may be quite different in the tworegions under consideration.

The implications of the previous discussionfor agricultural innovations are twofold. First,any change in the skills for labor commitmentof one sex-role category may effect otherswithin the same household; for example, in-creased emphasis on education for childrenmay require more work by adults. Second,individuals, including male household heads,should not be regarded as autonomouseconomic actors. Decision-making is usuallydone wi th the input and consent of spouses andother kin, who in many cases are the verypeople expected to change their labor patterns.Compliance wi th the head's wishes concerninginnovation is not automatic. During the 1960s inKenya, I found that agricultural extensionagents were addressing only men on the sub-ject of clean milk production; yet women wereresponsible for washing milk cans and proved

3. See Ames (1955).

reluctant to adopt new methods because itwould have increased their workload.

Where social stratification is marked, andlower class or caste laborers provide labor tolarge landowners, other impediments totechnological innovation may also occur.Epstein (1962, p. 64) notes, for instance, that inthe irrigated village she studied, spacing ofpaddy transplants (the Japanese method) wasnot accepted because it would upset thetechniques to which female labor was accus-tomed. Ten to 12 women form a team fortransplanting, weeding, and harvesting, and arepaid a fixed rate per job. The new techniqueswere resisted because they took much longer,and farmers were bound by customary ratesthat they could not increase.

Though the traditional division of labor inAfrica and India may result in constraints toproduction, a much more severe problem isseasonality. The following discussion concernsclimatic variation and strategies adopted byfarmers to even out labor use.

Seasonal i ty and Labor Use:Af r ica

Cleave (1974, p. 31) notes that until recently,discussions of labor use in agriculture havecentered on the existence or nonexistence ofunderemployment; now interest has turned tothe capacity of agriculture to absorb labor. Inthe case of both African and Indian agriculture,considerable latitude for increased labor inputsappears to exist. Cleave states, on the basis ofhis survey material, that in African agriculturetypicaIly only about 1000 hours per adult workeris spent in the field, or 4 to 6 hours per day forunder 200 days of the year. Additional time maybe spent in travel to fields, though this is truer ofWest than East Africa. The seasonal nature ofagriculture imposes limits to labor input:

On the one hand, if all available labor in anarea is fully employed at any period of peakdemand, then certainly labor may not bepermanently removed without affectingproduction, and increases in agriculturalproduction would be limited to off-peakperiods; on the other hand, if all employ-ment is concentrated in a limited period ofthe year, the total hours of effective workwhich can be put into and rewarded for

337

agricultural employment are l imited unlessthe agricultural system is susceptible tochange (Cleave 1974, pp 66-67).Cleave (1974, p 90) estimates that 15 to 50%

of the underutilization of labor in Africa is due toseasonality. The options, then, would be tochange cropping systems to spread out laborneeds or make more efficient use of existinglabor at peak seasons. Another alternative is tointroduce more agriculture related activities(public works, handicrafts) to absorb labor dur-ing non-peak periods.

In the tropics, water supply is the most sig-nificant variable affecting seasonality. Onegroup of African farms studied by Cleave arelocated in semi-arid regions, characterized by a unimodal wet season wi th only 4 or 5 monthsper year receiving more than 3 inches (75 mm)of rain, and 4 or 5 months with none at all. Thevillages are located in northern Rhodesia,where maize and groundnut are the main crops;in the eastern Sekoto Province of N ig eria, wheregroundnut produces a cash crop and milletand sorghum are grown for subsistence; andGambia, where groundnut is again a cash cropand rice a staple. While several surveys fromEast Africa are covered in Cleave's analysis,they unfortunately refer to humid and sub-humid cash-crop areas and not to regions ofsemi-arid farming (Cleave 1974, pp 73-74).

Cleave demonstrates a clear relationship bet-ween rainfall patterns and labor use, thoughthis is not the only variable involved. For exam-ple, in the Toro District of Uganda, which isclose to the Equator, adequate rainfall is avail-able 11 months of the year; farms are highlydiversified, and the difference between peakand slack seasons in agricultural employment issmall. "Conversely, in Chiweshe, Rhodesia, andSokoto Province, Nigeria, which approach thepolar limits of the area covered, periods occurwith no rain when crop activity effectivelyceases and agricultural work is confined to a l imited range of livestock and maintenancetasks" (Cleave 1974, p 128). Different rainfallpattterns may have a significant effect evenwithin a small geographical area. Cleave (1974,pp 128-9) notes that an additional month ofrainy season in northern Uganda, as comparedto northern Tanzania, allows' Uganda farmersgreater flexibility in avoiding a clash betweencotton and foodgrains. Although intense acti-vity fol lows the first rains throughout areas

where a definite dry season exists, he finds that" to a remarkable extent, weeding is a major callon labor in the farming year and the single mostimportant cause of seasonal pressures" (1974,p 129). He finds this to be true of cotton farms inTanzania, maize farms in Rhodesia, and farmsin Masii, Kenya, all due to the use of ox-drawnplows; in this case, the switch from labor-intensive hoeing to plowing results in the needfor more weeding later. Weeding is a majorproblem in the tropics, particularly in monsoonareas with a definite dry season, and weedingrather than harvesting is usually the cause ofpeak labor demands (Cleave 1974, p 130).

Cleave's surveys do not allow easy generali-zation concerning the role of cash crops inincreasing or decreasing seasonal variations inlabor use; he notes, however, that Africanfarming systems are consistently modif ied inresponse to labor conflicts by cultivators eitherchanging farm operations or the crop pattern. Innorthern Uganda, for example, farmers deli-berately delay cotton planting in order to giveprecedence to the staple food, millet. Earlymillet is necessary both because of high pricesjust before the main harvest and because it is animportant part of payment for communal andhired labor. Simultaneous planting of cottonand millet would lead to a clash of labor de-mand at the three stages of planting, weedingand harvest. He suggests that early modifica-tion of the farming system may result f romlabor conflict of staple and cash crops, evenwhen commercialization is not at an advancedstage (Cleave 1974, pp 131-132, 144).

In summary, Cleave's analysis of Africanfarming systems stresses their flexibility andgrowth potential. Farmers have not only addedcash crops to subsistence production, but havecommitted themselves to commercialization toa greater extent than usually assumed; in manyareas, family farm output has more than dou-bled, and Cleave suggests a per capita growthrate of around 2% per annum in real terms.Farmers have, however, increased productionsimply by cultivating more land and usingadditional family labor. In most cases onlytraditional cultivation techniques are used, andthough ox-drawn plows have been introducedto several areas, farmers have not adoptedother ox-drawn implements such as cultivators(Cleave 1974, pp 188-189).

The predominance of hand methods of pro-

338

duction means that labor is a l imiting factor; onthe one hand, commercialization has resulted inhigher inputs of family labor, but on the otherhand, hours spent on agricultural field work arestill low — only about 1000 hours per adult peryear, or 120 to 1604- to 6-hour fieldwork days.Farmers appear to rationally balance cost andreturn to effort, varying working hours accord-ing to seasonal conditions and modifying laborpatterns, farming systems, the sexual divisionof labor, and even food habits to relieve pres-sure on time and accommodate cash crops. "Lowtotal labor inputs in situations where farmoperations are organized to save labor is anapparent contradiction; it can be explained bythe highly seasonal nature of agricultural pro-duction and by the existence of alternative usesfor labor that have a positive value to the farmfami ly" (Cleave 1974, p 189). An inverse rela-tionship exists between t ime spent on farm andnonfarm activities (if resting is excluded);Cleave notes that even at times of maximumfarm activity (20 to 40 hours a week for adults), 3 to 4 hours a day may be spent on nonfarm work.There is no "absolute" surplus of labor, then,that can automatically be allocated to agricul-ture, but rather t ime and energy must be gradu-ally reapportioned as farmers' valuation ofactivities, changes, or labor demands fordomestic maintenance can be decreased. In-terestingly enough, Cleave finds no evidencethat ceremonial occasions or construction workinterfere with agricultural activities; rather,major constraints on labor use are food prep-aration and domestic chores for women (up to 4 hours per day) and school attendance by chil-dren at critical periods (Cleave 1974, p 191).Similar constraints are also true of India.

Seasonality and Labor Use:India

In India, as in Africa, what appears to be severeunderemployment of labor during slack periodsmay coexist wi th labor shortage in peakperiods, and seasonal ity is largely due to clima-tic conditions. Irregularity of rainfall dramati-cally affects agricultural labor use and formerlyled to widespread famines and epidemics. InMedak District, Andhra Pradesh, where theauthor carried out field research during 1970-71 and 1976-77, rainfall averages 35 inches per

year but the annual range is broad, makingdrought a constant threat (India, 1961, p 78). InAugust 1970, for instance, rains were veryheavy, causing flooding damage but resultingin a relatively good harvest. During the samekharif (rainy) season of the following year,however, the rains fell several months late;farmers prepared paddy seedbeds but werethen unable to transplant as village reservoirshad not filled. Only a few of the wealthierlandlords who owned field wells and pumpscould carry out operations as usual. Most of thedistrict's rice was lost, and the local administra-tion was forced to open ration shops andprovide work-relief programs for those nor-mally dependent on income from agriculturallabor. Extreme unpredictability of rainfall canthus seriously disrupt the agricultural cycle.

Table 1 summarizes the cycle of agriculturalactivity for Mallannapale village in MedakDistrict, where the author conducted researchduring 1970-71. This was a community of 733people cultivating both wet land and drylandand dependent on village tanks or reservoirs forits irrigation water. The total area of village andvillage lands was 1223 acres, and 23 guntas,4

but over half was waste and uncultivated. Of theland being farmed, 194.21 acres was wet land;277.24 acres was dry; and 29.05 consisted of dryconverted into wet. Wetland is classified as abi, tabi or do fasla ("two-season") depending onwhen it can be cultivated. Farmers sow abi landduring the rains of June and July, tabi duringOctober and November, while do fasla landproduces two crops a year. The latter is highlyprized and scarce, however. Most villagers de-pend on their dryland crops and small amountsof abi land, planted to paddy (Benson 1974, pp34-5). While variation wil l occur even betweencommunities in the same area in regard tocropping patterns and seasonal bottlenecks inlabor (see Dasgupta 1977, p 48), the sequencedescribed in Table 1 can be used as a startingpoint for discussion of labor use in a semi-aridregion of India.

4. There are 40 guntas to 1 acre. Figures after a decimal point indicate guntas in local practice; e.g.194.21 = 194 acres, 21 guntas. The above figuresare f rom the notebook of the Village Level Worker(the lowest-level extension agent in the develop-

ment program).

339

Table 1 . Agr icu l tu ra l cyc le fo r Ma l lannapa l lev i l l age , M e d a k D i s t r i c t , A n d h r aPradaah. 1 9 7 0 - 7 1 .

Month

June

July

AugustSeptember

OctoberNovember

December

January

FebruaryMarchApr i lMay

Act iv i ty

Plowing, sowing of paddy, maize, mil let,pulses, etc., preparat ion of seedbeds forrice, weed ing (kharif season).Transplanting paddy seedlings, weed-ing, guarding of maize.Weeding, guard ing of maize.Harvesting of maize and other dry landcrops.As above; beginning of paddy harvest.Complet ion of paddy harvest.

Preparation of seedbeds for secondpaddy crop; start of cane harvest; plant-ing of dry land crops (rabi season).Transplanting of paddy seedl ings; caneharvest and replanting.Weeding, water ing cane and paddy.Weeding.Weeding; second paddy harvest.Plowing for kharif crop begins.

First of all, a distinction between dryland andwetland crops should be noted. The former(millets, maize, pulses, broadcast paddy) arerelatively less labor-intensive than irrigatedcrops but rely more heavily on " intermit tent"family labor — women, children, the elderly — than the other. In her study of two villages in a ragi-growing region of Karnataka, for example,Epstein (1962, pp 212, 215) found that femalelabor was much more important in the cultiva-tion of dry than wet crops. In the dry village,Dalena, women weed and harvest millets anddo the winnowing, while men have only plow-ing duties. She states that the average laborrequirement for ragi (Eleusine corocana) was 20male labor days and 37 female labor days peracre; for one acre of jowar (Sorghum vulgare),10 male labor days and 14 female labor days.Because dryland crops could be tended withlargely female labor, men branched out intoother activities, including cultivation of paddy inneighboring villages, when irrigation arrived inthe area. A similar pattern occurs in parts ofAfrica where a high percentage of adult malesmigrate for work.

One important labor requirement of drylandcrops is protection from birds and animals

340

when the grain is ripening. During this period inMallannapalle (July-September), farm familiesbuild huts on stilts out in the maize fields andguard the plots with slingshots day and night.Women, children, and the elderly performmuch of this labor and also help wi th theharvest. Villagers frequently intercrop maizeand millet with pulses, eggplant, cucumber,groundnut, and beans, harvesting crops at dif-ferent times as they ripen.

While dryland crops are largely cultivatedwith the help of family members, wetlandcrops — mainly paddy and sugarcane — requiredifferent arrangements. Much more land prep-aration is necessary, t iming is crucial in paddytransplantation and weeding, and irrigation re-quires constant attention. Male labor is neededfor maintaining canals and ridges betweenfields, for plowing and levelling paddy plots, forirrigation, and for threshing. Women's groupsplay a major role in transplantation, weeding,and harvesting of paddy, while both men andwomen plant and cut sugarcane. Between Juneand November demand for labor is high sincecultivators are performing the same operationssimultaneously. In December, work begins onthe second paddy crop, and dryland crops cantheoretically be planted; in fact, most farmersdo not have water for a second paddy crop, andif dryland cultivation is attempted yields arepoor. Many cultivators, then, have relativelylittle agricultural work between December andMay or June, when plowing for the kharif cropbegins.

Sugarcane, a major cash crop in Mallan-napalle and the surrounding villages, compli-cates the farmer's schedule since it requiresconstant irrigation and takes 12 to 14 months tomature. The sugarcane harvest and replantingfollows the paddy harvest closely, extendingthe months of heavy labor; if sugar factoryprices are not good in a given year, farmers maychoose to crush their own cane and makejaggery, part of which is kept for householdconsumption and the remainder sold to localmerchants who market it in Hyderabad. Epsteinreports that f rom Wangala, the irrigated villagein her study, average labor needs for cane peracre were 178 male and 30 female workdays; bycontrast, paddy required only 97 male and 28female labor days. Farmers adjusted to the newdemands by contributing more family labor, butstill had to hire about half of it (Epstein 1962, pp

53, 62). The capital expenses involved in sugar-cane production, as well as the necessity ofowning sufficient land to produce both a subsis-tence and a cash crop, are d iscourag ing to manyMallannapalle farmers. By far the largest sugar-cane cultivators in 1971 were two well-to-dolanlords with 4 and 30 acres, respectively, andthey were major employers of agricultural laborin the village.

One important distinction in labor use be-tween African and Indian farmers has alreadybeen mentioned, namely that hired labor ismuch more common in India. African farmerssupplement input by household members withaid f rom visiting relatives and wi th communallabor, rewarded in kind, while hiring work forcash is not so common.

In India, smallholders and dryland cultivatorsdepend largely on family labor, while hiredlabor is needed for larger holdings and cashcrops. Commercialization affects labor supplyby displacing artisans and poor peasantry, whoare drawn to labor oppor tun i t ies in othervillages, while demand for labor rises wi th in-creased concentration of land. Greater pros-perity also results in withdrawal of women andchildren from the labor force for reasons ofstatus and education (Dasgupta 1977, pp10-12; Epstein 1962, p 72).

in Mallannapalle and other Telangana vil-lages, the main categories of labor are coolie orcasual labor; jetham pani or salaried work (aterm also used for government jobs); and gutha or job contract. Casual labor is hired by the fullor half day and paid in cash or in grain for har-vest work. Men plow and thresh, women andboys transplant, weed, and harvest, and bothmen and women do well cleaning and construc-tion work during the dry season. Almost allhouseholds supplement their cult ivation byworking for others at some t ime during the year.The importance of wage labor can be seen fromthe fact that 89% of the population between theages of 10 and 55 said they either customarilyperformed coolie activities or were working assalaried laborers. Casual labor may be used byany household that has sufficient land to afford(and require) it, but the larger landlords are ob-viously the most important employers. During1970-71 Mal lannapal le 's largest land lordemployed casual laborers on a nearly continualbasis, not only for cultivation but also to build a new well , clean out old ones, and lay a stone

flooring in his house. He had two servants withthe regular duty of calling persons for coolie(Benson 1974, pp 123-124).

Salaried or "permanent" labor (jetham pani) involves a longer term contractual relationshipbetween employer and employee, usually for a year at a time. While many households hirecasual labor during the height of the main ag-ricultural season, and may have their ownmembers do coolie work in turn, only a few ofthe wealthier families can afford permanentlabor — nine of 139 households in the case ofMallannapalle. The two large landlords men-tioned above were the largest employers with21 and 19 workers, respectively.

Landlords commonly obtain permanent laborby giving loans that must be worked off or re-paid; this is often difficult since the laborer mayneed additional advances from t ime to t ime.The work is considered rather demeaning sincea servant is subject to constant call by his mas-ter and may not leave the vi l lage wi thoutpermission. Women are hired to help wi thhousework or carry manure from the cattle shedevery day, while men cultivate, irrigate or carefor cattle; men also take turns guarding thelandlord's fields at night. While hereditary tiesbetween landlords and lower caste clients havebeen reported from many parts of India (cf.Epstein 1962, p 73), no evidence of such a sys-tem can be found in Mallannapalle. Of the 53workers employed as of August 1971, only fourhad worked for the same family for 10 or moreyears, while most had served for only 1 or 2 years. A look at labor rolls reveals that atten-dance is sporadic; servants frequently quit orrejoin. A number of villagers who formerlyworked for the larger landlords no longer do so.The relationship, then, is a contractual one(Benson 1974, p 126).

Although service is temporary, certain cus-tomary elements remain. The landlord usuallygives his laborers one meal a day, and buys newclothes for them once a year at the festival ofDassara; he is also expected to take responsibi-lity for the general health of his employees, andoften supports them in disputes (Benson 1974,pp 127-128). It is in the interests of the largelandlord, particularly an entrepreneurial one, tomaintain a permanent labor pool; one of thechief means of doing this is by establishing a debtor-creditor relationship. Although by 1977the Indian government was at tempt ing to

341

abolish "bonded labor" and simultaneously es-tablish rural bank branches that would givecredit to small-holders, it is not clear that thisinnovation wil l be successful.5

The third type of labor, gutha or job contract,is a form of piecework; employer and employeebargain over the price for a particular job, e.g.,weeding a field. The worker plays an indepen-dent role, which makes this a favored type ofemployment; also, since he or she is being paidby the job and not length of t ime, there is anincentive to finish more quickly.

Another type of labor, buthulu, or " loan , " alsoexists. Smallholders attempt as much as possi-ble to avoid hiring labor by using family mem-bers and by resorting to labor exchange incertain circumstances. For example, two far-mers may cooperate in alternately p lowingeach other's fields, or two families may jointlyrent a sugarcane crusher and pool their bullocksand labor to make jaggery. Cooperative laborseemed to be of minor importance in Mallan-napal le, however, perhaps because of theurgency of agricultural operations during thekharif season. Farmers are out plowing as soonas the first rain softens the soi l , and sinceeveryone is cultivating simultaneously, oxencannot be spared; even help to relatives seemeduncommon. Commercialization has probablyundercut cooperative arrangements to a certainextent, as it has in Africa, and affects nonfarmactivities as well .

In addition to agricultural work carried out atthe height of the season — between June andDecember or January — villagers also spendconsiderable t ime on nonfarm activities duringthe remainder of the year. By March, tempera-tures are rising as the hot season approaches,and for most farmers agricultural activity is min-imal. During this period cultivators repair theirhouses, construct new ones, clean out fieldwells and dig others, and do miscellaneousmaintenance jobs. Potters produce large num-bers of house tiles for construction; these mustbe purchased on a cash basis and are not as partof the customary exchange between land-owning and artisan households. As water levelsdrop in wells and tanks, some villagers supple-ment their income or diet by f ishing. Also at this

5. See Barry Michie (1978) for a discussion of thereasons vil lagers prefer moneylenders to banks assources of credit.

342

time local administration initiates public pro-jects such as the repair of reservoirs and roads,for which casual laborers are needed. The non-farm activities are complementary to agricul-ture, as in Africa, and do not compete wi thcultivation since they take place in a slack sea-son. Weddings, the most t ime-consuming of allceremonial obligations in India, are also mostfrequently held during the hot season (espe-cially May) rather than in winter.

