Education and allocative efficiency household income growth during rural reforms in China

www.elsevier.com/locate/econbase

Educationandallocativeefficiency:householdincomegrowthduringruralreformsinChina

DennisTaoYang*

DepartmentofEconomics,VirginiaPolytechnicInstituteandStateUniversity,3033PamplinHall,

Blacksburg,VA24061,USA

Abstract

ThispaperstudiesthecontributionofschoolingtoruralincomeinChinaduringfactormarketliberalizationbetween1986and1995.Therelaxationofcontrolspermittedfarmhouseholdstoreallocateproductiveinputsfromagriculturetononagriculturalactivities.Itishypothesizedthateducationfacilitatesthisadjustment.PaneldatafromSichuanprovincesuggestthatschoolingenhancedtheabilityoffarmerstodevotelaborandcapitaltononfarmproductiongiventheevidencethatless-than-optimumlevelsoftheseinputswereallocatedtononagriculturaluses.Duringthetransition,theexpansionofnonfarmactivitiescontributedsignificantlytohouseholdincomegrowth.D2004ElsevierB.V.Allrightsreserved.

JELclassification:J43;O15;O12

Keywords:Education;Marketliberalization;Inputallocation;Ruralincome;China

1.Introduction

PolicyreformsinChinahaverevitalizedtheruraleconomy.Sincetheinceptionofreforms,realruralpercapitaincomehasincreasedmorethanfourfold;earningsrosesharplybetween1978and1985,followedbyaperiodofcontinuedgrowth(SSBa,1998).Severalfactorshavecontributedtothisremarkableperformance.Theadoptionofthehouseholdresponsibilitysystem(HRS)andincreasesinstateprocurementpriceswereidentifiedasthemajorsourcesofincomegrowthpriorto1985,creatingaprofoundone-timeeffectonearningsthroughincreasedlaboreffortandpriceincentives(McMillanand

*Tel.:+1-540-231-7474;fax:+1-540-231-5097.E-mailaddress:deyang@vt.edu(D.TaoYang).

0304-3878/$-seefrontmatterD2004ElsevierB.V.Allrightsreserved.doi:10.1016/j.jdeveco.2003.12.007

138D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Zhu,1988;Lin,1992).Agriculturalresearchandtechnologicalchangearealsofoundtohavesignificantlyraisedcropyields(HuangandRozelle,1996;FanandPardey,1997).Whiletheircoverageextendstothecollectivizationperiod,thesestudiesareprimarilyconcernedwithproductivitygainswithinagriculture,especiallyduringtheearlyperiodofeconomicreforms.

Themainpurposeofthispaperistoexaminethesourcesofsustainedincomegrowthbetween1986and1995inabroadercontextoftheruraleconomythatincludesnonagriculturaldevelopment.Thefocusisonfarmers’responsestofactormarketliberalizationastheyexpandednonfarmproductionandontheroleoftheireducationinfacilitatingresourceallocationdecisions.Startingin1983,thegovernmentannouncedaseriesofpoliciesthatloosenedrestrictionsonlabormobilityoutofagriculture.Theregulatorychanges,includingpermissionforlong-distancetransport,marketingofcommodities,andemploymentinsmalltowns,encouragedfarmerstoestablishnonfarmbusinessesandseekoff-farmjobswithbetterpay.Atthesametime,farmhouseholdsalsodivertedfundsandcapitalequipmenttoindustrialandserviceactivitiesforhigherreturns.Duringthis10-yearperiod,thepercentageofrurallaborforceemployedintownshipandvillageenterprises(TVEs)increasedfrom12.8%to22.2%(SSBb,1996).In1986,thegrossoutputvalueofTVEswasabout88%ofthegrossvalueofagriculture,butin1995,theformerwasmorethanthreetimesthelatter.TheseindustrialdevelopmentsareanalyzedasamajorforcebehindrapidincomegrowthinruralChina.

Thehumancapitalapproachtoproductionefficiencyhaslongpostulatedthateducationmayenhancetheabilityoffarmerstoperceiveandinterpretmarketinformationsothattheycanbetterrespondtoeconomicdisequilibria(Schultz,1975).Inparticular,investmentineducationmayimprovefarmallocativedecisionsaswellasworkers’productionskills(Welch,1970).Returnstoeducationarehighwhenproductivelearningopportunitiescanbeexploited;theseopportunitiesareoftenassociatedwithtechnicalinnovationorchangesinthemarketandpoliticalregimes(Rosenzweig,1995).Whiletheseviewsaresupportedbyalargebodyofempiricalliteraturethatfindseducationtohavepositiveeffectsinmodern,dynamicenvironments,1fewattemptshavebeenundertakentoassessreturnsfromschoolingduringthetransitionofmarketandpoliticalsystems.2Hence,therecentpolicyreformsinChinapresentanunusualopportunityforexaminingtheroleofeducationinproductionduringthetransitionfromaplanningtoamarketeconomy.

Forinstance,schoolingisfoundtohavehigherreturnsduringperiodsoftechnicalchangebecauseitfacilitatestheuseoffertilizerandothermoderninputs(Huffman,1977;JamisonandLau,1982).Inaddition,duringtheGreenRevolution,moreschooledfarmerswereabletoachievehigherprofitsbecauseofmoreeffectiveadoptionofhigh-yieldseedvarieties(e.g.,FosterandRosenzweig,1996)andofbetterexecutionofproduction(e.g.,PittandSumodiningrat,1991).Schoolingalsoenhancesinformationskillsoftheruralpopulation(e.g.,StraussandThomas,1995),whichinpartimprovetheiropportunitiestoparticipateinruraltourbanmigration(e.g.,Schultz,1988).

2OneexceptionisOrazemandVodopivec(1995)whofindalargeincreaseincompensationtothemoreeducatedinSloveniaasitmovedfromthecentralizedeconomytothemarketeconomy.AnotherisLiandZhang(1998)whofindthateducationwasnotrewardedduringthecommuneperiodinruralChina,whereas,underthepostreformhouseholdfarmingsystem,therewerepositivereturnstofarmers’education.SignificantreturnstoeducationarealsoreportedforChina’spaperindustryinthepostreformera(Fleisheretal.,1996).However,thesestudiesdonotinvestigatethemechanismsthroughwhicheducationrealizesitsreturns,thefocusofthispaper.

1D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162139

Toclarifytheproductivevalueofeducation,thispapersetsupasimplemodelofprofitmaximizationinwhichthefarmhouseholdengagesinagriculturalandnonagriculturalactivities.3Undercentralplanningthatemphasizedlocalgrainself-sufficiency,factorsofproductionweredevotedexcessivelytoagriculture,resultinginresourcemisallocationswithhigherinputreturnsinthenonagriculturalsector.Therefore,asrestrictionsonfactormobilitywererelaxedduringreforms,ruralfamiliesincreasinglyreallocatedinputstowardsnonfarmproduction.Thecentralhypothesisproposedinthepaperisthatbettereducatedfamilieswouldadjusttothesechangesmorequicklythanlesseducatedfamilies.Theempiricalanalysisofthispaperuseshousehold-levelpaneldatabetween1986and1995fromtheSichuanprovincetoanalyzehowschoolingaffectsintersectoralinputallocationsandhowfactorutilizationinturndeterminesfarmprofits.ThepaneldataareconstructedfromtheRuralHouseholdSurveycollectedbyChina’sStateStatisticalBureau.Therichstructureofthedataenablescontrolforhouseholdfixedeffects,regionandtimespecificfactors,andendogeneityassociatedwithidiosyncraticshockstoindividualhouseholds.Thesefactorsarenotproperlytreatedinpreviousstudiesontheselectionofruralincomeactivitiesduetolimitationsofcross-sectionaldata(TaylorandYunez-Naude,2000;YangandAn,2002).Theempiricalfindingsofthispaperindicatethat,duringtheperiodoftransition,less-than-optimumlevelsoflaborandcapitalwereallocatedtononagriculturaluses.Moreimportantly,thefindingssuggestthatschoolingplayedacriticalroleinallocatingmoreoftheseinputstothefactor-scarcesector.Thus,schoolingwasanimportantfactorbehindtherapidexpansionofruralindustries,whichwasamajorsourceofsustainedincomegrowth.

Therestofthepaperisorganizedasfollows.Section2providesanoverviewofpolicyreformsinruralChinaandespeciallyofregulatorychangesgoverningfactormobility.Section3presentsasimplehouseholdmodelinwhichfarmersmakeallocativedecisionsregardinginputsacrossagriculturalandnonagriculturalproduction.Themodelillustratesthedistortioninfactorutilizationduetopolicyinterventions,expectedresponsesoffarmerstomarketliberalization,aswellastheroleofeducationduringmarkettransition.Section4discussesthedataset,theeconometricspecifications,andreportstheestimationresults.Section5presentsconcludingremarks.

2.PolicyreformsinChina

Priortothestartofreformsin1978,massivedistortionsintheallocationofresourcesexistedinChina’scentrallyplannedsystem.Thecumulativeeffectsfrompursuingaheavy-industry-orienteddevelopmentstrategysincethe1950sresultedinexcessiveallocationofcapitalassetsinurbanareas,andahighpercentageofthelaborforcebeingconcentratedinthecountryside.4Withintheruralsector,thisnational

ThisframeworkiscloselyrelatedtoTaylorandYunez-Naude(2000)andYangandAn(2002)whoanalyzetheeffectofeducationonthechoiceofincomeactivities.

4Themainenforcementmechanismsincludethestatecontrolofagriculturalproductionandprocurement,andrestrictionsonrural-to-urbanmigrationviaahouseholdregistrationsystem.SeeYangandZhou(1999)fordiscussionsonresourceallocationacrossruralandurbansectors.

3140D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

policystressedagriculturalproductionandlocalgrainself-sufficiency,astrategyrigorouslypursuedbyprominentleaderseversincethetragicexperienceoftheGreatLeapfaminebetween1959and1961.Beforethereforms,ruralindustrialactivitiesconcentratedonanarrowrangeofproducts,emphasizing‘‘fivesmall’’industries:ironandsteel,cement,chemicalfertilizer,hydroelectricpower,andfarmimplements.Enterprisesinthecountrysidewerenotorientedtowardsmarketandconsumerproductsandremainedsubsidiarytoagriculture(Findlayetal.,1994;Naughton,1996).In1978,onlyabout7%oftherurallaborforcenationwidewasinnonagriculturalemployment,generatingapproximately7%ofruralhouseholdearnings(SSBa,1988),alevelfarbelowthatofothercomparabledevelopingcountries(AndersonandLeiserson,1980).Duetorestrictionsonnonfarmproduction,capitalandlaborwerescarceandtheirreturnswerehighinthatsector,creatingopportunitiesforrapidexpansionalongwithpolicyreforms.

Market-orienteddevelopmentinruralChinastartedwithapackageofthreereforms:thereplacementofproductionteamswithhouseholdsasunitsofbasicproduction(HRS),officialincreasesinagriculturalproductprices,andtheliberalizationofmarketsforruralproducts.Thesereformsprovidedthenecessaryconditionsfortheboominruralindustrialdevelopmentstartinginthemid-1980s.

