a multilevel analysis of agricultural trade and ... · in these communities have important...

A Multilevel Analysis of Agricultural Trade and

Socioeconomic Inequality in Rural Mozambique∗

Julie A. SilvaUniversity of Florida

This study uses a mixed methodology to investigate the relationship between agricultural trade andsocioeconomic inequality in Mozambique at multiple geographical scales. The quantitative analysis examinesall rural regions of the country and a nationally representative sample of rural households, whereas thequalitative analysis focuses primarily on two case study villages. Research findings suggest that social andgeographical context greatly influence the relationships between agricultural trade and regional inequality.Trade-related effects on economic inequality—and the social status of traders—vary across Mozambicanregions. Results suggest that trade increases inequality in regions where traders have low social status, butreduces inequality in areas where they have high social status. These findings suggest that social and economicfactors work together to shape landscapes of inequality. Key Words: inequality, international trade, mixedmethodologies, Mozambique.

En este estudio se utiliza una metodologıa mixta para investigar las relaciones entre el comercio agrıcolay la desigualdad socioeconomica en Mozambique, a escalas geograficas multiples. El analisis cuantitativoexamina todas las regiones rurales del paıs y una muestra nacionalmente representativa de las viviendasrurales, mientras que el analisis cualitativo se concentra principalmente en dos pueblos. Los resultados de lainvestigacion sugieren que el contexto social y geografico influencia en gran medida las relaciones entre elcomercio agrıcola y la desigualdad regional. Los efectos relacionados con el comercio sobre la desigualdadeconomica, y el estatus social de los comerciantes, varıan en todas las regiones de Mozambique. Los resultadossugieren que el comercio aumenta la desigualdad en aquellas regiones en las que los comerciantes tienen unestatus social bajo, pero la reduce en areas en las que lo tienen alto. Estos resultados sugieren que los factoressociales y economicos funcionan conjuntamente para conformar el panorama de desigualdad. Palabras clave:desigualdad, comercio internacional, metodologıas mixtas, Mozambique.

W estern development institutions com-monly advocate trade liberalization to

spur economic growth in developing nations.Although protrade policies may generate ag-gregate welfare gains in low-income countries,not everyone benefits equally from liberal-ized trade (Dicken 1998; Kapstein 2000; Wade∗Funding support for this study was provided, in part, by dissertation research grants from the National Science Foundation (DDRI GrantBSC-0401776), the Association of American Geographers, and the Social Science Research Council Program in Applied Economics with fundsprovided by the John D. and Catherine T. MacArthur Foundation. The author thanks her advisor, Robin Leichenko, and committee membersMike Lahr, Joanna Regulska, Lyna Wiggins, and Richard Schroeder for all their help and support during this study. The author also expressesher gratitude to Stewart Duncan, Corene Matyas, and five anonymous reviewers for their comments and suggestions on earlier drafts of thisarticle. The author also thanks Nicholas Campiz for cartographic and graphic design assistance. All remaining errors and omissions are the soleresponsibility of the author.

2001; Leichenko and O’Brien 2002, 2008;O’Brien and Leichenko 2003). Important ques-tions remain about how these reforms affectinequality in emerging economies, especially atthe subnational level. Distinguishing the win-ners and losers of trade liberalization is of par-ticular importance in southern Africa, where

The Professional Geographer, 60(2) 2008, pages 174–189 C© Copyright 2008 by Association of American Geographers.Initial submission, June 2007; revised submission, October 2007; final acceptance, November 2007.

Published by Taylor & Francis, LLC.

Analysis of Agricultural Trade and Socioeconomic Inequality in Mozambique 175

Figure 1 Map of Africa.

most countries are extremely poor and alreadyhave high inequality (Evans 2001; Wobst 2003).This study investigates the effects of agricul-tural trade on inequality in rural regions ofMozambique, a country that aims to allevi-ate poverty by promoting agricultural exports(Figure 1).1

This study uses a conceptual framework thatdraws on neoclassical trade theory, new eco-nomic geography, and the work of AmartyaSen (1981, 1999) on entitlements and capabil-ities. Neoclassical trade theory claims that in-ternational trade drives levels of inequality, andpredicts that inequality in developing countrieswill decrease as trade levels rise (Wood 1994).Researchers working within the new economicgeography framework agree that trade can af-fect levels of inequality, but suggest that greatertrade volumes could increase inequality if cer-tain regions acquire permanent trade-based ad-

vantages over other areas (Krugman 1991). Iexpand on trade theory frameworks by incor-porating the nonmonetary dimensions of in-equality and the experiences of individuals.This aspect of the study draws on Sen’s (1999)capabilities framework, which states that thefreedom to engage in trade increases people’sopportunities and reduces social inequities.Thus, this study also investigates the relation-ship between trade and social inequality.

To benefit from the strengths of qualita-tive and quantitative research techniques, thisstudy employs a mixed methodology to addressthree research questions in rural Mozambique:2(1) What effects do agricultural trade andother factors have on income inequality atthe regional scale? (2) What factors influencehousehold participation in international ordomestic agricultural trade? (3) How does par-ticipation in trade-related activities—such as

176 Volume 60, Number 2, May 2008

petty trading of agricultural goods—influenceindividual experiences of socioeconomic in-equality? In investigating these questions, sta-tistical analysis is used together with case stud-ies, interviews, and field observations. The datacollection and empirical research were con-ducted sequentially so that findings at eachphase of the analysis could influence the de-sign of later phases. By assessing the effects ofagricultural trade at the subnational level, thisstudy contributes to our understanding of howsocial and geographical context influence ruraleconomies in the developing world.

The Case of Mozambique

Two key factors led to the selection of Mozam-bique as a case study for researching linkagesbetween inequality and agricultural trade. First,much of the debate surrounding the effectsof globalization centers on how much inter-national trade contributes to the rising lev-els of global and national inequality (Wood1994; Rodrik 1997; Anderson 2005). Mozam-bique has an export-led economic developmentpolicy that relies on increasing regional andinternational trade. In-depth empirical studieson Mozambique may well lead to insights onthe role international trade plays in generat-ing social and economic inequities in emerg-ing economies at multiple geographic levels(e.g., region, household, individual). Second,Mozambique has introduced protrade policiesin an environment where the majority of thepopulation lives in absolute poverty, peopleare highly vulnerable to natural hazards, andsmallholder farmers make up the majority ofthe agricultural sector. Such poverty and vul-nerability of farmers are common in the lessdeveloped countries, and therefore these fac-tors warrant inclusion in studies of the relation-ship between agricultural trade and inequality.

