relative real wages in swedish acriculture and industry, 1930–1950

17
RELATIVE REAL WAGES IN SWEDISH AGRICULTURE AND INDUSTRY, 1930-1950' PART I The growing interest in recent years in the relative income position of primary, secondary and tertiary producers has not yet produced much evidence to compare the real values (ignoring philosophical problems) of their money incomes. Only two relevant studies have been published.2 One relates to a country with large distances, the other to one with a freak- ishly small proportion of its population in agriculture, comparative congestion of population and an extremely high degree of urbanisation. The results of neither study can, therefore, be regarded as widely representative. Swedish experience provides a third case and has additional interest, partly because geographically, in economic structure and in extent of urban- isation it is typical of certain other countries, and partly because analysis is possible to allow for the possible shift in opportunity cost of acquiring self- supplies and payments-in-kind by farm employees, by alternative valuations based on rural retail prices or farm-gate prices.3 Several Swedish studies have been made of some of the problems involved and reliable statistics exist for more to be completed. This paper summarises some available results and adds another. The assessment of the compensation in money income necessary to maintain the same real income when a worker transfers from one occupation to another is clear in principle, although very difficult in practice. The problem is to decide by how much the money income of a worker moving from one occupation to another would have to vary so that his real income (excluding non-monetary considerations) would be the same in each situation. Difference in prices of identical commodities between the situations consti- tute no theoretical difficulty. The trouble arises because in the two situations commodities are consumed in different proportions. For this there are two reasons (ignoring differences in tastes). Firstly, prices relative to each other may, and usually will, be different in each case. In addition, differ- ences in consumption patterns will arise because of differences in the level of real income enjoyed by the worker before and after the transfer. In terms of indifference curve analysis his new budget can differ from his old one 1As part of an intensive study of relative incomes earned in agriculture and industry in various countries, in some cases going as far back as 1860, carried out by various members of the Agricultural Economics Research Institute, the author has made estimates of cost-of- living dierences between Swedish agricultural and industrial workers from 1870 onwards. This article summarises the most interesting (and statistically reliable) part of the estimates and places them in a wider context. 5Nathan Koffsky, Farm and Urban Purchasing Power in Studies in Income and WeaWi, Volume XI. National Bureau of Economic Research, 1949. and J. R. Beilerby, United Kingdom: Comparison of Urban and Rural Retail Prices, J .R.S.S. Vol. CXVI, Part 1, 1953. 'In Swedish wage statistics. payments-iti-kind are valued at farm-gate prices (the prices at which the produce could be sold to middle men buying from farm producers for sale in normal distributive channels)

Upload: g-r-allen

Post on 01-Oct-2016

220 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

RELATIVE REAL WAGES IN SWEDISH AGRICULTURE

AND INDUSTRY, 1930-1950'

PART I

The growing interest in recent years in the relative income position ofprimary, secondary and tertiary producers has not yet produced muchevidence to compare the real values (ignoring philosophical problems)of their money incomes. Only two relevant studies have been published.2One relates to a country with large distances, the other to one with a freak-ishly small proportion of its population in agriculture, comparative congestionof population and an extremely high degree of urbanisation. The resultsof neither study can, therefore, be regarded as widely representative.

Swedish experience provides a third case and has additional interest,partly because geographically, in economic structure and in extent of urban-isation it is typical of certain other countries, and partly because analysis ispossible to allow for the possible shift in opportunity cost of acquiring self-supplies and payments-in-kind by farm employees, by alternative valuationsbased on rural retail prices or farm-gate prices.3 Several Swedish studieshave been made of some of the problems involved and reliable statisticsexist for more to be completed. This paper summarises some availableresults and adds another.

The assessment of the compensation in money income necessary tomaintain the same real income when a worker transfers from one occupationto another is clear in principle, although very difficult in practice. Theproblem is to decide by how much the money income of a worker movingfrom one occupation to another would have to vary so that his real income(excluding non-monetary considerations) would be the same in each situation.Difference in prices of identical commodities between the situations consti-tute no theoretical difficulty. The trouble arises because in the two situationscommodities are consumed in different proportions. For this there are tworeasons (ignoring differences in tastes). Firstly, prices relative to eachother may, and usually will, be different in each case. In addition, differ-ences in consumption patterns will arise because of differences in the levelof real income enjoyed by the worker before and after the transfer. In termsof indifference curve analysis his new budget can differ from his old one

1As part of an intensive study of relative incomes earned in agriculture and industryin various countries, in some cases going as far back as 1860, carried out by various membersof the Agricultural Economics Research Institute, the author has made estimates of cost-of-living dierences between Swedish agricultural and industrial workers from 1870 onwards.This article summarises the most interesting (and statistically reliable) part of the estimatesand places them in a wider context.

5Nathan Koffsky, Farm and Urban Purchasing Power in Studies in Income and WeaWi,Volume XI. National Bureau of Economic Research, 1949. and J. R. Beilerby, UnitedKingdom: Comparison of Urban and Rural Retail Prices, J .R.S.S. Vol. CXVI, Part 1, 1953.

'In Swedish wage statistics. payments-iti-kind are valued at farm-gate prices (the pricesat which the produce could be sold to middle men buying from farm producers for sale innormal distributive channels)

Page 2: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

WAGES IN SWEDISH AGRICULTURE AND INDUSTRY 437

because of substitution or income effects, or because of both. The essenceof the present procedure is to eliminate the influence of income effects, andto measure only that of substitution effects. In other words, the quaesitum isthe budget which in the new situation at the new situation's prices will leavethe worker no better or worse off than he was in the old situation at the pricesthere ruling: this is the budget which will leave him on the same indifferencecurve. The relationship between the cost of this new budgetcalled here-after the compensating budgetand the cost of the old budget gives thecost of living difference associated with the two occupations.