In summary, Indian agriculture faces some ofthe same ecological constraints as African ag-riculture, particularly seasonality, and to a cer-ta in extent has evo lved s imi la r cop ingmechanisms, such as complementary nonfarmwork when cultivation is not possible. Duringyears of adequate rainfall observed by theauthor in Mallannapalle, the population ap-peared to be working actively and at a fairlyeven pace throughout most of the seasonal cy-cle; in fact, labor was scarce and difficult to hire.Part of this is probably due to the spread ofsugarcane as a cash crop and the introductionof new sugar factories in the district, althoughthe entrepreneurial activities of one or twohouseholds in Mallannapalle accounted for a large proportion of labor use. Commercializa-t ion, monetization, and accompanying changesin the organization of production have pro-ceeded to a much more striking extent in Indiathan in Africa, accentuating an already severesystem of social inequality. This majorsocioeconomic issue wil l be discussed below.

Commerc ia l i za t ion ,Labor Use, andImpl icat ions fo r Deve lopmen t

In his survey of rural labor use in India, based ondata from 133 Indian Agro-Economic ResearchCentres, Dasgupta (1977) argues that a closerelationship exists between the socioeconomiccharacteristics of a village and its labor utiliza-t ion pattern. Landlessness and unequal landdistribution negatively influence participation.Participation rates of the old (59+), women, andchildren (0-14) influence the overall participa-t ion rate, and the latter is inversely related toduration. "The larger the proportion of intermit-tent workers, the higher is the overall participa-tion rate, and the lower the duration of work."(Dasgupta 1977, p. 157) With the introduction of

cash cropping and irrigation, a shift from familylabor to hired labor takes place and there is a decrease in overall rate of participation of thevillage population. Dasgupta constructs twomodels of community labor use based on thisanalysis. Type A community is characterized bya lower overall participation rate, sharper classdivisions, and a higher proportion of 60% of thehouseholds in such villages do not own oroperate land at all. Type B (which fits much ofthe African data better) has a higher participa-tion rate, more emphasis on family labor, and isgenerally more egalitarian. Demand for labor power must thus be distinguished from de-mand for laborers.

The AERC surveys, however, were conductedduring the early 1960s, before the high-yieldingvarieties of crops and their associatedtechnologies had been introduced. Numerousstudies since then have pointed out that agricul-tural modernization, together wi th commer-cial izat ion and i r r i ga t ion , has createdsocioeconomic problems that prove detrimen-tal to "development" in any real sense. For thepurposes of this study I wi l l present a generalargument concerning the transformation oftraditional patron-client relationships in India,and wil l then illustrate it wi th case studies fromSouth India.

The argument is derived f rom Aruna Michie(1976), and concerns the transformation of" t radi t ional" socioeconomic forms as theyadapt to forces of "modernizat ion." She arguesthat "whi le certain behavioral patterns mayindeed persist, their content and the groupsinvolved are changed substantially." Altera-tions in production organization, due to agricul-tural modernization, destroy the traditional re-ciprocities of patron-client relationships andexclude the poorest as clients. "Insofar aspatron-client relationships performed the tasksof at least providing economic survial (thusincome distribution) and socio-political mobil i-zation, changes in the client base mean newpatterns of both redistribution and mobiliza-t ion . " (Michie 1976, p 1.)

Michie suggests that the process of change inpatron-client relations proceeds in the fol low-ing manner: (1) commercialization of agricul-ture alters the organization of production andthus traditional interdependencies; (2) recip-rocal obligations of patron-client relations aregradually destroyed; (3) the relationship be-

comes more contractual, less obligatory, andoften more coercive; (4) client bases are re-defined as government policies and inputsallow new forms of production organizationthat allow patrons to redefine their role to that ofbroker; (5) clients are redefined to exclude thepoorest, and elites (e.g. government officials)lose a traditional channel of information con-cerning the poorest economic stratum in ruralareas.

The above argument, it should be stressed,applies to specific tenurial conditions, i.e., situa-tions "where land is owned by a few, but iscultivated in small plots by tenants who often tillthe land in hereditary tenure, and may belandless themselves." It is not applied by Michieto either plantation economies or to fullypeasant-owned and operated agriculture (1976,pp. 6-7). In addition to tenant-landlord rela-tions, which required both the payment of cropshares or rents and personal services to thelandlord, the need for tools, handicrafts, andritual services formerly resulted in long-terminterdependencies between households oflandlords, tenants, and artisan and servicecastes. Though ultimate control rested in thehands of those who controlled land, economicinterdependence gave tenant and artisanhouseholds some bargaining power, and a rightto subsistence was at least recognized.6

Michie argues that agrarian patron-client re-lationships are grounded in the technology ofthe production process; if technological inputsresult in the erosion of economic reciprocities,then the leverage of the client is broken. Forexample, mechanization may not only increasethe scale of the production unit, but also resultin the displacement of labor; large landlordsfind tractors less of a management problemthan tenant families, and in fact this is often oneof the explicit intentions behind tractor pur-chase. As cash crops have become more profit-able and new inputs have been provided bygovernment, landlords have illegally displacedtenants who end up as day laborers. They nolonger have any right to a crop share, as thelandlord does not rely on them in the same way.A breakdown of relations with artisans has alsotaken place as prestigious machine-made goodscompete with local handicrafts and eliminate

6. See also Benson (1976).

343

this source of supplementary income. Becausein most cases there are no alternatives toagricultural labor in rural areas, and it is verydifficult for the poor to organize effective resis-tance, the coercive element in employer-employee relations increases.

Michie further suggests that resources for themodernization of agriculture, introduced by thegovernment, are only useful to farmers (and, I would add, the larger farmers); tenants andlaborers become largely irrelevant as clients.There is less emphasis on the redistribution ofwealth, more on aggrandizement, and patronsare transformed into brokers; i.e. dealers inextra-local goods, between parties not in directcontact — in this case, government agents andother farmers, as opposed to tenants and labor-ers. Political payoffs to former clients, in theform of land redistribution and other benefits,tend to be symbolic and minimal.

While Michie's analysis stems f rom her re-search in Rajasthan, and she applies it to" a single tenurial situation, similar consequencesof commercialization and technological changecan be seen elsewhere in India. First of al l , asshe points out, a large proportion of Indianfarmers are not in a position to benefit f romeconomies of scale: 87.5% of all farm familiesare either landless or operate farms of under 10acres (India 1967; quoted in Michie 1976, p. 7).Second, when tenants are rendered landless, orsmallholders lose their land to creditors, itcannot be assumed that their labor wil l beabsorbed elsewhere; often there is no place togo.

In her 1969 restudy of two South Indianvillages where she carried out economic re-search during the 1950s, Epstein (1976, p. 171)argues that increasing economic differentiationhas taken in rural Mysore (Karnataka) during thepast 10 to 15 years:

The wealthiest peasant farmers have be-come considerably richer. The jaggeryboom provided the possibility of a windfal lprofit to all farmers who owned more thantwo or three wet acres: the greater the caneacreage the greater the benefit a farmerderived from the soaring jaggery prices.This encouraged the wealthier and moreenterprising peasants to invest in more andmore wet land as well as in cane-crushersto process their crop into jaggery (Epstein1976, p 171).

344

Labor inputs for sugarcane have considerablyincreased: f rom 52 086 male labor days inWangala, the irrigated village, during 1955 to127 113 male labor days in 1969. Male inputhas more than doubled, wi th the averagenumber of yearly days per male in agricultureincreasing from 174 to 258. Female labor re-quirements in the same community increasedby 50% from 19 296 to 29 565; the annualfemale per capita workload actually fel l , f rom 73to 67 days, but Epstein notes that whi le 50% ofmale labor is provided by the household, 90%of female labor is hired. Farmers now use morefemale workers to reduce the costs of labor,since they are paid only half the male rate, whilethe women of well-to-do households havewithdrawn from cultivation.

In spite of this increase in labor input, markedseasonality continues to characterize the ag-ricultural system, and inflation has reduced thereal wages of laborers. Epstein (1973, p 172)notes that: "The case studies of the poor A. K.(scheduled caste) households in Wangala andDalena show that the greater a household'sdependence on rural cash wages for meetingnecessary expenditure, the greater has been thedeterioration in the standard in l iv ing." Thissituation is not unique to Karnataka; NationalSample Survey data indicate that the percen-tage of the national population below theminimum level of expenditure rose by 40%between 1960-61 and 1967-68 (Bardhan 1970,p 1245).

Another example of the consequence of ag-ricultural modernization is taken from my ownfield work in Andhra Pradesh (Benson 1974). InMallannapalle and other villages from ex-Jagir7

areas of the former Hyderabad State, mostcultivation is by owners and there are relativelyfew tenants or absentee landlords (cf. Khusro,1958). Although the majority of villagers pos-sess some land, it may not be sufficient forsubsistence needs,8 and an analysis of land-

7. A revenue assignment of a characteristic Mus l imtype; the holder of a jagir held the r ight to revenuecollection in one or more vil lages.

8. In 1971 the land records l isted 126 owners, ofwhich 111 were owner-cult ivators and 15 weretenants according to a 1970-71 agricultural cen-sus. A number of deeds have been taken out in thenames of sons and daughters, however, to evadeland ceil ing laws, when actually cult ivation is sti l ljoint, and many tenancies are held by oral contract.

holdings reveals marked economic differentialsbetween households (Table 2). Six out of 139households (five of them Redd is, members of a cultivating caste) have more than 20 acres,while 115 households own 5 or less. Redd is ownmore than 69% of the village wetland (98 acres),several times the amount held by other largecaste groups. Two of the largest Reddi land-lords were major employers of village labor in1970-71. One of them, whom I wi l l refer to asthe head of household 6 -7 , shared his budgetwi th another brother, a wholesaler resident inHyderabad, who played a moneylender role inMallannapalle and neighboring communitiesduring his frequent visits to the village. Duringthe last few years this household had beenexpanding its landholding by any means avail-able to it — purchase, taking land as collateralfor loans, and forcible occupation (Table 3). As a result of these tactics, by 1971 household 6-7was in effective control of 34.16 acres, mostlyirrigated, in addition to an inheritance of 68.28acres being disputed with a third brother. Inaddit ion, the same household leased another 12acres during 1970-71, and also controlled ap-proximately 100 acres in another village.

The expansive activities of 6-7 were due tothe interest of its head in two cash crops,namely paddy and sugarcane. His wholesalerbrother owned a share of a nearby sugar factoryto which the cane crop was sold. Since the headof 6-7 had the capital and familiarity withgovernment agents, he was able to take advan-tage of the new hybrid cane offered by the local

Table 2 . S i ze o f hold ings by househo ld i n Ma l -lannapaj le v l l laga, Andhra Pradesh,1 9 7 1 .

Acreage

0 - 56-10

11-1516-20Over 20

Total

Number of householdsa

11511526

139

a. When landholdlngs are joint but a family has split Intoseparate households, size of holding is estimated to be anequal share for each household.Source: Benson (1974, pp 113-114)

Table 3. Land acquisi t ions by Reddi house-hold 6 - 7 i n Mal lannapal le v i l lage,Andhra Pradesh. a

Occupied f rom other castesOccupied Reddi landPurchases from other castesCollateral for loan (Komiti)

Total

Wet

9.1911.126.15530

32.36

Dry

1.20

Total

9.1912.326.155.30

34.16

a. Refers to acquisitions for a number of years prior to 1971.Source: Benson (1974, pp 114-115)

Block Development Office. During the firstkharif season of my fieldwork (1970), thishousehold put approximately 3 acres into thenew hybrid cane; after its harvest, the cane wasused to seed a much larger area, giving house-hold 6-7 a total of about 30 acres the followingyear. Neighbors complained that this house-hold was acquiring all fields adjoining thesugarcane plots, occupying them by force ifnecessary; by November 1971, the head hadplans to trade for or purchase 6 additional acres.The village's water supply was also excessivelymonopolized for cane-growing by this house-hold. The other large Reddi landlord, 6-7 'seldest brother, was also expanding his sugar-cane area in Mallannapalle and another village,though not at the same rate.

Because of the lucrative nature of cash cropssuch as sugarcane, landlords find that it is moreprofitable to cultivate land, particularly irrigatedland, than to lease it to others, and may even letdryland remain idle. It is in their own interest tomaintain a potential labor pool with no alterna-tive sources of income. While household 6-7employed many laborers, it did so at low wages,and eliminated some labor needs entirely byusing its tractor to plow dryland crops andsugarcane. On several occasions agriculturallaborers attempted to strike against this house-hold, but were out-maneuvered by the head'seconomic power and his ability, through extra-village contacts with other landlords, to bring instrike-breaking laborers from other com-munities.9 As Michie suggests, the poor havelittle leverage.

9. In one case a Communist leader.

345

Conclusions and Suggest ionsfor Further Research

As the above discussion indicates, increaseddemand for labor power tends to adverselyaffect the lowest socioeconomic segment ofsociety in nonwestern countries. The process isundoubtedly slower in communities where landdistribution is relatively egalitarian to beginwi th (Dasgupta's "Type A" community, whichwould include tribal areas in India and many EastAfrican societies), and more rapid in areas wherevil lage land was tradi t ional ly monopo-lized by single castes or ethnic groups (Dasgup-ta's "Type B" community). This does not implythat economic differentiation has not takenplace betwen members of these once-powerfulcastes, or that middle-level castes have notacquired land in many areas; but entrepreneurialactivity is most possible for those who alreadyhave resources, and is often at the expense ofothers. Commercialization probably makes thisprocess inevitable, but planners should beaware of the social costs of their actions. AsCleave notes, development agents must payclose attention to the effects that their technicalrecommendations and innovations have onlabor demand, and should realize that small-holders require complete packages of innova-tions including credit and marketing facilities.At present, these tend to be monopolized by thewell-to-do.

While effective land redistribution in India ispolitically and economically unrealistic, someof the worst effects of commercialization andagricultural modernization could perhaps bemitigated by discouraging technology that dis-places labor, in favor of technology that is morelabor-intensive and reduces the seasonality ofagriculture. Another tactic would be to investi-gate the possibility of more cottage industries inrural areas; these could provide income duringseasons of agricultural inactivity and would beespecially appropriate for women, as well asgive laborers somewhat more bargainingpower (due to the presence of alternateemployment) wi th landlords.

In order to develop alternative developmentstrategies, further research is essential. Verylittle information exists concerning labor inputsfrom family members, especially women andchildren, although a number of studies suggestthat their work is quite important to the house-

346

hold economy. As Dasgupta (1977, p 19) states:It is important to know the relative impor-tance of various activities . . . how their d is-tr ibution by age-sex categories varies be-tween villages in response to differences invillage environmental conditions. Unfortu-nately, very few village surveys give suchdata; and even when they do, their defini-tions and methods of classification of'work' vary widely, partly because of social,cultural, and economic differences amongthe situations studied.Empirical research is necessary to explain the

variation that exists between labor inputs ofwomen and children of different socioeconomicclasses and ethnic groups in different ecologicalsettings. As Dasgupta (1977, p. 36) suggests,"To what extent cultural factors promote orhinder women's participation independently ofeconomic and institutional factors is a matter ofdispute." He notes that cropping patterns alsoinfluence the participation of different age-sexgroups in work, but very few Indian villagestudies document in detail the labor input re-quirements of different crops.

Since female work-groups handle many ofthe major agricultural operations in both Indiaand Africa, and at least in India commercial useof irrigated land seems to require lower-classfemale labor, the organization of these work-groups should be studied. Another problemwith time-allocation data is that even where itexists, it does not measure effort, which in turnaffects duration and productivity. Further re-search is necessary in all these areas.

References

A M E S , D. W. 1955. The economic basis of Wolofpolygyny. Southwestern Journal of Anthropo logy11: 391-403.

BAILEY, F. G. 1957. Caste and the economic frontier.Manchester, UK: Manchester University Press.

BARDHAN, P. 1970. Green revolut ion and agriculturallabourers. Economic and Political Weekly (SpecialNumber): 2 9 - 3 1 .

BARTH, F. 1960. The system of social stratif ication inSwat, North Pakistan. In Aspects of caste in SouthIndia, Ceylon and northwest Pakistan, ed. E. Ft.Leach. Cambridge; Cambridge University Press.

BENSON, J. 1974. Caste and power in a Telanganavillage. Ph.D. thesis, Brandels University, Waltham,Mass, USA.

BENSON, J. 1976. A south Indian jajmani system.Ethnology 15: 239-250.

BENSON, J. 1978. Author i ty , deference, and seclusion:male and female in Telangana. Paper presented atthe Seventh Annual Wisconsin Conference onSouth Asia, 3 - 4 November 1978.

BOSERUP, E. 1970. Women's role in economic de-velopment. New York: St. Mart in 's Press.

BROWN, J. K. 1970. A note on the divis ion of labor bysex. Amer ican Anthropologis t 72:1073-1078.

CHINAS, B. L. 1973. The Isthums Zapotecs: women 'sroles in cultural context. New York: Holt, Rinehart,and Winston.

CLEAVE, J. H. 1974. Afr ican farmers: labor use in thedevelopment of smallholder agriculture. New York:Praeger.

DASGUPTA, B. 1977. Vi l lage society and labor use.Delhi: Oxford University Press.

EPSTEIN, S. 1962. Economic development and socialchange in South India. Manchester, UK: ManchesterUniversity Press.

EPSTEIN, S. 1973. South India: Yesterday, today andtomor row. New York: Holmes and Meir.

FALLERS, L. A., Ed. 1964. The king's men. New York:

Oxford University Press.

GULATI , L. 1975. Female work part ic ipat ion: a study ofinter-state differences. Economic and Political

Weekly 10(1,2): 3 5 - 4 2 .

INDIA 1966. Census of India, 1961. Medak Districtcensus handbook. Hyderabad: Government of

Andhra Pradesh.

INDIA 1967. National Sample Survey: Seventeenth

Round, Report No. 176.

KHUSRO, A. M. 1958. Economic and social effects ofjagirdari abol i t ion and land reforms in Hyderabad.Hyderabad, India: Osmania University, Department

of Publications and University Press.

M A R T I N , M. K., and VOORHIES, B. 1975. Female of the

species. New York: Columbia University Press.M I N T Z , S. 1967. Pratik: Haitian personal economic

relationships. In Peasant Society, eds. Jack M.Potter, May N. Diaz, and George M. Foster, Boston:Litt le, Brown.

MICHIE, B. 1978. Baniyas in the Indian agrarianeconomy: a case of stagnant entrepreneurship.Journal of Asian Studies 37: 637-652.

MUELLER, E. 1977. The economic value of children inpeasant agriculture. In Population and develop-ment, ed. Ronald G. Ridker, Balt imore and London:Johns Hopkins University Press.

PALA, A. O. 1976. Afr ican women in rural develop-ment: research trends and priorities. AmericanCouncil on Education Overseas Liaison CommitteePaper 12.

RICHARDS, A. I. 1954. Economic development andtribal change; a study of immigrant labor inBuganda. Cambridge, UK: Heffer.

RYAN, J. G., GHODAKE, R. D., and SARIN, R. 1979.

Labor use and labor markets in semi-arid tropicalrural villages of peninsular India. This workshop.

SIMMONS, E. B. 1976. Economic research on women inrural development in northern Nigeria. AmericanCouncil on Education Overseas Liaison CommitteePaper No. 10.

SINHA, J. N. 1977. India: a demographic profi le. InIndia since independence, ed. S. C. Dube, NewDelhi: Vikas.

M A T S O N , W. 1958. Tribal cohesion in a moneyeconomy; a study of the Mambwe people of north-ern Rhodesia. Manchester, UK: Manchester Univer-sity Press.

N A G , M., WHITE, B. N. F., and PEET, R. C. 1978. An

anthropological approach to the study of theeconomic value of children in Java and Nepal.Current Anthropology 19: 293-306.