Mozambique adopted a structural adjust-ment program in 1987, following a brief pe-riod of socialism and centuries of Portuguesecolonial rule. The country’s structural adjust-ment was one of several austerity programssponsored by the International Monetary Fund(IMF) implemented in sub-Saharan Africa inthe 1980s and 1990s. IMF-sponsored programscommonly involve lowering trade barriers andopening markets, devaluing the national cur-

rency, privatizing state-owned enterprises, andcutting government subsidies and social ser-vices. In Mozambique, these reforms were in-tended to reverse declines in food consumptionand agricultural production in rural areas, re-duce the national debt and dependence on for-eign aid, and establish the conditions for eco-nomic growth (Pitcher 2000; Government ofMozambique 2001). In particular, the state aimsto increase rural incomes by fostering higherproduction of cash crops such as sugar, cot-ton, tobacco, paprika, and chickpeas, in whichMozambique’s agricultural sector appears tohave considerable export potential (Nathan As-sociates 2002; Walker et al. 2004).3

Structural adjustment in Mozambique hashad similar consequences as IMF-sponsored re-forms in other countries.4 The privatization ofstate enterprises led to the closure of state-runfactories and a dramatic reduction in the avail-ability of jobs in the formal sector (Mittleman2000). Access to health care and education de-creased, particularly for women and especiallyin rural areas (Knauder 2000; Sheldon 2002).The findings of Bowen (2000) suggest that pri-vatization has mainly benefited foreign-ownedmultinational corporations and Mozambique’sgoverning elite.

Mozambique faces considerable obstacles toachieving the necessary export growth to alle-viate poverty. Like many developing countries,Mozambique suffers from the adverse effectsof distorting terms of trade on agricultural pro-duction. Other problems include the absenceof risk-spreading mechanisms (e.g., insurance)and weak enforcement of contracts. In addition,connecting smallholders to markets has provenextremely difficult in rural regions. A poortransportation network and limited commod-ity price information translate into higher risksfor farmers as they turn to commercial crops.Moreover, the terms of IMF austerity programshave limited the state’s ability to mitigate somecommon negative effects of market liberaliza-tion, such as environmental degradation.

Uneven development in Mozambique haslong historical roots. Colonialism and civilwar contributed to widespread poverty inthe countryside. Colonial and postcolonialpolicies deliberately channeled development tothe southern region of the country, fosteringspatial disparities in income and well-being.5Other inequalities in Mozambique stem from


gender roles and gender relations that limitwomen’s opportunities to profit from manycommercial activities, including agriculturaltrade. Mozambican women are responsiblefor most agricultural tasks, but their access toland and control over the income from theseactivities varies across the country (Baden1997). Ethnic groups in southern regions ofMozambique, like the Shona, Tsonga, andChangan, have patriarchal gender relations.Males have more decision-making authorityand legal rights, such as the ability to enter intotrade contracts (Isaacman and Isaacman 1983).However, male migration to neighboringcountries led to many de facto female-headedhouseholds in southern Mozambique andgreater participation of women in the publicsector, although absent men still have a consid-erable amount of influence over how incomeis spent (Waterhouse and Vijfhuizen 2001;Arnaldo 2004). In the northern matrilinealsocieties of the Makua, Lomwe, and Chuwabo,wealth and residence pass through the femaleline. Women in these matrilineal commu-nities depend on their male relatives—nothusbands—for access to land, and elder womenin these communities have important cere-monial powers (Baden 1997). Thus wives andwidows have a stronger economic position intheir communities relative to southern women.

Studies of uneven development in Mozam-bique find some evidence that higher lev-els of market integration are associated withhigher levels of inequality. For example,Elbers et al. (2002) find that income inequalityis highest in the capital, Maputo, where mostindustry is located. Other studies have foundthat high levels of income inequality also existwithin rural communities (Heltberg, Simler,and Tarp 2001; Simler and Nhate 2002).Heltberg, Simler, and Tarp (2001) suggest thatmarket-oriented growth may increase inequal-ity in Mozambique by rewarding the small mi-nority who possess social and physical capital.6

Research Design

This study employs a mixed methodology tostudy both human experiences and broad re-gional economic trends, enabling a more com-prehensive analysis of the drivers of inequal-ity (Carvalho and White 1997; Tashakkori and

Teddlie 1998; Courgeau 2003). The study wasconducted between 2002 and 2004 using a se-quential process in which findings at each re-search phase influenced data acquisition andanalysis in subsequent phases (Figure 2). Forexample, macroanalyses informed the selectionof the case study sites and local fieldwork influ-enced the selection of variables for the econo-metric models. Empirical results were verifiedin follow-up interviews and discussions withMozambican farmers, allowing findings fromlater phases of the analysis to inform the inter-pretations of earlier results. Factors that can-not be easily quantified, such as social standingand community cohesion, were explored usingqualitative analyses. The study can be dividedinto five phases.

The first phase of the study involved ex-tensive background research and a geographicinformation system (GIS) analysis of social,economic, and physical factors to select a casestudy from among the less developed coun-tries. Building on the methodology outlinedin O’Brien et al. (2004) and O’Brien andLeichenko (2000), this study used GIS to con-struct poverty maps and vulnerability profilesfor countries. After the selection of Mozam-bique as the case study country, subnationalmapping, data analysis, reviews of secondaryliterature, and key informant interviews wereused to select two case study communitieswithin southern Mozambique (Figure 3). Field-work was then conducted in these two villages:one with a relatively high degree of marketaccess (Massavasse in Chokwe District) andone that is more isolated (Matidze in MabalaneDistrict). The geographic proximity of bothvillages to Mozambique’s capital, Maputo, aswell as their location in the fertile agriculturalzone along the Limpopo River make theseareas an appropriate case study for examiningthe effects of agricultural trade on inequality(Hermele 1988; Roesch 1988). A random sam-ple of sixty-three households was surveyed inthe two communities (thirty-three in Matidze,thirty in Massavasse) on topics includinghousehold demographics, economic activities,crops, migration and remittance patterns, andaccess to information, agricultural inputs, andsocial services. Interviews were conducted inthe local language or Portuguese—dependingon the language spoken by the head ofhousehold—and translated into English. The


Figure 2 Diagram of sequential research phases.

qualitative data were analyzed using NUD∗IST4 software (Richards and Richards 1997) toidentify recurring themes in the interviews,such as the benefits and costs attributed to

agricultural trade and other market-basedactivities.7

In the second phase of the study, a regionaltrade orientation database was compiled for

Figure 3 Map of case study villages in Mozambique.