However, the compensating budget cannot easily be identified. Atany time, the only budgets available will be those consumed by one set ofworkers in one situation and those consumed by others in another. Ignoringobjections to the practice of making interpersonal comparisons of utility,these budgets will most probably not lie on the same indifference curvefor the reason already indicated; that is, because of the influence of incomeeffects. In the dynamic economy with which we are now dealing, the impetusof change of labour from one occupation to another will be differences inreal income. Thus budgets obtained from workers in two different situationsare likely to lie on different curves.

Although a compensating budget cannot be measured absolutely itcan be defined in relation to other (existing) budgets. Consider two situa-tions A and I, and use the normal notation for prices (P) and commodities(Q). The problem is to find the budget Q', which equals in real terms the

PQ'budget QA, so that the ratio ,' can be obtained to measure the change

in money income necessary when the worker moves from A to I so that he. . . . P1Qis no better off than before. This ratio is not greater than the ratio

AQA'and will usually be less, because ,QA will be at least equal to, and moreusually greater than P1Q'J as a result of substitution influences as pricesrelative to each other change from situation A to situation L Thus is derivedthe upper limit of the difference of the cost of living between situations Aand I. When a person moves from A to I his money income will not need to

change by more than the ratio j, so as to leave him with the same real

income as before.Similarly, another measure can be derived of the cost of living difference

between the situations and the lower limit set to this measure. This measure is

AQ1Aand will be greater than the limit

If Q'A is equivalent to Q', in real terms, then the two limits can be usedtogether to provide the lower and the upper limits of the same measure ofthe cost of living difference, so that the actual difference in the cost of living

Page 3: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

438 THE BULLETIN

for workers enjoying the same real income in the two situations can be de-limited accurately. But this equality is unlikely by definition.

In practice it might be possible to choose budgets QA and Q1 that arevirtually equivalent. This obviously introduces an element of circularity intothe argument, which, however, does not invalidate the results. Thus, suppos-ing the average level of real income of workers in situation I is clearly greaterthan the average level of real income of workers in situation A, it might bepossible to approach equality in the cost of living comparison by choosingbudgets relating to relatively lowly paid workers in situation I and budgetsrelating to relatively well-to-do workers in situation A. But there is noway of indicating whether this equality is achieved,

Taking situation A as agriculture and situation I as industry, the foregoingdiscussion provides the following elements in the technique of making acomparison of the costs of living for workers engaged in each.

(x) An agricultural budget valued at prices confronting industrial workersexpressed as a ratio of its value at prices confronting agricultural workersgives the upper limit of the compensating variation of money income neces-sary, when a worker transfers from agriculture to industry, to maintain thesame level of real income.'

(z) An industrial budget valued at industrial prices as a ratio of its value atagricultural prices gives the lower limit of another measure of the compensa-ting variation.2

() Workers in each occupation will, on the average be earning a differentreal income, so that available budgets may give measures of the compensatingvariations at different levels of real income. Industrial workers will usuallyenjoy a greater real income than agricultural workers. As far as possible thisdiscrepancy should be eliminated by choice of budgets.

Further, budgets used for a comparison should be chosen to be as repre-sentative as nearly as possible of real income levels at the date to which thecomparison relates.

() In the case of Sweden, a special precaution is necessary. The pricedata used must reflect the prices faced on the average by all industrial workersand by all agricultural workers. For example, it is not satisfactory to useindustrial prices drawn solely from tb cities, or drawn predominantly fromthe larger cities. An account must be taken of the geographical distributionof the labour force.3

1 brevity, 'prices confronting industrial workers' becomes 'industrial prices'hereafter. Similarly the term 'agricultural prices' is used.

'Separate budgets should, in fact, be used for industrial workers living in differentlocations, for exactly the same reasons as it is necessary to distinguish between agriculturaland industrial budgets. Particularly, separate budgets ought to be distinguished forindustrial workers in rural areas and for those in towns. The inability to do this has probablynot introduced much error into any of the results given later.

'A large part of the industrial labour force is found in rural areas and smail towns.See Appendix B.

Page 4: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

WAGES IN SWEDISH AGRICULTURE AND INDUSTRY 439

H AVAILABLE ESTIMATES

Professor Erik Lindahi and Dr. Lars Lemne have made estimatesfor 1934 and 1941.1 The technique follows explicitly the method describedabove. Budgets obtained in 1933-4 have been valued at three sets of prices;farm-gate prices (rural retail prices, except that self-supplies or paymentsin-kind of food and fuel are valued at farm-gate prices), prices confrontingindustrial workers living in rural areas (rural retail prices) and pricesconfronting industrial workers in urban areas (urban retail prices).2 Threebudgets are distinguished for industrial workers for different levels of realincome: for farm labourers one budget is used. For one budget for eachgroup áomparisons are made including the effects of taxation; all budgets arecompared excluding allowance for taxation. The main results of the studyare shown in Table I.

The work of Dr. Lemne suggests a further comparison for I94x.In order to measure regional differences in the cost of living the SwedishSocial Board has, since before the first World War, made attempts to measurethe cost of purchasing a standard budget in various localities, and localitieshave been grouped on the basis of the differences in cost found.4 Using thecost data and information on the distribution of the industrial populations inTable z, the national average cost of purchasing the budget using the weightssuggested by the geographical distribution of the industrial labour force isfound as io6.8 per cent, of the average cost obtained by using the weightssuggested by the geographical distribution of the agricultural labour force.

This comparison is equivalent to measuring the lower limit of thecompensating variation, assuming that the level of real income of the agri-cultural workers concerned is extremely high. This is evident by the highlevel of money income (4350 Kr.) postulated by the budget.5

III. Owr. ESTIMATE: 1930-44The basis of the analysis is a division of Sweden into six types of localities

and of expenditure into seven groups of items. At any particular date thecost of each set of items is estimated for each locality group, and, by weightingthe results by statistics of the geographical distributions of agricultural and

1 JodbruhsbefoZkningens ¡evnadskostnader, a special report for the 1942 Committee onAgriculture, published as an official publication in 1944 (S.0.U. 1944: 1).