Thechangefromcommunestoahousehold-basedfarmsystembeganin1979andwasessentiallycompletedbytheendof1983.Thisinstitutionalchangeinducedstrongfamilyworkeffort,thusreducingthedemandforworkersonsmallChinesefarms.Moreimportantly,thehouseholdresponsibilitysystemenabledindividualstohaveincreasedcommandovertheirproductiveresources.Duringthesameperiod,thegovernmentalsoimplementedreformsinproductionplanninginwhichthestatereducedthenumberofproductionplanningtargets(orcategories).Oftheremainingtargets,fewweremandatory,andmanywereguidedbycomplementarypricesandincentiveschemes(Sicular,1988).Therefore,farmersnotonlyhadincentives,butalsocertainfreedominrelocatinglaborandcapitaltononfarmuses.

In1979,thegovernmentalsoimplementedlargeincreasesinstateprocurementpricesforagriculturalproducts.Quotapricesforgrain,oilcrops,cotton,sugarcrops,andporkwereincreasedbyanaverageof17.1%.Inaddition,thepremiumpaidforabove-quotasaleofgrainandoilcropswasraisedfrom30%to50%ofthequotaprices.Theweightedaverageincreasewas22.1%forallagriculturalproducts.5Ineffect,thesepriceadjust-mentsinjectedalargeamountoffundsintotheruraleconomy,whichcreatedademandforindustrialproductsandsuppliedtheflowfundsforcapitalinvestment,especiallynonfarmproduction.Finally,theliberalizationofruralmarketsnotonlyaccommodatedthesalesofnonfarmproducts,butalsofacilitatedthepurchaseofinputsfornonagriculturalactivities.Itisevidentthatthesethreereformswereinterrelated;eachhelpedreinforcetheimpactoftheother.

Consequently,bythemid-1980s,theeconomicbasisforacceleratedgrowthinruralindustrieswasalreadyembeddedinChina’sruraleconomy.Inputandoutputmarketshad

Fordetailsofthesepricechangesandagriculturalpriceadjustmentsinthefollowingyearsofreforms,seeSicular(1988).

5D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162141

emerged;householdswereconsciousoftheiralternativeopportunities;andtheyhadincentivestoquicklyallocateresourcestononagriculturalactivitiesthatwouldgeneratehigherreturnsthanthosefromfarming.ThisviewissupportedbytheempiricalfindingsofPutterman(1993),whoanalyzedintersectoralfactorallocationinfiveproductionteamsofDaheTownshipinHebeiprovince.Thestudysuggeststhat,in1985,themarginalproductivityofcapitalandlaborinthenoncropsectorexceededthelevelsinthecroppingsector,indicating‘‘overallocation’’ofresourcesinagriculture.

Thecatalystfortherapidexpansionofnonfarmproductionwasaseriesofpoliciesthatloosenedrestrictionsonlabormobilityandtheoperationofruralenterprises.Thepolicy’sevolutioncanbebrieflydescribedasfollows.

In1983,DocumentNo.1oftheCentralCommitteeoftheChineseCommunistParty(CCP)providedgeneralguidelinesthatencouragedtheemergenceofspecializedhouse-holdsandpraisedtheireffectivenessinmakingthebestuseoflimitedfundsandlabor.Skillfulworkersandcraftsmenwerepermittedtoleavefarmingandengageinavarietyofnonagriculturalactivities,includinglong-distancetransportandthemarketingofcommod-ities.Inaddition,thedocumentallowedcooperativeventures,aswellasruralindustrialandcommercialhouseholds,toemploylabor(Ash,1988).Inaccordancewiththeseliberaliza-tionmeasures,thestatecontinuedtonarrowtherangeofproductsforcompulsoryprocurement.Itwasaftertheinceptionofthisdocumentthatsomefarmersbegantoquitfarmingtotakeupjobsinproducttransport,goodsretail,orbusinessandhandicraft.InMarch1984,theCentralCommitteeoftheCCPandtheStateCouncilissuedthe‘ReportonCreatingaNewSituationinCommuneandBrigade-runEnterprises,’whichoutlinedanewdevelopmentstrategythattargetedindustryasthefocusoffutureruraldevelopment.Industrialdevelopmentwasexpectedtoprovideinputsforagriculture,absorbrurallabor,andhelpraiseruralearnings(Findlayetal.,1994).Thisstrategysharplycontrastedwiththeoldpolicyoflocalgrainself-sufficiencyinwhichruralindustrieshadonlyasubsidiaryroletoagriculture.

In1985,DocumentNo.1oftheCCPpermittedfarmerstoseekemploymentandestablishbusinessesinnearbytowns,iftheycouldprovidetheirownfoodgrainandwerefinanciallycapabletorunabusiness.Thislandmarkderegulationofficiallyrelaxedthecontrolsonlabormobilitywithinruralregimes;inthepast,farmershadtoliveandworkinvillageswheretheyheldhouseholdregistration.

Inadditiontotherelaxationofcontrolsonlabormobility,amajorreforminagriculturalproductionandprocurementhelpedtriggertherapidgrowthofruralindustries.Atthebeginningof1985,afterconsecutiveyearsofgoodcropharvests,thestateannouncedthatitwouldnolongersetanymandatoryproductionplansinagricultureandthatobligatoryprocurementquotasweretobereplacedbypurchasingcontractsnegotiatedbetweenthestateandfarmers(Lin,1992).Thelooseningoffarmingconstraints,togetherwiththeincreasedfreedominallocativedecisions,promptedfarmerstoadjusttheirproductiveactivitiesinaccordancewithprofitmargins.6In

However,itshouldbeacknowledgedthatinsomepartsofChina,implementationofcontractnegotiationsfellshortofthecontractrhetoric,inwhichmandatoryquotascontinuedafter1985.Asignificantpriceliberalizationoccurredagainin1992–1993,butcametoahaltin1994whenChinaexperiencedhighinflation.

6142D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

1985,thegrain-sownareaatthenationallevelfellby4%,outputby7%,cotton-sownareaby26%,andcottonoutputby34%(Sicular,1988).Incontrast,thenumberofTVEsmorethandoubledinthesameyear,andtheirtotallaborforceincreasedbymorethan30%,followingayearofstronggrowthin1984(SSBa,1988).Thesedramaticchangesinpoliciesandinfarmers’responsesmarkedthebeginningofsustainedexpansioninnonagriculturalactivities.73.Anillustrativemodel

Tobetterunderstandthemechanismsthroughwhichstateinterventionsleadtodistortionsinfactorallocationsandtherationalresponsesoffarmerstopolicychanges,Isetupafarmhouseholdmodelwithtwoactivities.Forsimplicity,theanalysisfocusesonlabormobilityregulationsandtheconsequencesofpolicychanges,takingotheraspectsofthereforms,suchastheadoptionofHRSandtheemergenceofruralmarkets,asgiven.Consequently,themodelaccordswiththeeconomicrealityofruralChinaintheearly1980sandthesubsequentchangesafterfactormarketliberalization.Moreover,themodelprovidesaframeworkthatalsoaccommodatestheroleofeducationinaffectingtheprofitabilityoffarmbusinessthroughresourceallocationdecisions.

Considerastaticprofitmaximizationprobleminwhichthehouseholdengagesinbothagricultural(a)andnonagricultural(n)activities,ja{a,n}.Assumethattheactivity-specificproductionfunctiontakesthefollowingform:

yj¼fjðxj;kj;ljÞ;

ð1Þ

whereyjistheoutput;{xj,kj,lj}arevariableinputsandquasifixedfactorsofcapitalandlaborusedinthejthactivity.8Imaketheusualassumptionsabouttheneoclassicalproductionfunctionfjinthatithasdiminishingreturnsineachoftheinputsandthethreefactorsarecomplementaryinproduction.Analytically,

Bfj

>0;BXB2fjB2fj

<0;and>0

BXmBXmVBX2formpmV;

ð2Þ

}indextheinputs.whereX={xj,kj,lj}and{m,mV

Theallocativedecisionsofthehouseholdinvolvethepurchaseofthevariableinputx,theutilizationofthisinput,aswellastheuseoffamilylaborandcapitalacrossthetwolinesofproduction.Followingconventionaltreatment,totalfamilylabor(l)andcapital

Itshouldbenotedthatobstaclestorurallabormobilitystillexisttodaydespitecontinuedimprovementssincetheearlyyearsofreforms.Forinstance,aruralworkercurrentlyemployedintheenterpriseofanothervillagedoesnotreceivetheallocationofhomesteadorotherhousingarrangements,evenifthejobispermanent,thusincurringhighcoststothemigrants.AsreportedinYao(1999),localprotectionismisalsoasignificantissue,inwhichvillageworkersoftenearnmuchhigherwagesthanoutsiders.Inaddition,insomeregions,localgovernmenthascontinuedtoimplementvoluntaryproductioncontractswithacertaindegreeofcoercion.Clearly,muchhasbeenimprovedregardinglabormobility,butfurtherreformsareneeded.

8Notethatlandisomittedinthemodel,althoughitcouldbeaddedtotheagriculturalproductionwithoutaffectinganyofthefollowinganalyticalresults.Thisomissionsimplifiesnotation,anditwillberelaxedintheempiricalanalysis.

7D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162143

assets(k)areassumedtobefixedinputs;theiraccumulationdecisionsareonlyaffectedbylong-termconsiderations.Becausethethreeinputsareusedinbothactivities,theysatisfythefollowingresourceconstraints:

l¼laþln;k¼kaþkn;x¼xaþxn:

ð3Þ

Theequalitiesintheequationsareconsistentwithfull-capacityusageandzerocostofadaptingtheinputstothealternativeactivities.Thevariablevaluesassociatedwitheachsectorreflecttheextentofactivityparticipation.

LettheaggregateprofitPbethesumofprofitsfromthetwoactivities,P=Pa+Pn.Themaximizationproblemofthehouseholdcanbewrittenas:

la;ka;x;xa

maxP¼payaþpnynÀwxx;

subjecttoconditions(1)–(3);{pa,pn,wx}representthepricesfortheagriculturalproduct,thenonagriculturalproduct,andthepriceofthevariableinput,respectively.Theoptimalsolutionstotheproblemscanbedenotedas{la*,ka*,x*,xa*}.Thesederiveddemandforinputsarefunctionsofhouseholdendowmentsandinputandoutputprices,satisfyingtheconditionsthatthemarginalreturnofxisequaltothemarginalcostofpurchase,andthatthemarginalvaluesforeachofthethreeinputsareequalizedacrossthetwosectors.ThesestandardoptimalchoicesunderthecompetitivesituationcanbeusedasareferenceforstudyingfarmhouseholdbehaviorinChina.Bytheearly1980s,Chineseruralhouseholdsstillcouldnotadoptoptimalproductionplansdespitethefactthatmarketshademergedandtheyhadreasonablecommandovertheirresourceswithinagriculture.Thisisbecauseofficiallytheywerestillnotpermittedtoengageinnonagriculturalproduction,reflectingtheolddevelopmentstrategythatemphasizedagricultureandlocalgrainself-sufficiency.Thesetoflabormarketdistortingpolicyinterventions(P)thatpreventedoptimallaborflowfromagriculturetononagriculturecanberepresentedbyabindingconstraint

laPÀla*>0;orln*ÀlnP>0

wherelaPcorrespondstooverutilizationoflaborinagricultureunderthepolicyinterventions.DenotingknPasfarmers’choiceofcapitalfornonagriculturalproductionwithpolicycontrols,itiseasytoshowthatundertheconditionsgivenabove

kn*ÀknP>0;

aresultsuggestingsystematicunderutilizationofcapitalinthenonfarmsector.