W O L F , E. R. 1966. Peasants. Englewood cliffs,NJ : Prentice-Hall.

347

Rural Labor Marke t s in West A f r i caw i t h Emphasis on t h e Semi -Ar id Tropics

Derek Byer lee*

Abstract

This paper provides an overview of rural labor markets in West Africa with emphasis on

the Semi-arid tropical (SAT) regions. A knowledge and understanding of these labor

markets is essential to an understanding of traditional cropping patterns and practices as

a basis for the design of new crop technologies appropriate to the farmers of the region.

The paper reviews empirical evidence on rural labor markets at a hierarchy of levels

beginning with household or family labor utilization, proceeding to the local or village

hired labor market, and finishing with the regional labor market integrated by rural-rural

and rural-urban migration. On the basis of this review, some generalizations about rural

labor markets in the SAT areas are drawn and implications for agricultural research

programs noted.

The purpose of this paper is to provide anoverview of rural labor markets in West Africawith emphasis on the Semi-Arid Tropical (SAT)regions. There are several characteristics of theSAT region of West Africa that set it apart f romother SAT regions and that have importantimplications in explaining labor market be-havior in the area. First, the SAT reg ion of WestAfrica is generally characterized by a communalland tenure system that ensures each memberof the community a right to the use of a piece ofland; land cannot usually be sold, althoughshort term access to land is often possible at low" ren ta l " values. In addit ion, the region as a whole has had a relatively low populationdensity — usually less than 50 persons/sqkm — although this is rapidly increasing andpockets of higher population density are nowcommon. These two combine to preclude theexistence of a landless laborer class and indeedin many cases result in a relatively equitabledistribution of land per capita.

Another factor of importance in analyzinglabor markets in the region is the lack of agricul-tural opportunities in the dry season, althoughnonfarm opportunities are available to provide

* Centro Internacional de Mejoramiento de Maiz Y Trigo, (CIMMYT), Mexico.

some dry-season employment. Irrigation de-velopment is in its infancy, and bottomlandssuitable for dry season cultivation are onlylocally available. Rainfall for perennial crops isinadequate, and the dry-season clearing ofbush characteristic of the humid tropics is notgenerally necessary in the savanna areas. How-ever, the humid coastal or forest areas of WestAfrica provide considerable dry seasonemployment in tree crop production, giving riseto opportunities for seasonal migration bet-ween the two areas.

The broad emphasis in this paper is on labormarkets in the semi-arid tropical or savannaareas of West Africa,1 but because of my ownresearch experience in the area and access toresearch of others, evidence is largely drawnfrom the Anglophone countries of Nigeria,Ghana, and Sierra Leone, including the forestzone. Land tenure and population pressure inthe forest areas are usually similar to thesavanna areas and the very fragmentary infor-mation on rural labor markets in savanna areas

1. Countries included in the SAT areas of West Africaare the Sahelian countries — Gambia, Senegal,Ma l i , Mauri tania, Upper Volta, Niger, andChad — and the northern parts of the coastalcountries of Guinea, Ivory Coast, Ghana, Benin,Togo, and Nigeria.

348

can be greatly augmented by cautious inclusionof evidence from the forest zone. Furthermore,as mentioned above, the rural labor markets inthe two ecological areas are often closely linkedby seasonal and permanent migration streams.Therefore I often refer to a comprehensivestudy of rural labor markets and migration weconducted inSierra Leone, which, except for thevery north, lies in the humid tropics. Here toowe gain some useful insights by noting differ-ences in labor use and labor market behaviorbetween the south and the drier north.

This paper reviews empirical evidence onrural labor markets at a hierarchy of levelsbeginning with household or family labor utili-zation, proceeding to the local or village hiredlabor market, and finishing with the regionallabor market integrated by rural-rural andrural-urban migration. On the basis of thisreview some generalizations about rural labormarkets in the SAT areas are drawn and impli-cations for agricultural research programsnoted.

Family Labor

Most evidence on family labor use in the SATregions of West Africa comes from multiple visitsurveys (usually twice weekly) of rural house-holds to elicit daily labor activities of eachfamily member over the whole year. The mostcomprehensive of these surveys have beenundertaken in the north of Nigeria by Norman(1973) and later by Matlon (1977). Here maleadults worked approximately 700 to 1300hours/year in farm and nonfarm work. Thesefigures are lower than observed in similar sur-veys in the humid areas of Sierra Leone, wheremales worked 1400 to 1500 hours/year. Becauseof Moslem tradit ion, female participation inagricultural field activities is minimal in north-ern Nigeria. In northern Sierra Leone femalesare an important part of the labor force, working900 hours/year (mostly in farm work) and con-tributing 39% of all labor inputs. In the southernregions of Sierra Leone females contribute aneven higher percentage of total labor. Olderchildren are also an important part of the laborforce. Matlon observes that boys 10 to 15 yearsold contributed 250 hours/year, compared withadult males wi th 680 hours/year in his study ofthree villages in Kano State, Nigeria.

A somewhat different approach recently usedby Tripp (1978) in the north of Ghana involvesvisiting a household (selected at random from a given sample of households) at a random pointin t ime and eliciting information on the ac-tivities of each household member at that pointin t ime. By continuing this approach over thewhole year, Tripp was able to construct a profileof use of time use by family members for thesample of households. Time allocated to workin his sample was 1500 hours for farm work and560 hours for nonfarm work for men and 1070hours in farm work and 670 hours for nonfarmwork for women. Marketing, food preparation,and household maintenance and other house-hold chores occupied another 1.8 hours/day formen and 4.4 hours/day for women. These f i -gures, based on observation rather than recall,suggest somewhat higher labor inputs thanpreviously estimated and also document theconsiderable time spent in household activities,particularly for women.

Where female and child labor are important,there is usually some specialization. For exam-ple in northern Sierra Leone and Ghana,females are the largest contributors togroundnut and traditional vegetable productionand to weeding and processing activities.Young males are important in pest-scaring andlabor-selling activities (Gunther 1978, Tripp1978). Nonetheless, all types of labor participatein all enterprises and most activities, and thereseems no reason to assume that sexual or agespecialization of labor is rigidly enforced exceptfor particularly strong religious or socialreasons, as among the Moslem Hausa ofNigeria. In fact, Tripp notes in northern Ghanaan increasing role of women in farm work;partly in response to labor shortages, womennow weed millet — traditionally a male activity.

High seasonality of labor use is of coursecharacteristic of this area. In the north of Nigeria53% of all annual labor inputs occurred in thefour busiest months of May to August, whileonly 15% occur red in the four slack months ofJanuary to April (Norman and Pryor 1978). Inthe peak month, July, adult males worked about140 hours/month compared with only 51hours/month in the slack months. In northernSierra Leone peak month labor inputs, again inJuly, are as high as 200 hours/month (Spencerand Byerlee 1976) and in northern Ghana,Tripp's data (1978) indicates that males work as

349

much as 250 hours/month in the wet season.At the peak month there is evidence of an

increase in the supply of labor f rom othermembers of the family. In Sierra Leone in thepeak month of July females contribute 43% ofall labor inputs compared wi th only 29% in theslack season (Gunther 1978). Matlon (1977) andNorman and Pryor found in northern Nigeria,where female agricultural labor is unimportantin all seasons, that similar increases in therelative proportion of labor provided by youngmales occurred in the peak season.

In addit ion, in the peak season labor is drawnfrom nonfarm to farm activities.2 Over the entireyear Matlon (1977) found that 2 1 % of all familylabor inputs were in nonfarm activities, largelylivestock and traditional manufacturing, trade,and services. However in the peak months ofMay to July this percentage falls to 10%. InSierra Leone; 1 1 % of family labor is employedin small-scale industry and trading on an annualbasis, but in the peak month only 2 percent oflabor is used in this type of nonfarm workcompared wi th 19% in the slack month ofFebruary (Spencer and Byerlee). In northernGhana, the percentage of work t ime spent oncrafts and trade by adult males decreases from54% in the dry season to 14% in the wet season.In all cases the absolute as well as percentagehours worked in nonfarm activities falls in thepeak period.

These figures only document some of theevidence leading to the general conclusion of a peak season labor bottleneck in most areas ofWest Africa. It is reasonable to hypothesize thatthe severity of this problem increases as wemove north into lower rainfall areas, becausethe flexibility to stagger plantings is reducedand, as a result, planting and weeding work aremore concentrated. There is indeed some evi-dence to support this. In Nigeria, the coefficientof variation (CV) of monthly labor use (60-70%)in the driest area, Sokoto, was among thehighest of the three areas analyzed by Normanand Pryor. Moreover, the adult males workedapproximately 30% more in the peak month inthis area than in the other two areas. In Sierra

2. The def in i t ion of nonfarm labor varies f r om studyto study. In this paper we general ly mean all workconducted away f rom former's f ields, e. g. agricul-tural processing, l ivestock, local crafts and t rad ing,labor sold out b u t excluding household work.

350

Leone, there is a very marked increase in thecoefficient of variation of family labor use,moving from 16 to 25% in the south to 40 to 50%in the north. Again the peak season labor inputsincrease, moving from an average of 140 to 150hours/month in the south to about 200 hours/month in the north, although in the dry season,labor inputs in the north are consistently lowerthan in the south, further increasing seasonalvariation in labor use (Spencer and Byerlee1976). Moreover, there is evidence of greatersubstitution of farm for nonfarm work in thepeak season in the north, again indicating a more severe peak season labor constraint.

Finally, disaggregation of household datareveals that labor use and allocation in the peakseason shows substantial variation even withina village. In the north of Nigeria, Matlon foundthat the peak season for low-income house-holds is less acute with adult males workingonly 93 hours/month compared wi th 153 hoursand 116 hours/month for the middle and highestincome group. Moreover, the lowest incomegroup participated to a greater degree in non-farm activities in the peak season, when 25% oflabor was still utilized in nonfarm work com-pared with only 3% for higher income house-holds. In part this may reflect a need of low-income food deficit households to continueearning cash incomes through nonfarm work topurchase food. In Sierra Leone, low incomehouseholds also had the lowest labor inputsduring the peak season as well as for the wholeyear (Gunther 1978).3 However the largest dif-ference occurs in the slack season where thehighest income households worked twice asmuch as the lowest income households largelybecause of their participation in relatively capi-tal intensive dry season activities such asfishing, small-scale industry, and tree crops. Incontrast to Nigeria, there was a tendency forhigh income households to spend more t ime inthe peak season on off-farm activities relative tolow income households, but these were largelyhigh returns activities such as small-scale in-dustries and f ishing, compared wi th the lowreturns activities, such as working on nearbyfarms, that are fol lowed by low-income house-holds in the peak season in Nigeria.

3. In both Mat lon and Gunther, households wereclassified into three income terciles for theanalysis.

The low labor input of low-income house-holds is clearly an important factor explainingthe low income status of these households.Matlon and Newman for example, show that inthe north of Nigeria low income householdsplant and weed basic food crops later thanhigher income households and also weed less-intensively leading to lower technical efficiencyof production. They explain part of this differ-ence in the t iming and intensity of labor inputsin terms of a liquidity constraint forcing lowincome households to work off the farm (seenext section) but this does not explain the lowtotal family labor input in the peak seasons,which might result f rom other factors such asnutrition or disease (Chambers 1978).

Hired Labor

Hired labor in traditional farming situations inSAT West Africa typically accounts for from10 to 20% of all labor use and is also thelargest production expense. In the north ofNigeria, it accounts for 57% of all cash expenses(Norman and Pryor); in Sierra Leone, 70% of allexpenses.

Three systems of hiring labor are commonlyused — the traditional exchange and com-munal work, daily wage work, and contractwork. The importance of traditional exchangeand communal work, where young males formgroups to work from farm to farm, or engage inindividual labor exchanges, varies from area toarea and is probably being supplanted bycommercial labor hiring. In the north of Nigeria,55% of labor was hired by the hour, 44% ona contract basis and only 1% on exchange(Norman and Pryor). In Sierra Leone, up to halfof hired labor is exchange labor and only 13% iscontract labor (Spencer and Byerlee). There arealso different methods of paying hired labor. InSierra Leone, 38% of hired labor (largely ex-change labor) was paid only meals, 38% mealsand cash, 3% cash only and the rest othercombinations of cash and kind payments(Spencer and Byerlee). Overall, cash pay-ments were only about 38% of all wage pay-ments since substantial meals were alwaysprovided. Cash payments are probably moreimportant and meal payments less important inthe north of Nigeria. Wages paid to contractlabor are effectively higher than daily

wages — over 100% higher in the case of SierraLeone and 69% higher in Nigeria (Norman andPryor).

The amount of hired labor varies seasonallyand appears to generally fol low the seasonallabor profile of family labor. However, in theaggregate there seems to be little tendency forthe proportion of hired labor to increase in thepeak season. To some extent this is to beexpected, given the lack of a landless laborerclass and the fact that most households experi-ence peak labor demands on their own farm atthe same time.

However, disaggregation of householdsshows some differences in use of hired labor byincome group. In the north of Nigeria, Matlon(1977) found that higher income householdshire a greater proportion of their labor. In SierraLeone, this is true to a small extent but, moreimportantly, Franzel (1978) shows that highincome households tend to hire a greater pro-portion of labor at the peak season of July/August compared with lower income house-holds, who hire in April/May and again atharvest in November — periods of inter-mediate demands on family labor inputs. Thisprobably reflects the greater cash availability ofhigher income households in the peak seasonas well as the fact that exchange labor is notgenerally used for peak season tasks such asweeding.

It is probably also true that lower incomehouseholds sell out more of their labor. At leastin the north of Nigeria, Matlon found that lowincome households sell out about 12% of theirtotal labor while middle and higher incomehouseholds sell out only 1 to 2%. Moreover, inthe peak season, low income households sellout relatively the same proportion of their laborand, in terms of hours, actually sell more labor.In Sierra Leone, the amount of labor sold outdoes not seem to vary much by income level,but this may be because all income groupsparticipate in exchange labor, which is impor-tant in this area.

Given the patterns of seasonal labor de-mands, some seasonal variation in wage ratesis expected. In fact, the only evidence availablefrom the savanna areas indicates that this is notthe case, perhaps because the cash constraintcoincides with peak labor demands, preventinghigh income households from offering higherwages or forcing lower income households to

351

continue seeking off-farm incomes (Norman1973).

Evidence from Sierra Leone, however, showssome quite important variations in wage rates.Analysis of variance showed that region, sex,and month all had highly significant effects onrural wages. Drier northern regions had thelowest wages and consistently in almost allregions the wage rate for females was 65 to75%of male wages, and wages for children less than15 years were only half of adult male wages.However, in the peak seasons there was a tendency for male/female wage differentials tonarrow. Peak wages occurred in May to Julywi th a secondary peak in the dry season fromJanuary to March, but the differences were notlarge, wi th May to July wages only about 25%higher than the low wage period of August toDecember. The May-July wage peak of coursecorresponds roughly to the period of peak labordemands. The high wages in the dry season areless easy to explain but they might be related togreater cash availability to both labor buyersand sellers. Both payments in cash and thevalue of meals vary significantly by month. Inparticular, there is a very clear tendency for theabsolute level of and proportion of cash pay-ments in wages to increase immediately afterharvest in December and remain high untilJune. In June the value of payments in-kind,particularly meals, increases until the "hungryseason" of August/September, when both cashand kind payments (Spencer and Byerlee 1976)are low. Thus the periods of high cash pay-ments correspond very closely to the periodof cash availability and vice versa. The highdry-season wage rate could then be due tosome households substituting hired labor, forfamily labor, using available cash to attractworkers who, wi th harvest in hand, wouldnormally consider this a period of relative lei-sure and reserve a high supply price for theirlabor.

Matlon (personal communication) also notesa relatively high wage in the dry season in theKano area of northern Nigeria, which he feelscan also be partly explained by the arduousnature of dry season tasks such as deep plowingof lowland fields by hand.

Finally, there is evidence that average wagerates paid are reflected in the marginal valueproductivity (MVP) of labor. Production func-tion studies in Nigeria by Matlon and by

352

Norman generally show the MVP of labor to beslightly above the wage rate, but not sig-nificantly so. There is, however, no evidenceavailable to show that the MVP of labor alsovaries by sex and season. Matlon did estimateMVPs of labor for households at different in-come levels and again finds at all levels that theMVP of labor brackets the wage rate. This thenis evidence that the labor market is fairlyefficient on average in allocating labor. On a seasonal basis there is evidence from linearprogramming models that the MVP of labor inthe peak months wi th labor inputs set at ob-served levels is three to four times above theprevailing wage rate (Norman and Pryor). Thisundoubtedly relates to the seasonal l iquiditysituation discussed above.

Migra t ion

The most important and widespread migrationin the semi-arid areas of West Africa has histori-cally been the seasonal migration to the forestzone during the dry season. The magnitude ofthis migration is not really known but has beenestimated to be as high as 3 mill ion persons peryear (Udo 1964). Generally migrants leave theirfarms in the savanna zone immediately afterharvest in November-December and returnagain for planting in April-May. This is, ofcourse, the dry season of low farm activity in thesavanna areas but a peak season for tree cropsin the forest areas, particularly coffee, cocoa,and oil palm.

Much of this seasonal migration is inter-national. For example, up to 65 to 70% of thehired farm workers in the Ivory Coast in the dryseason are from Upper Volta, Mali and Guinea.A 1960 survey showed that Upper Volta alonehad about 7% of its population, or 500 000people working in the Ivory Coast, over half asseasonal migrants (Songre 1973). Seasonalmigration within countries is also important.Beats and Menezes (1970) estimate that 200 000persons f rom northern Ghana migrate annuallyto the south to work in cocoa areas. In the highpopulation density area of northern Ghanawhere Tripp worked, 40% of males migrated tothe south in the dry season in which he con-ducted his survey (1976), and almost all maleshad migrated at some t ime in the past. In a farmsurvey in the Sokoto area of Nigeria, 45% of the

off-farm work was accomplished throughshort-term migration during the dry season(Norman and Pryor).

Typically, seasonal migrants are young maleswith no education, but much less information isavailable about the characteristics of thehouseholds and regions from which they origi-nate. There is evidence that villages with low-land areas available for cultivation in the dryseason produce fewer seasonal migrants(Norman and Pryor). There is also some evi-dence that high density population areas suchas the Moshi area of Upper Volta and the Sokotoarea of Nigeria produce more migrants.

Within a region it is tempting to hypothesizethat seasonal migrants originate in poorerhouseholds with need of additional cash in-comes and food. However we have found noavailable evidence of this linkage.

Seasonal migration has a number of potentialimplications for farm resource allocation anduse in the savanna areas. It wil l compete withagricultural activities that occur in or overlapwith the dry season. Skinner (1965) speculatesthat the failure to introduce the cash crop,cotton, in the Moshi region of Upper Volta is dueto a conflict of cotton harvesting with seasonalmigration in December (cotton is typicallyplanted and harvested after the food crops).4

Capital formation activities using labor in thedry season — such as bush clearing, well-digging and terracing — may also be reduced inareas of high seasonal migration. Skinner ar-gues that in the Moshi areas of Upper Volta newbush land is less frequently cleared, leading tomore intensive cropping and perhaps decliningfertility.

On the other hand, seasonal migration mayadd to household resources for farming. Cashreturned wi th migrants could help offset thecash constraint faced in most rural areas, al-though evidence from more permanent migra-tion streams in Sierra Leone indicates thatrelatively few cash remittances are used directlyin agriculture.5 Furthermore, households withat least one member absent for 4 to 6 months

4. Of course it is implici t ly assumed that returns toseasonal migrat ion are higher than for cottonproduct ion.

5. However, the cash returned by migrants mightindirectly make cash more available for farm in-vestments.

wil l experience a decline in home food con-sumption needs which might either increasethe marketable surplus (or reduce the fooddeficit) or lead to a change in farming systemsaway from food crops to more profitable cashcrops. Unfortunately, no evidence is availableon these points.