Table 1 Sources used to construct district-level trade orientation database

Sample sizeData set name Year Source (households)

Base Topographic and Land Use/LandCover Data (GIS data)

1999 Mozambican National Directorate ofGeography and Cadastre(DINAGECA)

N/A

National Housing and PopulationCensus

1997 Mozambican National Institute ofStatistics (INE)

N/A

National Household Survey on LivingConditions

1996–1997 Mozambican National Institute ofStatistics (INE)

8,289

Census of Agriculture and Livestock 1999–2000 Mozambican National Institute ofStatistics and Ministry ofAgriculture and Rural Development(INE/MADER)

23,423

136 rural districts in Mozambique using varioussecondary data sources (Table 1). Next, regres-sion models were estimated by means of ordi-nary least-squares (OLS) and as spatial errormodels (SEM) via maximum likelihood.8 In-equality is modeled as a function of trade orien-tation, agglomeration measures, income mea-sures, locational measures, and demographicfactors. This study employed spatial regressionanalysis to control for the possible presence ofspatial autocorrelation in the regional data set.The use of regression analysis also facilitatescomparability of this study’s findings withother research on the subnational influenceof trade on inequality (Sanchez-Reaza andRodrıguez-Pose 2002; Galiani and Sanguinetti2003; Edin, Fredriksson, and Lundborg 2004;Silva and Leichenko 2004; Aguayo-Tellez2006).9 Information gleaned from the firstphase of the study informed the selection of thevariables and the construction of the regressionmodel. For example, interviewees stresseddifferences between trading vegetable crops—where there were few formal contracts—andtrading cash crops such as cotton and sugar.This influenced the design of subsequentphases of the analysis by focusing the investiga-tion of agricultural trade on two separate typesof crops: domestically traded vegetables andinternationally traded cash crops. Thereforeregional orientation toward vegetable croppingwas treated as a separate category from othertypes of cash crop orientation to measure anydifferential effects on inequality.10

The third phase of the study consisted ofa second research trip to Mozambique to fol-low up with survey respondents in the case

study communities and conduct twenty-one in-formant interviews with government officialsand development specialists. These interviewswere conducted, in part, to verify findings fromthe regional analysis. They also sought to iden-tify the factors that influence household partic-ipation in commercial activities promoted bytrade-based growth policies.

The fourth phase of the study involvedfurther quantitative work using the Trabalhode Inquerito Agrıcola (TIA)—a nationally rep-resentative survey of agricultural Mozambicanhouseholds conducted in 2002 that covered theagricultural year 2001–2002, a typical year withregard to weather conditions and productionyields (Walker et al. 2004). Using bivariate tab-ular analysis, the study investigated the variouseconomic, social, and demographic character-istics of households located north or south ofthe Zambezi River. The analysis investigatedlinkages between household characteristics andparticipation in agricultural trade, and whethersignificant differences existed between andwithin northern and southern regions. House-hold comparisons were conducted within re-gions: Northern trading households were com-pared with northern nontrading households,and southern trading households were com-pared with southern nontrading households.Then household comparisons were conductedbetween regions: Southern trading house-holds were compared with northern tradinghouseholds, and southern nontrading house-holds were compared with northern nontradinghouseholds.

In the fifth phase of the study, follow-up interviews were conducted with survey


respondents in the southern case study sites.In addition, a case study site in northernMozambique—Rapale—was added to the anal-ysis. Fourteen small-scale farmers and eight keyinformants were interviewed there (Figure 3).11

The addition of a northern case study site al-lowed for a qualitative investigation of those re-gional differences in the relationship betweenagricultural trade and inequality indicated inearlier phases of the study. Interviews were con-ducted with northern and southern villagersabout how commercial activities, and agricul-tural trade in particular, impact economic andsocial standing of households. Finally, resultsfrom all phases of the analysis were synthe-sized to investigate the drivers of inequality inMozambique, how they vary within the countryand at different geographical levels of analysis,and the implications of these variations for thegovernment’s protrade policies.

Data Limitations

Although the use of a mixed-method, multi-scalar approach enriched the scope of this study,the analysis has two main limitations. First, theuse of large secondary data sources to constructthe cross-sectional, regional database mighthave introduced error into the regression anal-yses and bivariate analyses depending on the

accuracy of survey data. The reliability of socialand economic data gathered through extensivegovernment surveys is a common concernamong social scientists, especially given thefinancial constraints of data collection in lessdeveloped countries. However, the variety ofsecondary data sources that exist for Mozam-bique allowed me to employ triangulation(i.e., cross-checking the calculations derivedfrom one data set against those of another) toverify the reliability of secondary data sources.This technique indicated the robustness ofkey measurements derived from the availabledata sets and used in this study, such asmean incomes and percentages of householdsengaged in agricultural trade.

The second key limitation comes from thecross-sectional research design of the regres-sion model and bivariate statistics, which pre-vented the study from capturing the temporaldimensions of inequality. Longitudinal studiesthat can control for period fixed effects (suchas natural disasters) at the district level wouldenable an analysis of the impacts of changingtrade pressures on inequality over time. Unfor-tunately data limitations currently make suchstudies impossible. The qualitative componentof this study attempts to redress the lack ofregionally representative longitudinal data byassessing the relationship between trade and

Table 2 Descriptive characteristics of southern case study households

Massavasse (N = 33) Matidze (N = 30)

Category Variable description % of households N % of households N

Commercial Sells cash cropsa 36% 12 3% 1agriculture Sells vegetable cropsb 39% 13 37% 11

Sells basic food crops in emergenciesa 33% 11 87% 26Agricultural Sells crops in a regional marketb 52% 17 70% 21

marketing Sells crops locallyb 33% 11 43% 13Sells crops to intermediaries (guevas)a 0% 0 47% 14

Off-farm income Has off-farm wage incomeb 45% 15 40% 12sources Receives remittancesa 64% 21 30% 9

Demographics Male interview respondentb 42% 14 67% 20Female interview respondentb 58% 19 33% 10Interview respondent is literateb 45% 15 30% 9All eligible children in schoola 88% 29 70% 21

Access to Access to agricultural extension agenta 64% 21 3% 1services Holds title to landb 73% 24 83% 25

Has irrigated landa 52% 17 0% 0

aChi-square test for independence indicates the relationship between this variable and geographic region is statisticallysignificant at the 0.05 level.bChi-square test for independence not statistically significant at the 0.05 level.Source: Author’s calculations using data collected in the field, 2002–2004.


inequality from a different vantage point, usingin-depth qualitative data. Thus, the study pre-sented in this article captures the place-basedvariation in the effects of trade in Mozambiqueand provides an important baseline from whichto study future change.

Results

Attitudes Toward Agricultural TradeThe case study component of this researchfocused mostly on two rural villages in theLimpopo River Basin of southern Mozam-bique: Matidze in Mabalane District and Mas-savasse in Chokwe District (Figure 3). Theoverwhelming majority of surveyed house-holds (98 percent) engaged in some typeof agricultural production. Although mostinterviewees were basically subsistence agri-culturalists, many did engage in small-scaleagricultural trade and other commercial activ-ities (Table 2). Nearly half the households inMassavasse and one-quarter of the householdsin Matidze grew at least one crop intended forsale.

The survey results suggest that farmersviewed vegetable production as distinct fromcash cropping. In general, people expressed apreference for farming vegetables rather thaninternationally exported crops such as cotton.Even people who did not participate in eitheractivity said that—given equal opportunities—they would rather grow vegetables.