Therefore,priortofactormarketliberalizationduringtheperiodof1983–1985,Chinesefarmersoperatedunderasecond-bestscenario.Theirkeyproductiveresources,laborandcapital,wererestrainedfrombeingallocatedtononfarmactivitiesdespite

144D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

households’incentivesforintersectoralreallocation.Hence,whentheseriesofpoliciesliftedlabormobilitycontrols,farmerswouldrespondbyclosingtheinefficiencygapsoffactorallocation,(kn*ÀknP)and(ln*ÀlnP).

Whiletheaboveanalysisfocusesontheallocationofinputsacrossthetwosectors,theframeworkcanbeextendedtoaddresscapitalaccumulationdecisions.Ifthehouseholdispermittedtopurchasecapitalequipment,ratherthantreatingitasaquasifixedfactor,theeffectofpolicyinterventionsoncapitalaccumulationcanalsobeanalyzed.Thetotalcapitalstockwithpolicyrestrictions,kP=knP+kaP,woulddependoncumulativecapitalinvestmentsinthetwosectorsovertime,anditwouldfallshortofmarketoptimallevelwhencapitalhasstrongercomplementaritywithlaborinnonag-riculturalsector,i.e.,kPB2fa/BkaBla.InChina,landperfarmislimited,formingaconstrainingfactorinagriculture,whichsuggeststhatthiscomple-mentarityconditionislikelytohold.Consequently,thesetoflabormarketpolicyinterventionswouldcauseunderinvestmentincapital.Inadynamicframework,becausenewlypurchasedequipmentdoesnotfullydepreciateinthecurrentperiod,reducedcapitalspendingineachperiodwouldresultinaless-than-optimalleveloftotalcapitalstockforthehousehold.

Inthispaper,Idonotsystematicallyexplorecapitalaccumulationdecisionsovertimebecauseofthefollowingtworeasons.First,capitalassetsownedbyruralfamiliesinChinainthemid-1980swerelargelydeterminedbyfactorsgoingbeyondtheexogenousvariablesintheabove,simplemodel.AsdocumentedbyWen(2000),assetspreviouslyownedbycollectiveunitsweredividedanddistributedtoindividualfamiliesduringthedisbandmentofcommunesbetween1982and1984basedprimarilyonthesizeofthefamilypopulation,and,insomeregions,onthesizeoffamilylaborforceorothercriteria.Theempiricalanalysisofthepapercovers1986–1995.Datalimitationsonlocalregulationsregardingdivisionofassetsdonotallowproperinvestigationsintopastassetaccumulation.Second,fromatheoreticalperspective,capitalaccumulationdecisionsarecloselyrelatedtooccupationalchoicesoffamilymembers.Forinstance,self-employ-mentinnonfarmactivitieswouldusuallyrequiremorecapitalinvestmentrelativetohavingwagejobsinlocallabormarkets.Formallyincorporatingtheseconsiderationsintothecurrentframeworkwouldrequireadifferentmodellingstrategy,whichwouldgobeyondthescopeofthecurrentpaper.Therefore,throughoutthepaper,Ifocusonintersectoralinputallocationdecisions,takingthecapitalassetsofthehouseholdasgiven.

Theabovestaticmodel,whilehighlightingfactormisallocationduetopolicyinterventions,hasnotyetincorporatedtheroleofeducationintheadjustmentprocesswithmarketliberalization.Theideathatschoolingmayenhancetheefficiencyoffarmerstorespondtoeconomicdisequilibriaandtoexplorelearningopportunitiesiswellknown(e.g.,Schultz,1975;Rosenzweig,1995).Inparticular,educationmaycontributetoprofitabilitythrougha‘‘workerproductivityeffect’’andan‘‘allocativeeffect’’(Welch,1970).Theformerreferstoeducation’seffectontechnicalefficiency,andthelatterreferstotheskillsofobtainingandusinginformationformanagerialdecisions,includingtheallocationofinputstoalternativeuses.ApplyingthesethesestoruralreformsinChina,weexpectthat,oncepolicyrestrictionswereloosened,moreeducatedfarmerswouldhavebetterinformationaboutresourcemisallocationsandtheir

D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162145

schoolingwouldfacilitatequickeradjustmentsinallocatingmorecapitalandlabortononagriculturalproduction.

Twoissuesariseregardingthespecificationofeducationinproduction.Thefirstpertainstothemeasuresofeducationtoapproximateallocativeskillswithinthehousehold.Theapproachtakenincorporatestheideathatfamilymembershaveincentivestoshareinformationwitheachothersothatthememberwiththehighestschoolinglevelismostlikelytocontributetomanagerialdecisions.Thefactthatfamilymembersoftenlivetogetherandjointlyengageinproductioncertainlyfacilitatesinformationsharingandcollectivedecisionmaking.Hence,thispaperdistinguishesthehighestlevelofschoolinginthehousehold(sh)fromtheaverageschoolingofotherfamilyworkers(sa)asproxiesforallocativeskills.9Inthecontextofthispaper,theallocativeeffectofschoolingismodelledthroughinputallocationsacrossagriculturalandnonagriculturalsectors.Therefore,ifthereiscentralizeddecisionmaking,wewouldexpectthatshplaysamoreimportantrolethansainaffectingresourceallocations.Laterempiricalanalysiswillalsoinvestigatewhetherfamilyschoolingenhancesfarmefficiencythroughotherchannelsincludingtheworkerproductivityeffect.

Thesecondissuepertainstounobservedcharacteristicsofthehousehold.Forinstance,unobservedabilityandmanagerialskillsspecifictotheithfarm(mi),whichislikelycorrelatedwiththeschoolingattainmentoffamilymembers,mayhaveaneffectonfarmprofitsindependentofschooling.Omittingtheseunobservedvariablesareomittedinempiricalanalysismaycausebiastotheschoolingestimates.Therefore,carefulandexplicitspecificationsforfamilyunobservedvariablesarecalledforinboththeanalyticalmodelandempiricalanalysis.

Itshouldbenotedthattheadjustmentprocesstowardsanewequilibriumafterpolicyreformsmaylastforalongperiodoftime.Theestablishmentofnewnonagriculturalbusinessmayrequiremonthsorevenyearsofpreparation.Thescaleofbusinessoftengrowsgraduallytowardsoptimum.Employmentinthelabormarketalsoinvolvescollectionofinformationand,perhaps,long-termplanninginaccordancewiththeongoingfarmingactivities.Therefore,takingintoaccounttheallocativeroleofeducationduringtheperiodofadjustment,theoptimizationproblemforhouseholdiforyeartleadstothefollowinginputdemandfunctions:

nPkit¼knPðlit;kit;pt;sit;mi;FitÞ;

ð4Þð5Þ

and

nPlit¼lnPðlit;kit;pt;sit;mi;FitÞ:

Thesuperscript‘‘P,’’asnotedearlier,referstochoicesofinputsduringtheadjustmentperiodinwhichtheinfluenceofpolicyinterventionsisstillnotfullycorrected.The

HypothesesconcerningtheeffectsofworkerschoolingcompositiononfarmefficiencyareproposedandtestedinYang(1997).Empiricalfindingsbasedonsmall-scalefarminginChinaindicatethateducationalreturnscomeprimarilyfromthehighestfarmschoolingthroughallocativeeffects.ThisresultisconsistentwiththefindingsbyFosterandRosenzweig(1996)thatwhetheranyoneinthehouseholdhasaprimaryeducationisagoodpredictorfortheadoptionofhigh-yieldseedvarietiesduringtheearlystagesoftheGreenRevolutioninIndia.Also,seeBasuetal.(2001)forevidenceofintrahouseholdexternalityofliteracyforBangladesh.

9146D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

explanatoryvariablesthatdeterminetheallocationofcapitalandlabortononagriculturalactivitiesinEqs.(4)and(5)includeaggregatelaborandcapitalendowment{lit,kit},pricevectorpt,avectoroffamilyschoolingsit=(shit,sait),unobservedmanagerialabilityandothertime-invariant,family-fixedfactorsmi,andavectorofhousehold/localcharacter-isticsFit,includinglocalweatherconditions,familylandendowment,workerexperience,geographicfeaturesandagriculturaltaxburdens.Thesehousehold/localvariablesmayormaynotchangewithtime.Notethatthesizeoffamilylaborforce,capitalstock,andtheschoolingvariablesmaydifferovertime,althoughtheirchangesareassumedtobedeterminedbylong-termconsiderationsandareexogenoustoperiod-specific,intersectoralinputallocations.Toclarifytheallocativeroleofeducation,notethatifthesectoral

n*nPn*nPallocationsofinputsarealreadyattheoptimum,i.e.kitandlit,sitwouldhave=kit=litn*nPn*nPnoroleinaffectinginputdemand.But,wheninefficiencygaps(kit-kit)>0and(lit-lit)>0exist,wewouldexpectthatsitispositivelyassociatedwithinputallocationstothenonfarmsector.Inotherwords,bettereducatedfamilieswouldadapttooptimalinputusesmorequicklythanlesseducatedfamilies.

Theeffectofinputadjustmentsonearningsisrevealedinthehouseholdnetprofitfunction,whichisdefinedastotalsalesminustheexpendituresonvariableinputsxt.Netprofitforhouseholdiinyeart,denotedasVit,canberepresentedasfollows:

nP

Vit¼Vðlit;dnPlit;kit;dkit;pt;sit;mi;FitÞ;

ð6Þ

nPnP,dkit}aresharesoflaborandcapitaldevotedtononagriculturalproduction.where{dlitThisprofitfunction,togetherwithEqs.(4)and(5),encompasstheideasthateducationmayaffectinputallocationsandthatthoseallocationsinturnmayaffectfarmprofits.Morespecifically,ifschoolingfacilitatesinputadjustmentandless-than-optimallevelsofcapitalandlaboraredevotedtononfarmuses,wewouldexpectapositiveassociationbetweensitnPnPnPnPand{dlit,dkit},andbetween{dlit,dkit}andVit.10Notethattocapturetheeffectsofschoolingonprofitsotherthansectoralinputuses,Ihavealsospecifiedsitintheprofitfunctionconditionalonfactorallocationdecisions.ApositiveassociationbetweensitandVitwouldimplyadditionalcontributionofeducationtoefficiencyontopoftheallocative

nPnPeffectsalreadyembeddedinthefactorshares.Because{dlit,dkit}arechoicevariables,propertreatmentforendogeneityiscalledforinempiricalimplementation.