In addition to seasonal migration, there areimportant, more permanent migration stre-ams.6 Rural-rural migration is one means ofequilibrating interregional differentials in laborproductivity in agriculture. There is evidencethat rural-rural migration has indeed partlyplayed this role. Associated with seasonalrural-rural migration has been a tendency formigrants over t ime to settle in the destinationareas where wages and year-long employmentopportunities are better. As a result, about onemill ion foreigners, largely f rom Upper Volta andMali, now reside in the Ivory Coast (World Bank1979).

In Sierra Leone, we estimated on the basis ofcross-sectional data an elasticity of about 2.5 ofthe rate of permanent rural-rural migration wi threspect to interregional rural wage differentials.However, a wage differential of almost 2 to 1 still existed between the highest and lowestwage regions, despite a considerable f low ofmigrants from the lowest to highest wageregions. It is likely that ethnic and land tenurebarriers considerably restrict the amount ofpermanent rural-rural migration and, as a re-sult, pockets of high population density anddeclining soil fertility are common. In manyareas strangers can still gain access to landthrough arrangement wi th local leaders but it ismore difficult to obtain permanent user rightsand even more so where ethnic differencesexist (Eicher et al. 1970).

Rural-urban migration has been growing inimportance throughout the region and prob-ably now exceeds rural-rural migration in num-bers in most countries.7 It is also likely that thepattern observed in Sierra Leone is fairly gener-ally applicable in West Africa. There migrantsdif ferentiated into two wel l -demarcated

6. In this paper we define permanent migrants asthose w h o move to a new locat ion for a period ofover 6 months. This excludes seasonal migrants.

7. For evidence f rom Ghana, Sierra Leone and UpperVolts see Caldwell (1969), Byerlee et al. (1976), andGregory (1971), respectively.

353

streams (Byerlee et al. 1976). On the one sidewere migrants without formal schooling whooriginated in poorer households and poorer regions of the country. On the other side weremigrants with at least primary schooling whocame from higher income households and re-gions of the country.

The large number of rural-urban migrants isconsidered a serious problem by many gov-ernments in the area because of the increases inurban unemployment and urban social prob-lems associated with this migration. There hasbeen considerable research on rural-urbanmigration but it has largely focused on docu-menting the migration streams and their impl i-cations for urban areas. From the perspective ofthis paper, the impact of this type of migrationon farming systems and crop technologies isthe critical issue, and here there is little evidenceavailable. Although the rate of rural-urban mig-ration has been high, rural populations in mostareas continue to increase. However, becauseof the selectivity toward young people 15 to 30years old, the dependency ratio has increasedand probably aggravated the seasonal laborbottleneck in many areas.

Conclusions on RuralLabor M a r k e t Behavior

Throughout West Africa family labor is the mostimportant resource in household productionand hired labor the largest production expense.At the peak season — usual ly Ju l y andAugust — labor is generally a bottleneck in ag-ricultural production in the sense that there is a relatively high MVP of labor at this t ime. There isa widespread and competitive labor market toallocate labor within households, villages, re-gions, and even across international bound-aries. At peak seasons additional householdlabor resources — w o m e n and ch i ldren — often play a relatively more important role. Inaddition labor f rom nonfarm activities is drawninto agricultural product ion. Labor is hired(from local labor markets) at average wagerates that reflect the marginal productivity oflabor on an annual basis. Seasonal and perma-nent rural-rural migration link widely separatedlabor markets , par t i cu la r l y be tween thesavanna and coastal areas.

However, there are also some factors that

364

lead to inefficiencies in labor allocation. In someareas such as the north of Nigeria, where fieldwork by females is l imited, the sexual divisionof agricultural work may lead to a more seriousproblem of seasonal labor bottlenecks. Further-more, there is growing evidence that the buyingand selling of labor is intimately linked to thecash flows of rural households, particularly be-cause the peak labor requirements generallycoincide with the period of acute cash shor-tages.8 To the extent that there are im-perfections in the credit market — and thereundoubtedly are — these wil l be translated intoimperfections in the labor market. At the peaklabor demand period, the demand for hiredlabor is constrained by lack of available cashwhile supply remains steady because of theneed for some households (particularly lowerincome households) to obtain cash. As a result,the variation in wage rates is relatively smallcompared with the likely variation in the MVP oflabor.

There is also evidence of considerable varia-tion of patterns of labor use and allocationbetween households and regions. Lower in-come households tend to be labor sellers bothlocally and to other rural regions through sea-sonal and permanent migration, while higherincome households tend to be labor buyers,particularly at the peak period. Across regions,rainfall and population density affect rural labormarkets. Drier northern areas have a morepronounced seasonal labor peak and a longerslack period for seasonal migration. More de-nsely populated areas generally have higherrates of both seasonal and permanent rural-rural migration.

Looking to the future, rural population pres-sure wil l undoubtedly increase, placing moreemphasis on land shortages and declining fertil-ity and further differentiating households intolabor sellers and buyers. In areas of low popula-tion pressure, the increasing dependency ratioth rough higher fer t i l i ty , age-selective out-immigrat ion, and the spread of rural educationwi l l probably aggravate the seasonal laborbottleneck. At the same t ime, seasonal north-

8. Whether wages are paid in cash or k ind is notimportant because this is also the "hung ry " periodof food shortage for household consumpt ionforcing many households to use cash to purchasefood on the market (King and Byerlee 1977).

south migration may decline partly because ofincreased political problems of internationalmigration (as has already happened to someextent) and more emphasis on food crop versuscash crop production in coastal countries as a result of increasing food deficits (e.g. in Ghana).

Impl icat ions for Agr icu l tura lTechno logy

Clearly the seasonal labor bottleneck has im-portant implications for understanding currentfarming systems and technologies and for in-troducing new ones. To some extent currentsystems use enterprise combinations that havecomplementary seasonal labor needs. Forexample, Franzel (1978) shows that all majorenterprise combinations except one in ruralSierra Leone included enterprises wi th com-plementary seasonal labor profiles (i.e. diffe-rent peak labor requirements). Likewise cottonis planted in the north of Nigeria after foodcrops to minimize labor competition eventhough earlier planting dates are recommended(Norman and Pryor). However, the ability tocombine enterprises and stagger planting datesundoubtedly declines as rainfall and the lengthof the growing season decrease. The one enter-prise combination in Sierra Leone with highlycompetitive labor inputs only occurred in thenorth (Franzel 1978).

The existence of the seasonal laborbottleneck opens the possibility of labor-savingtechnology at this period. Tractor mechaniza-tion of land preparation has proved costly(Byerlee et al. 1976) and in any event does notbreak the weeding bottleneck. Herbicides offerpotential to do this but this imposes a cashexpense at the period of acute cash shortageand is also complicated by the widespreadinterplanting of several crops. Oxen power forboth land preparation and weeding may offerthe greatest potential wi thout greatly increas-ing cash costs (Norman and Pryor). Biological-chemical technologies that raise yields ofcourse add labor at harvest t ime but at the sametime usually require some labor at the peakseason even if in the form of better manualweed control. There is a danger that onlyhigher-income farmers wi th cash to hire labormay be able to take advantage of thesetechnologies unless adequate credit or input

subsidies are provided to low income farmers(Matlon 1977).

Technologies that utilize labor in the slackseason (e. g. capital improvements to the land,irrigation, longer harvest periods for higheryielding technologies) must consider the op-portunity cost of labor in off-farm activities andseasonal migration which may provide returnsup to three times higher than in agriculture.9

Impl icat ions for Researchon Rural Labor M a r k e t s

It is clear from the evidence presented in thispaper that there is a wide range of labor marketsituations in the SAT reg ions of West Africa andthat the design of agricultural research prog-rams in the region must take into account thelocal situation in the labor market. Most evi-dence presented in this paper was obtainedfrom "cost route" studies to obtain daily data onlabor inputs throug hout the year. While this hasprovided a useful quantification and under-standing of the labor allocation process it isdoubtful that this methodology can be repli-cated over a wide area of the SAT regionsbecause of the t ime and resource costs ofimplementing it.

Recent experience by Collinson et al. (1978) inareas of East Africa where labor shortages are a critical factor in understanding current farmingsystems and technologies and introducing newones suggests that much information can beobtained from well focused one-contact sur-veys based on a systematic familiarization withfarming systems in the area prior to the survey.Information has been successfully obtained onlabor force participation, busy seasons, wagerates, labor hiring and migration, and mostimportantly, farmers' make decisions in thelight of these labor market characteristics.Where more detailed labor input informationfor specific tasks and crops is required, moreintensive interviews of a small sample of far-mers may be necessary.

9. For example, the agricultural wage rate in Ghana isnow reported to be US $5/day compared w i th US$1/day in the Sahelian countries.

355

References

BEALS, R. E., and MENEZES C. F. 1970. Migrant labor

and agricultural output in Ghana. Oxford EconomicPaper 2 1 : 109-127.

BERG, E. 1965. The economics of the migrant laborsystem. In Urbanization and migrat ion in WestAfr ica, ed. Hilda Kuper. Berkeley: University ofCalifornia Press.

BYERLEE, D., EICHER, C. K., LIEOHOLM, C., and SPENCER, D.

S. C. 1976. Rural employment in tropical Afr ica:Summary of f indings. Afr ican Rural Economy Work-ing Paper 20, Michigan State University, East Lans-ing, Mich , USA.

BYERLEE, D., TOMMY, J. L, and FATOO, HABIB, 1976.

Rural-urban migration in Sierra Leone: Determinantsand policy implications. African Rural Economy Paper13, Michigan State University, East Lansing, Mich,USA.

CALDWELL, J. C. 1969. African rural-urban migrat ion.Canberra: Austral ian National University Press.

CHAMBERS, R. 1978. Seasonal dimensions to ruralpoverty. Institute for Development Studies, Univer-sity of Sussex, Brighton, UK.

COLLINSON, M., et al. 1978. Demonstration of an interdis-cipl inary approach to planning adaptive agriculturalresearch programs; Serenje District, Zambia ReportNo. 3, CIMMYT, East Afr ican Regional EconomicProgram, Nairobi, Kenya.

EICHER, C., ZALLA, T., KOCHER, J., and W I N C H , F. 1970.

Employment generat ion in Afr ican agriculture. Insti-tute of International Agricul ture research report 9,Michigan State University, East Lansing, Mich, USA.

FRANZEL, S. 1978. Enterprise choice, enterprise com-binations and income distr ibut ion among smallfarmers in Sierra Leone. Unpubl ished M. S. Thesis,Michigan State University, East Lansing, Mich, USA.

GREGORY, J. 1971. Migrat ion in Upper Volta. Afr icanUrban Notes 61(1): 4 4 - 5 2 .

GUNTHER, H. 1978. Empirical evidence on factorsaffecting female and household labor al location inrural Sierra Leone. M. S. Thesis, Michigan StateUniversity, East Lansing, Michigan, USA.

HILL, P. 1972. Rural Hausa — a vi l lage and a sett ing.London: Cambridge University Press.

KING, R. P., and BYERLEE, D. 1977. Income dist r ibut ion,

consumpt ion patterns, and consumpt ion linkages inrural Sierra Leone. Afr ican Rural Economy Paper 16.Michigan State University, East Lansing, Mich , USA.

M A T L O N , P. J. 1977. The size distr ibut ion, structureand determinants of personal income among far-mers in the nor th of Nigeria. Ph.D thesis, CornellUniversity, Ithaca, NY, USA.

MATLON, P., and N E W M A N , M. Production efficiency

and the distr ibut ion of income among tradit ionalAfr ican farmers. Unpubl ished Draft Paper, Depart-ment of Agricultural Economics, Michigan StateUniversity, East Lansing, Mich, USA.

N O R M A N , D, W. 1973. Economic analysis of agricul-tural product ion and labor uti l ization among theHausa in the north of Nigeria. Afr ican RuralEconomy Paper 4, Michigan State University, EastLansing, Mich, USA.

NORMAN, D., and PRYOR, D. H. 1978. The small farmer

in Hausaland. Unpubl ished Draft Paper. KansasState University, Manhattan, Kans, USA.

SKINNER, E. 1965. Labor migrat ion among the Moshi ofUpper Volta. In Urbanization and migrat ion in WestAfr ica, ed. Hilda Kuper. University of CaliforniaPress.

SONGRE, AMBROISE. 1973. La emigracion en Masa de

Al to Volta; Realidades y efectos. Revista Interna-cional de Trabajo 87: 231-245.

SPENCER, D. S. C , and BYERLEE, D. 1976. Small farmers

in West Afr ica: A descriptive analysis of employ-ment, incomes and productivi ty in Sierra Leone.Afr icanRural Economy Working Paper 19. MichiganState University, East Lansing, Mich, USA.

TRIPP, R. 1978. Economic strategies and nutr i t ionalstatus in a compound farming sett lement of North-ern Ghana. Unpubl ished Ph.D. thesis, ColumbiaUniversity, New York, NY, USA.

UDO, R. 1964. Rural migrat ion and problems of ag-ricultural labor in Western Tropical Africa. In Spatialaspects of development, ed. R. Hoyle. London.

WORLD BANK. 1979. Focus on migrat ion. Wor ld Bankreport, Washington, DC, USA.

356

Labor Use and Labor Marke ts in Semi-Ar idTropical Rural Vi l lages of Peninsular India

James G. Ryan, R. D. Ghodake , and R. Sar in *

Abstract

Comparisons of household data from six villages in South India with experimental

watersheds show that prospective crop-, land-, and water-management technologies for

semi-arid tropical (SAT) India appear to offer considerable scope for increased

employment, particularly on Vertisols. Wide seasonal variations in wages, daily labor

market participation rates, and probabilities of employment exist for both males and

females. Hence there seems good scope for designing technology to capitalize on low

opportunity cost periods. Labor opportunity costs are calculated for use in project

analyses. Women presently comprise about half the total agricultural labor use in the

south Indian SAT villages. Women participate in these daily labor markets more than

men, although their wages and chances of obtaining employment are much less. There

was little evidence that the "dual labor market" hypothesis operates in these sexually

segmented daily labor markets.

Most developing countries of the semi-aridtropics (SAT), particularly in Asia, have a rela-tive abundance of labor resources in propotionto capital and land. Statistics on this apparentabundance are usually only available (if at all) interms of national or regional annual aggre-gates, as pointed out recently by McDiarmid(1977, pp 9-10, 18, 29, 54-55), K. Bardhan(1977), Brannon and Jessee (1977, pp 13-15).Even these statistics are often not reliable,particularly for the rural areas, where problemsof seasonal unemployment are most acute, asrevealed in Rudra's comprehensive study(1973). It is imperative to derive better measuresand understanding of the demand and supplyparameters of rural labor markets, particularlyin India, where 70% of the labor force areclassified as agricultural workers.

With a more precise knowledge of seasonallabor markets parameters and behavior, it wil lbe possible to more expl ici t ly designtechnologies and projects to capitalize onperiods when opportunity costs are relativelylow. This wil l enhance the basic economic viabil-ity of the technologies and projects, and hence

* Economist and Leader, Research Fellow,Research Technician, ICRISAT.

and

their acceptability, as well as enable incomes ofthe agricultural labor force to be increasedduring seasons when it is constrained.

This paper is an attempt to bridge some of thegaps in our knowledge by studying the laborutilization patterns and labor market behaviorof a sample of 240 labor and cultivator house-holds in Maharashtra and Andhra PradeshStates of SAT peninsular India. This region haslargely been neglected in this field of research.

The first section of the paper describesthe data on which the analyses are made;the second section examines the labor usepattern of the 180 cultivator households. In-cluded in this is a discussion of regional andfarm size differences in labor use, differencesbetween males and females, differences acrossseasons, and comparisons made with labor useprojections under the prospective watershed-based soil and crop management technologybeing researched at ICRISAT Center. The thirdpart of the paper discusses seasonal participa-tion rates, probabilities of employment, wagerates and opportunity costs for male and femalelabor in rural labor markets and how these varyacross regions, farms and between sexes.Some tests of Sen's (1966) dual labor markethypothesis are made, as well as McDiarmid's(1977) hypothesis that rural labor markets are

357

more "compet i t ive" in peak seasons. A conclu-sion fol lows.

Data

The data for this study are drawn f rom sixvillages in the SAT of peninsular India in whichintensive socioeconomic studies by theEconomics Program at ICRISAT have beenunder way continuously since May 1975. Sincethat t ime resident investigators have been in-terviewing at 2 to 4 week intervals, a stratifiedrandom sample of 30 cultivator (small, mediumand large size) and 10 labor households. Detailsof labor utilization of each family member andof hired personnel were obtained. These datarelated to both on-farms and off-farm activitiesas well as to household use.1

The six villages were purposefully selected torepresent three broad agroclimatic zones ofSAT peninsular India. Aurepalle and Dokurvillages in Mahbubnagar District of AndhraPradesh were selected to represent the Alfisollow (713 mm) and uncertain rainfall areas inAndhra Pradesh, Karnataka and Tamil Nadu.Shirapur and Kalman villages in Sholapur Dis-trict of Maharashtra represent the deep andmedium-deep Vertisol, low (691 mm) and un-certain rainfall areas of Maharashtra, Karnatakaand Madhya Pradesh. The villages of Kanzaraand Kinkheda in Akola District of Maharashtrawere chosen as typical of the relatively high(819 mm) and more assured rainfall areas ofnorthern Maharashtra and Madhya Pradeshhaving medium-deep Vertisols.2

The present analysis of farm labor utilizationand rural labor markets in these six villagesapply to the agricultural year 1975-76. Data forsubsequent years is presently being analyzed.

Labor use data f rom the watershed-basedresearch being conducted by the ICRISAT Farm-ing Systems Research Program since 1974 has

1. The on-farm and off- farm activities were col-lected on a 2-to-3 week full recall basis. Householdt ime al location of each household member wascollected only for the day immediately precedingeach interview.

2. For a detailed descript ion of the methodology,vi l lages, and the complete range of informat ionsee Binswanger et al. (1977), Jodha et al. (1977)and Binswanger and Jodha (1976).

been used as a basis of comparison wi th thecurrent village situation. The basic concept inthe prospective technology involves manage-ment of the soil and water on a small catchmentor watershed basis ranging in size from 1 to 50ha. High-yielding varieties (HYVs) of crops areshown wi th improved fertilizers and crop man-agement on broadbeds which are establishedbetween furrows. The broadbeds and furrowsare established with improved animal-drawnimplements on a graded slope between 0.4 and1.0%. This is to enable excess runoff f romheavy rainfall storms during the rainy season tobe guided slowly across the natural grade(usually 1.5 to 2%). In this way rainfall penetra-tion into the root profile of crops is increasedand soil conservation improved.

We have selected several of the experimentaltreatments which have been found promisingby the scientists of ICRISAT's Farming SystemsProgram to compare with the village data. Theexperimental treatments have been matchedwith villages having similar soil types in themanner shown in Ghodake et al. (1978, Table 9).

Farm Labor Ut i l i za t ion

In this section we wil l discuss the patterns oflabor use under the technologies and croppingpatterns presently employed in these six vil-lages. These results are presented in muchgreater detail in the paper by Ghodake et al.(1978).

In these comparisons we confine ourselves tolabor used in crop-related activities. We excludelabor used for domestic work, animal husban-dry, marketing, handicrafts, etc.

Regional and Village Comparisons

We observe a wide variability across the sixvillages in total labor use per hectare of cultivatedland, whether expressed on a net or grosscultivated area basis (Table 1).3 The Alfisol

3. Paired t-tests of the differences in for tn ight lylabor use in each vil lage for 1975-76 comparedw i th 1976-77 showed that only in Mahbubnagardid labor use signif icantly increase in 1976-77.The ranking of vil lages in terms of labor use perhectare in the 2 years was also the same.

358

Table 1. H u m a n labor use in man-equiva lent hours by f a r m aiz a groups in six S A T vi l lages ofpeninsular India ( 1 9 7 6 - 7 6 ) .

