您好,欢迎来到吉趣旅游网。
搜索
您的当前位置:首页2014 Institutional determinants of vertical integration in China

2014 Institutional determinants of vertical integration in China

来源:吉趣旅游网
CORFIN-00810;NoofPages16

JournalofCorporateFinancexxx(2014)xxx–xxx

ContentslistsavailableatScienceDirect

JournalofCorporateFinance

journalhomepage:www.elsevier.com/locate/jcorpfin

InstitutionaldeterminantsofverticalintegrationinChina☆

JosephP.H.Fana,1,JunHuangb,⁎,RandallMorckc,2,BernardYeungd,3abcdCUHKBusinessSchoolandInstituteofEconomics&Finance,TheChineseUniversityof,,ChinaInstituteofAccountingandFinance,ShanghaiUniversityofFinanceandEconomics,Shanghai,ChinaUniversityofAlberta,NationalBureauofEconomicResearch,CanadaNationalUniversityofSingaporeBusinessSchool,Singapore

articleinfoabstract

Wherelegalsystemsandmarketforcesenforcecontractsinadequately,verticalintegrationcancircumventthesetransactiondifficulties.But,suchenvironmentsoftenalsofeaturehighlyinterventionistgovernment,andevencorruption.Verticalintegrationmightthenenhancereturnstopoliticalrent-seekingaimedatsecuringandextendingmarketpower.Chinaoffersasuitablebackgroundforempiricalexaminationoftheseissuesbecauseherlegalandmarketinstitutionsaregenerallyweak,butnonethelessexhibitsubstantialprovince-levelvariation.WereportthatChinesefirmsinthe2000'saremoreverticallyintegratedthantheU.S.firmsinthe1990s.Wefindthatverticalintegrationismorecommonwherelegalinstitutionsareweakerandwhereregionalgovernmentsareoflowerqualityormoreinterventionist.Further,firmsledbyinsiderswithpoliticalconnectionaremorelikelytobeverticallyintegrated.Finally,verticalintegrationamongpoliticallyunconnectedfirmsisassociatedwithelevatedpercapitaGDPlevelandgrowth,whileverticalintegrationamongpoliticallyconnectedfirmsisunrelatedtolocaleconomyperformance.

©2014ElsevierB.V.Allrightsreserved.

Articlehistory:

Received4December2013

Receivedinrevisedform23May2014Accepted25May2014AvailableonlinexxxxJELclassification:L22P14G38P16

Keywords:

VerticalintegrationRentseeking

InstitutionaldevelopmentGovernment

1.Introduction

Thispaperexplorestherelationshipofverticalintegrationtoinstitutionaldevelopmentanditsimpactoneconomyperformance.Wedefineverticalintegrationascommoncontroloveradjacentproductionstages.Inaneconomywithperfectmarkets–thatis,withnotransactioncosts–verticalintegrationisintrinsicallyinefficientbecauseitpreventsspecializationand

☆WearegratefulforhelpfulcommentsfromPedroDalBó,ZhiwuChen,KaiLi,HaroldMulherin,JoanneOxley,IvanPng,PabloSpiller,TracyWang,LarryWhiteandotherparticipantsinseminarsandconferencesattheAmericanFinanceAssociation,theChinaInternationalConferenceinFinance,HarvardBusinessSchool,theHassSchoolatUCBerkeley,theNationalBureauofEconomicResearch'sChinaWorkingGroupWorkshop,NewYorkUniversity,andtheUniversityofCopenhagen.JosephFangratefullyacknowledgesfinancialsupportfromtheInstituteofEconomics&FinanceofCUHK.JunHuangthankstheNationalNaturalScienceFoundationofChina(No.71102136,No.71272008andNo.71372038),theMOEProjectforKeyResearchInstitutesofHumanitiesandSocialScienceinUniversities(No.11JJD790008),theShanghaiPhilosophyandSocialScienceFoundation(2013BGL008),theInnovationProgramoftheShanghaiMunicipalEducationCommission(14ZS078),andtheProgramforInnovativeResearchTeamofShanghaiUniversityofFinanceandEconomics.RandallMorckthankstheSSHRCandtheBankofCanadaforpartialfunding.BernardYeungacknowledgespartialfundingfromtheLallySchoolofManagementandTechnologyatRensselaerUniversity,wherehewasavisitingscholarinJan2008.