Insummary,thecentralhypothesesconcerningtheadjustmentprocessandschoolingeffectsonfarmefficiencycanbestatedasfollows.Inaspecificyearduringtheperiodoffactormarketliberalization,wewouldexpect:

(1)thatless-than-optimallevelsofcapitalandlaborareallocatedtononfarmactivities

(misallocationofresources);

Analternativeempiricalstrategytodemonstratethatpolicyinterventionsindeedresultedinbindingconstraintsoninputallocationsistoestimatethemarginalreturnsofcapitalandlaborinthetwosectors.Evidenceonhigherreturnsinnonagriculturewouldconfirmthebindingconstraints.However,itcanbeshown

n*nP*nPthatthebindingconstraintsinfactorallocations,(knitÀkit)>0and(litÀlit)>0,arenecessaryandsufficient

conditionsforapositiveassociationbetweenfactorsharesinnonagricultureandoverallhouseholdprofit.Thus,

nPnPempiricalevidencethathigher{dlit,dkit}actuallyraisedVitwouldimplythattheresourceconstraintswereactuallybinding.

10D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162147

(2)thatgivenhouseholdendowments,theallocationsofcapitalandlaborto

nonagriculturalactivitiesarepositivelyrelatedtotheschoolingofhouseholdworkers,inparticularthehighestlevelofschoolingattainment(allocativeeffects);and

(3)thatgiveninputallocations,thehouseholdprofitispositivelyrelatedtotheschooling

attainmentoffamilyworkers(workerproductivityandotherallocativeeffects).Hypotheses(2)and(3)canbereadilyexpressedinanalyticalforms.TakingpartialderivativesofEq.(6)withrespectivetotheschoolingvariables,themechanismsandsourcesofschoolingreturnsarederivedas:

BVitBVitBdnPBVitBdnPBVitkitlit

¼nPþnPþ;BsitBdkitBsitBdlitBsitBsit

ð7Þ

wherethefirsttwotermsareassociatedwithallocativeeffectsofsectoralcapitalandlaboruses,andthelasttermwithallothereffectsofschooling.

Iwilluseapaneldatasetcovering1986–1995,aperiodimmediatelyafterlabormarketpolicyreforms,todescribethechangesinfarmperformanceandtotesttheabovehypotheses.

4.Empiricalanalysis4.1.Data

Thedatausedforthisstudy,whichwerecollectedbyChina’sStateStatisticalBureau(SSB),arefromtheRuralHouseholdSurvey(RHS)fortheSichuanprovincefor1986–1995.11Sichuan,themostpopulousprovinceintheinlandofChina,ishistoricallypraisedasthe‘‘landoffishandrice’’becauseofitsfavorableclimaticconditionsforfarming.Thesurveyrotatesafractionofthesampledhouseholdseachyear.Thedataconsistoftwopanels,onefor1986–1989andtheotherfor1991–1995,thatareconstructedfromtheoriginal,completesample.12Anumberofadjustmentswererequiredinordertomakethedatasuitableforthisstudy.TheDataAppendixprovidesdetailedinformationonsourcesandadjustments.Here,Ireportonlyasummarydescriptionofthedataset.

Inthisstudy,agriculturalactivitiesincludecropping,animalhusbandry,forestry,fishery,andsidelineproduction,abreakdownthatisconsistentwiththestandarddefinition

Thisnationalsurvey,startedin1952,consistsoflargerandomsamplesandrecordsofdetaileddiaryinformationonproduction,incomes,andexpenditures.Datasince1986areincomputer-usableform,buttheyhavenotbeenreleasedtothepublic.IhavelimitedaccesstotheSichuandatathroughacollaborativeprojectwithresearchersattheSSB.

12ThetwoseparatepanelsreflectthefactsthatSSBstartedacompletenewsamplein1991,andthatthe1990dataIreceivedfromSSBwascorrupt.Asecondattempttorestorethedatabyadifferentmeansofdatatransferwasalsonotsuccessful.

11148D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Table1

SummarystatisticsofrealpercapitaincomeandmajorinputsYear

Realper

capitaincome(yuan)(1)501507509486(187)(198)(224)(186)–653(288)674(305)620(337)661(343)710(368)

Landperfarm(mu)(2)7.76.96.56.66.26.16.26.15.9

(8.9)(5.6)(4.9)(5.4)–(4.8)(4.7)(4.9)(5.0)(4.9)

Laborforceperfamily(3)2.62.62.62.72.62.62.62.62.6

(1.1)(1.0)(1.1)(1.1)–(0.9)(1.0)(1.0)(1.0)(1.0)

Capitalstockperfamily(yuan)(4)(643)(697)(1058)(1181)–

1046(1359)1101(1682)1337(2124)1554(2466)1613(2520)502551590646

Numberofobservations(5)809798770788–15191516151115161518

1986198719881989199019911992199319941995

(1)Thefiguresincolumn1arein1986prices.Thedeflatorusedistheconsumerpriceindexofruralresidents(SSBb,1996).Yuan=$0.125.(2)Incolumn2,mu=0.165acre.(3)Figuresincolumn4areinnominalpricesbecauseofnoappropriatepricedeflator.(4)Figuresinparenthesesarestandarddeviations;thesameappliestoTable2–4.

of‘‘agriculture’’inChinesestatistics.Nonagriculturalactivitiesconsistofavarietyofproduction,rangingfromindustrytohandicrafts.Theprofitfromeachlineofproductionisequaltorevenueminusvariablecosts.Wageemploymentinnonagriculturalactivitiesisspecialbecauseitdoesnotincurvariablecosts.Toouradvantage,theRHSrecordstheutilizationofcapitalandlaborbyindustry.Therefore,wecanaggregatethefactorallocationsintoagriculturalandnonagriculturalactivities.

Table1reportssummarystatisticsofincomeandmajorinputsownedbythefarmhouseholds.Realpercapitaincomeforthesamplerosefrom501yuanin1986to710yuanin1995,indicatingsustainedgrowthaftertheinitialburstinearningsbetween1978and1985.13DespitethesmallscaleofChineseagriculture,landperfarmdeclinedduringtheperiod,whilethenumberofworkersperfamilystayedconstant.Incontrast,thevalueofcapitalequipmentincreasedovertime,afactconsistentwiththenationaltrend.

Thetwopaneldatasetshave,respectively,approximately8and15hundredhouse-holdsforstatisticalanalysis.Duetomissinginformationforsomefamiliesincertainyears,thenumberofobservationsisnotexactlythesameforeachoftheyearsbecausetheyear-specificobservationswithmissinginformationaredeleted.Thenumberofobservationsreportedin[column(5)ofTable1]isthesampleusedforthefollowingempiricalanalysis.

ThefiguresreportedinTable2onactivity-specificallocationsofinputsrevealthat,ingeneral,thehouseholdsdevotedincreasinglymorelaborandcapitaltononagriculturalactivitiesovertime.Duringthe10-yearperiod,theshareoflaborwithnonagriculturalworkasitsmainoccupationincreasedbyabout10%points.Accordingly,theshareof

ForSichuan,ruralrealpercapitaincomeroseby115.7%between1978and1985;anditaccomplished40.8%ofgrowthbetween1986and1995(SSBb,1996).Theincreaseof41.7%inrealincomeforthesamplehouseholdsbetween1986and1995isrepresentativeoftheprovincialincomegrowth,despitethefactthatthereisnoticeablediscontinuityacrossthetwopanels.

13D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Table2

Allocationoflaborandcapitalinagriculturalandnonagriculturalactivities(householdaverage)Year

No.ofworkersinagriculture(1)2.35(1.04)2.25(0.96)2.27(1.01)2.35(1.07)

–

2.23(0.94)2.21(0.96)2.19(1.01)2.09(1.01)2.10(0.98)

No.ofworkersinnonagriculture(2)0.28(0.44)0.34(0.49)0.35(0.49)0.32(0.46)

–

0.34(0.49)0.37(0.53)0.44(0.60)0.54(0.67)0.52(0.65)

Laborsharein

nonagriculture(3)10.813.013.312.0–13.114.216.820.619.8

Capitalinagriculture(4)

178.6(360.0)181.8(214.0)171.7(254.8)184.7(254.3)

–

286.4(454.7)289.5(483.6)306.3(461.0)367.1(650.0)423.0(1089.2)

Capitalinnonagriculture(5)

(501.8)(630.6)(997.7)(1121.5)–

770.6(1156.5)819.4(1513.0)1031.6(2015.8)1187.2(2282.0)1192.0(2225.6)323.0368.8419.2461.5

149

Capitalshareinnonagriculture(6)64.467.071.071.4–72.973.977.176.473.8

1986198719881989199019911992199319941995

capitalequipmentfornonfarmusesalsoincreasedbycloseto10%.Itappearsthatnonagriculturalproductionismuchmorecapital-intensivethanfarming.14Table3containsinformationonschoolingandlaborforceexperience.TheRHSreportsthelevelofschoolingcompletionfortheruralworkersinsteadofyearsofschoolingattained.Iorganizethecompletionlevelsintofourcategories:illiterateandsemiilliterate,elementaryschool,secondaryschool,andhighschoolandplus(seetheDataAppendixfordetails).Itisevidentfrom[columns(1)–(4)ofTable3]thattheeducationalattainmentofthelaborforceimprovedoverthe10-yearperiod.Thepercentageofworkerswhowereilliterateorsemiilliteratefellfrom31.4in1986to14.9in1995.Thisfallintheilliteracyrateandtheriseinthepercentageofworkerswithmiddleandhighschooldegreesaremainlyduetotheentranceintothelaborforceofmoreeducatedworkers,theexitofolderworkerswithlessschooling,andotherchangesinfamilydemographics,suchasdeathsandmigration.Withineachsubperiod,thereweresomechangesinthehighestandaveragelevelsofschoolingwithinhouseholds.Forthe1986–1989period,21%ofthehouseholdsreportedchangesinthehighestlevelofschooling,while57%ofthehouseholdshadchangesinaverageschooling.Forthe1991–1995period,thecorrespondingpercentageswere24and59.Alsonotethattheaverageschooling,reportedinthelevelofcompletion,issignificantlybelowtheaveragehighestlevelofeducationofthehouseholds,indicatingschoolingvariabilitywithinfamilies.Theaverageexperienceofthelaborforce,definedas(age-schooling-7),isstableovertime.