District/Village

MahbubnagarAurepalle

Dokur

SholapurShirapur

Kalman

AkolaKanzara

Kinkheda

Farm sizegroup

SmallMed iumLargea

All Farmsb

SmallMediumLargeAll Farms

SmallMediumLargeAl l Farms

SmallMed iumLargeAll Farms

SmallMediumLargeAll Farms

SmallMed iumLargeAl l Farms

Per hectare ofnet cult ivated

land

242320650***540

209312521022**1156***

613426309*380

222266169*2 1 1 * * *

475472406425

562449484483

Per hectare ofgross cult ivated

land

2423135 5 1 * * *479

1566989920* *9 9 4 * * *

550370278*338

211241158*195** *

448441389404

540438447453

a. Aster isks used fo r the " l a rge f a r m " category Indicate large f a r m f igures are signi f icant ly di f ferent f r o m the respect ive smal lfa rm f igures.

b. Aster isks used fo r the "a l l f a r m " category Indicate values for the concerned vi l lage are signi f icant ly di f ferent f r o m therespect ive values of the other v i l lage In the same reg ion.

* * * , ** a n d * represent s igni f icant di f ferences at the 1, 5, and 1 0 % levels, respect ively.

villages in Mahbubnagar District — with paddy,groundnut, castor, sorghum, pearl millet andpigeonpeas predominating — use two to threetimes more labor per hectare than the villages inSholapur and Akola Districts. The primaryreason for this seems to be greater use ofirrigation in Mahbubnagar, particularly in Dokurvillage, where 45% of its net cultivated land isirrigated from tanks and wells.

Although the two Sholapur villages havemore irrigation than those in Akola, (10 vs 2%),

they use about one-third less labor per hectare.According to Jodha (1977), some 60% of thecropped land in the Sholapur villages is leftfallow in the rainy season and cropped in thepostrainy season only, compared with less than5% left fallow in Akola in the rainy season.Postrainy season crops require less labor thanrainy-season crops. In the Akola villages, mixedcropping predominates (70 to 84%) whereas inSholapur it is primarily sole cropping (57 to86%). Sole crops have a lower labor require-

359

ment per hectare. Another factor leading tomore labor use in Akola is the dominant rolecotton plays in the cropping pattern. InSholapur, sorghum is the dominant crop, fol-lowed by chickpea and safflower.

The ranking of villages with respect to laboruse is similar whether expressed on a net orgross cultivated ha basis. This suggests that theintensity of cropping either does not contributesubstantially to differences in labor use inten-sity in these six villages or other factors such ascropping patterns and resource endowmentdifferences must offset the effects of croppingintensity differences as postulated by Vyas(1964).

The cotton-growing villages in Akola usedmuch more hired labor as a proportion of theirtotal requirements (73 to 82%) compared wi ththe other villages, where foodgrains generallypredominate (42 to 58%).

Farm Size Differences

In three of the six villages (Dokur, Shirapur, andKalman) small farms used significantly morelabor per hectare than large farms (Table 1).This is in line wi th previous work by Rudra(1973), Bharadwaj (1974) and Bardhan (1973).However, in the two cotton-growing Vertisolvillages in Akola there were no signifi-cant differences in labor use per ha on thesmall and large farms. In Aurepalle village largefarmers used significantly more labor per hec-tare than small farmers, mainly because ofmore irrigation (10% vs 0%), bullocks, andimplements on the large farms.

Are these differences between small andlarge farm labor use due to a differential in theopportunity costs of labor between these twocategories, as postulated by Agarwal (1964),Mazumdar (1963 and 1965) and Sen (1966)?Opportunity costs are defined here as the cur-rent wage rate times probability of employmentin the daily labor market. In a subsequentsection on rural labor markets in this paper anattempt wil l be made to test this hypothesis.

Small farmers in general hired less labor as a proportion of their total labor use than did largefarmers. Hired labor small farms ranged from29% in Aurepalle to 68% in Kanzara; for largefarms it ranged f rom 45% in Shirapur to 9 1 % inKinkheda. Hence, although large farmers reliedmore heavily on hired labor than small farmers,

the use of hired labor by the latter was by nomeans insignificant. The larger the farm size,the higher the proportion of hired female laborin total labor use. Large farmers employed 36%of all hired labor in Kalman village and 92% inAurepalle; these represent the min imum andmaximum values.

Labor

In Dokur, Shirapur, Kanzara, and Kinkheda vil-lages, the total female labor use on cropsexceeded total male labor. In Aurepalle andKalman there was slightly more total male laborused. These high proportions of female laboruse on agricultural land far exceed the 20%figure cited for Asia by Boserup (1970). Littlework on crop activities is performed by children.Increasing amounts of cotton and irrigationseem to imply increasing employment poten-tials for women in these villages. In all sixvillages, the proportion of female labor hired ismuch higher than the proportion of male laborwhich is hired. In the Mahbubnagar and Akolavillages 80 to 90% of total female labor is hired.In Sholapur the figure is somewhat less (60 to70%). In the Mahbubnagar villages and inShirapur village in the Sholapur area, less than30% of total male labor is hired. In Kalmanvillage, also in the Sholapur region, the figure is50%, whereas in the Akola villages more than66% of male labor is hired.

The work of Binswanger and Shetty (1977)showed that new technologies such as use ofherbicides in the SAT can have adverse con-sequences on the demand for female laborrequired for handweeding. Almost all hand-weeding is done by hired females, the mostdisadvantaged of all the labor categories in SATIndia. The present analysis of the overall pictureof female labor use in these villages suggeststhat any changes that affect female labor wi l llargely affect hired females, particularly in irri-gated and/or cash crop situations.

Of the total hired labor used in the Aurepalle,Dokur and Shirapur villages, 63 to 88% con-sisted of females. In the Akola villages the malesand females shared almost equally in total hiredlabor use. Males always represented the largestproportion (ranging from 64 to 90%) of the totalfamily labor utilization, especially in theMahbubnagar villages. The range was from 64to 90 percent amongst the villages.

360

Seasonality

In all six villages the intraseasonal variability oftotal labor use per hectare, as measured by thefortnightly coefficients of variation, was sig-nificantly higher on small farms than on largefarms (Table 2). The variability of per hectarelabor use seemed to be negatively related to thenumber of different crops and/or crop mixturesgrown on farms as an increase in the number ofcrops helps in evening out labor demands andresults in lower CVs. In all cases small farmersgrew significantly fewer crops than did largefarmers (Table 2), even though in most casesthe former have been shown by Jodha (1977) tohave a higher proportion of their cultivated areasown to mixed or intercrops.

The higher CVs on small farms occurred inspite of the higher mean level of labor use theyhave in Dokur village and the two Sholapurvillages. Hence small farm intraseasonal varia-bility is quite large in comparison to that of largefarms. The reason for this could be that they

choose cropping patterns that free them forwage employment off their own farms at irregu-lar intervals. We wil l show later that they par-ticipate in the daily labor market in and aroundthese villages in a significant manner.

The correlation coefficients between totalfortnightly labor use per hectare on large farmsin each village and the amount of hired labor perhectare in the corresponding fortnight were allaround 0.90 or greater, and they were sig-nificantly different from zero at the 0.1 percentlevel.4 This suggests that hired labor demandchanges at least in proportion to total seasonal

4. The correlations for small farms were lower thanthose for large farms, but still signif icantly dif fe-rent f rom zero and greater than 0.80 in three of thesix villages. The correlation coefficients betweenfortnight ly total and fami ly labor use for allfarms were also signif icantly posit ive in allvillages except in Kinkheda, but these valueswere always lower than the corresponding valuesbetween total and hired labor use.

Table 2. Coef f ic ient of var ia t ion (%) in fo r tn igh t l y labor use in six S A T vi l lages of peninsular India( 1 9 7 5 - 7 6 ) .

District/

Vil lage

MahbubnagarAurepal le

Dokur

SholapurShirapur

Kalman

AkolaKanzara

Kinkheda

Farm size

Small

154(1.8)a

151(1.6)

213(4.2)

147(4.9)

146(2.1)

137(2.4)

Medium

150(2.5)

175(3.2)

157(7.4)

113(11.0)

122(3.2)

120(2.5)

Largeb

9 7 * * *(7.1)***

142*(4.0)***

139* * *(8.7)*

95 * *(11.1)***

9 4 * * *(7 .1)***

9 5 * * *

(5.6)**

Al l farms c

133(3.9)

156(2.9)

170(6.8)

1 2 1 * * *(8.9)*

121(4.1)

117(3.5)

a. Figures in parentheses are the average n u m b e r of di f ferent sole crops and/or di f ferent crop mix tures on these fa rms.

b. Aster isks here Indicate " l a rge t e r m " f igures are s ign i f icant ly di f ferent f r o m the respect ive " sma l l f a r m " f igures.

c. Asterisks here indicate s igni f icant dif ferences exist be tween vi l lages In the same Distr ict .

* * * , * * and * represent s ign i f icance at the 1, 5 , a n d 1 0 % levels, respect ively.

361

labor demand. The correlation between totalfortnightly labor use and the percentage ofhired labor used in the same fortnight was notnearly as high as that mentioned above. In fourout of six villages the range in r values for allfarms was 0.34 to 0.46, wi th three significant at 5 percent and one at 10 percent. In Kalman andKanzara there was no significant correlation.Hence in four villages, whenever seasonal labordemand rises, hired labor's share in this alsorises. Contrary to this, in five out of six villages,correlation coefficients between total labor andproportion of family labor to total labor usewere significantly negative while in one villagethere was insignificant negative correlation.

Existing and ProspectiveTechnologiesThe labor used per hectare of net cropped landunder the prospective broadbed and furrow,improved crop and watershed managementtechnologies researched at ICRISAT Center was100 to 500% greater than the existing labor useobserved in five of the six villages (Table 3).5

(The exception was the highly irrigated Dokurvillage where existing average labor use ex-ceeded projected requirements of the newtechnology under Alfisol conditions.) This wasnot the case in Aurepalle which has much lessirrigation. The new technology doubles the pre-sent village labor requirement.

On the Vertisols the potential for increasedlabor demand under the new technology seemsgreater than on Alfisols, particularly in Sholapurdistrict where labor use could increase by morethan four times.6 In the Akola Vertisol region

5. When labor use was expressed per gross croppedha the prospective technologies generated bet-ween 10 and 140% more than existing vi l lagetechnology (excluding Dokur vi l lage, which hadmore labor use per gross cropped ha due to thepredominance of irr igated paddy).

6. The Sholapur vi l lage compar ison w i th the newtechnology at ICRISAT Center Hyderabad shouldbe viewed w i th some caut ion, however. AsVirmani et a l . (1978, pp 2-7) show, the prob-abil i ty of successfully g rowing rainy seasoncrops at Hyderabad is signif icantly higher than inSholapur. As the new technology pr imar i ly de-pends for its superiori ty on rainy season croppingfo l lowed by postrainy season cropping, it is notclear that in Sholapur this wi l l be possible wi thoutconsiderable risk.

there is scope for almost triple labor demandwith the new technology. These increased de-mands apparently wi l l occur without substan-tially affecting the CVs of fortnightly labor useper hectare throughout the year, providing, aswas the case here, that it is based on more thanone cropping system (Table 3).7

The extent of apparent surplus labor availableafter attending to farm crop activities is anindication of the degree to which laborbottlenecks are likely to arise under differentsituations. These apparent surpluses have beencalculated for both the existing village and theprospective technology scenarios (Table 3).They are computed as the sum of the fortnightlydifference between the apparent supply ofable-bodied man-equivalents per hectare in thevillage (including cultivator and labor house-holds) and the fortnightly requirements for cropactivities. Apparent supplies have been calcu-lated assuming 14 days of work a fortnight at 7 hours per day.

Considering other labor commitments, thepressures on available annual village laborfrom the prospective technologies would ap-pear to be greatest in the Vertisol villages ofAkola and Sholapur. In the former region, theapparent surplus would drop from 82 to 45%with the prospective technology and in thelatter from 90 to 53%. The fall in the case of theMahbubnagar villages would only be from 83 to75%.

When we focus on the three peak fortnightswi th the apparent surplus calculation we obtaina picture of the effects of the prospectivetechnologies on seasonal labor bottlenecks(Table 3). These seasonal peaks would appearto be greatest under the prospective technologyin the case of the Sholapur and Akola Vertisolregions. From an apparent surplus labor posi-t ion of 80 and 60% in the existing three peakperiods in these Sholapur and Akola villages,we would move to deficit situations of approx-imately 120 and 80%, respectively. The surpluswould fall from around 60% to zero in the caseof the Mahbubnagar villages.

7. The CVs of the individual cropping systems laborrequirements under the new technology weremost ly much higher than those presently existingin the vil lages (Ghodake et al. 1978, Tables 9 and10).

362

363

The three major peak periods would shiftunder the prospective technology in the vil-lages, in Mahbubnagar (Fig 1) they would movefrom July, August, and November to August,October, January, and November. In the case ofSholapur (Fig 2) the move would be f romOctober, January, and February at present toSeptember, October, and January. In Akola (Fig3) the existing peaks in August, October, andNovember would change to September,November, and January.

The major bottleneck under the new technol-ogy would appear to be at harvest and thresh-ing t ime of the first crop in either the sequentialcrop or the intercrop system. In addit ion, therewould be bottlenecks at weeding t ime after theinitial sowing. The scope for spreading out theharvesting/threshing period to alleviate thestrain on labor at that t ime would appear to begreatest under an intercrop system, wherethere is less pressure to harvest the first croppromptly than wi th a sequential crop system,where the first crop must be removed before the

second can be sown.8 If, as Ryan et al. (1978)show for ICRISAT Center, intercropping is moreprofitable and less risky than sequential crop-ping, then the superiority of a new soil, waterand crop management technology based on anintercrop system seems evident.

R u r a l L a b o r M a r k e t s

In this section we discussed the functioning ofdaily rural labor markets in the six villages, par-ticularly attention is given to the extent to whichable-bodied people attempt to participate in thedaily labor market throughout the year, whatwages they receive, and with what probabilitythey obtain employment. This wi l l be examinedseparately for males and females from the four

8. Relay cropping is a possibil i ty but there is a risk ofdamage to seedlings if harvesting of the f irst cropoccurs after sowing of the second crop.

Standard fortnight

Figure 1. Average seasonal human labor availability and use with existing and prospective

watershed-based technology on Alfisols.

364

Exis t ing —Aurepal le V i l l a g e , Mahbubnagar (1975-76)

Prospect ive watershed-based technology at ICRISAT Center[Average of Sorghum-ratoon sorghum sequential and p igeonpea /se ta r iaintercrop —excluding threshing (1976-77)]Prospect ive watershed-based technology — inc lud ing threshing (1976-77)

Ex is t i ng -Sh i rapu r V i l l age , Sholapur (1975-76)

Prospective watershed-based technology at ICRISAT Center[Average of maize-chickpea sequential and p igeonpea/maizeintercrop —exclud ing threshing (1976-77)]

Prospective watershed-based technology —inc lud ing threshing (1976-77)

6 8 10 12 14 16 18 20 22 24 26 2 4 6 Apr May June July Aug Sept Oct Nov Dec Jan Feb Mar

Standard fortnight

Figure 2. Average seasonal human labor availability and use with existing and prospective

watershed-based technology on deep Vertisols.

categories — labor, small, medium, and largefarmers — using data collected on a 2- to 4-week recall basis throughout the 1975-76 year.9

One key measure to be derived f rom theabove statistics is the seasonal opportunity costof labor, measured as OCt where:

(1) OCt = Wt P(Et)and Wt = actual equil ibrium wage rate in

fortnight t.

9. For details see Subrahmanyam and Ryan (1976,pp, 11-16). Most estimates reported here areannual averages. However the comparisons ac-ross farm size groups, vi l lages, and sexes werealso made separately for "peak" and "s lack"labor seasons. In almost all cases results for theslack seasons were consistent w i th the annualaverage.

P(Et) = probability of obtaining dailywage emp loymen t inperiod t.

It is important to know the opportunity cost oflabor in different seasons as some of the pros-pective technologies for rainfed areas, such aswatershed-based land and water managementtechnologies that are under investigation atICRISAT and in the All India Coordinated Re-search Project for Dryland Agriculture, requireengaging labor in capital-creating activitiessuch as building tanks, wells, bunds, levellingland, etc. For these it is necessary to know theseasonal opportunity cost of labor in order todetermine the appropriate time for carrying outsuch activities. More importantly seasonal op-portunity cost data are required to enableproper benefit-cost analyses of these types oftechnologies.

365

Exis t ing-Kanzara V i l l age , Ako la (1975-76)

Prospective watershed - based technology at ICRISAT Center[Average of ma i ze - ch i ckpea sequent ia l and p igeonpea/ maizein te rc rop -exc lud ing threshing (1976-77)]Prospect ive watershed - based technology - inc lud ing threshing (1976-77)

6 8 10 12 14 16 18 20 22 24 26 2 4 6

A p r May J u n e J u l y A u g Sept O c t Nov Dec J a n Feb Mar

Standard f o r t n i g h t

Figure 3. Average seasonal human labor availability and use with existing and prospective

watershed-based technology on medium-deep Vertisols.

In many instances data can be obtained onseasonal rural market wage rates, but there isno way of knowing how close these are to theseasonal opportunity costs of labor, which areat the heart of questions related to labor supplyanalysis and the value of household productionand t ime. McDiarmid (1977, pp 9-10, 18, 29,54-55), refers to the necessity of local or re-gional measures of seasonal opportunity costs(as opposed to the countrywide single mea-sures he derives) under conditions wheremin imum wage laws and religious and lan-guage constraints prevent free labor mobil i ty.These circumstances generally prevail in theparts of the Indian SAT studied here. Theopportunity cost equation (1) above is equiva-lent to the first of the three elements of McDiar-mid's shadow wage rate — namely the " i m -mediate cost" or the alternative production thatwould have occurred in the opt imum availablealternative employment. Subrahmanyam andRyan (1976, pp 9-10) felt the opportunity costmeasure in (1) is preferable to other alternativessuch as wages of attached servants, marginalvalues productivities (MVPs) derived fromproduction function, or linear programmingstudies.

The interval between interview rounds varied

366

from village to village and from household tohousehold. In addit ion, it was not possible todate the days of employment and involuntaryunemployment.10 For these reasons a smoo-thing procedure was used whereby the frac-tions of participation rate, employment andinvoluntary unemployment during the roundinterval were assigned to each day in the roundand then summed up over standard fortnights.Details of this can be found in Subrahmanyamand Ryan (1976, pp 11-16).

Participation Rates

The participation rate is defined as the propor-tion of days in the period in which a personattempted to f ind a job on the daily labor marketin and around the village. The numerator in-cludes the days the person was successful in

10. Investigators asked each household memberhow many days they obtained wage employmentsince the last interview and how many days theyt r ied but were unsuccessfu l . The lat ter rep-resented days of invo luntary unemployment .Wage rates referred to the average of the wagesreceived by all respondents in each category dur-ing the round.

f inding wage employment plus the days he wasnot. The latter are subsequently referred to asinvoluntary unemployment days. The particip-ation rate does not include days of work on theirown farm.

Participation rates for males were sig-nificantly lowest in the two Mahbubnagar vil-lages and generally highest in the two Akolavillages (Table 4). From Appendix Table 1, it canbe seen that interdistrict participation rateswere all significantly different for males as wel las females. For only 7% of the t ime in Aurepalledid males endeavor to f ind a job on the dailylabor market. In Kanzara, on the other hand, themales participated almost 50% of the t ime. Thewhole-sample participation rates for all villagesare reduced by the meager participation bylarge farmers and, to a lesser extent, by mediumfarmers. Participation by labor and small-farmhouseholds is generally much higher. For malesthe highest average participation rate during1975-76 for both the labor and small farmgroups was in Kanzara at 0.87 and 0.70 respec-tively, while the lowest was in Aurepalle at 0.18and 0.14.

Females participated substantially more (al-ways significant at 0 .1% level) than males inMahbubnagar villages and in Kinkheda while inother villages males participated significantlymore than females. Dokur registered the high-est participation on the average for females at0.61. In this village even females from largefarms participated 36% of the t ime, whereas inboth labor and small-farm households femalesparticipated 82% of the t ime, the highest of allvillages. This indicates the influence of paddyirrigation on demand for female labor. It is suchas to lead to substantial labor market particip-ation by females, even those from the large-farm group.

The lowest female participation came in theSholapur villages at around 0.25. The rangewithin these two villages, in which foodgraincrops predominate, was a high of around 0.47for females from labor households to a low of0.08 for those from the large-farm group.

