Structural Studies Division RETURNS TO SOCIAL CAPI

RETURNSTOSOCIALCAPITALAMONGTRADERSbyMarcelFafchampsandBartMintenMSSDDISCUSSIONPAPERNO.23MarketsandStructuralStudiesDivisionInternationalFoodPolicyResearchInstitute2033KSt.N.W.Washington,D.C.20006U.S.A.July1998Contact:CarolynRoperTel:202/862-5600orFax:202/467-4439MSSDDiscussionPaperscontainpreliminarymaterialandresearchresults,andarecirculatedpriortoafullpeerreviewinordertostimulatediscussionandcriticalcomment.ItisexpectedthatmostDiscussionPaperswilleventuallybepublishedinsomeotherform,andthattheircontentmayalsoberevised.ContentsAbstract..................................................................1Section1.ConceptsandTestingStrategy.........................................5Section2.TheData.........................................................7Section3.ReturnstoSocialNetworkCapital.....................................11Section4.SocialCapitalandModesofTransaction................................16Conclusion...............................................................20References...............................................................22Tables...................................................................25TablesTable1.DependentVariablesandRegressorsTable2.ModesofTransactionTable3.EffectofSocialCapitalonTotalSalesTable4.EffectofSocialCapitalonGrossMarginsTable5.EffectofSocialCapitalandModeofTransactiononTotalSalesTable6.EffectofSocialCapitalandModeofTransactiononGrossMarginsTable7.DeterminantsofModesofTransactionTableA1.Instruments:FirstSetTableA2.Instruments:SecondSetReturnstoSocialCapitalAmongTradersMarcelFafchamps†andBartMinten††Abstract1UsingdataonagriculturaltradersinMadagascar,thispapershowsthatsocialcapi-talhasalargeeffectonefficiency.Betterconnectedtradersareshowntohavesignificantlylargersalesandgrossmarginsthanlessconnectedtradersaftercontrollingforphysicalandhumaninputs.Theanalysisindicatesthatthreedimensionsofsocialnet-workcapitalshouldbedistinguished:relationshipswithothertraders,whichhelpfirmseconomizeontransactionscosts;relationshipswithindividualswhocanhelpintimesoffinancialdifficulties,whichinsuretradersagainstliquidityrisk;andfamilyrelationships,whichreduceefficiency,possiblybecauseofmeasurementerror.Socialnetworkcapitalenablestraderstodealwitheachotherinamoretrustworthymannerbygrantingandreceivingcredit,exchangingpriceinformation,andeconomizingonqualityinspection.Schoolingiscorrelatedwiththeuseofsuperiormodesoftransactionbutneedstobecomplementedbysocialnetworkcapital._______________†DepartmentofEconomics,StanfordUniversity,Stanford,CA94305-6072.Tel.:(650)[e-mail].††InternationalFoodPolicyResearchInstitute,2033KStreet,NW,WashingtonDC20006.Tel.:(202)[e-mail].1WebenefittedfromconversationsandcommentsfromJeanClaudeRandrianarisoa,ElianeRalison,ManfredZeller,GauravDatt,andfromseminarparticipantsatIFPRI.SpecialthanksgotoOusmaneBadianeforobtainingfinancialsupportforthisresearch.WeacknowledgefinancialsupportfromtheUnitedStatesAgencyforInternationalDevelopment.2Socialscienceshavelongrecognizedtherolethatsocialcapitalplayinfacilitatinghumaninteraction(e.g.,Coleman(1988),Putnam,LeonardiandNanetti(1993),Grano-vetter(1985)).Unlikehumancapital,however,whichnowisseenasafundamentaldimensionofmosteconomicprocesses,theconceptofsocialcapitalhasyetbeenlittleusedineconomics(e.g.,NarayanandPritchett(1996),Barr(1997),Fafchamps(1998),LundandFafchamps(1997))andisstillregardedwithsuspicionbymany.Thispapercontributestothedebatebyprovidingevidencethatsocialcapitalhasalargesignificanteffectontheperformanceofeconomicagentsbeyondthoseofphysicalandhumancapi-tal.Wedemonstratethatcertaintypesofsocialnetworksaremorevaluablethanothersandwethrowsomemuchneededlightonthepossiblechannelsthroughwhichsocialcapitalaffectseconomicefficiency.Oneofthereasonswhyeconomistsarewearyofusingthetermsocialcapitalisthatitsmeaningisimprecise.Fromaneconomist’spointofview,thereareatleasttwomean-ingsofthephrasethatmustbeclearlydistinguished.Thefirstmeaningseessocialcapitalasa’stock’oftrustandanemotionalattachmenttoagrouporsocietyatlargethatfacili-tatetheprovisionofpublicgoods.ExamplesofthisdefinitionofsocialcapitalcanbefoundintheworksofColeman(1988)andPutnam,LeonardiandNanetti(1993):Cole-man(1988)arguesthatkidsperformbetterinschoolwhenparentsgetinvolvedrunningtheschool;Putnam,LeonardiandNanetti(1993)arguethathistoricaldifferencesinlev-elsoftrustbetweenindividualsaccountforthedivergingeconomicexperiencesofnorth-ernandsouthernItalybecauseitaffectedfirms’abilitytocontractwitheachother.Greif(1994)makesarelatedpointwithrespecttomedievaltradersonbothsidesoftheMedi-teranean.FurtherexamplescanbefoundintheworksofPlatteau(1994),Gambetta(1988),Fukuyama(1995),andothers.3Asecondmeaningseessocialcapitalasanindividualassetthatbenefitsasingleindividualorfirm;thismeaningissometimesreferredtoassocialnetworkcapitaltoemphasizethatagentsderivebenefitsfromknowingotherswithwhomtheyformnet-worksofinterconnectedagents.Laboreconomistsandsociologists,forinstance,havelongrecognizedthatknowingpotentialemployershelpspeoplefindajoband