Forempiricalanalysis,Iwillalsoexaminewhethergeographicenvironmentsinfluencefactorallocationsandhouseholdincomegrowth.Table4containsinformationonthegeographicfeaturesofthesample(plain,hillyareasvs.mountainousregions)and

Thestatementissubjecttoonecaveatthatthedefinitionadoptedforoccupationmayunderreporttheextentoflaborparticipationinnonagriculturalactivities.Aworkerisclassifiedasinnonagricultureif,accordingtothesurvey,hisorher‘‘mainoccupation’’isanonfarmindustry.Thisdefinitionsystematicallyunderrepresentspart-timeparticipationinnonfarmactivitiesforsomefarmers.Unfortunately,thereisnootherinformationinRHSthatcouldremedythisdatalimitation.

14150D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Table3

SchoolingandexperienceofthelaborforceYear

Educationofthelaborforce(%)IlliterateandPrimarysemiilliterate(1)school(2)1986198719881989199019911992199319941995

31.4(30.7)30.9(30.3)29.0(29.4)28.8(29.9)

–

16.7(26.0)16.7(26.3)15.2(24.6)14.7(24.5)14.9(25.1)

43.0(33.3)43.5(32.9)44.0(32.3)43.5(32.6)

–

45.9(33.0)45.4(33.1)45.5(33.0)44.7(32.9)43.8(32.9)

Middleschool(3)22.5(27.8)22.6(27.5)24.0(28.4)24.6(28.6)

–

32.4(32.3)32.8(32.1)33.6(31.9)34.5(32.2)35.5(32.6)

LevelofLevelofAverageworkaveragehighestexperience(7)High

school+(4)education(5)education(6)3.03.02.92.94.95.05.65.95.6

(11.3)(10.9)(10.9)(11.0)–(15.5)(15.6)(16.3)(16.7)(16.0)

1.9(0.5)1.9(0.5)2.0(0.5)2.0(0.5)–2.2(0.5)2.2(0.5)2.3(0.5)2.3(0.5)2.3(0.5)

2.5(0.7)2.5(0.7)2.5(0.7)2.5(0.7)–2.7(0.7)2.7(0.7)2.7(0.7)2.8(0.7)2.8(0.7)

23.1(6.5)23.3(6.4)23.3(6.7)23.4(6.5)–

21.2(6.7)21.6(6.7)21.9(6.5)22.5(6.8)23.5(7.2)

(1)Schoolingattainmentreportedincolumns5–6isanarithmeticaverageofschoolingcompletionlevels,whereilliterateandsemiilliterateisrepresentedaslevel1,primaryschoolaslevel2,middleschoolaslevel3,andhighschoolandplusaslevel4.(2)Experienceisapproximatedas(age-schoolingyears-7).

compulsoryleviesonagriculture,whichreflectpolicyenvironmentsforthehouseholds.Itshouldbenotedthatforeachsubperiod,thesmallvariationsinthepercentageofhouseholdsbelongingtoageographictypereflectthefactthatthenumberofobservationsisnotexactlythesameforeachyear.Somefamilieswithmissingyear-specificinformationaredeletedfromthesampleforthecorrespondingyears.Moreover,notethatthesecondpaneldrawsahigherpercentageofhouseholdsfromplainareas,whichinpartexplainsthedifferencesinsamplecharacteristicsacrossthetwotimeperiodsasrevealedinTables1–3.Therefore,cautionisneededwhenmakingacross-panelcomparisons.

4.2.Econometricspecification

Toestimatetheinputallocationfunctionsforlaborandcapitaltononagriculturalactivities,asinEqs.(4)and(5),Iusethefollowingparametricspecification:

XX

lnXit¼aXdlnditþallnlitþaklnkitþ

X

c

ccXX

bXshshitþbsasaitþbxxait

þ

X

a

aaX

cXggitþcffitþ

X

r

rr

cXdDtþmiþeit;

ð8Þ

nPnP,kit},andXa{lnP,knP}whichdenotesinput-specificparameters.TowhereXita{litsimplifynotation,Iremovethesuperscript‘‘np’’from{lnP,knP}inthefollowinganalysis,giventhefactthatthecorrespondingparametersareforinputdemandfunctionsforthenonfarmsector.InEq.(8),{dit,lit,kit}areaggregateland,laborandcapitalendowments,

crespectively,forhouseholdiinyeart;shitaredummyvariablesrelatingtothecompletionlevelforthemosteducatedfamilyworker:c={elementary,secondary,highschooland

D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Table4

SummarystatisticsofgeographicandpolicyvariablesYear

Geographicfeaturesofsamplehouseholds(%)Plainareas(1)

1986198719881989199019911992199319941995

6.35.75.95.8–16.515.816.416.416.4

Hillyareas(2)60.161.462.061.6–55.856.155.154.353.8

Mountainousareas(3)33.432.831.932.4–27.627.928.429.229.7

151

Compulsorylevieson

farming(yuan)(4)31.9(27.8)40.8(33.2)47.7(60.2)63.9(56.3)

–

103.6(103.4)107.6(117.6)103.6(144.8)139.9(179.3)155.1(173.7)

plus},withelementaryschoolasthereferencegroup;15saitistheaveragecompletionlevelofschoolingforotherfamilyworkers;andxaitistheaverageworkexperienceforall

afamilyworkers.Thedummyvariablesgitrepresentgeographicareas:a={plain,hilly,mountainousregions},withplainareasasthereferencegroup.Thevariablefitstandsforcompulsoryfeesimposedbythegovernmentonfarming.Thisestimationequationalsoincludeshouseholdfixedeffects(mi)relatingtofactorssuchasunobservedmanagerialabilityandlandquality,aswellasregion-andtime-specificeffects(Dtr)relatingtofactorssuchastime-varyingweathershocksattheregionlevelandlikelypricevariationsbothovertimeandspace.16Anadditionaladvantageofhavingtime-regionaldummiesisthattheycancontrolforchangesinfactordemandinthenonfarmsector.ThesuperscriptrinDtrindexes23prefecturesfromwhichthesamplehouseholdsweredrawn.Theerrortermeitisassumedtorepresenttheeffectsofremainingomittedvariables,whichareindependentoftheexplanatoryvariablesandareindependentlyandidenticallydistributed.

17XXXcXXXaXXr{ad,al,aXk,bsh,bsa,bx,cg,cf,cd}areparameterstobeestimated.

Thetwoactivity-specificfactordemandfunctionswillbeestimatedconditionalontotalfactorendowmentsforthehouseholdineachperiod,namelyland,laborandcapital.Accordingly,theparameterestimateswouldyieldmeaningfuleconomicinterpretations.Forinstance,theparameterakkwouldrepresentthepropensitytoinvestinnonfarm

kproduction:ak=1isassociatedwithaneutralstrategy,whileakk>1impliesahighpropensityofinvestment.Thatis,foragivenpercentageincreaseintotalcapital,ahigher

Theilliterateandsemiilliterategroupismergedwiththegroupwithelementaryeducationbecausethepercentageofhouseholdsbelongingtotheformercategoryislow,around7%forthe1986–1989panel,anddroppingbelow3%forthe1991–1995panel.Therefore,thefollowinganalysisfocusesontheefficiencyofthehouseholdswithelementaryeducationrelativetothesecondaryandhighschoolandplusgroups.

16ThistreatmentforpricevariationsisnecessarybecausetheRHSdoesnotcontainpriceinformationforvariousagriculturalandnonagriculturalproducts.Theapproachtakenisconsistentwiththefactthattherelikelyexistcompetitivemarketsinlocalregions,butnotnecessarilyacrossregions.

17Forthisandthefollowingempiricalfunctions,thesuperscripttoaparameterindicatesassociationwithaparticulardependentvariable.Schoolcompletionlevelc,geographictypeindicatora,andregionindexrarealsoincludedinthesuperscript.Thesubscriptsrefertotheindependentvariablesassociatedwiththeparameters.

15152D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

percentagewillbeallocatedtothenonfarmsector.Similarinterpretationsalsoapplytoall.Moreover,becauselandisonlyusedinagriculture,andiflandisacomplementaryfactor

lktocapitalandlabor,wewouldexpect{ad,ad}<0,implyingthatabundantlandendowmentwouldhavetheeffectofretainingtheothertwoinputsinagriculture.

Moreimportantly,theestimationresultswillshedlightontheeffectofschoolingattainmentonfactorallocationstononagriculturalactivities.Ifthehighestlevelof

cschoolinginthehousehold(shit)isareasonableproxyforallocativeskills,andschoolingfacilitatesanarrowingoftheinefficiencygapsduringtheadjustmentperiod,wewould

lckcexpect{bsh,bsh}>0.Thatis,familieswithmiddleandhighschooleducationsystematicallytomakebetterinputdecisionsthanprimaryschooledfamilies.

TheprofitfunctioninEq.(6)thatincorporatestheeffectsofinputallocationdecisionswilltakethefollowingempiricalform:

vnPvnPvv

lnVit¼avdlnditþallnlitþhllndlitþaklnkitþhklndkitþ

X

c

vc

bvcshshitþbsasait

þbvxxaitþ

X

a

av

cvaggitþcffitþ

X

r

cdjrDtrþmiþlit:

ð9Þ

Theparametersfortheeducationvariablesandfactorsharesareofparticularinterest.

nPkitvcv,dnP},{bsh,bsa}>0First,becausefactorallocationdecisionsarealreadyembeddedin{dlitwouldimplyproductivevalueofeducationontopofitsallocativeroleincapitalandlaborusesacrossthetwosectors.Thisadditionalvaluemayincludetheuseofmoderncropvarieties,fertilizer,otherallocativedecisions,aswellasworkerproductivityeffects.Second,whenallocationsoflaborandcapitalarealreadyattheoptimumacrossthetwosectors,marginalchangesinthesharesoflaborandcapitalaroundtheirsamplemeanswould

vvhavenoimplicationsforprofits,i.e.,hlv=hvk=0.However,wewouldexpect{hl,hk}>0,ifthesamplehouseholdswerestillintheprocessofmakingadjustmentstowardsthenewequilibrium.Whenthesetwoparametersarepositive,theroleofeducationinreallocat-ingresourcestononagriculturalactivitieswouldcontributetothefarmhouseholdearnings.

While{hlv,hvk}willindicatesystematicmisallocationsoflaborandcapitaltononfarmsector,itwouldbeinterestingtofurtherexaminewhethertheextentofinefficiencydeclineswiththedeepeningofeconomicreforms.Weexpectthat,ceterisparibus,theestimatedparameterswillbecomesmallerovertimeasthegapsofdistortionaregraduallyclosedupwithadjustments.However,ifnonfarmopportunitiesimprovecontinuouslywithrapidgrowthandstructuraltransformation,theneedforinputrelocationtowardsthatsectormaypersistforanextendedperiod.Toinvestigatetheefficiencyoffactorallocationsovertime,IwilladdtothebasicmodelinEq.(9)interactiontermsofthe

nPnPtimeperiods(1986–1989and1991–1995)withlndlitandlndkit,thusallowingdifferentparameterestimatesacrossthetwopanels.Iftheestimatedparametersarepositivebutsmallerforthelatterpanel,theywouldimplyimprovementsinintersectoralfactorutilizationwithmarketliberalization.