There was a significant amount of seasonalvariation in participation of males and females,particularly in Aurepalle (Table 4). The CVs offortnightly male participation ranged from a high of 48% in Aurepalle to a low of 14% inDokur and Kanzara. For females the range wasfrom 6 1 % in Aurepalle to 10% in Dokur. Again

the effect of extensive paddy irrigation in Dokuris reflected in a much more steady labor forceparticipation throughout the year.

The mean fortnightly participation rates ofmales and females from labor households werealways greater than those from the small-farmhouseholds. The differences were significant atthe 1 % level using paired t-tests in five out of sixvillages for males and two of six for females(Appendix Tables 2 and 3). The labor group alsohad higher participation rates than medium-farm households for males and females. Thesedifferences were statistically significant at the1% level in five of six villages for males and in allvillages for females. Except for Shirapur, malesand females from small-farm households par-ticipated significantly more than those f rommedium-farm households.

The simple correlation between the partici-pation rates of males from labor and from small-farm households was positive and significantin four of the six villages (Appendix Table 4).Aurepalle and Kalman had negative correla-tions, the latter not being statistically sig-nificant. In five of six villages there was nocorrelation between the participation of themale members of the labor households in themarket and that of those from medium farmhouseholds. For females (Appendix Table 5) thecorrelations were significantly positive in threecases between participation of those from thelabor and small-farm categories. The otherthree village correlations were not significant.Only two correlations were significantly posi-tive between labor and medium-farm femaleparticipation rates; one was significantly nega-tive and the others were not significant.

These results suggest, as expected, thatmales and females from labor households par-ticipate more in the labor market than thosefrom cultivator households. Those from thelabor group tend to enter the labor market at a similar t ime to those from the small-farm group.However, there does not seem to be as muchcompetition between participants f rom thelabor group and those from medium-sizedfarms, especially among males.

Probabilities of Employment

The probability of employment was calculatedas the number of days a person was successfulin obtaining wage employment as a proportion

367

Table 4 . Average par t ic ipa t ion ra tes , emp loymen t probabi l i t ies, oppor tun i ty costs and w a g e ra tesfor adu l ts in rura l dai ly tabor markets in six S A T vi l la ges of peninsular India, 1 9 7 5 - 7 6 .

District/Village/

Sex

MahbubnagerAurepalle

MaleFemale

DokurMaleFemale

SholapurShirapur

MaleFemale

KalmanMaleFemale

AkolaKanzara

MaleFemale

KinkhedaMaleFemale

Participationrates

Mean

0.07*** a

0.27

0.25***0.61

0.38***0.25

0.29***0.22

0.48***0.39

0.30***0.36

CV

(%)

4861

1410

2922

1917

1422

2220

Probabilityof employment

Mean

0.710.69

0.76***0.82

0.70***0.49

0.92***0.77

0.82**0.77

0.88***0.91

CV

(%)

2320

3020

1237

1118

1230

127

Daily oppor-tunity cost

MeanRs

1.77***

1.03

1.97***1.58

1.80***0.68

2.50***1.08

2.05***1.41

2.52***1.41

CV

(%)

2526

3530

1247

3128

1535

1922

Daily wagerates

MeanRs

2.50***1.49

2.59***1.93

2.57***1.39

2.72***1.40

3 .72** *1.83

2.86***1.55

CV

(%)

3316

1715

1318

1914

1112

1119

a. Asterisks indicate s igni f icant di f ferences be tween male end female f igures of the same vi l lage:* * Signi f icant a t 1% level

* * * Signi f icant a t 0 . 1 % level .

of the number of days in the period he tried.11

In the two drought-prone, predominantly food-grain producing villages of Sholapur region andin Kanzara village which has relatively assuredrainfall wi th a predominance of cotton, maleshave a significantly better chance of obtainingdaily wage employment than females. In Au-repalle village there is no significant difference.On the other hand, in Dokur and Kinkhedafemales had a significantly higher probability ofemployment than males (Table 4). In Shirapur

11. In calculating the probabi l i ty of employment wehave not dif ferentiated between employment onother farms, in nearby urban areas, for private orgovernment employers.

village females succeed in finding employmentonly 50% of the t ime, compared with males,who are successful 70% of the time. In Shirapurnot only are average female probabilities ofemployment low but their fluctuation through-out the season is particularly high also, with a CV of 37%, whereas for males it is much less(12%). The range in employment probabilitiesfor females was from a low of 0.2 in October1976 to a high of 0.7 in November 1975. Therange for males was from 0.5 in October 1976 to0.8 in June 1976. In Kalman village of the sameregion, employment probabilities for bothmales and females were substantially better at0.92 and 0.77, respectively, than in nearbyShirapur. Seasonal variations were also less.

368

This illustrates the difficulties of generalizingresults from individual villages to the region inwhich they are located.

The most buoyant daily labor markets appearto be in the two cotton-growing Akola villages(Appendix Table 1). In Kanzara both males andfemales succeed about eight times out of 10 infinding a job, while in Kinkheda they succeednine times in 10. Seasonal fluctuations are notsubstantial except for Kanzara females, whoseCV is 30%.

Males in Mahbubnagar are successful infinding off-farm employment about three timesin four attempts; Dokur females are successfuleight times out of 10, but in Aurepalle femalessucceed only seven times out of 10. The sea-sonal variation in job probabilities also seemshigher for males than for females in these twovillages, where irrigated paddy is important.The range for males is from a low of 0.3 to a highof near 1.0, whereas for females the comparablefigures are 0.5 to 1.0.

Paired t-tests showed that for males therewas a mixed picture with respect to the relation-ship between probabilities of employment forthe labor group and the small- and medium-farm groups (Appendix Table 2). In three vil-lages there was no significant difference bet-ween the mean probability of employment forthe labor and small-farm groups. In two villagesthe probability for small-farm males was sig-nificantly greater than for males from the laborgroup, but in one village the reverse was true. Inthree villages the probability of employment formales from the medium-size category was sig-nificantly less than that of the labor group, andin the other three there was no significantdifference. In four villages the small-farm maleshad a significantly better chance of finding a jobthan medium-farm males. In the remaining twovillages there was no significant difference.

Females from small farms in three of thevillages had significantly better employmentprobabilities than those from labor households.In one village the reverse was true but in theother two there was no significant difference(Appendix Table 3). Females from the mediumfarm group in three villages also had sig-nificantly better chances of finding a job thanthose from the labor households. There wasno significant difference in the other threevillages. Females from the small and mediumfarm group did not differ statistically in their

employment probabilities in four out of the sixvillages. In Dokur the latter had a better chancewhile in Shirapur the former did.

What the above pattern suggests is that ingeneral males from labor households have a better chance of being successful in findingdaily wage employment vis-a-vis their counter-parts in the cultivator households than do theirspouses. Hence it is not only true that femalesgenerally are no better off, and often worse off,than males in terms of daily labor market em-ployment opportunities, but in addition femalesfrom the poorest houeseholds, namely thelabor group, are often the worst off of all.

It seems that probabilities of employmentthroughout 1975-76 for females from the labor,small-, and medium-farm group in each villagetended to move more together than those of themales. This is suggested by the fact that all 12correlations were positive and significant forfemales while for males only six were (Appen-dix Tables 4 and 5). This suggests that femalesfrom the different socioeconomic groups withinthese villages tend to be competing for similarjobs more so than males.

The overall average probability of involuntaryunemployment12 for males in these six villagesduring 1975-76 averaged around 0.20, while forfemales the average was about 0.26.13 Theseestimates of the extent of unemployment are farin excess of the rates derived in the 1961 censusin India of 0.005 and 0.001 for males andfemales, respectively. They are also muchhigher than in the twenty first round of theNational Sample Survey in 1966-67 of 0.018 forrural males and 0.045 as reported in Sen (1975,pp 115-134). Our average involuntary un-employment rates compare with Mehra's figurefor disguised unemployment in the total ag-

12. The probability of involuntary unemployment iscalculated as 1 minus the probabil i ty of employ-

ment.13. The average probabilities of involuntary un-

employment for males were 0.12 and 0.39 dur ingpeak and slack periods, respectively, whi le forfemales, the corresponding figures were 0.11 and0.50. These indicate that during peak periods theprobabilit ies of involuntary unemployment formales and females become almost equal whi leduring slack periods the difference increases,affecting female labor employment more ad-versely.

369

ricultural labor force in India of 0.17 as reportedin Sen (1975, p 130), but are about double thosederived by Krishna (1973) using NSS data onrural workers who are idle but wil l ing to workmore. Krishna points out that his figures areminimum estimates.14

A number of models of rural-urban migration,such as those by Harris and Todaro (1970) andMcDiarmid (1977, pp. 75-76), assume the prob-ability of employment in rural areas to be 1. Lal(1974) reports a value of 1 as the ratio of themarket to the "social wage rate" in a number ofstates in India. These are often used as implicitweights in social benefit-cost analysis of pro-jects. The results from these six villages and theabove studies suggest that calculations ofopportunity costs of rural labor based on prob-abilities of employment of 1 may be overesti-mated. This may explain the paradox of increas-ing rural-urban migration in India in spite ofapparently increasing amounts of urban un-employment. Ignoring rural unemployment insuch models may be the villain of the piece.

Wage Rates and Opportunity Costs

Average female daily wage rates in these sixvillages in 1975-76 were about 56 percent ofthose for males and also were significantlydifferent at the 0 .1% level. Male wages aver-aged Rs 2.83 per day and females Rs 1.60(Table 4). Male wages were generally highest inthe two cotton-growing Akola villages. Kanzarahad the highest of Rs 3.72 and Aurepalle thelowest at Rs 2.50.15 Females from Dokur hadthe highest average wage rate of Rs 1.93 perday. The lowest female wages were in theSholapur villages, where they averaged onlyRs 1.40 per day.

Rodgers (1975) contended that the degree ofseasonal wage variability is less than that of

14.

15.

The various studies cited don't always use similardefinit ions of unemployment and this is oneexplanation for discrepancies. They are alsoaggregate measures involving counts of peopleunemployed on their survey day or week. Themeasures in th is study are taken over the who leyear for the same people.These are expressed in nominal terms only.Deflating by a foodgrain price index in the twostates would no doubt br ing these wages closertogether.

employment variability; we find more evidenceto support this in the case of females than we dofor males. Rodgers contends that wages aremore "sticky" than employment due to provi-sion of meals by employers and interseasonal"guarantees." We find in five of the six villagesCVs of wages are less than the CVs of theemployment probabilities for females. Formales it is true in three villages but not in theother three. Similar to Raj (1959) though, we didfind a general tendency for lower averageemployment probabilities to be associated withhigher CVs of wages. In line with Sethuraman(1972) and McDiarmid (1977, pp 48-49), wefound that whenever there was a statisticallysignificant correlation between fortnightly dailywages and probabilities of employment for thelabor and small-farm households, it was in mostcases positive (Table 5). However, this was truein only one of the 12 cases for males and in fourof 12 for females. On the other hand there was atleast one case of small farm males in Aurepallewherein the correlation coefficient value wassignificant and negative which is in line withRodgers (1975). Hence the evidence on thisrelationship is still quite weak.16

Males from the labor households have sig-nificantly lower average wage rates in four ofsix villages compared to those of small farmers,and in one of six compared to medium farmers.In one village wages of male workers f rommedium farms were significantly less thanthose from the labor category (Appendix Table2). There seems to be no significant differencesin the average wages of males from the small-and medium-farm categories.

Except in Aurepalle, there does not seem tobe any major difference between averagewages earned by females from the labor andsmall-farm households (Appendix Table 3). InAurepalle the latter have significantly higherwages than the former. In Kalman villagefemales from the labor group have significantlyhigher average wages than those from themedium-farm group. In the other five villagesthere are no significant differences. In Aurepalleand Kinkheda females from the small-farm

16. Bardhan (1977) provides an excellent review ofthe available Indian literature on the relationshipsbetween labor supply, demand and wage deter-minat ion.

370

Table 5. S imp le corre la t ion coef f ic ients bet-ween fo r tn igh t ly dai ly wages andprobabi l i t ies of emp loyment in sixS A T vi l lages of peninsular India,1 9 7 5 - 7 6 .

District/Village

MahbubnagarAurepal leDokur

SholapurShirapurKalman

AkolaKanzaraKinkheda

Labor

households

Males

0.07-0 .01

-0 .170.20

0.000.39**a

Females

0.210.46**

0.170.11

0.070.27

Small-farm

households

Males

- 0 . 6 2 * *0.02

-0.170.19

0.00-0 .09

Females

-0 .160.38*

0.39*0.23

0.50**0.10

a. ** and * indicate significance at the 1 and 5% levels,respectively.

group average significantly higher wages thanthose from medium farms.

From these results it also seems thatmales from the labor group, though theygenerally participate more in the rural dailylabor market, do not tend to receive higherwages as a result of their being more continu-ously availability throughout the year. In fact,there is evidence from these villages to thecontrary. Hence it seems that small-farm malesin these villages can hire daily workers fromlabor households at the same time that they hirethemselves out, and be better off for it. Thesame is not true for females. This no doubthelps explain the relatively high levels of laborhiring observed on the small farms by Ghodakeet al. (1978 pp 17-20).

There were significant positive correlationsbetween male participation rates and pro-babilities of employment in four of the sixvillages. For females this was true in only twovillages, and in two others the correlation wassignificantly negative. This suggests that malesthere have a better chance of obtaining a job themore they participate in the labor market. This isnot so for females. Hence in these villagesi t wou ld seem there are no s t rong

"discouraged-worker" effects operating formales as they tend to withdraw from the labormarket when employment probabilities arelow.

Opportunity cost of labor for males averagedRs. 2.26 per day during 1975-76, which was90% higher than that for females (Rs 1.20). Inspite of the mixed picture for employmentprobability in these villages, the opportunitywage figures showed consistency, maleshaving significantly higher opportunity wagesthan females (Table 4). The CVs of opportunitycosts were in general higher than for both dailywages and probabilities of employment. Therewould therefore seem to be considerable scopefor designing technologies that specifically aimat capitalizing upon periods when labor oppor-tunity costs are low. This would enhance theprofitability of the technology as well as createemployment in slack/unremunerative periods,with consequent redistributive benefits. Theperiods of the year when opportunity costs areat their maxima and minima are shown in Table6. Peak periods are generally associated withoperations such as harvesting; transplanting,preparatory tillage, sowing, and weeding.

To test the labor market dualism hypothesisof Sen (1966), that the imputed price of labor tosmall farmers is lower than the actual price oflabor to large farmers, paired t-tests were per-formed on the fortnightly wage rates of thelabor category (by sex) compared to thefortnightly opportunity costs of labor f rom thesmall-, medium-, and large-sized farms.17

Medium and large farmers employ more than80% of all hired labor in these villages. Presum-ably then the best measure of their cost of laboris the wage rate received by the labor house-holds. If we find that these wage rates aresignificantly higher than the opportunity costsof labor from the smaller farm categories, thislends support to the dual labor markethypothesis.

In three of the six villages the differencesbetween wage rates of males from laborhouseholds and small-farm male opportunitycosts are significantly positive. In two, thedifferences are not significant and in one thedifference is significantly negative (Table 7). In

17. For details about this hypothesis,al. (1978).

see Ghodake et

371

Table 6 . Peek and s lack labor per iods for adul ts in six S A T vi l lages of peninsular Ind ia , 1 9 7 6 - 7 6 .

District/Vil lage Category

MahbubnagarAurepal le Males & Females

Dokur Males & Females

SholapurShirapur Males

Females

Kalman Males

Females

AkolaKanzara Males

Females

Kinkheda Males

Months

Dec-Jan

Nov-Jan

Apr i l -MayJu l y -Aug

Sept

Dec-Feb

Jan-Mar

May

Mar -Apr

Nov

Mar

Aug -Sep tNov

Oct-Dec

March

Apr i lJune-Ju ly

Peak period

Major operations

Harvesting and threshingsorghum, pearl mil let,castor

Harvesting and threshingsorghum, nursery bedpreparation paddy, paddytransplanting

Preparatory t i l lage,p loughing, sowingpearl mil let, mesta,mungbean

Harvesting and threshingpearl mil let, mesta,mungbeanSowing & harvestingwheat, sorghum, chickpea,saff lower

Harvesting & threshingwheat, sorghum, chickpea,safflowerPreparatory t i l lage,ploughing

Harvesting 81 threshingwheat, sorghum, chickpea,saff lowerHarvesting & threshingpearl mil let, mesta

Harvesting cot ton; harvest -ing & threshing pigeonpeaPreparatory t i l lage, sowingwheat, chickpea, harvestingsorghum, groundnut

Harvesting & threshingsorghum, groundnut, cottonHarvesting cotton

Preparatory t i l lageSowing, interculturingcot ton, sorghum, pigeonpea,mungbean.

Slack periodmonths

Feb-Apr i l

Feb - J u n e

Dec , Feb -March

Apr - A u g

Aug - O c t

Oct - Dec -Jan

Apr. -Sept ,

Apr - J u n e

Aug-Oc tMay

Continued

372

Table 6 Continued

District/Village Category

Females

Peak period

Months Major operationsSlack period

months

Nov-Dec Harvesting & threshing,sorghum, groundnut; sowingwheat and chickpea.

May Field cleaning Feb-MarSept-Dec Harvesting & threshing

sorghum, groundnut;weeding cotton.

Table 7 . Paired t - tests o f d i f fe rences b e t w e e n for tn igh t ly w a g e ra tes o f labor households andoppor tun i ty labor costs of sma l l , med ium and large farms in s ix S A T vi l lages of peninsularIndia, 1 9 7 5 - 7 6 .

District/Village

MahbubnagarAurepal le

Mean differences (Rs/day)Paired t-values

DokurMean differences (Rs/day)Paired t-values

SholapurShirapur

Mean differences (Rs/day)Paired t-values

KalmanMean differences (Rs/day)Paired t-values

AkolaKanzara

Mean differences (Rs/day)Paired t-values

KinkhedaMean differences (Rs/day)Paired t-values

Smallfarms

0.050.34

0.695.87**

0.393.64**

-0 .32- 2 . 7 2 *

0.020.19

0.382.58*

Males

Mediumfarms

0.210.73

0.232.08*

0.906.18**

0.130.53

0.332.52*

0.918.30**

Largefarms

a

0.331.68

0.662.63*

-0 .40- 4 . 1 5 * *

0.603.32**

1.378.12**

Smallfarms

0.325.15**

0.315.34**

0.626.74**

0.355.85**

0.575.54**

0.112.99**

Females

Mediumfarms

0.151.94

0.254.82**

0.696.74**

0.427 . 4 1 * *

0.454.65**

0.0040.07

bargefarms

-0 .08-1 .09

0.263.28**

0.685.12**

0.286.36**

0.323.12**

0.122.89**

** significant at 1% level* significant at 5% level

a. No participation by large farm members.

373

the case of females, in all six villages the wagerates of the labor group significantly exceed theopportunity costs of family female labor f romthe small-farm households. It thus appears thatfor these villages the dual labor markethypothesis receives only partial support in thecase of males and full support in the case offemales.

A paradox remains, however, in the case oflarge farmers who participate in the labor mar-ket. Table 7 shows that in three out of sixvillages the fortnightly wage rates of male laborfrom the labor group are significantly higherthan the opportunity labor costs of male laborf rom large farms. In the case of females, five outof six villages showed positive, significant dif-ferences between labor wage rates and oppor-tunity labor costs of females from large farms.Hence for those large farms whose membersenter the labor market the dual labor markethypothesis holds equally well. That is, wagerates paid by large farmers for hired laborexceeds the opportunity costs of their ownfamily labor. It seems clear the hypothesis is atbest an oversimplification of the operation ofthe labor market in these villages.

Judging by the generally low and/or nonsig-nificant correlations of males and female wagerates, there would appear to be considerablesegmentation in the male and female labormarkets in five of the six villages studied here

(Table 8). The notable exception is the highlyirrigated village of Dokur.