These estimates are, in principle, defective on two counts although neither is likelyto be serious in practice.

The price statistics for rural and urban areas are averages and it is not clear that theseare obtained by weighting returns from individual districts according to their importancein terms of population. Since industrial workers in ¡ural areas live in the relatively morecostly regions and agricultural workers in the less costly, the real cost of living differencemay be understated.

The 1934 budget is also used for 1941. Since increasing real income means an increasein the value share in expenditure of urban 'exports ' and so causes price relatives to moveto the disadvantage of agriculture, the recorded faU from 1934 to 1941 in the cost of livingdifference will be understated.

* LOner och ievnads*osternader i sverige, an unpublished study made available to thewriter.

'The procedure is called 'dyortsgruppermg'.¡ See n. L, p. 444.

Page 5: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

Sour

ceT

able

16

of r

epor

t.

TA

BL

E I

Val

ue o

fPe

rcen

tage

Exc

ess

ofPe

rcen

tage

Exc

ess

ofB

udge

t in

Val

ue o

f B

udge

t at U

rban

Ret

ail P

rice

s ov

er V

alue

at

Val

ue o

f B

udge

t at R

ural

1934

, tax

esex

clud

ed, a

lFa

rm-g

ate

Pric

esR

ural

Ret

ail P

rice

sR

etai

l Pri

ces

over

Val

ua a

tFa

rm-g

ate

Pric

esru

ral r

ttaü

Tax

esT

axes

Tax

esT

axes

Tax

esT

axes

pric

esIn

clud

edE

xclu

ded

Incl

uded

Exc

lude

dIn

clud

edE

xclu

ded

Kr.

1934

1941

1934

1941

7934

7941

1934

1941

1934

1941

1934

1941

Ear

m L

abou

rers

' Bud

get

1928

22.1

14.2

24.6

17.5

13.5

7.1

15.5

9.5

7.5

6.6

7.8

7.3

Indu

stri

a) W

orke

ri' B

udge

t(1

)22

7621

.114

.712

.59.

37.

74.

9(2

)27

8614

.912

.919

.815

.79.

48.

113

.510

.35.

04.

55.

64.

9(3

)32

3119

.014

.412

.59.

35,

84.

8

Page 6: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

Sources: Cost of budget statisticsBetSnkande angoende dyartsgrupperingen, p. 231.Population statisticsLemne's unpublished report (Based on population census,)Nom. Groups A and B correspond approximately to Agricultural and Mixed Communes.

C and D to Industrial Communes, E to Municipalities. F to Cities with less than 20.000population, G and H to Cities with between 20,000 and 100,000 population, and I to citieswith over 100,000 population. (See below, Part III.)

industrial populations,' coefficients have been derived showing the cost ofeach budget item valued at industrial prices as a ratio of the cost at agricul-tural prices. These coefficients have then been combined according toestimated expenditure patterns for agricultural and industrial workers at thedate concerned on two bases, the first a retail price valuation and the seconda valuation based partly on farm-gate prices. For each date, therefore, fourmeasures are derived of the ratio of all expenditure valued at industrial pricesto the same expenditure valued at agricultural prices.

In more detail the six locality-types are (a) Agricultural and Mixed Com-munes, (b) Industrial Communes, (c) Municipalities, (d) Cities with less than20,000 population, (e) Cities with population between 20,000 and xoo,000,and (f) Cities with populations over ioo,000. The seven budget items dis-tinguished are Food, Housing, Fuel, Light, Clothing, Local Taxes, andMiscellaneous Expenditure. Within each group, except fuel, fixed budgetsare used. Thus, within the groups, the expenditure patterns of agriculturaland industrial workers are assumed the same, the necessary information beingusually taken from family budgets relating to industrial workers.'

The result of the combination of the coefficients into one or other averagewill approximate to one of the limits to the cost of living difference defined inthe theoretical discussion. Using weights derived from agricultural budgetswill give the upper limit of the compensating variation at one level of realincome. Industrial budget weights will provide the lower limit of the com-pensating variation at another level of real income. These levels of real incomewill not usually coincide although budgets have been chosen, whereverpossible, relating to levels of money income which will bring the levels ofreal income towards equality.

'For the method of estimating these distributions see Appendix B.2 In practice, the theoretical imperfection is not important. Evidence given later

(Appendix A), shows that in the case of food no serious errors arise through ignoring thispart of the substitution effect.

Cost ofLivingGroup

RelativeCost of Budget

(S1oc.ho1m = 1000)

Percentage DistributionWorkers

Agricultural

ofin 1940

Industrial

A 786 61 22B 814 19 16C 842 12 12D 869 5 9E 885 2 10F 902 1 13G 936 7

H 949 2I 974 9

WAGES IN SWEDISH AGRICULTURE AND INDUSTRY 44'TABLE 2

Data for Estimating Cost of Living Difference in 1941, using 'Dyortsgruper' Statistics

Page 7: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

TA

BL

E 3

Cos

i of

Mai

n Il

enis

of

Bud

get E

xpen

ditu

re in

193

0, 1

941

and

1944

by

Loc

aliti

es a

nd f

or A

gric

ultu

ral a

nd I

ndus

tria

l Wor

kers

(Cos

t in

Agr

icul

tura

l and

Mix

ed C

onzm

unes

100.