Giventhestructureofthemodel,thetwofactorsharesareendogenousvariables,e.g.,idiosyncraticshockstoindividualhouseholds(suchasbeingluckyinlandinganonfarmjob),wouldaffectbothfactorsharesandhouseholdprofit,violatingtheorthogonality

D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162153

conditionoftheerrorterm.Therefore,itisnecessarytoestimateEq.(9)usinginstrumentalvariables.Fromthemodelspecifications,itisalsoevidentthatlaggedfactorsharesmayaffectinputallocationdecisionsforthecurrentperiod,butnotthecurrentprofitlevelofthehousehold.Therefore,Iusefactorshareslaggedbyoneperiodasinstruments.Consequently,inadditiontocontrollingforfamily-fixedeffects(mi)andregion/timespecificeffects(Dtr),thepredictedfactorsharesfromthefirst-stageequationswillbeusedinestimatingEq.(9).4.3.Estimationresults

Table5reportsfixedeffectestimatesofinputdemandfunctionsfortheallocationsofcapitalandlabortononagriculturalactivitiesbasedonpooleddataofthe1986–1989and1991–1995panels.Forreferenceofcomparison,Ifirstfitthetwoinputdemandfunctionswitharandomeffectsprocedurebasedontheassumptionthattheeffectsofunobservedcharacteristics,suchashouseholdmanagerialskillsandtime/regionfluctuationsinweatherandprices,areindependentoftheincludedexplanatoryvariables.Then,tobeconsistentwiththeanalyticalmodelpresentedearlier,Iestimatethefunctionswithafixedeffectprocedure,controllingforbothhousehold(mi)andtime/region(Dtr)heterogeneity.Thesetwoproceduresyieldquitedifferentestimates.Toselectappropriatespecifications,theresultsoftheHausmanteststronglyrejecttherandomeffectmodels,suggestingthattheunobservedhouseholdandtime/regioneffectsaredependentontheexplanatoryvariables.Assuch,thefollowingdiscussionswillconcentrateonthefixedeffectresults.Thenegativecoefficientsestimatedforthelandvariableareconsistentwiththeviewthatlandendowmentraisestheproductivityofcapitalandlaborinagriculture,thus

Table5

EstimatesofinputdemandfunctionsforcapitalandlaborinnonagriculturalactivitiesExplanatoryvariables

Dependentvariablesln(capitalinnonagriculture)Fixedeffectsestimates(1)

ln(land)

ln(totallabor)ln(totalcapital)

Middleschool,highestHighschool+,highestAverageeducationAverageexperience

Averageexperience2(Â1000)Hillyareas

Mountainousareas

Leviesonagriculture(Â1000)R2À0.190**(0.869)À0.306**(0.073)1.383**(0.024)0.100*(0.055)0.177*(0.099)0.017(0.039)À0.005(0.013)À0.018(0.243)0.373**(0.169)0.879**(0.198)0.117(0.111)0.747

ln(laborinnonagriculture)Fixedeffectsestimates(2)À0.058*0.523**0.048**À0.0250.131*À0.003À0.029**

0.239À0.269**À0.457**0.299**0.721

(0.032)(0.056)(0.018)(0.043)(0.077)(0.031)(0.010)(0.190)(0.132)(0.155)(0.086)

Bothregressionsincludehouseholdandregion/timedummies.Samplesizeis8480.Standarderrorsaregiveninparentheses.

*Significantat10%level.**Significantat5%level.

154D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

reducingtheoutflowofthesetwoinputsfromfarming.Theelasticityofcapitalinvestment

nPinnonfarmactivities,asrepresentedbyakk=Blnk/Blnk,indicatesthata1%increaseintotalcapitalisassociatedwitha1.38%increaseinnonagriculturaluses.Therefore,newcapitalisincreasinglybeingdirectedtowardsruralindustriesandservices.Agriculturestillappearstobethemainsectorforlaboremploymentbecausetheelasticityofnonagricul-turallaborallocation,all=BlnlnP/Blnl,is0.52.

Withregardtoeducation,thefindingssupportthehypothesisthatallocationsofcapitalandlabortothenonfarmsectorarepositivelyrelatedtotheschoolingattainmentofthehousehold.Usingthehighestlevelofeducationasaproxyforhouseholdallocativeskillsandcontrollingforhouseholdaggregatecapitalassets,themiddle-schooledfamiliesdevote10%morecapitaltononfarmusesthanthereferencegroup,theprimary-schooledhouseholds.Householdshavingmemberswithahighschoolorcollegeeducationinvestevenmorecapitalinnonfarmuses17.7%higherthantheprimary-schooledhouseholdsand7.7%higherthanthemiddle-schooledhouseholds.Moreover,thedatayieldevidencethathouseholdshavinghighschoolandcollege-educatedmembersallocate13.1%morelabortononfarmactivitiesrelativetothereferencegroup,althoughmiddle-schooledfamiliesdonotappeartoallocatemoreworkersawayfromagriculture.Thereisalsoevidenceofcentralizeddecisionmakingonthefarmssupportedbythefactthattheaverageeducationoffamilyworkersexcludingthehighestattainmentdoesnotcontributesignificantlytoeithercapitalorlaborallocations.Moreover,theresultsindicatethat,whiletheexperienceofworkersdoesnotsignificantlyinfluencetheintersectoraldistributionofcapital,olderfarmerswithmoregeneralworkexperiencearelesslikelytoparticipateinnonfarmwork.Thisfindingisconsistentwiththestandardimplicationsofhumancapitaltheory.18Withregardtogeographicandpolicyvariables,theestimatesindicatethatleviesonagriculturediscouragefarmingactivities,thushavingtheeffectofencouragingfactorallocationstononfarmuses,althoughthecoefficientforcapitaldoesnotreachtheconventionallevelofsignificance.Geographiclocationalsohasasignificantimpactonsectoralinputallocation.Beinginhillyandmountainousregionsincreasescapitalinvestmentinnonfarmuses,aresultthatisconsistentwiththeviewthatadversegeographiccharacteristicsmayhaverelativelylessnegativeeffectonreturnsforcapitalinthenonfarmsector.Incontrast,beinginhillyandmountainousregionsisassociatedwithlesslaborutilizationinnonagriculture,whichmayreflectthefactthatgeographicallydisadvantagedlocationspresentlessnonfarmopportunities.

Table6reportsestimationresultsfortheprofitfunctioninEq.(9)anditstwovariantspecifications.Themodelincolumn(1)isabaselinecaseexcludinglaborandcapitalshares,andthemodelincolumn(3)includesinteractiontermsofthetwotimeperiodswithfactorsharesinordertoinvestigatechangesinresourcemisallocationovertime.Sincethefactorsharesarechoicevariables,theirlaggedvaluesareusedasinstrumentsformodels

Becausetheexperiencemeasureishighlycorrelatedwiththeageoflaborers,thefindingisalsoconsistentwithacohortpreferenceeffect.Thatis,thenewcomerstothelaborforcepreferleavingthefarmindependentoftheirworkexperience.Unfortunately,Iamnotabletodisentanglethepreferenceandhumancapitaleffectsdueinparttothelackofdataonfarmvs.nonfarmexperience.Ithankarefereeforpointingouttheissueregardingcohortpreference.

18D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Table6

EstimatesofprofitfunctionsExplanatoryvariables

ln(totalland)ln(totallabor)

ln(laborshareinnonagriculture)ln(laborshareinnonagriculture)Âfirstperiod

ln(laborshareinnonagriculture)Âsecondperiodln(totalcapital)

ln(capitalshareinnonagriculture)ln(capitalshareinnonagriculture)Âfirstperiod

ln(capitalshareinnonagriculture)ÂsecondperiodMiddleschool,highestHighschooland+,highestAverageeducationAverageexperience

Averageexperience2(Â1000)Hillyareas

Mountainousareas

Leviesonagriculture(Â1000)R2Dependentvariable=ln(householdnetprofit)Fixedeffects(1)0.105**(0.017)0.185**(0.031)

––

–0.084**(0.029)

––

–

0.0300.014À0.0070.016**À0.339**

0.0620.137*À0.197**0.719

(0.023)(0.041)(0.017)(0.006)(0.103)(0.070)(0.082)(0.047)

IVFixedeffects(2)0.137**(0.018)0.138**(0.037)0.178**(0.038)

–

–0.027(0.030)0.215**(0.048)

–

–

0.019À0.043À0.0000.022**À0.352**

0.1030.152À0.269**0.722

(0.023)(0.042)(0.017)(0.006)(0.103)(0.073)(0.091)(0.049)

155

IVFixedeffects(3)0.135**(0.019)0.136**(0.037)

–

0.182**(0.044)0.179**(0.038)0.035(0.043)

–

0.309**(0.071)0.193**(0.049)0.018À0.0400.0030.022**À0.344**

0.1070.149À0.265**0.722

(0.023)(0.042)(0.016)(0.005)(0.102)(0.072)(0.091)(0.048)

Allregressionsincludehouseholdandregion/timedummies.Samplesizeis8480.Standarderrorsaregiveninparentheses.

*Significantat10%level.**Significantat5%level.

(2)and(3).Table7reportsthefirst-stageregressionsinwhichthelaggedfactorsharesappeartohavestrongexplanatorypowerforthecurrentperiodfactorallocations.19SpecificationtestsarealsoperformedinwhichtheHausmanteststatisticsstronglyrejecttherandomeffectmodelinfavorofthealternativefixedeffectspecification.Therefore,thefollowingdiscussionswillconcentrateontheIVfixedeffectestimates.

Incolumn(2)ofTable6,family-ownedassets—land,laborandcapital—allcontributetohouseholdearnings,althoughthecoefficientforcapitalisnotstatisticallysignificant.Controllingforthelevelofthesequasifixedfactors,iflaborandcapitalarealreadyoptimallyallocatedacrossthetwouses,anyintersectoralredistribution,asreflectedin{dlnP,dnPk},wouldhavenoeffectonprofits.Thepositiveandstatisticallysignificantcoefficientsestimatedfortheseparametersindicatethat,duringtheperiodofadjustmentinruralChina,theallocationsofthetwoinputshavenotyetreachedtheoptimum.Inparticular,ceterisparibus,a10%increaseintheshareoflaborinnonfarmactivitieswould

TheF-valuesareF(2,6190)=18.13inthecapitalsharefunctionandF(2,6190)=11.11inthelaborsharefunction,bothrejectingthenullhypothesesat1%significancelevelthatthetwocoefficientsfortheinstrumentsarejointly0.