⁎Correspondingauthorat:InstituteofAccountingandFinance,ShanghaiUniversityofFinanceandEconomics,No.777GuodingRoad,YangpuDistrict,Shanghai200433,China.Tel.:+862165904822.

E-mailaddresses:pjfan@cuhk.edu.hk(J.P.H.Fan),huangjun@mail.shufe.edu.cn(J.Huang),randall.morck@ualberta.ca(R.Morck),byeung@nus.edu.sg(B.Yeung).1Tel.:+85239437839.2Tel.:+17804925683.3Tel.:+6565163015.

http://dx.doi.org/10.1016/j.jcorpfin.2014.05.0130929-1199/©2014ElsevierB.V.Allrightsreserved.

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

2J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

createsintra-organizationalcomplexityandpolitics(MilgromandRoberts,1988).Inaddition,integratingverticallyrelatedbusinessesunderthecommonownershipinducesorganizationalcosts,includingthehighercostsofmonitoringworkersandmeasuringperformanceinthemorecomplexorganizations(AlchianandDemsetz,1972;Klein,2014).However,avarietyofconvincingreasonscanexplainverticalintegrationinrealeconomies.

First,transactioncostsindealingsbetweensuppliersandcustomerscaninduceverticalintegration.Wellknowninformationasymmetriesandagencyproblemscaninduceanticipatableopportunisticbehavior,andsodistortinvestment.Iflegalinstitutionsandmarketforcesaretoodysfunctionaltoletfirmscontractaroundsuchtransactionsproblems,verticalintegrationisaplausiblesolution(KhannaandPalepu,1999;Kleinetal.,1978;Williamson,1975).

Second,weaklegalinstitutionsandmarketdisciplinaryforcescanreflectunderlyingpoliticaleconomyinfirmities–suchasexcessiveregulatoryburdens,politicalinterference,orcorruption–thatcanhindermarkettransactions.Verticalintegrationcanbearationalresponsetocircumventsuchbureaucraticimpedimentsandsecure,e.g.,reliablefactorsupplies(Stigler,1951).Butthisneednotendthestory,forsucheconomiesinducepoliticalrentseeking(Krueger,1974).Ifregulatoryburdens,politicalinterference,orevencorruptionbecomeimportantfactorsinday-to-daybusiness,firmsinvestresourcesininfluencingthemby“lawyeringup”,lobbying,orevenbribinginfluentialpublicofficials.Oncefirmsacquiretheabilitytoinfluencegovernmentofficials,lobbyingforregulatoryorotherbarrierstocompetitionisanaturaluseofsuchinfluence(Tullock,1965).Verticalintegrationreflectstheexpansionofarent-seeker'sboundaries.Inaddition,averticallyintegratedmonopolyisunambiguouslymoreprofitablethanasequenceofverticallyrelatedsingle-marketmonopolies(Spengler,1950).

Empiricalstudiesofverticalintegrationpredominantlyexamineagivenindustryinagivendevelopedeconomyforverticalintegrationasaresponsetoexpectedopportunisticbehavior(Chipty,2001;Fan,2000;Joskow,1987;Masten,1984;MonteverdeandTeece,1982;Mulherin,1986;Ohanian,1994;Teece,1976andothers).4Relativelyfewstudiestestforalinkbetweenverticalintegrationandinstitutionaldevelopmentorexaminetheimpactofverticalintegrationoneconomy-levelperformanceinthecontextofadevelopingeconomy.Yetthesearepreciselywherelegalsystemsandmarketforcesareknowntobeespeciallyproblematic(DeSoto,2000).Studiesofverticalintegrationinsucheconomies–wherecontractingsolutionsareunreliable,politicalrent-seekingisrife,andcorporategovernanceislargelyunsupervisedbyshareholders–shouldprovidethemostpowerfulevidenceofitseconomicfunctions.5AnotableexceptionisAcemogluetal.(2009),whousecross-countrydatatoshowthatverticalintegrationissignificantlygreaterincountrieswithhighercontractingcostsyetgreaterfinancialmarketdevelopment.Cross-countrystudies,thoughimmenselyvaluable,areunavoidablyvulnerabletoomittedvariableproblems,aswellasdatacomparabilityandmeasurementproblems.6Forexample,innumerablelatentfactorslinkedtoculture,history,politics,orevenlanguagecanaffecthumanbehaviorandhencetheorganizationoftheeconomyaswellasinstitutionaldevelopment.