0)

1930

Food

1941

1944

Hou

sing

1930

194

1-4

Lig

ht19

30 1

941-

4 19

30Fu

el19

4119

44C

loth

ing

1930

-41

1944

193

0T

axes

1941

1944

Mis

c. I

tem

s.19

30-4

1 19

44

Agr

icul

tura

l and

Mix

ed C

om-

mun

es...

......

100.

0 10

0.0

100.

0 10

0.0

100.

0 10

0.0

100.

0 10

0.0

100.

0 10

0.0

100.

0 10

0.0

100.

0 10

0.0

100.

0 10

0.0

100.

0In

dust

rial

Com

mun

es ..

....1

01.6

101

.6 1

00.4

104

.3 1

04.3

100

.0 1

00.0

108

.6 1

08.6

113

.9 1

00.4

100

.297

.697

.699

.1 1

00.5

100

.7M

unic

ipal

ities

.....

....1

02.4

102

.4 1

00.8

116

.8 1

16.8

100

.0 1

00.0

122

.9 1

22.9

134

.999

.8 1

00.4

105

.0 1

06.8

109

.7 1

00.4

100

.4T

owns

und

er 2

0.00

0 po

pula

tion

103.

0 10

3.9

101.

5 13

1.6

136.

093

.393

.7 1

30.8

131

.9 1

42.0

100

.2 1

00.7

104

.7 1

06.3

115

.9 1

00.7

100

.8T

owns

20,

000-

100,

000

popu

-la

tion

......

......

104

.6 1

04.2

100

.8 1

62.4

160

.491

.095

.9 1

36.2

136

.7 1

50.0

103

.2 1

01.3

107

.3 1

10.0

121

.7 1

01.1

101

.4T

owns

ove

r 10

0.00

0 po

pula

tion

108.

2 10

5.6

102.

4 22

7.0

214.

191

.683

.3 1

70.1

132

.9 1

35.7

100

.8 1

01.1

107

.5 1

10.2

110

.4 1

03.2

103

.1V

alue

of

Inde

x fo

r al

l Agr

icul

-tu

ral W

orke

rs...

...10

0.3

100.

4 10

0.1

101.

3 10

1.3

99.9

99.9

102

.1 1

02.1

103

.3 1

00.1

100

.099

.799

.8 1

00.2

100

.1 1

00.1

Indu

stri

al W

orke

rs...

...10

2.9

102.

7 10

0.9

134.

8 13

6.3

96.8

95.4

123

.4 1

18.3

123

.710

0.5

101.

9 10

3.3

106.

510

1.1

1930

100.

610

1.0

1941

100.

710

1.0

Cos

t to

Indu

stri

al W

orke

rs a

sPe

rcen

tage

of

Cos

t to

Agr

i-cu

ltura

l Wor

kers

......

102.

6 10

2.3

100.

$ 13

3.1

134.

6%

.9p3

.5 1

20.9

115

.9 1

19.7

100.

5 10

2.2

103.

5 10

6.3

101.

019

3010

0.5

100.

919

4110

0.6

100.

9D

itto,

allo

win

g fo

r se

lf-s

uppl

ies,

etc.

, to

farm

wor

kers

at f

arm

-ga

te p

rice

s...

......

121.

1 11

3.8

112.

116

5.6

158.

8 16

4.0

Page 8: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

WAGES IN SWEDISH AGRICULTURE AND INDUSTRY 443

For x.i and 1944 no new calculations of price spreads have been neces-sary. A'vailable statistics are reproduced in Table 3.1 The calculations for1930 are shown in Appendix A, together with alternative price estimates forself-supplies and payments-in-kind for farm labourers at all dates.

Using the population weight suggested in Appendix B, agricultural andindustrial average coefficients of the price indices have been obtained, asgiven in Table 3. Combination of these coefficients give the followingresults.2

TABLE 4Percentage Excess of Cost of Living of Industrial Workers over Cost of Living of Agricultural

Workers

IV. FiRM. ESTIMATE 1930-44

In reconciling the foregoing estimates the results from the Lemne-Lindahlcalculations (Part JI(A)) have been taken as being, most likely, slightly toolow, and those of Part III as being, perhaps, slightly too high. There are tworeasons why the former may be too low.

1They are taken from Bet4lnkande angaende dyortsgrupperingen augivet av 1943 arsdyortskoinmi ¿te (S.O.U. 1945:32. Chapter 5). The patterns of expenditure used to measurethe cost variations by districts are based on a hypothetical budget which postulates amoney income of 4,570 kr. This use of a budget referring to a level of real income verymuch higher than any attainable by agricultural workers will underestimate the urban-ruralprice spreads confronting industrial workers on the same level of real income as agriculturalworkers recorded, since urban industrial workers on low incomes would 'import' a greaterproportion of their expenditure from rural areas than would those with higher incomes.

'The weights used were derived from the following budgets, chosen from varions alter-natives. to reflect satisfactory levels of real income. Budgets A and C being used for 1930and B and D for 1941 and 1944.

Sources: Budget A. Statistisk arsbok for sverige. 1948, p. 221. Budgets B and C, ford-bruketsbefolkningens levnadshostnader. p. 59. Budget D, Bet5nkande angaende levnadskosinad.sindex (S.O.U. 1943:8). p. 73.

NOTES. Budget B has self-supplies valued at rural retail prices. No agricultural budgetrelating to a date later than 1934 is available. Budget B reflects a level of real incomeenjoyed by few agricultural workers in 1934. By 1941 it would be much more typical.