19156D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Table7

First-stageinstrumentalvariableestimatesforcapitalandlaborsharesinnonagriculturalactivitiesExplanatoryvariables

ln(laggedlaborshareinnonagriculture)ln(laggedcapitalshareinnonagriculture)ln(land)

ln(totallabor)ln(totalcapital)

Middleschool,highestHighschool+,highestAverageeducationAverageexperience

Averageexperience2(Â1000)Hillyareas

Mountainousareas

Leviesonagriculture(Â1000)R2ln(capitalshareinnonagriculture)ln(laborshareinnonagriculture)Fixedeffectsestimates(1)0.010*(0.006)0.085**(0.014)À0.103**(0.024)À0.164**(0.041)0.241**(0.014)0.058*(0.032)0.097*(0.057)0.003(0.022)À0.006(0.007)0.043(0.140)0.139(0.097)0.435**(0.114)0.047(0.063)0.615

Fixedeffectsestimates(2)0.104**(0.012)0.110**(0.030)À0.072(0.050)0.473**(0.087)0.071**(0.029)0.001(0.067)0.239**(0.120)À0.011(0.005)À0.023(0.016)À0.015(0.294)À0.419**(0.204)À0.642**(0.239)0.395**(0.133)0.711

Bothregressionsincludehouseholdandregion/timedummies.Samplesizeis8,480.Standarderrorsaregiveninparentheses.

*Significantat10%level.**Significantat5%level.

raisehouseholdincomesby1.8%,whereasa10%increaseintheshareofcapitalinnonfarmactivitieswouldraiseincomesby2.2%.Thesefindingsprovidesupportforthehypothesisthatless-than-optimallevelsofcapitalandlaborwereallocatedtoruralindustriesandservicesduringtheperiodoftransition.CombiningtheseresultswiththefindingsinTable5,weseeevidencethatmoreschooledhouseholdsallocatedmoreinputstononagriculture,andthattheseadjustmentsinturnraisedfamilyearnings.

Inadditiontotheallocativeroleofeducationembeddedinthefactorshares,schoolingvariablesarealsoincludeddirectlyintheprofitfunctiontocaptureotheraspectsoftheireffectsonearnings.However,thosecoefficientsarenotstatisticallysignificant.Wealsofindthatexperiencecontributestoincomethroughaconcaveschedule.Controllingaverageexperienceat0,themarginalvalueof1yearofexperienceis2.2%ofannualhouseholdprofit;atsamplemeanlevelofexperience,themarginalvalueis0.6%oftheprofit.

Conditionalonfamilyendowmentsandresourceallocations,householdswholiveinhillyandmountainousregionsarenomoredisadvantagedthanhouseholdslivinginplainareas,accordingtothedata.Thisfindingcouldbesamplespecific,however,becausealargepercentageoffarmhouseholdsintheSichuanprovince(seeTable4)liveinhillyandmountainousareaswheretheclimateandsoilconditionsaregenerallygood.Consistentwithexpectations,compulsorytaxationonfarmingunambiguouslyreducesoverallhouseholdearnings.

[Column(3)inTable6]presentstheresultsofinteractingfactorsharesinnonagriculturalactivitieswiththetwotimeperiods,thusallowingchangesintheextentofresourcemisallocationovertime.Theestimatedcoefficientsforallvariablesremainstable,andtheestimatesforthefactorsharesarebothpositiveandsignificant,confirmingtheexistenceofresourcemisallocation.Morespecifically,theestimatesshowthattheelasticityofprofit

D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162157

withrespecttocapitalsharesinnonagriculturalactivitiesisreducedconsiderably.Inthefirstperiod(1986–1989),ceterisparibus,a10%increaseintheshareofcapitalinnonfarmproductionwouldraisetotalincomeby3.1%,butitloweredto1.9%inthesecondperiod(1991–1995).Thisresultimpliesthatpolicyreformshadreducedtheseverityofcapitalmisallocationovertime.Forthelaborshares,thecoefficientsforthetwoperiodsarenotstatisticallydifferent.ThisresultisconsistentwiththepossibilitythatrapideconomicstructuralchangestakingplaceinruralChinaduringtransitionmayhavecreatedcontinueddemandforlaborinthenonfarmsector,andthattheadjustmentshadnotfullyrespondedtothechanges.Itwouldbeinterestingtoinvestigatethechangesoveralongerperiodoftime.UsingestimatesreportedinTables(5)and(6),decompositionanalysisasoutlinedinEq.(7)canbeconductedtoshedlightonthesourcesandmechanismsofschoolingreturns

nPtofarmprofits.BecauseBlnknP/Bs=bskcasinEq.(8)andBlndnPwhenaggregatek=Blnk

nPnPkccapitalstockisheldfixed,itfollowsBlndk/Bs=Blnk/Bs=bs,wheres=(sh,sa).Similarly,BlndlnP/Bs=BlnlnP/Bs=blcs.Therefore,applyingEq.(7)totheempiricalestimationwithlogarithmicspecifications,themarginalvalueofschoolingcanbecomputedas:

BVˆvÂbˆkcþhˆvÂbˆlcþbˆv;¼hks1ssBsnPˆvˆvˆvwhereBlnV/BlndnPk=hk,BlnV/Blndl=hl,andBlnV/Bs=bsfromEq.(9);theparameters

withhatsareestimatedvalues.Thefirsttermcapturestherateofschoolingreturnsfromintersectoralcapitalallocation;thesecondtermrelatestotherateofschoolingreturnsassociatedwithintersectorallaborallocation;andthelastexpressionistherateofreturnsassociatedwithallothersourcesofschooling’seffectsonprofitability.

EmpiricalresultsreportedinTable5and[column(2)ofTable6]suggestthatforthissampleofChinesefarms,theeffectsofeducationonprofitabilitycomeprimarilyfromoptimallyallocatingcapitalandlaboracrossthetwosectors.Ifstatisticallyinsignificantestimatesareignored,whenthehighestlevelofeducationofthehouseholdisraisedfromprimarytomiddleschool,thereturnwouldbe2.15%(=0.10Â0.215)ofthetotalhouseholdprofit.Thisreturnisattributabletomoreefficientcapitalallocations.Ifthehighestlevelofeducationisraisedfromprimarytohighschool,thefamilyearningswouldincreaseby6.13%;thiseffectisattributabletomoreefficientcapital(3.8%=0.177Â0.215)andlabor(2.33%=0.178Â0.131)allocations.TheseresultscorroborateearlierempiricalfindingsreportedbyYang(1997)basedoncross-sectionaldatafromruralChinathatschoolingreturnscomeprimarilyfromallocativedecisionsandthatthehighestlevelofschoolingcontributesthemosttofarmefficiency.

Finally,datareportedinTable2indicatethatlaborsharesinnonfarmactivitiesincreasedby83.8%from10.8to19.8between1986and1995,andcapitalsharesincreasedby14.5%from64.4to73.8.Basedonestimatespresentedin[column(2)ofTable6],thesechangesinintersectoralinputallocationswouldresultinan18%increaseinfarmhouseholdearnings.Therefore,therapidexpansionsinnonfarmactivitieswouldaccountforapproximately43.6%ofthetotalfarmincomegrowthbecauserealearningsgrewby41.2%inthis10-yearperiod(seeTable1).Whilecautionmustbegiventothese

158D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

estimates,duetothefactnotedearlierthatthereisanissueofdatacomparabilityofthetwopanels,thesesampleestimatesareneverthelessbroadlyconsistentwithprovinciallevelinformation.20Althoughthemodelandresultspresentedinthepaperarenotsuitedtotraceoutmanydynamiceffectsofthechanges,themechanismsconcerningtheeffectsofeducationoninputallocationandprofitsareclearlyrevealed.Schoolingsignificantlyreducestheinefficiencyofintersectoralinputallocations;theincreaseinefficiencyinturncontributestofarmhouseholdincomegrowth.

5.Concludingremarks

ThispaperusespaneldatawithricheconomicanddemographicinformationtoinvestigatethesourcesanddeterminantsofsustainedincomegrowthinruralChinabetween1986and1995.Thiswasaperiodofmarketadjustmentwhentherelaxationoflabormobilitycontrolsinducedruralfamiliestoreallocatetheirproductiveinputsfromagriculturetononagriculturalactivities.Thefindingssuggestthatschoolingplaysacriticalroleinraisingtheefficiencyoffarmerstorespondtochangingmarketconditions.Householdswithbetter-educatedmembersactedmorequicklyindevotingmorecapitalandlabortononagriculturalactivitiesthatyieldedhigherreturns.Asaresultoftheseallocativeeffects,educationcontributedsignificantlytothesustainedruralincomegrowth.

TheabovefindingshavewiderimplicationsthansimplyimprovingtheunderstandingofaspecialperiodofincomegrowthinruralChina.Thecentrallyplannedsystemcreatedmassivemisallocationofresourcesbothwithintheruralsectorandacrossrural–urbanregions.Whilepastreformshavegreatlyimprovedallocativeefficiencywithintheruraleconomy,Chinaisstillfacinglong-term,arduousstructuraladjustmentsacrossthesectors.21Mobilityofresourceswillbeakeyaspectofthisprocess;consequentlyruralschoolingwillhaveahighvalueduringthetransition.22Inviewofthefactthattheconditionsofmanyruralschoolshavedeterioratedduringrecentorganizationalchanges,publicattentionandinvestmentininfrastructureisimperative.Toalargeextent,thisstudyofChinaalsomirrorstheexperienceofotherdevelopingcountries,inwhichruralpeoplemustfacetheselectionofincomeactivitiesandtheprospectofleavingagriculture.Schoolingplaysacriticalroleintheseallocativedecisionsforraisingincomesinthecurrentperiodaswellasforthefuture.

Attheprovinciallevel,rurallaborsharesinnonfarmactivitiesincreasedfrom12.8%in1986to22.2%in1995,anincreaseof73.18%(SSBb,1996).Inaddition,thegrowthinrealpercapitaincomeofthesampleiscomparablewiththeprovinciallevelgrowth(seeFootnote13).Unfortunately,aggregatestatisticsdonotcontainsufficientinformationforcomputingcapitalutilizationdevotedtoagriculturevs.nonagriculture.

21Forinstance,Johnson(2000)forecaststhattheagriculturallaborforceinChinawilllikelytofallbyasmuchas63%between1997and2030.Therefore,theruralnonfarmsectorandurbanregionswillfacealong-termchallengeofabsorbinglargenumberofworkersreleasedfromagriculture.

22ThisviewiscorroboratedbyZhao(1997)whofindsthatseniorhighschooleducationraisestheaccessibilityofruralpeopletourbanformalemployment.

20D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162159

Acknowledgements

IwouldliketothankBarryNaughton,TiejunWen,ShukaiZhao,XiaodongZhu,conferenceparticipantsinNewResearchonEducationinDevelopingCountriesatStanfordUniversity,andespeciallyAlbertParkandtwoanonymousrefereesforvaluablediscussionsandsuggestions.IamalsogratefultoXianZude,ShengLaiyun,WangPingpingandotherresearchersatRuralSurveyOrganizationofChina’sStateStatisticalBureaufordatasupport,andtotheCenterforResearchonEconomicDevelopmentandPolicyReformatStanfordUniversityforfinancialsupport.