A key difference between farming vegeta-bles and cash crops involved the differing risksand commitments associated with each croptype. Cash cropping required contracts andconnections to large companies. Under theterms of cash cropping contracts, farmers weregiven seeds in exchange for agreeing to selltheir produce to a particular company (oftenat below-market rates). The cost of the seeds—and any other agricultural inputs advanced tothe farmer—was later subtracted from the valueof the crop. Cash cropping was described as amale activity due to the need for these farm-ers to enter into binding contracts and qual-ify for credit. Vegetable trading could be doneinformally in small amounts, even if farmerslacked the connections to have a stall in the lo-cal market. Farmers could sell vegetables out-side their homes, or to intermediaries (called

guevas) who later resold the goods to markettraders. Women commonly participated in veg-etable cropping in southern Mozambique andguevas were often female.

Farmers described vegetable sales as themost viable economic activity for householdsin the Limpopo River Basin that wished tofarm commercially. A key benefit was theability of households to eat produce that theycould not sell. Lack of access to cold storagerequired that produce be sold quickly or eaten.Some interview respondents stated that theywould give away what they could not usethemselves. Food gifts functioned as a form ofsocial insurance, as they could expect the favorto be returned in the future.

Other factors also contributed to the pref-erence for vegetable cropping. Respondentsstated that prior experiences with cash cropproduction—especially cotton—had been neg-ative. Rural agriculturalists were forced to growcotton under Portuguese occupation and par-ticipate in collective agricultural schemes un-der the socialist government (Urdang 1989;Abrahamsson and Nilsson 1995; Isaacman1996). People also described the terms of cashcropping contracts as unfair, making the ac-tivity risky and unprofitable. Even householdsthat did not grow cash crops expressed famil-iarity with market prices for them. The pricesof cash crops had been decreasing but pricesfor vegetables were rising. Urbanization ap-pears to be a factor in the growing demandfor vegetables.

Regional Patterns of InequalityThe qualitative research already described re-vealed strong preferences for growing vegeta-bles, in contrast to the government’s prefer-ences for export crop production. The nextphase of the study addressed the first re-search question and investigated whether thetwo types of trade had differential effects oninequality across Mozambique. The relation-ship between trade and inequality was analyzedwithin the Heckscher–Ohlin (HO) framework,a widely used version of trade theory that pre-dicts higher levels of trade within a develop-ing country will be associated with lower lev-els of inequality. The model was estimatedseparately for districts north of the ZambeziRiver and those districts to the south, dueto evidence of structural change in the data


set (i.e., linear regression parameters were notequal across northern and southern districts).12

The regression results, reported in greater de-tail in Silva (2007), suggest a relationship be-tween inequality—as measured by the Ginicoefficient—and agricultural trade, but the di-rection of the relationship is mixed.13 The find-ings of Silva (2007) suggest that cash cropping isassociated with low inequality in the north, al-though vegetable trading has no significant ef-fects on inequality there. In contrast, vegetablecropping is associated with higher inequality inthe south, but cash cropping apparently has nosignificant effect on inequality there.

The differential effects of trade may beexplained, in part, by regional variationin economic development, social structures,and physical geography. Historical factorshave contributed to spatial disparities withinMozambique. For example, colonial policieschanneled development into the south of thecountry because, in part, of the region’s prox-imity to South Africa. As a result, the south gen-erally has higher levels of development, includ-ing better infrastructure such as roads, healthfacilities, and schools. Southern Mozambiquealso has a long tradition of male labor migra-tion to South Africa, which gives some house-holds access to remittance income. Householdsoften use this income to purchase farming in-puts and increase their agricultural productivity(Hermele 1992). Although Mozambican tradepolicy promotes export crop production inboth areas of the country, liberalization reformshave taken place unevenly. Private investmenthas largely happened in the southernmostprovinces of the country. The effects of trade oninequality may also vary regionally accordingto gender roles. Widowed or divorced womenin the north may be able to continue commer-cial agricultural production (because they retainaccess to their land), leading to less inequalitybetween them and their married counterparts(Waterhouse and Vijfhuizen 2001). However,women in southern Mozambique often lose ac-cess to agricultural land after being widowed ordivorced. Physical geography also contributesto regional differences. Due to climate and soilconditions, the north of the country has muchhigher agricultural potential than the south.However, the Zambezi River splits the countryphysically and there is no central bridge withinMozambique. Thus the Zambezi River acts as

a barrier to regional economic integration byinterrupting transportation routes and isolatesthe north from the more globally linked south-ern region of the country.

The inequality-dampening effects of cashcropping in the north were unexpected giventhat southern farmers—who generally havebetter access to markets—described the activityas risky and often unprofitable. The inequality-enhancing effects of vegetable trade in thesouth also run counter to expectations, becausefarmers described vegetable cropping as a rela-tively accessible activity for poorer householdsand a way for these households to raise theirincomes. To explore these curious results, thenext phase of the study investigated the typesof households that participated in vegetable andcash cropping, and considered what other fac-tors could be complicating trade’s effects oninequality.

Household DynamicsBecause vegetable cropping orientation wassignificant only in the southern model, and cashcropping orientation was significant only in thenorthern model, findings were inconclusiveas to whether the varying relationships be-tween different types of agricultural trade andinequality have regional explanations. An alter-native possibility is that economic and demo-graphic differences between households withinregions influence which types of crops aregrown, and that this, in turn, impacts inequal-ity. Further qualitative work was conducted toaddress the second research question and inves-tigate what factors influence household partici-pation in different agricultural activities. Theseinterviews also provided the opportunity toshare results from the earlier phases of the anal-ysis and get local interpretations of quantitativefindings from farmers involved in the study.

In interviews discussing the regional regres-sion analysis, southern farmers supported theinterpretation that vegetable trade increased in-equality, noting that the scale at which farmerscould engage in the activity was determined bytheir access to resources. For example, house-holds that had access to irrigated land couldproduce more vegetables and become wealth-ier than households that relied on rain-fed agri-culture. The demand for irrigated land far ex-ceeds availability, even in Massavasse, which islocated near a large irrigation scheme. Thus


historic rights to irrigated land (or the abil-ity to rent it) coupled with opportunities toengage in the vegetable trade appear to havethe tendency to increase and reinforce inequal-ity. Southern producers also speculated that theinequality-dampening effects of cash crop tradein the north could arise from the low marketprices for cash crops. Farmers believed that cashcropping would not increase inequality if thosewho produced the crops did not make a profit.Most farmers’ explanations thus did not focuson north–south differences, but on the differentreturns for vegetable and cash crops.

Key informant interviews with governmentofficials and rural development specialists sup-ported the view that farming in the southrather than the north did not bring commer-cial benefits, such as greater profits, to mostfarmers. For most southern farmers, marketaccess and prices were much the same as fornorthern farmers. The interviews revealed, forinstance, that government heavily focuses itsefforts to improve market access on a few ar-eas in the country that are seen as having themost agricultural potential. Although the southreceives more government and private invest-ment, development is spread very unevenlywithin this region. Therefore most southernfarming households face the same challenges tomaking a profit as do northern ones (e.g., poortransportation structure, limited access to in-puts and services). Rural producers across thecountry are largely left on their own when itcomes to commercializing and developing link-ages to export markets.