1930 1941 1944

Using only retail prices for valuation:Agricultural Budget ............ 5.6 5.2 4.6Industrial Budget ............ 6.5 6.0 5.7

Using farm-gate prices for valuation of self-supplies. etc.Agricultural Budget ............... 17.0 12.8 12.1Industrial Budget ............... 16.6 13.9 13.6

PerAgricultural

A

Cent of TotalBudgets

B

ExpenditureIndustrial

CBudgets

DDate of Expenditure 1933-4 1934 1933-4 1941

Food 51.0 52.8 45.3 46.8Housing 9.1 9.3 12.8 10.4Light 1.0 1.0 1.0 1.0Fuel 4.5 3.5 3.9 5.9Clothing 12.6 12.3 13.3 11.6

2.9 Not included Not included 5.1Miscellaneous 18.9 21.1 23.7 19.2Average Expenditure of Workers

covered by Enquiry (kr.) 1,688 1,928 2,560 3,100

Page 9: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

444 THE BULLETIN

Firstly, their treatment of local taxation shows the level of taxation aslower in urban than in rural areas. This seems an untenable result, whichis a view supported by the statistics of Betnkande angaende dyortsgrupper-ingen. It is true that the higher taxation in urban areas is partly the cause ofhigher real incomes enjoyed. But, the increase in taxation arising from transferfrom work in agriculture to work in industry in urban areas is unavoidable,and for that reason can safely be regarded as a cost-of-living difference.Similarly the Lindahi-Lemne assessment of complete equality of prices inurban and rural areas for miscellaneous expenditure is unsupportedly atodds with the preferred estimate from Betänkande angaende dyortsgrupper-ingen.

Secondly, their method of simple distinction between urban and ruralareas, makes insufficient allowance for the detailed geographical locationsof the two labour forces. The retail prices facing agricultural workers inrural areas will be less than, not equal to, those confronting industrial workersin rural areas, for the reason mentioned earlier.1 The urban retail prices areprobably too low, because from the sources used they appear to be unweightedmeans based on quotations from individual towns. Thus, the returns from thelarger, less numerous, towns will not be insufficiently represented in termsof industrial workers living in them.

The slight contrary bias of the calculation of Part III is probably due tothe overweighting of the larger towns in the estimate of the distribution ofthe urban industrial population.2 Further, there is the imperfect allowancefor the substitution effect, since fixed budgets are used within each main item ofexpenditure. However, an effort has been made to limit this error in calcu-lating the co-eflicients for food and fuel (where it could be most serious), sothat any resulting over-estimate of the cost-of-living difference is small.

The various estimates for assessing the limits of the compensating varia-tion, using either only retail price valuations or farm-gate price valuationsare given in Table 4. Budget comparisons with and without taxes are madebecause in the Lemne-Lindahl study the industrial budget which clearlycomes closest in real terms to the agricultural workers' budget is valued onlywith taxes excluded. The inclusion of the comparisons without taxationconsequently permits a more exact fixing of the compensating variations.

The procedure adopted has been to assess first the i4.i limits by using allestimates for that date, except that obtained in Part III by using the industrialbudget (this clearly gives an upper limit for a standard of real income muchhigher than that enjoyed by agricultural workers). Then all available serieshave been used to carry back the estimate to 1930 and 5934, and forward to1944, the rate of change recorded by the budget excluded for 294! beingconsidered more reliable than its absolute values. The results are given inTable 4.3

See n. 2, p. 440. See Appendix B.Although a mean value is given for the limits it is, of course, recognised that in principle

there is not usually one unique value for the cost-of-living difference.

Page 10: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

TA

BL

E 4

Sum

mar

y E

stim

ates

. Cos

t-of

-Liv

ing

Dif

fere

nces

. with

Fin

al R

esul

ts.

1930

-194

4.

Perc

enta

ge E

xces

s of

Cos

t of

Bud

gets

at I

ndus

tria

l Pri

ces

over

Cos

tat

Agr

icul

iurc

i Pri

ces

w -t

Ret

ail P

rice

Val

uatio

ns O

nly

1930

1934

1941

1944

Farm

-Gat

e Pr

ice

Val

uatio

ns f

orFo

od a

nd F

uel i

n Fa

rm L

abou

rers

Bud

gets

1930

1934

1941

1944

o, - (n

Ow

n E

sti,n

ate

Incl

udin

g T

axes

(Upp

er L

imit

'i L

ower

Lim

it5.

66.

55.

26.

04.

65.

717

.016

.612

.813

.912

.113

.60

With

out T

axes

fUpp

er L

imit

t Low

er L

imit

5.7

6.5

5.2

6.1

17.1

16.6

12.8

14.0

C)

tUpp

er L

imit

7.7

3.3

13,8

10.1

Incl

udin

g T

axes

'Low

er L

imit

4.0

3.7

939.

0L

emne

-Lin

dahl

Est

imat

esU

pper

Lim

it6.

74.

415

.012

.0W

ithou

t Tax

es't

Low

er L

imit

(Bud

get 2

)5.

84.

711

.79.

9

Lm

ne E

stim

a fe

With

out T

axes

ILow

er L

imit

(Bud

get 1

)L

ower

Lim

it5.

44.

36.

813

.59.

4o,

Sugg

este

d L

imits

f U

pper

Lim

it5.

57.

45.

04.

417

,015

.512

.011

.7of

Cos

t-of

-Liv

ing

iLow

er L

imit

5.5

6.0

4.9

4.4

(12.

0 to

13.3

9.4

9.4

Dif

fere

nces

14.5

)M

ean

5.5

6.7

5.0

4.4

15.0

15.0

10.7

10.5

Page 11: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

THE BULLETIN

It is clear from Table 4 that although conclusions which are broadlycorrect can be drawn, exact measurement of the cost-of-living differencescan not be made. The worthwhile conclusions seem to be as follows.

(i) With retail price valuations only, the increase in money incomenecessary to leave a worker with the same real income on moving fromagriculture to industry would be of the order of 6 per cent. in the early193os, falling to just over 4 per cent. by the middle of the 194os.

(z) If a worker ignored the effort necessary to acquire self-supplies theadjustment would be around r per cent. at the beginning of the period andio per cent at the end; in brief, two and a half times the compensation necessaryon the assumption of retail price valuations.