Studyingprovince-leveldatafromChinamitigatesthesecriticisms,atleasttosomeextent;forthesefactorsarelargelycommonacrossprovinces.Duetovariationinproximitytotheoutsideworldandinpre-liberalizationconditions,differentprovincesstandatsubstantiallydifferentlevelsofinstitutionaldevelopment.ByWesternstandards,inthe1990sand2000'slegalsystemsinmanypartsofChinaseemsclerotic,politicalrent-seekinglooksrampant,andstatecontrolseemspervasive(Allenetal.,2005;CheandQian,1998;CullandXu,2005;Fanetal.,2007;Montinolaetal.,1995).Buttheseproblemsaremuchworseinsomeprovincesthaninothers(QianandWeingast,1996,1997).Againstacommonculturalbackground,thisvariationininstitutionaldevelopmentprovidesausefullaboratoryforexploringhowbusinessesorganizethemselvesinresponsetotheirinstitutionalenvironments.

WeuseChina'sinput–output(IO)tabletomeasuretheprevalenceofverticalintegrationamongpubliclylistedfirms(excludingfinancialfirms)ineachofChina'svariousprovinces,specialadministrativeregions,andautonomousregions—whichwecollectivelycall“provinces”forbrevity.7Controllingforpotentialassetspecificity(Kleinetal.,1978;Williamson,1979)andotherplausiblefactors,wefindverticalintegrationmorecommoninregionswithweakerlegalsystems,worselocalgovernment,andlessdevelopedmarketeconomies.Weinterpretthisasconsistentwithverticalintegrationasacorrectivemeasuretocircumventinstitutionallacunaandamanifestationofpoliticalrentseeking.

Frombiographicalinformation,wedefineeachfirm'stopexecutives,whomwecallCEOsregardlessoftheirChinesetitles,asapparatchiksiftheyarenoworhaveeverbeenCommunistPartyorstateofficials.Wefindapparatchik-runfirmstobeunusuallypronetoverticalintegration.Alternatively,wemeasurepoliticalconnectionsasaccesstotheprivilegeofindustryentryandtobankloans,bothofwhicharedefactorationed.Thesefurtheranalysesagainshowthatmorepoliticallyconnectedfirmsaremoreverticallyintegrated.Theseresultsareconsistentwithatleastsomeverticalintegrationarisingtoexpandreturnstopoliticalrent-seeking.

SeeLafontaineandSlade(2007)forasurveyoftheliterature.

Parallelconceptsareinvestigatedintheforeigndirectinvestmentliterature,whichfindsthatmultinationalfirmspreferfullownership(i.e.,integration)oftheirforeignsubsidiarieswheretransactiondifficultiesintrinsictothenatureofthebusinessarehigh(e.g.,Henisz,2000).Unfortunately,suchstudiesshedscantlightonintegrationasaresponsetoweaklegalormarketinstitutionsbecauseconfoundinginterpretationsarise.Forexample,foreignanddomesticfirmshavedifferentcapabilitiesfordealingwithawiderangeofinstitutionalinfirmitiesbeyondweaklegalandmarketinstitutions.Thesedifferencesconstraintheirrolesasacquirersandacquired(seee.g.FeenstraandHanson,2005).6Forexample,industriesthatareverticallyrelatedintheUnitedStatesmightnotbeelsewhere,wheredifferenttechnologiesrequiredifferentinputs.Also,contractingcostmeasure,tobecomparable,mustbesimple—forexample“thenumberofstepsrequiredincollectingadebt.”Yet,asimilarstepmightbemoreonerousinonelegalsystemthananother.Acemogluetal.(2009)findmeaningfulresultsdespitesuchchallenges.7ThismethodologywasdevelopedbyFanandLang(2000),andusedbyShahrur(2005),FanandGoyal(2006),Acemogluetal.(2009),andAcemogluetal.(2010).