Findings from the key informant interviewssuggest that the differential effects of trademay be due to differences between householdswithin each region. Evidence from furtherquantitative analysis using the TIA survey ofrural households supports these qualitativefindings. Results of bivariate tabular analysesindicate that there were many significantdifferences between the northern and southernhouseholds with regard to participation incommercial agriculture, physical conditions,access to services, and demographic charac-teristics (Table 3). However, chi-square testsof independence found that, in general, thecharacteristics of cash and vegetable croppinghouseholds did not differ by region.14 A fewnotable differences include (1) southern veg-etable cropping households were more likely to

have access to off-farm wage income than theirnorthern counterparts, χ2(1, N = 758) = 12.07,p > 0.01; (2) female-headed households weremore likely to participate in vegetable croppingin the south than in the north, χ2(2, N = 758)=10.57, p > 0.01; and (3) a minority languagewas more likely to be spoken in a northern veg-etable cropping household than in a southernone, χ2(1, N = 758) = 59.83, p > 0.01.

Despite these differences, findings suggestthat northern and southern vegetable farm-ers form a fairly homogeneous group with re-gard to literacy levels, disease burdens, accessto services, and proximity to markets. Theresults from difference of means tests lendfurther support to the hypothesis of the ho-mogeneity of the group of small-scale com-mercial agriculturalists, as these tests indicatethat the mean per capita consumption of veg-etable farmers and cash crop producers, respec-tively, did not significantly differ across regions(Table 4).15 However, some economic differen-tiation existed within the same region betweencommercial agriculturalists and other (non-participating) households (Table 5). North-ern cash-cropping households had higher meanper capita consumption than their non-cash-cropping neighbors. In contrast, the mean percapita consumption for southern cash-croppinghouseholds was not significantly different fromthat of non-cash-cropping households. A sim-ilar set of relationships hold for vegetable-cropping households.

Lived Experiences of InequalityGiven the intraregional differences betweenhouseholds that participate in agricultural tradeand those that do not, the final phase of theanalysis addressed the third research ques-tion, examining how crop-trading householdsfit into their local social systems, and how thisaffects trade’s relationship to inequality. Thisphase of the research involved a third round ofinterviews with southern households as well asinterviews in a new case study site, Rapale, innorthern Mozambique. Interviews with north-ern and southern farmers centered on (1) in-centives and disincentives for participating inagricultural trade, and (2) the status of com-mercial agriculturalists in their communities.The findings from these interviews suggest thatthe context of agriculture trade—and the social


Table 3 Descriptive characteristics of northern and southern rural households

North (N = 2,692) South (N = 1,913)

% of Northern % of SouthernCategory Variable description households N households N

Commercial agriculture Sells cash cropsa 10% 269 2% 31Sells vegetable cropsa 19% 511 13% 247

Labor market participation Has off-farm wage incomea 13% 353 23% 437Demographics Male-headed householdsa 78% 2,106 74% 1,417

Female-headed household (notwidowed/not divorced)a

18% 476 20% 386

Female-headed households(widowed/divorced)a

4% 110 6% 110

Speaks minority languagea 47% 1,270 26% 492Household head is literatea 2% 58 6% 112

Health Prolonged illness in household(>2 years)a

5% 135 6% 124

Death in householda 7% 196 10% 200Physical infrastructure Market in villageb 28% 750 28% 530

Accessible by paved roadsa 14% 366 23% 438Access to services Access to agricultural

extension agentb15% 410 14% 264

Member of agriculturalassociationb

4% 111 4% 84

Receives price informationa 36% 982 24% 453Holds title to landa 9% 235 21% 409Has irrigated landa 1% 30 3% 51

Climate and location Experienced droughta 30% 799 54% 1,039Experienced flood

(2002–2002)a27% 723 51% 975

Located in coastal areaa 15% 414 9% 180

aChi-square test for independence indicates the relationship between this variable and geographic region is statisticallysignificant at the 0.05 level.bChi-square test for independence not statistically significant at the 0.05 level.Source: Author’s calculations using data from Trabalho de Inquerito Agricola 2002 (MADER 2002).

status of traders—varies greatly between thetwo regions.

Findings from the local analysis suggest thatmaterial or monetary inequality exists and isnoticed, even in the poorer and more isolated

regions of the country. However, not all de-scriptions of inequality related to income andassets. Custom and tradition were strong dis-incentives to engaging in commercial agricul-ture (of both vegetables and export cash crops).

Table 4 Comparison of daily per capita consumption between regions

North South t tests

Variable category M SD M SD t value p

Cash croppingDaily per capita consumption of participating

households6,625 2,3181 4,480 7,686 0.51 0.61

Daily per capita consumption of nonparticipatinghouseholds

3,175 14,492 5,340 39,906 2.24 0.03

Vegetable cropping

Daily per capita consumption of participatinghouseholds

5,901 29,895 5,990 9,519 0.05 0.96

Daily per capita consumption of nonparticipatinghouseholds

2,978 9,567 5,228 42,269 2.13 0.03

Note: Values given in Mozambican Metacais.Source: Author’s calculations using data from Trabalho de Inquerito Agricola 2002 (MADER 2002).


Table 5 Comparison of daily per capita consumption within regions

Participating Nonparticipatinghouseholds households t tests

Variable description M SD M SD t value p

Cash croppingSouthern per capita consumption 4,480 7,686 5,340 39,906 0.52 0.61Northern per capita consumption 6,625 23,181 3,175 14,492 −2.43 0.02

Vegetable CroppingSouthern per capita consumption 5,990 9,519 5,228 42,268 −0.64 0.53Northern per capita consumption 5,901 29,895 2,978 9,567 −2.18 <0.001

Note: Values given in Mozambican metacais.Source: Author’s calculations using data from Trabalho de Inquerito Agricola 2002 (MADER 2002).

Subsistence farming—regardless of whether itwas a successful enterprise—warranted a greatdeal of respect in all case study sites, and peo-ple said that subsistence farmers had higher so-cial standing than people who did not farm.Village leaders, teachers, and nurses were alsoconsidered to be in positions of authority andwere considered to be better off than other vil-lagers, even if their incomes were not signifi-cantly higher than those of other communitymembers.16

In both northern and southern Mozam-bique, strong preferences for subsistence farm-ing and self-sufficiency often led people to favorsubsistence agriculture even when other eco-nomic opportunities were available. Intervie-wees in southern Mozambique noted that par-ticipating in small-scale commercial activitiesdid not necessarily improve their food secu-rity, but forced them to specialize and lessenhousehold self-sufficiency with regard to farm-ing. For example, southern survey respondentsreported that profits from participating in com-mercial activities were often negligible. More-over, commercial farming had high social costs.Interviewees gave the highest social esteem tothose who met all of their food needs from theirown farms. Households lose respect and socialstanding in the community as they move awayfrom growing their own food toward purchas-ing food grown by others.