() Every effort has been made to discount the effect of quality differencesin housing normally available in urban and rural districts. The problem ofhousing always jeopardises the accuracy of urban-rural cost-of-living com-parisons. If errors have arisen through the insufficient exclusion of qualitydifferences all of the foregoing estimates would be 3 per cent too high.'

() In the early 1930's the gap between money wages in agriculture andindustry was extraordinarily wide when compared with the situation of thewar years and since. Even the most sanguine allowance for urban-ruralcost-of-living differences does not significantly change the position. This isshown by the following comparison, based on wage statistics given in theyearly publication Lönestatistik arsbok for sverige.

G. R. ALLEN.

Agricultural Economics Research Institute,Parks Road.

1This is the extreme limit, assuming that comparable housing costs are the same in ruraland urban areas. For various reasons (e.g. lower building wage rates in rural areas, theextent to which timber is used in Swedish housing) this is a doubtful assumption for theperiod considered. The paradoxical results of Table 4, where some' lower' limits are in facthigher than their corresponding upper limits arises because the housing expenditure itemin the budgets used do not eliminate the effects of quality differences.

'This series starts in 1936. It has been carried back by an index number based onstatistics which are more limited in scope than those compiled from 1936 onwards, yet stillreliable for showing the extent of change.

1930 1934 1941 1944 1949

Annual Industrial Earnings, adult males(kr.) ... 2,865 2.715 3,615 4,260 6,077

Annual Agricultural Earnings, adult males(kr)' ............... 1,270 1,171 1,986 2.573 4,450

Agricultural Earnings x MaximumCompensating Variation' (kr.) ... 1,460 1,347 2,185 2,830

Agricultural Earnings x Probable 'Com-pensating Variation' (kr.) ...... 1,346 1,241 2,065 2,676

Page 12: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

1930 4 18 11 2 3

(A few towns from the northern Swedish counties have been excluded, sincethese counties are not important from the viewpoint of the hired agriculturallabour market). Urban areas with less than 5,000 population have been takenas being municipalities-a reasonable assumption as far as price levels isconcerned. Towns between 20,000 and 100,000 population have been dis-tingrished into two groups to ensure more accurate weighting according to thegeographical distribution of the population. These are combined in the finalpresentation of results.

In the fuller unpublished study on which this summary is based, twopossible defects in the statistics were examined, but considered unimportant.These were (1) the small size of the sample, especially for municipalities, and(2) geographical unrepresentativeness of the sample.

(1) FoodRepresentative food items, according to Budget C. of footnote 1,p. 442,

are covered. By value they are 84 per cent of the food budget. Since noinformation is available for rural areas, the price spreads between theseand municipalities in 1941 have been assessed for 1930. This clearlyis a conservative assumption. When an agricultural budget (Budget B offootnote 1, p. 442) is used, the index is not changed substantially, as shownabove. This aspect of the substitution problem can be ignored.

Population (Under 5,0OO 100)

Commodity5,000-20,000

20.000-40,000

40,000100,000

100,000and over

Weight for AllFood Index

Budget C Budget BFresh Milk 101.0 101.5 96.5 105.4 19 23Butter 98.5 98.8 97.3 96.1 9 7Cheese ...... 104.7 102.7 108.5 109.2 3 3Margarine 100.1 100.1 100.5 100.9 5 7Fresh eggs 98.7 100.0 98.6 105.7 5 3Potatoes 106.7 109.1 105.3 110.1 4 5Wheat Flour ., 100.2 99.5 98.8 97.0 5 8Rye Flour...... 101.2 101.2 100.9 98.5 3 5Wheat Bread 96,0 105.0 119.8 112.8 7 2Rye Bread 102.3 100.9 102.0 101.7 2 2Beef and veal 100.0 104.3 106.7 113.9 10 7Pork and Bacon 103.8 105.4 100.8 116.6 11 13Fish 103.7 111.1 108.2 124.5 3 3Coffee 100.6 100.7 102.3 98.8 6 7Sugar 98.0 98.2 96.7 93.0 8 8AU Food (Weights

from IndustrialBudget)...... 100.6 102.3 101.8 105.7 100 103

All Food (Weightsfrom AgriculturalBudget)...... 101.1 102.1 100.6 105.0

WAGES IN SWEDISH AGRICULTURE AND INDUSTRY 447

APPENDIX APRICE SPREADS IN 1930

Food, Light and FuelThe source of the prices used to value these budget items in the different

localities is Detaljpriser och indexberükningar aren 1913-30 (Swedish officialpublication, 1933). Thirty-eight localities are covered in the analysis, as beiow.

Population SizeUnder 5,000- 20,000- 40.000- Over5,000 20,000 40,000 100,000 100.000

Page 13: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

448 THE BULLETINLightOnly paraffin prices are available for urban areas. The cost in rural areas

has been taken as the same as in the municipalities. As late as March 1936,only some 53 per cent of houses in rural areas had electricity. (S/irskilda folk-nïkningen 1935-6, Vol. III, p. 99). For farm workers the percentage would be,presumably, much less. The use of paraffin prices is, therefore, justified, atleast in assessing the upper limit of the cost-of-living difference.

FudThe fuels considered are birch and pine woods, anthracite and coke. Gas

is omitted. (In 1942, some 13 per cent of total expenditure on heating,assuming all gas was used for heating purposes, was spent on gas. Source:Betár.kande angaende levnadskostnadsindex, 1943, p. 44). The price statisticsgiven below show the need to make some allowance, albeit partly arbitrary, forthe substitution effect. Weighting the price data according to averagenational consumption patterns will overstate the real cost-of-living differences.