AppendixA

TheRuralHouseholdSurvey(RHS)organizedandcollectedbyChina’sStateStatisticalBureau(SSB)hasbeenthemajorandcontinuedsourceofstatisticalinformationforruralChina.ChenandRavallion(1996)providedetaileddescriptionsaboutthesurveydesign,samplingissues,administration,andcontentsoftheSurvey.ThispaperusesthedatafromtheSichuanprovincecovering1986–1995.Notethatstartingin1993,SSBchangedthequestionnaireforRHS.Variableswereextendedfrom640to1412,coveringadditionalinformation.Thispaperusesvariablesthatcanbefoundinbothsetsofquestionnaires.Thefollowingparagraphsdescribethemethodadoptedforconstructingthetwopaneldatasets,variabledefinitions,andadjustmentsthatarenecessarytomakethedatasuitableforeconometricanalysis.A.1.Constructionofpaneldata

In1986,RHSinSichuanstartedwithaboutfourthousandruralhouseholds.Ineachofthefollowingyears,SSBofficialsreplacedsomehouseholdswithnewsamples.Thegeneralguidelinessuggestedtherotationofabout20%ofthesampleeveryyear,butinpractice,therateofrotationmaydiffersignificantlyfromthegeneralrule.Akeyfeatureoftherotationisthatthereplacedhouseholdandthereplacinghouseholdalwayssharethesamehouseholdcode,whichisa10-digitnumberthatdistinguishestheprovince,thecounty,thevillage,andthehouseholdnumberwithinthevillage.However,noinformationisavailableregardingsamplerotation.In1991,SSBcompletelydroppedtheoldsampleandstartedwithanentirelynewsample.Overtheyears,thesizeofthesamplegrew,reachingapproximately6000householdsin1995.

Mygoalistoconstructtwopanels:the1986–1989panelconsistingofhouseholdsthatweresurveyedforfourconsecutiveyearsduringtheperiod(seeFootnote12forinformationabout1991data)andthe1991–1995panelconsistingofhouseholdsthatstayedinthesurveyforfiveconsecutiveyears.Iusethreesetsofinformationtosortoutthepanels:thehouseholdcode,theageofthehouseholdhead,andtheageofthehead’sspouse.Onayear-to-yearbasis,fortwohouseholdswiththesamefamilycode,Imatchtheageofthehouseholdheadinthefirstyearsamplewiththeageoftheheadinthesecondyearsample.Iftheageofthehead’sspousealsomatcheswiththeageofthespouseinthesecondyearsample,Idesignatethesetwofamiliesasthesamehouseholds.Otherwise,if

160D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

theageofonepairdoesnotmatch,Iconsequentlydropthehousehold.Thismethodisappliedtotheentiresample.Becausetheprobabilityofmatchingtheageoftwodifferentcouplesiscloseto0,giventheadditionalinformationonhouseholdcode,thesortedpanelisthereforeveryreliable.Table1reportsthenumberofhouseholdsavailableforstatisticalanalysisforthetwosubperiods1986–1989and1991–1995.A.2.Incomeandvariablecostsinagriculture

ConsistentwiththestandarddefinitioninChinesestatistics,agriculturalactivitiesincludecropping,animalhusbandry,forestry,fishery,andsidelineproduction.RHSreportsinformationonthesalesandthevariableproductioncostsforeachoftheseactivities.

A.3.Incomeandvariablecostsinnonagriculturalproduction

Nonagriculturalactivitiesincludehandicraft,industry,construction,transportation,commerce,foodretail,servicesandothernonagriculturalactivities.RHSalsoreportsinformationonthesalesandthevariableproductioncostsforeachofthesenonfarmactivities.RHSreportswageinformationseparatelyfromtheaboveself-employedactivities.Forthe1986–1992survey,wageearningsincludeincomes‘‘fromcollectives,economicunions,andlaboremployment.’’Forthe1993–1995survey,laborearningsisthesumoflaborincomesfrom‘‘collectiveorganizations,enterprises,andotherworkunits.’’A.4.Householdprofit

Thisvariableisdefinedasthesumofnetincomesfromagriculturalandnonagriculturalsources.A.5.Land

RHSrecordsthreetypesofland:cultivatedland,mountainousland,andwaterareasforfishponds.Thelandareausedinthispaperisthesumofallthreetypes.A.6.Capital

Thefixedassetsarereportedintheir‘‘originalvalue.’’Agriculturalcapitalincludesdraftanimals,largeandmiddle-sizefarmingtools,andmachineryforcropping,forestry,fisheryandanimalhusbandry.Nonagriculturalcapitalincludesindustrialmachinery,transportequipment,householdsforproductionandothernonagriculturalfixedassets.A.7.Labor

Laborisrecordedaccordingtotheircapacityas‘‘whole’’or‘‘half’’workers.Thepaperusesthesumofthesecapacitymeasurestoapproximatethehouseholdlaborforce.

D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162161

A.8.Schoolingandexperience

Educationisreportedincompletionlevelsratherthanyearsofschooling.Thesixlevelsareilliterate/semiilliterate,primaryschool,middleschool,highschool,technicalschool,andspecializedcolleges.Becausethereareonlyfewworkersbelongingtothetwohighestcategories,Igroupthemwiththehighschoolgraduatesas‘‘highschoolandplus.’’Toapproximateworkexperience,thesixcompletionlevelsarefirstconvertedintoyearsofschoolingwiththedesignationof3,5,8,10,11,and13yearsofeducationforeachcategory.Then,experienceiscomputedas(age-schooling-7).A.9.Leviesonagriculture

Thisitemincludesgovernmentextractionsthroughcompulsoryprocurementandothertypesofagriculturallevies.

References

Anderson,D.,Leiserson,M.W.,1980.Ruralnonfarmemploymentindevelopingcounties.EconomicDevelop-mentandCulturalChange28(2),227–248.

Ash,R.F.,1988.Theevolutionofagriculturalpolicy.ChinaQuarterly116,529–555.

Basu,K.,Narayan,A.,Ravallion,M.,2001.Isliteracysharedwithinhouseholds?TheoryandevidenceforBangladesh.LabourEconomics8(6),649–665.

Chen,S.H.,Ravallion,M.,1996.Dataintransition:assessingrurallivingstandardinsouthernChina.ChinaEconomicReview7(1),23–56.

Fan,S.G.,Pardey,P.G.,1997.Research,productivity,andoutputgrowthinChineseagriculture.JournalofDevelopmentEconomics53,115–137.

Findlay,C.,Watson,A.,Wu,H.X.,1994.RuralEnterprisesinChinaSt.Martin’sPress,NewYork.

Fleisher,B.,Dong,K.Y.,Liu,Y.H.,1996.Education,enterpriseorganization,andproductivityintheChinesepaperindustry.EconomicDevelopmentandCulturalChange44(2),571–587.

Foster,A.,Rosenzweig,M.,1996.Technicalchangeandhumancapitalreturnsandinvestments:evidencefromtheGreenRevolution.AmericanEconomicReview86(4),931–953.

Huang,J.K.,Rozelle,S.,1996.Technologicalchange:rediscoveringtheengineofproductivitygrowthinChina’sruraleconomy.JournalofDevelopmentEconomics49,337–369.

Huffman,W.E.,1977.Allocativeefficiency:theroleofhumancapital.QuarterlyJournalofEconomics91,59–77.

Jamison,D.T.,Lau,L.J.,1982.FarmerEducationandFarmEfficiency.JohnsHopkinsUniv.Press,Baltimore.Johnson,D.G.,2000.AgriculturaladjustmentinChina:problemsandprospects.PopulationandDevelopmentReview26(2),319–334.

Li,T.Y.,Zhang,J.S.,1998.ReturnstoeducationundercollectiveandhouseholdfarminginChina.JournalofDevelopmentEconomics56,307–335.

Lin,J.Y.,1992.RuralreformsandagriculturalgrowthinChina.AmericanEconomicReview82,34–51.

McMillan,J.W.,Zhu,L.J.,1988.TheimpactofChina’seconomicreformsonagriculturalproductivitygrowth.JournalofPoliticalEconomy97,781–807.

Naughton,B.,1996.GrowingOutofthePlan:ChineseEconomicReform1978–1993.CambridgeUniv.Press,Cambridge,UK.

Orazem,P.F.,Vodopivec,M.,1995.Winnerandlosersintransition:returnstoeducation,experience,andgenderinSlovenia.WorldBankDevelopmentReview9(2),201–230.

162D.TaoYang/JournalofDevelopmentEconomics74(2004)137–162

Pitt,M.M.,Sumodiningrat,G.,1991.Risk,schoolingandthechoiceofseedtechnologyindevelopingcountries:ameta-profitfunctionapproach.InternationalEconomicReview32,457–473.

Putterman,L.,1993.ContinuityandChangeinChina’sRuralDevelopment:CollectiveandReformErasinPerspective.OxfordUniv.Press,NewYork.

Rosenzweig,M.R.,1995.Whyaretherereturnstoeducation?AmericanEconomicReview85,153–158.Schultz,T.W.,1975.Thevalueofabilitytodealwithdisequilibria.JournalofEconomicLiterature13,827–846.Schultz,T.P.,1988.Educationinvestmentandreturns.In:Chenery,H.,Srinivasan,T.N.(Eds.),HandbookofDevelopmentEconomics,vol.1.North-Holland,NewYork.

Sicular,T.,1988.Agriculturalplanningandpricinginthepost-Maoperiod.ChinaQuarterly116,671–705.SSBa,StateStatisticalBureauofChina,1988–1998.StatisticalYearbookofChina.ChinaStatisticalPublisher,Beijing.

SSBb,StateStatisticalBureauofChina,1996.ChinaRegionalEconomy:AProfileofYearsofReformsandOpening-Up.ChinaStatisticalPublisher,Beijing.

Strauss,J.,Thomas,D.,1995.Humanresources:empiricalmodelingofhouseholdandfamilydecisions.In:Srinivasan,T.N.,Behrman,J.R.(Eds.),HandbookofDevelopmentEconomics,vol.3.North-Holland,Amsterdam,pp.1883–2005.

Taylor,J.E.,Yunez-Naude,A.,2000(May).Thereturnsfromschoolinginadiversifiedruraleconomy.AmericanJournalofAgriculturalEconomics82,287–297.

Welch,F.,1970.Educationinproduction.JournalofPoliticalEconomy38,35–59.

Wen,T.J.,2000.RuralChina:AStudyoftheBasicEconomicSystem(inChinese).ChinaEconomicPress,Beijing.

Yang,D.T.,1997.Educationinproduction:measuringlaborqualityandmanagement.AmericanJournalofAgriculturalEconomics79,764–772.

Yang,D.T.,An,M.Y.,2002.Humancapital,entrepreneurship,andfarmhouseholdearnings.JournalofDevel-opmentEconomics68(1),65–88.

Yang,D.T.,Zhou,H.,1999.Rural-urbandisparityandsectorallaborallocationinChina.JournalofDevelopmentStudies35(3),105–133.

Yao,Y.,1999.RuralindustryandlabormarketintegrationineasternChina.JournalofDevelopmentEconomics59,463–496.

Zhao,Y.H.,1997.LabormigrationandreturnstoruraleducationinChina.AmericanJournalofAgriculturalEconomics79,1278–1287.