Northern households appeared to have morefinancial and social incentives to participate inagricultural trade, particularly cash cropping.Because of the abundance of fertile agricul-tural land, northern households said they couldmaintain self-sufficiency via subsistence farm-ing and also participate in commercial agricul-

ture. In the north, cash cropping was seen as ahigh-income occupation, providing householdswith a strong incentive to participate if theycould. Vegetable cropping also appeared to bepopular in areas close to the larger cities inthe north. In the south, however, several otherincome-earning activities seemed to be avail-able. Southern interviewees were more likelythan their northern counterparts to unfavorablycompare commercial agricultural productionwith forms of formal employment. Intervie-wees in the south noted that people with formalemployment earned enough to maintain fam-ily farms using employed farm hands. The for-mally employed were thus able to retain respectin the community by remaining self-sufficientwith regard to household food production.However, small-scale, informal trading tooktime away from subsistence farming and typi-cally did not generate enough profit for tradersto hire replacement labor to work family farms.

The results also suggest that the social costsof market participation—including vegetablesales—are particularly high for women. Sev-eral female respondents in the south said thatselling vegetables in the market diminishedtheir social standing because it is an undig-nified activity. Informal market vendors, whoin southern Mozambique are overwhelminglywomen, reported being treated disrespectfullyby clients. Gender discrimination may play arole in the difficulties faced by female traders.The fact that both formal and informal marketsare generally unpoliced spaces may contributeto this phenomenon. Another possible expla-nation for this poor treatment is that vendorsengaged in the direct trade of services or prod-ucts for cash are operating outside traditionally


governed social structures (and thus breakinggender norms). For example, before the growthof the cash economy, goods and services wereexchanged according to the principal of ajudamutuo, or mutual help, where community mem-bers received goods in kind for help providedduring the harvest season. Petty trading andstreet vending have to a considerable extent re-placed these forms of exchange, and are muchless regulated by tradition.

Although most of the women in the southerncase study sites stated that they lost respect intheir community when they marketed agricul-tural produce (mostly vegetables), few womenin the north mentioned similar experiences.This was most likely related to the role womenplayed in commercial activities. Women in thesouthern case study communities play a muchmore visible role in the market economy thando women in the northern site. In the southernsites, vegetable sellers tend to be women in boththe formal and informal markets. In markedcontrast, few women participate in markets inthe northern case study site. When asked, malefarmers in the north stated that their wivesnever took produce to the markets. Moreover,women do not tend to participate in off-farmactivities. Gendered divisions of labor and so-cial norms appear to prevent them from engag-ing in these activities.

Conclusion

Using a sequential, mixed-methods approach,this analysis has looked closely at the effectsof agricultural trade on inequality in Mozam-bique. Results from this study suggest a com-plex relationship between agricultural trade andinequality in rural regions. Qualitative find-ings suggest that the commercial activities thatpeople are often able and inclined to do dif-fer from those promoted by the government’sexport-led development strategy. For example,southern farmers prefer small-scale vegetablefarming to cash cropping. Quantitative find-ings at the regional level indicate that trade hasdifferential effects on inequality. Regression re-sults suggest that vegetable crop orientation isassociated with higher inequality in the south,whereas cash cropping orientation is associatedwith lower inequality in the north. Later re-search phases investigated the reasons for thedifferential effects of trade, both across and

within Mozambican regions. Findings suggestthat anticipated profitability, not regional loca-tion, plays a major role in determining whichcrops households chose to produce. Anotherkey finding of the investigation is that tradeappears to increase inequality in areas wheretrading has a low social status, but does not in-crease inequality in areas where it has a highsocial status. This is the case even when partic-ipation in high-status trade increases mean percapita consumption (e.g., cash cropping in thenorth) or when participation in low-status tradedoes not (e.g., vegetable cropping in the south).

The results of this study point to the diffi-culties in using either neoclassical trade the-ory or new economic geography to predictthe effects of trade on inequality in rural re-gions of less developed countries. The pre-dictions of neoclassical trade focus on eco-nomic mechanisms—such as wages—as driversof trade-induced inequality, which may be un-able to explain the dynamics in Mozambiquewhere employment, formal or otherwise, isscarce (Wood 1994). The predictive powers ofnew economic geography in the Mozambicancontext may also be limited given that it bestfits places where economic sectors are relativelyadvanced (Krugman 1991). This is not the casefor the least developed countries, and it maytake decades before the regional dynamics ofthe Mozambican economy can be explained byeconomic trade theories developed for indus-trial and postindustrial nations. New economicgeography does, however, account for the factthat initial advantages tend to be reinforcing.This may be a critical factor in explaining in-equality in places where colonialism, corrup-tion, civil conflict, and ethnic strife have playeda strong role in determining which regions re-ceive initial investments.

Taken together, the study’s findings implythat social and geographical contexts greatlyinfluence the relationship between agriculturaltrade and regional inequality. For example,case study interviews revealed that peopleoften felt that commercial activities—such asselling vegetables in markets and other typesof informal sector work—damaged their socialstanding in the community, even when theirincomes were improved or unchanged. Thefindings of this study support the suggestionsof Sen (1992, 1997, 1999), Chakravorty(2005), and others that understanding income


inequality requires knowledge of the socialstructures and processes of change in particularplaces. Discovering the mechanisms by whichthese social drivers of inequality work meritsfurther attention. A better understanding ofthe social determinants of inequality wouldgive the Mozambican government greaterability to combat inequality via social policywithout drawing back from those export-basedstrategies that succeed in reducing absolutepoverty. �

Notes

1As this study focuses on agricultural production, ur-ban areas and households were excluded from theanalysis.

2For a series of helpful discussions on the use of mixedmethods in geography, see Rocheleau (1995) andthe special 1999 issue of The Professional Geographer(volume 51, issue 1) on multimethod research inpopulation geography.

3A more extensive discussion of Mozambican cropsis contained in Bias and Donovan (2003).

4For a more nuanced description of structural ad-justment programs and their impacts in Africa, seeMohan et al. (2000) and Mittleman (2000).

5Although a detailed discussion of the roles thatMozambique’s colonial history and civil war playedin shaping inequality are beyond the scope ofthis article, interested readers should see Sheldon(2002), Mittleman (2000), Bowen (2000), Isaacman(1996), Abrahamsson and Nilsson (1995), andUrdang (1989).

6In my use of the phrase social capital, I am drawingon the work of authors like Brouwer and Nhassengo(2006), Schuurman (2003), Sen (1999), and Rose(1998), who approach social capital (i.e., access toand strength of social relations) at the individuallevel to link the concept to uneven access to oppor-tunities and exposure to risks.