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx3

Finally,weareinterestedintherelationshipbetweenverticalintegrationandregionaleconomydevelopment.Ifverticalintegrationisasecondbestsolutiontotransactiondifficultiesunderweakinstitutions,itwouldcorrelatewithimprovedeconomyperformance.Onthecontrary,ifverticalintegrationisforrentseeking,itsimpactoneconomicperformanceislesscertain.Weexaminethisbyaveragingtheextentofverticalintegrationacrossconnectedandunconnectedfirmsbasedinagivenprovincetoconstructprovince-levelmeasuresoftheprevalenceofverticalintegration.TheresultsshowthatmoreprevalentverticalintegrationaccompanieshigherprovincialpercapitaGDPlevelsandgrowthratesinprovinceswhoseverticallyintegratedfirmsarepredominantlyrunbynon-apparatchiks.Incontrast,widespreadverticalintegrationisnotsignificantlyrelatedtopercapitaGDPlevelsorgrowthratesinprovinceswhoseverticallyintegratedfirmsarepredominantlyapparatchik-run.Totheextentthatouranalysisissubjecttocaveatslikeomittedvariablesandalternativeinterpretations,thefindingsherearetentative.

Thepaperproceedsasfollows.Section2developsourhypotheses.Section3describesthesample,discussesempiricalmeasures,andprovidesdescriptivestatistics.OurmainresultsarereportedinSection3.4.3.Section4concludes.2.Theeconomicsofverticalintegration

Byverticalintegration,wemeantheintegrationofactivitiesbelongingtodistinctlyidentifiableseparatebutverticallyrelatedindustries.Toillustrate,considerWeiqiaoGroup,thelargesttextilecompanyinChinainthe2000s.Thegroup'sprimarybusinessisproducingtextiles.Butitalsogrowsitsowncotton,makesthecottonintoyarn,weavesandfabricatestheyarnintotextiles,imprintsthetextileswithcolorsbeforesellingthemtocustomers,andoperatesitsownelectricityplanttopoweritsotheroperations.WeiqiaoexemplifiesadegreeofverticalintegrationcommonplaceinChina,butrareindevelopedmarketeconomies.Whatarelikelyexplanationsandeconomicimplications?2.1.Transactioncostsandverticalintegration

Markettransactionsbetweenseparatelyrunverticallyrelatedfirmscanbecostly.Coase(1937)positsthattransactionsoccurwithinafirm(verticalintegration)ifthecostofarm'slengthtransactionsbetweenspecializedfirmsexceedsthatofcoordinatingmultipleactivitieswithinonefirm.Coase'sinsightbegataninfluentialliteratureonthecostsoftransactionsbetweenverticallyrelatedbusinesses(e.g.,Kleinetal.,1978;Lucas,1978;andWilliamson,1973,1975).

Kleinetal.(1978)andWilliamson(1979)pointoutthatanticipatedopportunisticbehaviorstemmingfromassetspecificity,lowtransactionfrequency,anduncertaintyassociatedwiththetransactionsinquestionleadstoinefficiency.Considertwoindependentunitsengaginginupstreamanddownstreamproduction,eachmakingaspecializedinvestmentinperiodonetoprepareforproductioninperiodtwo.Onceaninvestmentismade,theresultingassetcannotbeusedforanotherpurposewithoutasubstantiallossinitsvalue.Thiscostofadaptingtheassettootheruses,calledassetspecificity,caninducetheothertransactingpartiestoextractrentsexpost(Kleinetal.,1978).InthecontextofChina,forexample,onemanifestationofsuchbehaviorisonlowsupplyofqualitythatmakesithardtosafeguardexposttroublestoverticallyrelatedfirms.Thissortofopportunisticbehaviorispossiblebecauseofinformationasymmetryaroundthetransaction—eitheragenuinelyexogenousunanticipatedchangeoracalculatedrent-extraction.Thisforeseeablepossibilityinducestime-inconsistentbehaviorthatreducestheexpectedreturntospecializedinvestment,andthereforecurtailsit,resultinginadeadweightlosstotheeconomy.Anothersourceofinefficiencyarisesbecausethevalueofoneverticallyrelatedbusinessdependsontheother—onitseffortlevelandflexibilityaswellasitswillingnesstoshareinformationandtocoordinateemploymentandinvestmentstrategies.

Iftheseinefficienciesarelarge,integratingtheverticallyrelatedfirmsintoasinglecompanymaybelesscostlythanmarkettransactionsbetweenseparatefirms(Coase,1937).Sincethecostsofarm'slengthtransactionsareduetoweaklegalsystemsrenderingcontractsunenforceable,andpropertyrightsthereforeunprotected,reorganizingthebusinessesunderacommonowneralleviatestheseproblems(Alchian,1965).