Males and females tend to participate atsimilar times, particularly in Aurepalle and inthe Sholapur villages. They also experiencesimilar movements in their chances of obtain-ing a job throughout the year (Table 8). Intheory, then, one would expect wage rates ofmales and females to also move together iffemales were able to shift in and out of taskssimilar to those done by males as their respec-tive wage rates begin to move apart. The extentto which this does not occur in five of these sixvillages is an indication of the degree of appa-rent segregation of their male and female labormarkets.

C o n c l u s i o n s

We found no overriding consistent evidence inthese six villages that the small farms use morelabor per hectare than the large farms. Smallfarms did employ a surprisingly large amountof hired labor (29 to 68%), but it was less thanthat employed by large farms (41 to 91%).

Female labor represented more than half thetotal labor employed in most of the six villages,and was more significant where cotton andirrigation were prevalent. Up to 90% of totalfemale labor was hired in. Hence technologies

Table 8 . S imp le cor re la t ion coef f ic ients b e t w e e n ma la an d female par t ic ipat ion ra tes , prob-abi l i t ies of e m p l o y m e n t and w a g e rates in s ix SAT v i l lage s of paninsular Ind ia , 1 9 7 5 - 7 6 . a

District;Village

MahbubnagarAurepal leDokur

AkolaKanzaraKlnkheda

Participationrates

0.47*b

0.22

0.40*0.56**

0.210.26

Probabilitiesof employment

0.76**0.95**

0.70**0.68**

0.82**0 .59**

Wagerates

-0 .160.76**

- 0 . 2 40.30

0.100.46**

a. Data for all respondents has been pooled here, including that from large farm households who do not participate much in thelabor market

b. ** and * indicate significance at the 1 and 6% levels, respectively.

374

SholapurShirapurKalman

that affect demand for female labor in thesevillages wil l largely affect the most disadvan-taged of all females — those entering the dailylabor market. Females represented 50 to 90% ofthe total hired labor force used. Males domi-nated the family labor used on these farms (64to 90%).

Small farms had much higher CVs offortnightly labor use than large farms and thisseemed to be inversely related to the number ofdifferent crops grown, with small farms havingfewer. This may allow the small farmers to havemore scope for off (own) farm work, as theywere found to participate significantly in thedaily labor market.

In comparisons of the labor requirements ofthe prospective watershed-based technologybeing researched at ICRISAT and current laboruse in these villages, it was found that, except inthe highly irrigated village of Dokur, the newtechnology had a potential for increasing laborrequirements by 100 to 500% with little alter-ation of the intraseasonal CVs of labor, providedit is based on more than one cropping system.The new technology would generate such vas-tly increased labor demands in a number ofpeak periods, particularly in the Vertisols, thatapparent supplies of labor existing in thesevillages would be exceeded. The periods ofharvesting, threshing and, to a lesser extent,sowing and weeding of the first crop wouldentail the largest potential labor bottlenecks.Intercrop systems offer better scope for spread-ing the harvesting/threshing peaks, as theyhave more flexible timeliness demands.

There is a wide degree of variability in dailylabor market participation rates both amongand within villages. Dokur village, with exten-sive irrigation, has high and stable participationrates compared the others. Males and femalesfrom labor households participate more thanthose from cultivator households. Those fromlabor and small-farm households tend to enterthe daily labor market at similar times duringthe year. Those from labor and medium-sizedfarms do not.

Average probabilities of involuntary un-employment for these six villages in 1975-76were 0.20 and 0.26 for males and females,respectively. These were highest in thedrought-prone, predominant ly foodgrain-producing village of Shirapur in the Sholapurregion. There was large seasonal variation in

employment probabilities, especially in thatvillage. The Akola cotton-growing villages hadthe most buoyant labor markets. Peak periodsfor employment varied from village to village.

The males from labor households have a better chance than the females of obtaining a job compared with their counterparts in cul-tivator households. Thus females from thepoorest socioeconomic group are the mostdisadvantaged in terms of employment oppor-tunities. Females from all households tend tocompete in the labor market at the same t imeduring the year.

Female wage rates were some 56% those ofmales on an average (Rs 2.83 vs Rs 1.60 perday). There was no evidence that wage ratesand employment probabilities were negativelyrelated. In some villages there was evidence of a significant positive relationship, while in onlyone village was it significantly negative, hencegiving weak support to McDiarmid's (1977, pp48-49) hypothesis that labor markets are more"competi t ive" in peak wage periods.

There was no evidence to support the conten-tion that those males who participated more inthe daily labor market had significantly greaterwages than those who participated less; in fact,in several villages the reverse was true. Thissuggests that there may be factors such asdifferences in education, skill, and nutrition thataffect wages. More research on this aspect isrequired, embracing recent advances in theliterature on the economics of human capitalapplied to the developing country situation. Nodoubt the fact that many times small farmerscan hire labor at less than their own opportunitywage influences their decisions to be bothbuyers and sellers of labor at the same time.

Average opportunity costs of male labor (Rs2.26 per day) were some 90% greater than thoseof female labor (Rs. 1.20) in these villages. CVsof opportunity cost were quite large throughoutthe season (12 to 47%). Hence there seems to begood scope for design of technology thatstrategically uses low opportunity cost periodsto enhance the economic viability of thetechnology and generate additional incomesfor disadvantaged groups at times when jobopportunities and/or wages are low.

What is the appropriate cost of labor to use inbenefit-cost analyses where this apparent ex-cess labor pool would be called upon atstrategic times of the year? According to Sen

376

(1975, pp 89-91), the social cost of labor is zeroin a surplus labor economy if there is nosub-optimality of savings. On the other hand ifall weight is on investment as opposed toimmediate consumption, and there are no sav-ings out of wages, the social cost of labor isgiven by the market wage rate. As the actualposition with respect to Sen's conditions liessomewhere in between these extremes, it isprobable that the true opportunity cost of laborlies between zero and the market wage. Thecalculated opportunity costs of wages in thispaper make the adjustment in the right direc-t ion, although as Sen (1975, p 102) points out,this is a planned arbitrary adjustment. It iscertainly preferable to the unplanned arbitraryguesswork involved in alternative methods ofcalculating shadow prices, which involvemathematical programming models and thedetermination of competitive equilibria.

This probability adjustment to market wages,in fact, moves us towards the equil ibrium wageand towards Sen's concept of the social cost oflabor. Whether the adjustment overshoots orundershoots the mark requires knowledge ofthe elasticity parameters of demand and supplyof labor. There is a dearth of such estimatesavailable, particularly of supply parameters andseasonal estimates. Until we have these itseems the probability-adjusted wage rates arethe best measure we have of the social cost oflabor. However, assuming the elasticity of Iaborsupply is not zero then when wages are ad-justed downwards by the probabilities (if < 1)in this way, the amount of surplus labor orinvoluntary unemployment available does notstay constant. Hence if one uses theprobability-adjusted opportunity wages as thesocial cost of labor, one should not at the sametime assume the available surplus labor pool tobe the same as that at the prevailing (full)market wage rates.

We found weak support in these six villagesfor Sen's (1966) dual labor market hypothesis. Itwould seem to be an oversimplification of theoperation of the labor markets in these villages.We also found evidence of significant amountsof segmentation between the male and femalelabor markets in these villages.

Acknowledgment

The authors are grateful to T. Balaramaiah, S. S.

Badhe, V. Bhaskar Rao, M. J. Bhende, N. B. Dudhane,and K. G. Kshirsagar, the investigators w h o wereresponsible for the data collection in the six vil lageson wh ich this s tudy is based. They also thank MitaSandilya and M. Asokan for computat ional assis-tance, and M. von Open and D. Jha for their criticalcomments on an earlier draft of this paper.

References

AGARWAL, R. 1964. Size of holding and product iv i ty:Further comments. Economic Weekly 16(47):1850-1851.

BARDHAN, K. 1977. Rural employment , wages andlabor markets in India: A survey of research III.Economic and Political Weekly 12(28): 1101-1118.

BARDHAN, P. K. 1973. Size, Productivity, and Returns toscale: An analysis of farm level data in Indianagriculture. Journal of Political Economy 81 (6):1370-1386.

BHARADWAJ, K. 1974. Note on farm size and productiv-ity, Economic and Political Weekly Review of Ag-riculture 9(13): A11-A24 .

BINSWANGER, H. P., and JODHA, N. S. 1978. Manual of

i ns t ruc t ions f o r economic inves t iga to rs inICRISAT's V i l l age -Leve l S tud ies . ICRISATEconomics Program Vil lage Level Studies Series II,Patancheru, A.P., India.

BINSWANGER, H. P., JODHA, N. S., R Y A N , J. G., and VON

OPPEN, M. 1977. Approach and hypotheses for thevil lage level studies of ICRISAT. ICRISAT EconomicsProgram occasional paper 16, Patancheru, A.P.,India.

BINSWANGER, H. P., and SHETTY, S. V. R. 1977.

Economic aspects of weed control in semi-aridtropical areas of India, ICRISAT Economics Prog-ram, occasional paper 13, Patancheru, A.P., India.

BOSERUP, E. 1970. Woman's role in economic de-velopment. London: Al len and Unwin.

BRANON, R. H., and JESSEE, D. L. 1977. Unemployment

and underemployment in the rural sectors of theless developed countries. Occasional Paper SeriesNo. 4, Economics and Sector Planning Division,Office of Agricul ture, Technical Assistance Bureau,USAID, Washington, D.C.

GHODAKE, R. D., RYAN, J. G., and SARIN, R. 1978.

Human labor use w i th exist ing and prospectivetechnologies in the semi-arid tropics of peninsularIndia, ICRISAT Economics Program progress report1, Patancheru, A.P., India.

HARRIS, J., end TODARO, M. 1970. Migrat ion, un-

employment and development: A t w o sectoranalysis. American Economic Review 60 (1) :126-142.

378

JODHA, N. S. 1977. Resource base as a determinant ofcropping patterns, ICRISAT Economics Programoccasional paper 14, Patancheru, A.P., India.

JODHA, N. S., ASOKAN, M., and RYAN, J. G. 1977. Village

study methodology and resource endowments ofthe selected vil lages, ICRISAT Economics Programoccasional paper 16, Patancheru, A.P., India.

KRISHNA, RAJ . 1973. Unemployment in India.Economic and Political Weekly 8 (9):475~484.

LAL, D. 1974. Social prices for non-traded com-modit ies. Government of India. Planning Commis-sion Technical Paper 12, New Delhi, India.

MAZUMDAR, D. 1963. On the economics of relativeefficiency of small farmers, Economic Weekly (Spe-cial Number): 1259-1263.

MAZUMDAR, D. 1965. Size of farm productivi ty: A problem of peasant agriculture. Economica (NewSeries): 161.

M C D I A R M I D , O. J. 1977. Unskil led labor for develop-ment: Its economic cost. Balt imore and London:Johns Hopkins University Press, (for World Bank)

R A J , K. N. 1959. Employment and unemployment inthe Indian Economy: problems of classification,measurement and policy. Economic developmentand cultural change. 3(3): 258-278.

RODGERS, G. B. 1975. Nutr i t ional ly based wage deter-minat ion in the low income market. InternationalLabor Organization, Wor ld Employment Prog-ramme, Geneva, Switzerland.

RUDRA, A. 1973. Marginalist explanation for more

intensive labor input in smaller farms: Empiricalverif ication. Economic and Political Weekly 8 (22):989-994.

RYAN, J. G., SARIN, R., and PEREIRA, M. 1979.

Assessment of prospective soi l , water, and cropmanagement technologies for the semi-arid tropicsof peninsular India. This Workshop.

SEN, A. K. 1966. Peasants and dual ism wi th or w i thoutsurplus labor, Journal of Political Economy 74:425-450.

SEN, A. K. 1975. Employment, technology and de-velopment, Oxford: Oxford University Press.

SETHURAMAN, S. V. 1972. Seasonal variations in un-employment and wage rates: Implications for ruralworks programme. Economic and Political Weekly7(24).

SUBRAHMANYAM, K. V., and RYAN, J. G. 1976. Analysis

of the rural labor market in Kanzara vi l lage,Maharashtra, India. Presented at the AgriculturalDevelopment Council Workshop on HouseholdStudies, 3 - 7 Aug 1976, Singapore.

VIRMANI, S. M., SIVAKUMAR, M. V. K., and REDDY, S. J.

1978. Climatological features of the semi-aridtropics in relation to the Farming Systems ResearchProgram. Presented at the International Workshopon the Agroclimatological Research Needs of theSemi-Arid Tropics, 23 -24 Nov. 1978, ICRISAT,Hyderabad, India.

VYAS, V. S. 1964. Agricultural labor in four Indianvillages, Sardar Patel University, Vallabh Vid-yanagar, Gujarat. (Mimeographed.)

Appendix Tab le 1. Pai red t-tests of d i f ferences in part ic ip at ion rates and probabi l i t ies of emp loy -men t in dai ly labor market tor adul ts in d i f ferent S A T distr icts. a

Sex

Participation rateMale

Female

Probability of employmentMale

Female

Mean difference

Mahbubnagar-Sholapur Mahbubnagar-Akola

- 0 . 1 9 * * * a - 0 . 2 3 * * *[21.63]b [16.431

0.22*** 0 . 1 1 * * *[16.92] [10.58]

- 0 . 1 6 * * - 0 . 1 6 * * *[3.43] [4.95]

0.07 - 0 . 0 7 * *[2.07] [3.62]

Sholapur-Akola

- 0 . 0 6 * *[3.61]

- 0 . 1 4 * * *[9.731

- 0 . 0 5 *[2.67]

- 0 . 1 9 * * *[6.30]

a. *** Signi f icant at 0 . 1 % level

* * Signi f icant a t 1% level

* S igni f icant at 5% level

b. Bracketed f igures are 't ' values.

377

378

Append ix Table 4 . S imple corre lat ion coeff ic iants o f par t i c ipat ion rates, probab i l i t ies o f employ-m a n t , and wage rates fo r males f r o m dt f ferent households i n six S A T villages).

Village

Labor-Small farm

Participationrate

- 0 . 4 8 * a

0.44**0.40*

- 0 . 3 10.44*0.43*

Probabilityof employment

0.300.89**0.290.35

* *0 .58* *0.78**

Wagerate

-0 .410.75**0.49**0.67**0.35

-0 .11

Labor-Medium farm

Participationrate

-0 .410.100.15

- 0 . 4 0 *0.340.15

Probabilityof employment

0.020.80**0.11

-0 .080.46**0.78**

Wagerate

- 0 . 6 9 * *0.230.24

-0 .070 .41*

-0 .003

a** Significant at 1% level; * Significant at 5% level.

Append ix Table

Village

Aurepal leDokurShirapurKalmanKanzaraKinkheda

5. S imp le corre lat ion coef f ic ients of part ic ipat ion rate s, probabi l i t ies of employ-m e n t , and w a g e rates for females f r o m d i f ferent household s in six S A T vi l lages.

Labor-Small farm

Participation Probabilityrate of employment

0.83**a

- 0 . 240.33

-0 .220.39*0.42*

0.620.97**0.73**0.40*0.79**0 . 8 3 "

a** Significant at 1% level; * Significant at 5% level.

Wagerate

0.53**0.77**0.250.120.59**0.77**

Labor-Medium farm

Participationrate

0.88**0.05

- 0 . 6 0 * *0.120.180 .61* *

Probabilityof employment

0.82**0.92**0.64**0.74**0.77**0.72**

Wagerate

0.380.85*0.280.64**0.190.72**

379

AurepalleDokurShirapurKalmanKanzaraKinkheda

Discussant 's Comments

V . N . M i s r a *

My observations on these papers are in no wayto detract from the quality of the papers. Theyare well written and encompass several disci-plines such as economics, sociology, and an-thropology. However, I want to raise someissues discussed in the papers, in relation tomethodology as well as with regard to theimplications of the findings for policy purposes.

The paper by Benson is a good review ofdifferent issues relating to labor use based onstudies conducted by several authors. How-ever, the statement that "commercialization ofagriculture displaces labor" is too general. Infact, there is evidence to show that commer-cialization of agriculture has increased the de-mand for labor.

Benson's conclusion about the role of cottageindustries in providing alternative or supple-mentary employment is now well-recognized.However, the basic problem is the competitive-ness of cottage industries vis-a-vis manufactur-ing industries. Despite our emphasis on cottageindustries in several five-year plans, they havenot evolved as a viable alternative.

Two observations can be made on Byerlee'spaper. One is related to methodology while theother has some bearing on the functioning ofthe labor market.

Byerlee observes that in West Africa " thelabor market is fairly efficient in allocatinglabor". This finding is based on the comparisonof marginal value productivities (MVPs) oflabor estimated through production functionanalysis. This value is slightly higher than pre-valent wages, but there is no significant differ-ence between the two. Thus, his conclusiondrawn, that the labor market functionsefficiently appears to be correct. However, theauthor then proceeds to make estimates of theMVP of labor based on a linear programingmodel. In this case the values are found to bethree to four times higher than the prevailingrates. Hence one obtains the impression

* V. V.India.

Giri Institute of Regional Studies, Lucknow,

that the labor market is efficient when the MVPcalculation is based on the production functionbut that it is inefficient when the MVP of labor isbased on a linear programing model. How canwe reconcile this? The estimates of MVP oflabor based on production function analysisshould not be compared with those derivedfrom a linear programing model because theformer is positive, whereas the latter is norma-tive in the sense that it gives an estimate of MVPat an opt imum level. The estimates of MVP oflabor based on a linear programing modelshould only be compared with wages thatwould exist at an equil ibrium or opt imum posi-tion in order to judge the efficiency of labormarket.

Another finding by Byerlee is that wage diffe-rentials of almost two to one still exist betweenthe highest and lowest wage regions, despite a considerable f low of migrants f rom the lowestto the highest wage regions. This is difficult toexplain. It would be expected that whensufficient labor migrates to the highest wageregions, and is available in the job market, thatthe wages in these areas would become lower.There are however, two necessary conditionsfor making wages a function of the supply oflabor: free mobility of labor among employersand also the availability of alternative employ-ment. There is no information on these twoaspects in the paper. Thus, it becomes difficultto explain wages and the persistance of diffe-rentials under conditions of considerable mig-ration. Byerlee has tried to explain the wagedifferentials by considering the demand side ofthe labor market. In addition to this, there areother factors such as the financial capacity ofcultivators, size of holding, cropping patternand extent of irrigation, which supposedlyinfluence the demand for labor and hencewages. If Byerlee would have taken these fac-tors into consideration also, he would havebeen able to better explain the wage differen-tials.

Ryan and his colleagues have presented anexcellent paper. It covers all the important

380

aspects of the labor market. It is so comprehen-sive that it raises many debatable issues. Thefindings are very important for research as wellas with regard to policy implications.

It is observed by Ryan et al. that intensity ofcropping does not contribute substantially todifferences in labor use intensity in the sixvillages. This finding is quite different from theresults of the study of Shakuntala Nehru. In herpaper, which is based on farm managementdata, it was found that intensity of croppingcontributes to differences in labor use. Since thefinding of Ryan et al. is quite different from whatone would expect, it seems worthwhile to com-pare the per hectare use of labor given in Table 1 of their paper with the intensity of cropping forthe six villages. These latter data are taken fromanother paper by Ghodake et al. (1978, p 7) citedby the authors. There does not appear to bemuch difference in the cropping intensity in theSholapur and Akola villages. In the former, therange is from 105 to 115, whereas in the latter,the variation in the cropping intensity is verylow, being in the range 103 to 108. The result isthat in the villages of these two districts, inten-sity of cropping has not really been the impor-tant factor accounting for the differences inlabor use. However, the position is entirelydifferent in the villages of Mahbubnagar dis-trict. In Aurepalle, per hectare labor use in-creases with increases in farm size. A similartrend exists with increasing intensity of crop-ping. In the case of Dokur village, the position isreversed in the sense that the per hectare laboruse decreases with increases in farm size. How-ever, in this case, intensity of cropping alsodecreases with increases in farm size, againresulting in a positive relationship betweenlabor use per hectare and intensity of cropping.From this comparison it seems that whereverthere is sufficient variation in the intensity ofcropping, it does account for differences inlabor use. On the other hand, wherever thereare only small variations the intensity of crop-ping cannot account for differences in labor use.