7Interviews conducted in phases three and five of thestudy used a similar procedure and the data wereanalyzed in an identical fashion.

8Before estimating the regression model, I performedseveral specification tests to determine the bestmodel. The results of spatial autocorrelation testscalculated for each model grouped determined ifthe regression was estimated via OLS or as a SEM.The SEM is a spatially autoregressive model used tocontrol for spatial autocorrelation in the residualsand thus the error term is assumed to be made upof a vector of spatially lagged errors, an autoregres-sive coefficient, and a vector of independent randomerrors. See Silva (2007) for a more detailed descrip-tion of the regression model and results. The results

from all spatial diagnostic tests are available fromthe author on request.

9There are other ways in which one could conductregional analyses, such as multilevel analyses.However, standard multilevel models do not easilyallow for controlling spatial dependency. Some re-cent work has started to develop techniques forcombining autoregressive models and multilevelmodels within ecology (Wu and David 2002; Thog-martin, Sauer, and Knutson 2004) and regional sci-ence (Parent and Riou 2005). The use of spatialhierarchical analyses remains an important futureendeavor to see what light they can shed on tradeand inequality linkages.

10Cash crops are comprised of cotton, sunflower,sugar, tobacco, soybeans, sisal, tea, rice, and ginger.Vegetable crops are horticultural crops, includingmelons and beans.

11Because of the 2004 Mozambican presidential elec-tion, I was unable to access village registries andselect a random sample of northern households.Therefore a snowball sampling technique was usedto select households for interviews.

12The Chow test result, given by the F statistic, wassignificant at the 95 percent confidence level. Theresults are available from the author on request.

13Silva (2007) exclusively discusses, in greater detail,the modeling of income inequality in Mozambiqueas a function of trade orientation and other factorsusing a regression analysis. This article presents amore holistic account of the relationship betweentrade and inequality by drawing on both qualitativeand quantitative methodologies, and synthesizingthe findings from both approaches.

14Chi-square results tables are available from the au-thor on request.

15The use of consumption values as a proxy for in-come is common in studies of developing coun-tries where subsistence activities are widely preva-lent and income can be difficult to quantify.

16In both Massavasse and Matidze, interviewees de-fined “better off” as being more socially respectedand having an easier life than others in the commu-nity.

Literature Cited

Abrahamsson, H., and A. Nilsson. 1995. Mozambique:The troubled transition. London: Zed Books.

Aguayo-Tellez, A. 2006. Income divergence betweenMexican states in the 1990s: The role of skill pre-mium. Growth and Change 34 (2): 255–77.

Anderson, E. 2005. Openness and inequality in de-veloping countries: A review of theory and recentevidence. World Development 33 (7): 1045–63.

Arnaldo, C. 2004. Ethnicity and marriage patternsin Mozambique. African Population Studies 19 (1):143–64.


Baden, S. 1997. Post-conflict in Mozambique—Women’s special situation, population issues, andgender perspectives—To be integrated into skillstraining and employment promotion. Geneva: Inter-national Labor Office. http://www.bridge.ids.ac.uk/Reports/re44c.pdf (last accessed 23 January2008)

Bias, C., and C. Donovan. 2003. Gaps and op-portunities for agricultural sector developmentin Mozambique. Ministry of Agriculture andRural Development, Directorate of EconomicsResearch Paper Series, Research Report 54E.Maputo, Mozambique: Ministry of Agricultureand Rural Development.

Bowen, M. L. 2000. The state against the peasantry.Charlottesville: University Press of Virginia.

Brouwer, R., and J. Nhassengo. 2006. Aboutbridges and bonds: Community responses to the2000 floods in Mabalane District, Mozambique.Disasters 30 (2): 234–55.

Carvalho, S., and H. White. 1997. Combining thequantitative and qualitative approaches to povertymeasurement and analysis: The practice and thepotential. World Bank Technical Paper No. 366.Washington, DC: World Bank.

Chakravorty, S. 2005. Fragments of inequality:Social, spatial, and evolutionary analyses of incomedistribution. New York: Taylor & Francis.

Courgeau, D., ed. 2003. Methodology and epistemologyof multilevel analysis: Approaches from differentsocial sciences. Dordrecht, Netherlands: KluwerAcademic.

Dicken, P. 1998. Global shift. 3rd ed. New York:Guilford.

DINAGECA (Direccao Nacional de Geografia eCadastro). 1999. Base topografica simplificada euso e coberatura da terra (Base topographic andland use/land cover data). Maputo, Mozambique:DINAGECA.

Edin, P., P. Fredriksson, and P. Lundborg. 2004.The effect of trade on earnings: Evidence fromSwedish microdata. Oxford Economic Papers 56 (2):231–41.

Elbers, C., P. Lanjouw, J. Mistiaen, B. Ozler,and K. Simler. 2002. Are neighbours equal?Estimating local inequality in three developing coun-tries. IFPRI FCND Discussion Paper No. 147.http://www.ifpri.org/divs/fcnd/dp/papers/fcndp147.pdf (last accessed 23 January 2008).

Evans, D. 2001. Identifying winners and losers insouthern Africa from globalization: Integratingfindings from GTAP and poverty case studies onglobal trade policy reform. Paper prepared for theFourth Annual Conference on Global EconomicAnalysis, Purdue University, West Lafayette, IN.

Galiani, S., and P. Sanguinetti. 2003. The impact oftrade liberalization on wage inequality: Evidencefrom Argentina. Journal of Development Economics72 (2): 497–513.

Government of Mozambique. 2001. Action plan for thereduction of absolute poverty (2001–2005). Maputo,Mozambique: Government of Mozambique.

Heltberg, R., K. Simler, and F. Tarp. 2001. Publicspending and poverty in Mozambique. UnitedNations University/World Institute for Develop-ment Economics Research Discussion Paper No.2001/63, UNU-WIDER, Helsinki, Finland.

Hermele, K. 1988. Land struggles and social differ-entiation in southern Mozambique: A cast study ofChokwe, Limpopo 1950–1987. Research ReportNo. 82, Scandinavian Institute of African Studies,Uppsala, Sweden.

———. 1992. Stick and carrot: Political alliances andnascent capitalism in Mozambique. In Authoritar-ianism, democracy and adjustment, ed. P. Gibbon,Y. Bangura, and A. Ofstad, 169–86. Uppsala,Sweden: Scandinavian Institute of African Studies.

INE (Instituto Nacional de Estatıstica). 1998.Inquerito nacional aos agregados familiares sobre ascondicoes de vida, 1996–1997 (1996–1997 nationalhousehold survey on living conditions). Maputo,Mozambique: INE.

———. 1999. Recenseamento geral da populacaoe habitacao 1997 (1997 national housing andpopulation census). Maputo, Mozambique: INE.