Assetspecificity,wherebyassetsinplacecannotbereassignedtootheruses,anduncertaintyaboutthevalueofcontractsarefundamentaldeterminantsoftransactioncosts.Assetspecificityoftendependsonaccesstoalternativesuppliersandcustomers.Forexample,afactorybuiltinacitywithinadequateroadandraillinkstoothercitiescanbeatthemercyoflocalsuppliersandcustomerswhodemandchangesinthetermsofcontracts.Butafactoryinacitywithefficienttransportationlinkstoothercitiescandobusinesswithmoredistantsuppliersorcustomersiflocalonesbecometooirksome.Ofcourse,iflocalcustomersandsupplierscanbetrustedtoperformastheyhaveagreed,efficienttransportationlinksarelesscritical(see,e.g.,Joskow,1987).

Thus,ourstartingpointisthatafirmislikelytobemoreverticallyintegrated,ceterisparibus,ifitstransactionsaresubjecttohighperformanceuncertaintyanditislocatedinaregionwithpoortransportationinfrastructure.2.2.Institutionaldeterminantsofverticalintegration

Whileintegrationintoasinglefirmcouldbeasolutiontothesetransactiondifficulties,itsdrawbackisthatextensiveverticalintegrationrenouncesthebenefitsofspecialization(Smith,1776),dilutingmanagement'sfocusandmonitoringincentive(Klein,2014),andeveninducing“territorial”politicalconflictswithinthecompany(MilgromandRoberts,1988).

2.2.1.Legalsystem

Manytransactiondifficultiescanbeavoidediftheperformanceoflocalcustomersandsupplierscanbeguaranteedwitheffectivelegalcontractsthatcanbeenforcedquickly,cheaply,andreliably.Contractsthatclearlylayouteachparty'srightsandPleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

4J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

obligationsineachsetofcircumstancescansubstantiallyreducethefeasibilityandgainsfromexpostbargainingtoextractrents.Reliablecontractingcanalsoreduceshirking,stipulateeffort,governinformationsharing,andpredeterminedegreesofcooperation.Carefullydraftedlegalcontractscanpreciselystipulatelegaldevelopmentandprohibittheirinfringement.Theseconsiderationssuggestthatverticalintegrationshouldbemoreextensiveinfirmslocatedinregionswithweakerlegalinstitutions.2.2.2.Marketforces

Marketforcescanalsodisciplineopportunisticbehavior,renderingintegrationunnecessary.Highlyvisibletransactionsprovidemarketparticipantinformationaboutthebehavioroffirmsandtheirinsiders.Importunateshirking,inadequateperformance,anddisdainforlegaldevelopmentallengenderbadreputationsandwarnawayfuturepotentialbusinesspartners.Acquiringareputationasanopportunist,whorenegesonpromisesandmanipulatesbusinesspartners,canbecomeaseriousliability.Inafreemarket,wherepeoplecanchoosetheirbusinesspartners,anyfirmthatplanstoremaina“going-concern”choosestobehavereliablyandhonorably(KleinandLeffler,1981).

Extensivegovernmentinterventioncanweakenmarketforcesbyforcingfirmstotransactwithpoliticallyfavoredfirms.Forexample,heavy-handedregulationinChinadetersentryintotheelectricitygenerationbusinesstoprotectState-ownedenterprises.Thesesporadicallyrenegeoncommitmentstosupplyelectricity,apracticequiteprevalentinthe1990sandnotunknowninthe2000s—perhapsbecauseofsimpleincompetence,orperhapsinattemptstoextractbribesforreliablepowersupplies.Sincenoalternativepowersuppliersarepermitted,electricityusersmustacceptbehaviorthatcouldnotpersistinanopenandcompetitivemarketforelectricity.

Verticalintegrationcanagainprovideawayout.Thetextilesbusinessdiscussedabove,Weiqiao,optedforverticalintegrationbyconstructinganin-houseelectricityplant.ThecheapesttechnologyoperatedonascalelargerthanWeiqiaorequired,sothefirmfounditselfwithmoreelectricitythanitneeded.Forbiddenfromsellingpower,Weiqiaobeganrefiningaluminumtomakeprofitableuseofitsexcesselectricity.