Fuel Prices in 1930. (Prices in Towns Under 5000...100)

Details on the national consumption pattern in towns are available for 1913(Cost of Living in Sweden, pp. 161-2). For 1920, a year in which agriculturallabourers' real incomes were high, statistics are available showing the fuelconsumption pattern for a small sample of agricultural labourers and small-scale farmers. (Levnadskosnaderna a landsbygden i sverige vid ar 1920, p. 83.372 workers are covered by this enquiry. The statistics used above refer to225 workers, the others having been excluded since they obtained earningshigher than the average for agricultural wage earners at this date. Theirinclusion would not influence the statistics appreciably.) These latterstatistics may be taken to represent the fuel consumption pattern of allworkers living in rural areas enjoying the same level of real income as agricul-tural workers, and the others to show the expenditure pattern of all workersagain on the same level of real income as agricultural workers (in view of thedates this is probably roughly correct), in urban areas. In 1930 just under onequarter of urban industrial workers would live in towns with under 20,000population. Using these fractions as weights, and carrying out the adjustmentso that a reasonable relationship exists in the changes of percentage expendi-ture on anthracite between different localities, the following results are obtain-ed, the last two lines being used to combine the price statistics for urban areas.Estimated Distribution of Expenditur. on Fuels by Localities for Agricultural Workers

and Industrial workers Enjoying the same Level of Real Income

For the same reason as with food and light, the index of the cost of the fuelbudget by urban districts has been extended to rural areas to assuming, againconservatively, the same spread of the index as in 1941 between them andmuiucipalities.

per cent

Rural Areas 25 72 3 oAll Towns, of which 32 45 12 11Towns under 20,000 36 49 8 7Towns over 20.000 30 43 14 13

Birch Pine and F ir Anthracite Coke

5.000-20,000 population 105.5 109.2 100.5 97.820.000-40,000 108.1 111.8 99.5 91.6population40,000-100.000 population 118.8 146.3 99.0 93-9Over 100,000 population 144.5 165.7 90.9 90.9

Birch 'Pine and Fi, Anthracite Coke

Page 14: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

Average for both types1 Room + Kitchen 2 Rooms + Kitchen StatisticsFromReport Estimated

FromReport Estimated

FromReport Estimates

5,000 10,000 105.1 112.1 108.610,000 20,000 (110.3 114.9 (112.620,000 40,000 . 125.7 .c 131.3 135.2 142.8 130.5 . 137.040,000-100,000 J 1142.3 156.2 1149.2

100,000 and over 192.4 208.2 200.3

WAGES IN SWEDISH AGRICULTURE AND INDUSTRY 449

HousingAn index for 1930 has been derived by interpolating between the 1941

index and estimates for 1924 based on Hyresräkningen tir 1924, Volume I.(Swedish official statistics, 1925).

The calculations for 1941 appear to cover all standards of housing signi-ficant for the budget which forms the basis of the cost-of-living calculations inthat year, with adequate allowance for the effect of quality differences betweendistricts. For earlier dates, the cost of renting two types of accommodationare examined. These are houses of one room plus kitchen, and those of tworooms plus kitchen, and the statistics for each have been given equal weight-ing. According to the Partial Census of Population of 1936, at that date 39per cent of all dwellings in Sweden consisted of at most one room plus kitchen,and 29 per cent of two rooms plus kitchen. For rural areas, the figures were,respectively, 34 per cent and 29 per cent and for urban areas 46 per cent and28 per cent (Statislisk arsbok for sverige, 1948, p. 229). For earlier dates thesewould clearly be a greater proportion of single roomed houses.

Considerable allowance for quality difference can be made when usingthe 1924 statistics, by excluding statistics for housing with central heating,since this service will tend to be found to a relatively greater extent in urbanareas, especially in the more densely populated and will be associated withhousing of better quality in other respects. It is not clear whether thestatistics as used exclude other quality differences, but these are probablynot important, in that their development has been mainly associated with theexpansion in the rate of urban house construction since the later 1920's. Itseems safe, therefore, to assume that quality differences do not play a significantpart in influencing the range of the indices for 1924.

The 1924 statistics derive from a large scale enquiry relating to 220localities including all important urban areas and some rural areas mostlyadjacent to towns. (An estimate, not published here, shows that inclusionof these unrepresentative rural areas does not distort the results). The level ofrents is distinguished for four groups of localities in the following Table. Forthe first three the statistics as given in the report have been used; but forthe group of towns with more than 100,000 population an average rent hasbeen calculated by weighting the returns for each town according to theirrelative sizes. The necessity of obtaining a weighted average for this grouparises from the wide disparity of both rents and size of towns in the group.The weights used are, Stockholm 76, Göteborg 44, and Malmö. The thirdgroup was broken down by using the rent estimates for 30 individual towns,covering 15 with populations under 20,000, 13 with populations between20,000 and 40,000, and the 2 towns with populations between 40,000 and100,000. (The reason for reducing the size of the sample slightly was to effecta comparison with results for the identical towns in 1913 for the purpose ofthe fuller enquiry into cost-of-living differences).

Cost of Renting Accommodaion in 1924 by Size of Loca3ity(Rent in areas with Less than 5,OOó 100.0)

Page 15: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

450 THE BULLETIN

The index has been regrouped, as required for comparison with 1941,according to the total urban population in 1924 in each group. The result isas follows.

Cosi of Renting Accommodation in 1024 and 1041 in Urban Areas by Size of Locality(Rent in municipalities = 100.0)

Once again, no information is available for rural areas and the 1941estimate has been used for 1930. This has probably caused some under-estimation of the spread of rents in 1930.

Clothing and Miscellaneous ExpenditureIn default of other information the indices for 1941 have been used,

Local TaxesThe index for 1941 measures the locality variation in the amount of tax

which would be paid by low grade state employees. From 1930 to 1941 localtaxes rose more in towns than in rural areas. This is shown by Dr. Lemne(in his unpublished study mentioned earlier), who records a rise of local taxesin towns of 36 per cent between 1930 and 1941 relatively to the rise in ruralareas. The estimate for 1930 has been obtained by deflating the 1941 indexvalues for urban areas according to this change.