INE/MADER (Instituto Nacional de Estatıstica &Ministerio da Agricultura e Desenvolvimento).2001. Censo agro-pecuario 1999–2000 (1999–2000census of agriculture and livestock). Maputo,Mozambique: INE/MADER.

Isaacman, A. 1996. Cotton is the mother of poverty: Peas-ants, work, and rural struggle in colonial Mozambique,1938–1961. Portsmouth, NH: Heinemann.

Isaacman, A., and B. Isaacman. 1983 Mozambique:From colonialism to revolution, 1900–1982. Boulder,CO: Westview.

Kapstein, E. B. 2000. Winners and losers in theglobal economy. International Organization 54 (2):359–84.

Knauder, S. 2000. Globalization, urban processes,urban problems, rural disadvantages: Evidence fromMozambique. Burlington, VT: Ashgate.

Krugman, P. 1991. Geography and trade. Cambridge,MA: MIT Press.

Leichenko, R. M., and K. L. O’Brien. 2002. Thedynamics of rural vulnerability to global change:The case of southern Africa. Mitigation andAdaptation Strategies for Global Change 7: 1–18.

———. 2008. Environmental change and globalization:Double exposures. New York: Oxford UniversityPress.

MADER (Ministerio da Agricultura e Desenvolvi-mento Rural). 2002. Trabalho de inquerito agrıcola2002 (National agriculture survey 2002). Maputo,Mozambique: MADER.

Mittleman, J. H. 2000. The globalization syndrome:Transformation and resistance. Princeton, NJ:Princeton University Press.


Mohan, G., E. Brown, B. Milward, and A. B.Zack-Williams, eds. 2000. Structural adjustment:Theory, practice and impacts. London: Routledge.

Nathan Associates. 2002. Mainstreaming trade: Apoverty reductions strategy for Mozambique. Maputo,Mozambique: Nathan Associates.

O’Brien, K. L., and R. M. Leichenko. 2000. Dou-ble exposure: Assessing the impacts of climatechange within the context of economic global-ization. Global Environmental Change 10 (3): 221–32.

———. 2003. Winners and losers in the context ofglobal change. Annals of the Association of AmericanGeographers 93 (1): 89–103.

O’Brien, K., R. Leichenko, U. Kelkar, H. Venema,G. Aandahl, H. Tompkins, A. Javed, S. Bhadwal,S. Barg, L. Nygaard, and J. West. 2004. Mappingvulnerability to multiple stressors: Climate changeand globalization in India. Global EnvironmentalChange 14: 303–13.

Parent, O., and S. Riou. 2005. Bayesian analysis ofknowledge spillovers in European regions. Journalof Regional Science 45 (4): 747–75.

Pitcher, M. A. 2000. Transforming Mozambique: Thepolitics of privatization, 1975--2000. Cambridge,U.K.: Cambridge University Press.

Richards, L., and T. Richards. 1997. NUD∗IST,Version 4. Melbourne, Australia: QualitativeSolution & Research Pty. Ltd.

Rocheleau, D. 1995. Maps, numbers, text, and con-text: Mixing methods in feminist political ecology.The Professional Geographer 47 (4): 458–66.

Rodrik, D. 1997. Has globalization gone too far?Washington, DC: Institute for InternationalEconomics.

Roesch, O. 1988. Rural Mozambique since theFrelimo Party Fourth Congress: The situationin the Baixo Limpopo. Review of African PoliticalEconomy 15 (41): 73–91.

Rose, R. 1998. Getting things done in an anti-modern society: Social capital and networks inRussia. Socal Capital Initiative Working Paper 6,World Bank, Washington, DC.

Sanchez-Reaza, J., and A. Rodrıguez-Pose. 2002.The impact of trade liberalization on regional dis-parities in Mexico. Growth and Change 33 (1): 72–90.

Schuurman, F. J. 2003. Social capital: The politico-emancipatory potential of a disputed concept.Third World Quarterly 24 (6): 991–1010.

Sen, A. 1981. Poverty and famines: An essay on entitle-ment and deprivation. Oxford, U.K.: Clarendon.

———. 1992. Inequality reexamined. Cambridge,MA: Harvard University Press.

———. 1997. On economic inequality. 2nd ed. Oxford,U.K.: Clarendon.

———. 1999. Development as freedom. New York:Anchor.

Sheldon, K. E. 2002. Pounders of grain: A his-tory of women, work, and politics in Mozambique.Portsmouth, NH: Heinemann.

Silva, J. A. 2007. Trade and income inequality in aless developed country: The case of Mozambique.Economic Geography 83 (2): 111–36.

Silva, J. A., and L. M. Leichenko. 2004. Regional in-come inequality and international trade. EconomicGeography 80 (3): 261–85.

Simler, K., and V. Nhate. 2002. Poverty, inequalityand geographic targeting: Evidence from small-area estimates in Mozambique. Paper preparedfor the conference on Understanding Povertyand Growth in Sub-Saharan Africa, Center forthe Study of African Economies, St. Catherine’sCollege, University of Oxford, Oxford, U.K.

Tashakkori, A., and C. Teddlie. 1998. Mixedmethodology: Combining qualitative and quantitativeapproaches. Thousand Oaks, CA: Sage.

Thogmartin, W. E., J. R. Sauer, and M. G. Knut-son. 2004. A hierarchical spatial model of avianabundance with application to cerulean warblers.Ecological Applications 14 (6): 1766–79.

Urdang. S. 1989. And still they dance: Women, war,and the struggle for change in Mozambique. NewYork: Monthly Review Press.

Wade, R. 2001. Global inequality: Winners andlosers. The Economist, 28 April 2001, 72–74.

Walker, T., D. L. Tschirley, J. Low, M. PequeninoTanque, D. Boughton, E. Payongayong, andM. T. Weber. 2004. Determinants of rural income,poverty, and perceived well-being in Mozambique in2001–2002. Maputo: Ministry of Agriculture andRural Development, Economics Directorate.

Waterhouse, R., and C. Vijfhuizen, eds. 2001.Strategic women, gainful men: Gender, land, andnatural resources in different rural contexts in Mozam-bique. Maputo: Eduardo Mondlane University(UEM)/Action Aid.

Wobst, P. 2003. The impact of domestic and globaltrade liberalisation on five southern African coun-tries. Journal of Development Studies 40 (2): 70–92.

Wood, A. 1994. North-south trade, employment, andinequality: Changing fortunes in a skill-driven world.Oxford, U.K.: Clarendon.

Wu, D., and J. L. David. 2002. A spatially explicitapproach to modeling complex ecological systems:Theory and applications. Ecological Modelling135:7–26.

JULIE A. SILVA is an Assistant Professor in theDepartment of Geography and Center for AfricanStudies at the University of Florida, Gainesville, FL32611. E-mail: [email protected]. Her research in-terests include uneven development, internationaltrade, poverty, and the global criminal economy.

a multilevel analysis of agricultural trade and ... · in these communities have important...

Documents