Another point is related to segmentation ofmale and female labor markets. Based on lowand nonsignificant correlations of wage rates ofmale and female labor, it has been observed byRyan et al. that there appears to be segmenta-t ion in the male and female labor markets.Female labor is a non-competing group withmales for the obvious reason that certain types

of agricultural operations are generally done bymales only. However, having proved the seg-mentation of labor markets for males andfemales, it is difficult to understand what hasprompted the authors to compare the probabi-lity of employment for females wi th that ofmales.

Ryan et al. found that hired labor demandchanges at least in proportion to total seasonallabor. This observation is based on the correla-tion coefficients between total labor use and theabsolute amount of hired labor. The coefficientsbetween total labor and proportion of familylabor to total labor have been found sig-nificantly negative. Based on this it is concludedthat if disadvantaged groups such as landlesslaborers and small farmers are to benefit f romadditional employment created by the newtechnology in these villages, then this evidencesuggests one should not necessarily designthem to have stable seasonal labor use pat-terns. This conclusion contradicts the inferencedrawn earlier in the paper that there wouldseem to be considerable scope for designingtechnologies that specifically aim at capitalizingupon periods when labor opportunity costs are low. This would enhance the profitability of thetechnology as well as create employment inslack/unremunerative periods wi th consequentredistributive benefits. The stabilization in de-mand for labor need not mean a stabilization ata low level, which seems to be proposed by theauthors in suggesting technology developmentto take advantage of low opportunity costs. Ifthese opportunity costs are to be used as a basisfor designing new technology then such,technology may turn out to be very profitablebecause of the low opportunity costs. However,how is providing employment at such lowopportunity costs going to benefit agriculturallaborers? It seems that employment would becreated per se, rather than to really improve theincome level of agricultural laborers.

381

Chai rman 's Summary

J . K a m p e n *

In the discussion of Benson's paper, a questionwas raised on how relief projects to providejobs during non-peak periods can be financedwi thout tax ing or otherwise obta in ing thenecessary resources from agriculture. It waspointed out that in India resources for such pur-poses are not raised directly from the agricul-tural sector but are obtained from general re-sources.

Issue was taken wi th Benson's generalizationregarding women's specialization in maintain-ing subsistence crops, and it was pointed outthat in the Sudan these activities are executedas cooperative activities between men andwomen. It was agreed that while this may betrue the division of tasks still exists (e.g. menclearing brush and women weeding).

The generalization of migration aspects inByerlee's paper without considering relation-ships between migration, off-farm activities andon-farm activities, particularly in the context ofthe African countries where migration occursfrom dryer to less dry areas was questioned.The author pointed out that the considerableliterature on migration f lows does not clearlyidentify such relationships. He said that rural-rural migra t ion is most impor tant to low-income food-deficit households and is moreimportant than rural-urban migration in thecontext of new agricultural technologies andthe implementation of innovations in WestAfrica.

Labor bottlenecks and constraints aspects ofnew technology were discussed at length. Theoptimistic view was expressed that after a lapseof 5 to 10 years when new technologies wouldactually be implemented on farmers' f ields,such constraints would have disappeared be-cause of increasing populations. Therefore, theneed for ICRISAT's research strategy to be out-put oriented and to give less attention to issuessuch as labor bottlenecks was stressed.

In the Sudan in areas w i t h ex tens ive

* Farming Systems Program, ICRISAT.

mechanized agriculture, lean period wages areonly 20% of peak period wages. There is a danger that if one tries to smooth out the sea-sonal peaks, migrant and landless laborerwould be adversely affected.

A question was raised about the policy con-clusion in the paper by Ryan et al. regarding thedesign and transfer of new technologies duringlow opportunity cost periods, and implicationsthereof on wage rates. This is an area wheremore research is needed. Furthermore, becausefew operations are done separately by malesand females on a piece wage rate basis, it isdifficult to measure the productivity differen-tials between them and hence attempt to exp-lain why their wages really differ.

The use of village survey research findings interms of feedback to ICRISAT's overall researchstrategy was discussed. It was pointed out thatthis is a continuous process; for example laboruse information in relation to possible solutionsfor anticipated bottlenecks has been assembledas well as the application of village opportunitycost measures in the assessment of newtechnologies. This information in turn is beingutilized by the Farming Systems Research Prog-ram scientists in developing technologies.

The necessity to do operational participationanalysis in order to understand participationdifferences according to sex and/or operationwas emphasized. It was suggested that the sub-stantial contribution of female labor can bepartly explained by male migration in many in-stances.

A discussion ensued about ICRISAT's inter-disciplinary teamwork to generate technologythat wil l meet the requirement of efficient use ofseasonally available labor at low opportunitycost. I t was po in ted out that at presentwatershed-based research in collaboration wi thother physical and biological scientists is un-derway in villages. This activity helps to ascer-tain the potential constraints related to labor orother factors; it wil l also provide opportunitiesto verify the low opportunity wage periods.

Caution was suggested wi th regard to the useof oppor tun i t y wages in eva luat ing new

382

technology. There is a need to take into accountthe possibil i ty of enforcement of m in imumwages. Peak period differentials in case ofmales and females, if relevant and significant,should also be considered. It was suggestedthat minimum wage laws, if effective, wouldprobably have the effect of increasing un-employment and the net effect on real oppor-tunity cost may well be neutral. However, actualeffects depend on demand and supply elas-ticities of labor, and little is known about these.

It was felt that the health-nutrition-pregnancyrelationship was not adequately discussed inthe papers. Additional information in this areawould help explain which people do not workand why, and if ICRISAT should explore the pos-sibility of attracting the attention and involve-ment of those responsible for public healthprograms. The linkages between these health as-pects with labor employment should be clearlyestablished. It was noted in response that the IC-RISAT Economics Program's village-level studiesinvolved a diet, nutrition and health evaluationproject in collaboration with the Home ScienceCollege of Andhra Pradesh Agricultural Univer-sity and the National Institute of Nutrition inIndia. These studies were started in late 1976and the data have been collected; the finalf indings, implications and their interpretation inexplaining labor employment patterns have notyet been published.

In response to lively discussion, it was statedthat there is no conflict between a researchfocus on agricultural production or productivityof the natural resource base and an Institutecharter emphasizing benefits to small farmersand the rural landless poor. Appropr ia tetechnologies that do not displace rural labormust be designed for agricultural development;if increased and more stable production are tobe achieved and the quality of life enhanced forthe less affluent people of the SAT.

In relation to the objectives of the workshop itappears that the papers presented emphasizedthe need for further socioeconomic researchand surveys, as wel l as gu ide l ines andmethodologies for such studies in nationalprograms and by ICRISAT. Unfortunately, therewere few suggestions with regard to the designof improved technology and agriculturalpolicy, particularly in the African setting. Thisconclusion would seem to be relevant to thefocus and orientation of future socioeconomicresearch and surveys.

383

Appendix 1

F r e n c h A b s t r a c t s a n d T e x t s( R e s u m e s e t T e x t e s e n F r a n q a i s )

387

388

389

390

391

392

393

394

395

396

397

Moussa Fal l *

398

399

400

401

Tableau 2. Techniques culturales par statut.

Statut deI'exploitant

Chef de carreChef de menageSourgaNavetaneFemme

Ensemble

Nombrede

parcelles

5017781586

246

Numerode pluiede semis

1,281,181.371,001,69

1,43

Parcellesbinees2 fois

de plus(%)

7876946658

75

Nombremoyen debinages

2,342,592,462,071,91

2,23

Nombremoyen de

desherbages

1,801,411,781,671,86

1,78

Parcellesavec

engrais(%)

6241155321

29

Tableau 3. Utilisation da I'angrals par statut.

Statutde I'exploitant

Chef de carreChef de menageSourgaNavetaneFemme

Parcellesavec engrais

(%)

3241155321

Dose moyenned'engrais

(kg/ha)

6039155019

Dose sur parcelleavec engrais

(kg/ha)

11695949489

402

403

404

Jacques Faye*

405

406

407

408

409

410

411

412

413

414

M. Le Moigne*

415

416

417

418

419

420

421

422

423

Appendix 2

L i s t o f C o u n t r y P a p e r s

List of Count ry Papers

Central and West AfricaSocioeconomic Aspects of Semi-Arid Tropical Regions of Nigeria

Perspectives on AgriculturalDevelopment in Senegal

Problems of Controlling DroughtEffects in Upper Volta

Constraints to AgriculturalDevelopment in the Semi-AridTropics of Chad

Socioeconomic Aspects of theSemi-Arid Tropics of Niger

Discussant's CommentsChairman's Summary

East AfricaSocioeconomic Problems in BotswanaAgriculture

The Role of Agriculture in theSudanese Economy and Problems in theSemi-Arid Tropics in Western Sudan

Discussant's CommentsChairman's Summary

Central and South AmericaSome Characteristics of theAgricultural Sector of Mexico

The Agricultural Economy ofNortheast Brazil

Discussant's CommentsChairman's Summary

IndiaConstraints in Adoption of NewTechnology in the Semi-Arid Tropicsof Haryana

Some Development Issues in theSemi-Arid Tropics of Rajasthan

Performance of Agriculture in theSemi-Arid Tropics of UttarPradesh: An Inter-District Analysis

Socioeconomic Research in Tamil Nadu

G . O . I . Abalu andB. D'Silva

M. Fall

K. Talata

P. S. Dedjebe

G. Numa

H. P. BinswangerB. Harriss

V. Amann

H. A. Z. M. Salih

M. von OppenB. F. Johnston

A. T. Fernandez

G. M. Calegar

V. S. DohertyV. Rajagopalan

I. J. Singh

S. S. Acharya

V. N. Misra andA. Joshi

V. Rajagopalan

427

Some Constraints to Farming Efficiencyin the Semi-Arid Tropics of Gujarat

Dryland Agriculture in Maharashtra:A Microlevel Study of the Economicsof Cropping "Patterns and Practices

Agroeconomic Features of Semi-AridTropical India

Discussant's Comments

Chairman's Summary

A. S. Patel

R. G. Patil, S. B. Dangatand M. P. Dhongade

S. L. Bapna, D. Jhaand N. S. Jodha

N. S. Jodha

B. Reddy

428

Appendix 3

P a r t i c i p a n t s a n d O b s e r v e r s

Participants

Abalu, G. O. I.Head, Department of Agri. Economics and Rural

SociologyInstitute for Agricultural ResearchAhmadu Bello UniversityZaria, Nigeria

Acharya, S. S.Associate ProfessorDepartment of Agri. EconomicsUniversity of UdaipurCampus: Jobner 303 329District JaipurRajasthan, India

Amann, VictorChief Agricultural EconomistDivision of Planning and StatisticsMinistry of AgriculturePrivate Bag 003Goborone, Botswana

Anderson, J. R. 1

Deputy DirectorBureau of Agri. EconomicsP. O. Box 1563, BartonCanberra ACT 2601, Australia

Asokan, M.Research Technician, ICRISATPatancheru P. O. 502 324Andhra Pradesh, India

Bapna, S. L. 2

Economist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Barah, B. C. 3

Economist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

1. Present address: Dean, Faculty of Economics, Uni-versity of New England, Armidale, NSW, Au-stralia.

2. Present address: Indian Institute of Management,Vastrapur, Ahmedabad, Gujarat, India.

3. Present address: Department of Economics,School of Social Sciences, Central University of

Hyderabad, "Golden Threshold" Nampally Sta-t ion Road. Hyderabad 500 001, AP, India.

Benson, Janet (Ms)Professor, Dept of Sociology & AnthropologyKansas State University, ManhattanKansas 66506, USA

Benoit-CattinAgroeconomistInstitut de Recherches Agronomiques Tropicales

(IRAT)B.P. 503534032 Montpellier CEDEX, France

Berg, ElliotProfessor of Economics and Research ScientistCenter for Research on Economic DevelopmentThe University of Michigan, Ann ArborMichigan 48106, USA

Bhaskar Rao, V.Technical Assistant, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Bhende, M. J.Research Technician, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Biggs, S. D. 4

Economist, CIMMYTC/o ICRISAT, 23 Golf LinksNew Delhi 110 003, India

Bigot, YvesInstitut de Recherches Agronomiques Tropicales

(IRAT)B.P. 5035, 34032 Montpellier CEDEX, France

Binswanger, H. P. 5

Principal Economist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Byeriee, DerekEconomist, CIMMYTP.O. Box 6-641Londres 40, Mexico 6 DF, Mexico

4. Present add ress : Ins t i tu te o f Deve lopmen tStudies, Br ighton, BNI 9RE Sussex, United King-dom.

5. Present address: Employment & Rural Develop-ment Division, The World Bank, 1818 H StreetNW, Washington DC 20433, USA.

431

Calegar, G . M.EMBRAPARua Presidente Dutra160 Caixa Postal 23, 56.300 Petrolina PE, Brazil

Cancian, FrankSchool of Social SciencesUniversity of California, IrvineCalifornia 92717, USA

Chambers, R. J. H.The Institute of Development StudiesUniversity of Sussex, BrightonSussex BNI 9RE, UK

Chaudhari, A. K.Economist, Indo-UK Dry Farming Project (ICAR)Jawaharlal Nehru Krishi Vishwa VidyalayaCampus: College of Agriculture, Indore 452 001Madhya Pradesh, India

Davies, J. C*Principal Entomologist and LeaderCereals Program, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

de Lafforest, J.(French Translator/Interpreter)TRANSLA-lnter LIMITED29 Villiers Street, StrandLondon WC 2N 6ND, UK

Dedjebe, PierreDivision d'Etudes AgronomiquesB.P. 441Ndjamena, Chad

Dillon, J. L.Pro-Vice ChancellorUniversity of New EnglandArmidale, NSW, Australia

Doherty, V. S.Principal Social Anthropologist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Eicher, C. K.Department of Agri. EconomicsMichigan State UniversityEast LansingMichigan 48824, USA

* Presently Director for International Cooperation.

Fall, M.Centre National de Recherches Agronomiques

(CNRA)BP 51,Bambey, Senegal

Faye, J.Head, Unite ExperimentaleInstitut Senegalais de Recherches Agricoles (ISRA)BP 3120Dakar, Senegal

Ghodake, R. D.Economist, ICRISATPatancheru P. O. 502 324Andhra Pradesh, India

Gross, J. P.(French Translator/Interpreter) TRANSLA-lnter LIMITEDAvenue Boileau 131040 Bruxelles, Belgium

Harriss, Barbara (Ms)School of Development StudiesUniversity of East AngliaNorwich NR4 7TJ, UK

Hiremath, K. C.Department of Agri. EconomicsCollege of Agriculture, University of Agri. ScDharwar Campus, KrishinagarDharwar 580 005Karnataka, India

Jha, D. 6

Visiting Senior Economist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Jodha, N. S. 7

Senior Economist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Johnston, B. F. 8

International Institute for Applied System AnalysisA-2361, Laxenburg, Austria

6. Present address: Professor of Ag r i . Economics, Di-vision of Agr i . Economics, Indian AgriculturalResearch Institute, New Delhi 110 012, India.

7. Present address: Agr icu l tu ra l Economist , IITA/USAID/Tanzanian Nat ional Research Project,l longa Agr icul tural Research Institute, PrivateMai l Bag, Kilosa, Tanzania.

8. Present address: Stanford Food Research Insti-tu te , Stanford, California, USA.

432

Le Moigne, M.*CEEMAT, Parc de Tourvoie92160 Antony, France

Talata, KafandoMinistere de Plan et de la CooperationOuagadougou,Upper Volta

Kampen, J. A.Principal Land & Water Management Engineer,ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Kanwar, J. S.Director of Research, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Krantz, B. A. 9

Principal Agronomist and LeaderFarming Systems Research Program, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Kshirsagar, K. G.Research Technician, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Mc Ginnis. R. C. 10

Director of International Cooperation, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Misra, V. N.V. V. Giri Institute of Regional StudiesLucknowUttar Pradesh, India

Morris, W. H. M.Professor, Department of Agri. EconomicsPurdue University, West LafayetteIndiana 47907, USA

Norman, D. W.Professor, Department of EconomicsKansas State University, ManhattanKansas 66506, USA

* Contr ibuted a paper but could not attend the

workshop.9. Present address: 760, Mulberry Lane, Davis, Ca.

95616, USA.10. Present address: Dean, Faculty of Agriculture,

University of Manitoba, Canada R3T 2N2.

Numa, GilbertEconomist, Institut National de Recherches

Agronomiques du Niger (INRAN,) BP 429Niamey, Niger

Patel, A. S.Reader, Department of EconomicsS. P. UniversityVallabh Vidyanagar, District KairaGujarat, India

Patil, R. G.Head, Department of Agri. EconomicsMahatma Phule Krishi VidyapeethRahuri, District AhmadnagarMaharashtra, India

Parthasarathy Rao, P.Technical Assistant, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Rajagopalan, V.Head, Department of Agri. EconomicsTamil Nadu Agricultural UniversityCoimbatore 641 003Tamil Nadu, India

Raju, V. T.Economist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Rastogi, B. K.Sr. Agricultural EconomistAll-India Coordinated Research Project for Dryland

AgricultureAmberpet, Hyderabad 500 013Andhra Pradesh, India

Balwanth Roddy, H.Administrative Staff College of IndiaBella Vista, Hyderabad 500 475Andhra Pradesh, India

Ruthenberg, Ham (deceased)Institut fur Landwirtschaftliche BetriebslehureUniversitat Hohenheim,7000 Stuttgart 70,Postfach 106, West Germany

Ryan, J. G.Principal Economist & LeaderEconomics Program, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

433

Sayed Hamid Abu Zeid Mohmed SalihDepartment of Soil ConservationLand Use and Water ProgrammingMinistry of AgricultureKhartoum, Sudan

Sanders, J. H.*Economist, CIATCali, Colombia

Sarin, R.Research Technician, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Subba Rao, K. V.Research Technician, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Singh, I. J.Department of EconomicsHaryana Agricultural UniversityHissar, Haryana, India

Swindale, L. D.Director General, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Thierstein, G. E.Principal Agricultural Engineer, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Traore, BakaryTafawa Balewa (Post Graduate) HallUniversity of IbadanIbadan, Nigeria

Uzureau, ClaudeDirector, CEEMATParc de Tourvoie92160 Antony, France

Virmani. S. M.Principal Agroclimatologist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

von Furer-Haimendorf,Consultant, ICRISATPatancheru P.O 502 324Andhra Pradesh, India

* Contr ibuted a paper but could no t attendworkshop.

the

von Oppen. M.Principal Economist, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Observers

Balaramaiah, T.Technical Assistant, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Bidinger, P. D. (Ms)Research Scholar, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Charreau, C.Resident DirectorICRISAT West Africa Project28 Rue Thiers, B. P. 3340Dakar, Senegal

Collins, NormanThe Ford Foundation55, Lodi EstateNew Delhi 110 003, India

Dagg, MathewThe Ford Foundation55, Lodi EstateNew Delhi 110 003, India

Deshpande, A. L.Technical Assistant, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Hallsworth. E. G.The International Federation of Institutes

for Advanced Study (IFLAS),Private Bag No. 2 P.O. Glen Osmond, South Australia 5064

Germain, Adrienne (Ms)The Ford Foundation320 East, 43rd StreetNew York, NY 10017, USA

Harvey, JimSocial ScientistICARDA, AleppoSyria

Krishna. K. V.Technical Assistant, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

434

Krishna Gopal, C.Research Technician, ICRISATPatancheru P.O. 502 324Andhra Pradesh, India

Kuhnen, FrithjofDirectorInstitute for Rural DevelopmentUniversity of Gottingen, Buesgenweg 2,D-3400 Gottingen (Weende)Federal Republic of Germany

Le RicollalsORSTOM. 24 Rue BoyardParis 8, France

Perrauit, P. T.12050 Blvd de I'AcadieApartment 2 Montreal, Canada H3M 2T9

Rajendran, S.Technical Assistant, ICRISATPatancheru P.O. 502 324 | Andhra Pradesh, India

435

ICRISATInternational Crops Research Institute for the Semi-Arid Tropics

ICRISAT Patancheru P.O.

Andhra Pradesh, India 502 324