FoodValuation of Self-Supplies at Farm-Gate Prices

From JordbruhsbefoThningens levnadskostnad.er (p. 42 and pp. 53-4), the followingtable is compiled.

V,.lue of Agricultural Labourer's Food Budget

AAt farm-gate orretail prices, as

appropriate

BAt rural

retailprices

Aaspercentege

of B

The first percentage has been used for 1930 and the second for 1941 and 1944,

FuelIn 1934 fuel expenditure (at farm-gate prices) was 61.6 per cent of total

expenditure on fuel and light while in 1941 light and fuel were valued at 146kr. when fuel was valued at farm-gate prices, and at 179 kr. when rural retailprices only were used. (Jordbruksbefolkningens l.evnadskostnader, p. 42 andp. 54). Applying the percentage to the first of the 1941 values, fuel in 1941would cost 90 kr. at farm-gate prices and 123 kr. at rural retail prices. Thesefigures suggest that the ability to obtain supplies at farm-gate prices wouldreduce the farm labourers' cost of purchasing fuel by 27 per cent in 1941, adeflating factor which has also been used for 1930.

Valued at Receipts in kind 247.41934 prices Purchases with cash 631.2

Total 878.6 1037.4 84.7Valued at Receipts in kind 479.01941 prices Purchases with cash 1087.0

Total 1566.0 1742.0 89.9

1924 1941

Towns under 20,000 110.6 116.4Towns 20000-100.000 140.0 137.3Towns over 100,000 200.3 183.3

Page 16: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

In 1930, 58.4 per cent of the male industnal labour force lived in ruralareas (Thomas, op. cit., p. 181). By using Census data relating to all industrialworkers in 1930 and 1940, the figure for 1940 is estimated 52.3 per cent. Thebreakdown of this group between the three groups of rural communes hasused much of the same material as was employed above. In 1930, in agri-cultural communes 13 per cent of the total population was dependent onindustry, in mixed communes 24 per cent and in industrial communes 44cent (Population Movements, p. 330). For 1910 estimates the 1930 estimateshave been adjusted according to the variation in the proportions of thepopulations not dependent on agriculture between 1910 and 1930 to give,respectively for each type of commune, 12 per cent, 20 per cent and 40 percent. Interpolated percentages for 1925 are 13, 23, and 43. Thus, the pro-portion of the total industrial labour force which lives in agricultural andmixed communes has been estimated as follows, 1910 44 per cent, 1925 41per cent and 1930 49 per cent.

The distribution of the urban industrial labour force has been taken ascorresponding to the distribution of the total urban population between townsof different size. (These statistics were derived from Statistisk arsbok, 1948and Thomas, op. cit. p. 43). There is probably some over-assessment of theimportance of the larger towns, since generally they contain a lower proportionof their population in secondary occupations and a greater proportion intertiary activities than do the smaller. The error is clearly not great.

Estimates on similar lines to the foregoing have been made back to 1890.There is no need or space to show them here. One conclusion from the longerperiod analysis is, however, important. Between the years 1890 and 1930, the

1905-14 19 15-20 192 1-30

Per centAgricultural Communes 58.9 56.1 56.6Mixed Communes ...... 22.4 23.6 22.9Industrial Communes 17.2 18.4 18.4Towns 13 1.7 1.9Stockholm 0.2 0.2 0.2

WAGES IN SWEDISH AGRICULTURE ANY) INDUSTRY 45'

APPENDIX BLoCATIoN or THE AGRICULTURAL AND INDUSTRIAL POPULATIONS

The starting point is the distribution of the total Swedish population inthree periods, 1905-14, 1915-20 and 1921-30, as between the three types ofrural commune (agricultural, mixed and industrial), towns excepting Stock-holm, and Stockholm. (Source, Population Movements and Industrialisation,Institute for Social Sciences, University of Stockholm, p. 331). For 1910and 1930 the percentage of rural populations dependent on agriculture for thethree types of commune are given in D. S. Thomas' Social and EconomicAspects of Swedish Population Movements (p. 371) as follows: 1910, agricul-tural communes 72 per cent, mixed communes 54 per cent and industrialcommunes 30 per cent: 1930, 70 per cent, 44 per cent 24 per cent respec-tively. Interpolated estimates have been made of these percentages to applyto the total population figures just mentioned, to give the following result.

Location of AgrscuUural Population

Page 17: RELATIVE REAL WAGES IN SWEDISH ACRICULTURE AND INDUSTRY, 1930–1950

42 THE BULLETIN

proportion of the total industrial labour force which lived in industrialcommunes was constant at 35 per cent. This percentage has been taken for1940. Thus, for 1940 given the proportion of the total industrial labour forcein urban areas, the proportion of it in agricultural and mixed communes wasfound as a remainder item. The following suznmaiises the results of the analysis.

Final Estinsaies for 1930 and 1940 of Geographical Disfribu*ion of Male Agriculiural andIndusfrial Labour Forces

Erratum. BULLETIN, Vol. 15, No. 12. T. Schulz: A'Human Needs' Diet. Page 434, Diagram. The per-centage figure in the upper half of the diagram referring topossible non-food expenditure out of an income of 155s.should read 58 (and not 56).

PerAgriculture1930-40

cen*Industry

1930 1940

Agricultural and Mixed Communes 80 23 19Industrial Communes 18 35 33Municipalities 1 2k 2Towns under 20.000 Population

with1 11 12

8 7Towns 20-40,000 PopulationTowns with 40-100,000 Population 2k 5Towns under 100,000 Population 18 20