AluminumsmeltingisnotobviouslyrelatedtoWeiqiao'scoretextileoperations,butrequiresvastamountsofelectricity.Sequentialverticalintegrationofthissortcanlooklikeunrelateddiversification,butisnot.Again,thestate'ssuppressionofmarketforcesisresponsible(Alchian,1965).Suchmulti-stageverticalintegrationwouldmakenosensehadWeiqiaobeenabletosellitssurpluselectricity.

IntheUnitedStates,theseconsiderationsareprimarilyofhistoricalinterest.Forexample,theUnitedStatesretainedwartimepricecontrolsonmanygoodsintotheearlypostWorldWarIIera.Peacetimeeconomicfundamentalsleftsomeofthesepricestoolowtocoverproductioncosts,andcausedfirmstoceaseproducingtheprice-controlledgoods.Thisleftdownstreamusersoftheprice-controlledgoodscontendingwithextremeshortagesofcriticalinputs.Sinceofferinghigherpricesinarm'slengthtransactionswasillegal,customerfirmsmergedwiththeirsuppliers(Stigler,1951).Oneunitoftheverticallyintegratedfirmcouldthencompensatetheotherforitscostswithoutviolatingthepricecontrollaws,sincenoactual“sale”occurredatanyspecifiable“price”.

Butin21stcenturyChina,thesesameconsiderationsarise.Heavy-handedregulationinducesverticalintegrationbothbyunderminingmarketforcesthatotherwisewouldencouragemarkettransactionsbetweenverticallyrelatedfirmsandbyinducingdistortionsthatdisruptintermediategoodsmarkets.Hence,ourthirdhypothesisisthatafirmismorelikelytoadoptaverticalintegrationstrategywheremarketsareheavilyregulatedand/orunderdeveloped.

2.2.3.Theothersideofthecoin

OurconsiderationwillbeincompletewithoutpayingattentiontothecontributioninAlchianandDemsetz(1972)whicharguesthatjointteamproductionfacesameteringproblem:itreferstothechallengeinmeasuringeachinput'scontributionandthusappropriatelyincentivizingandrewardingeachinjointproduction.Anowner–managercanorganizeandmonitortheworkersinafirm,thereforedirectlyhandlingthemeteringproblemandimprovingteamproductivity.Klein(2014)linksthistothetransactioncosttheoryofthefirmandpointsoutthatthemeteringandmonitoringproblemisgreaterthemorecomplicatedisafirm.Verticalintegrationcertainlyraisesafirm'scomplexityandthusthepossibilityofinefficientproduction.

Inotherwords,theconcernsraisedinKleinetal.(1978)areaboutthecostsofusingthemarketforverticallyrelatedactivitiesstemmingfromthebargainingandcoordinationproblemswhichinturnstemmingfromindividualfactorowners'tendencytoextractquasi-rentsandshirk.AlchianandDemsetz(1972)focusontheroleofanentrepreneurandhisfirm,asdefinedinCoase(1937),inovercomingcontractingproblemsarisingfromthemeasurementissue.

However,theentrepreneur'smonitoringcapacityislimitedintherealworld,whichinturnlimitsthesizeofafirm.Moreover,theentrepreneur'sincentiveisalsolimited.IntheAlchianandDemsetz(1972)discussion,theentrepreneurisaresidualclaimant,amanager–ownerwhoimposesnoagencyproblemsonherself.However,inreality,wheremanagementandownershipareoftenseparate,themanager'sobjectivecandeviatefromthatofdiffusedresidualclaimants,inducingcostsofincentivealignment(JensenandMeckling,1976).

InthecontextofChina,SOEmanagers'aresurelynotownersofthefirms,andthereforesubjecttotheincentivealignmentissueandassociatedagencycosts.Likewise,privatelyownedfirmsmayhaveaconcentratedcontrolproblemleadingtoresourceextractionforprivatebenefits.Moreover,themarketandlegalinstitutionsinthecountryaregenerallyweak.Althoughtheweaklegalinstitutionsandmarketforcesraisethecostofusingthemarketandthereforeincreasethebenefitofintegrationbetweenverticallyrelatedbusinesses,theyalsoraisethecostofusinginternalorganizationasimpliedbytheanalysisinAlichianandDemsetz(1972)andKlein(2014).WhethertheChinesefirmsbecomemoreorlessverticallyintegratedinresponsetothecountry'sweakinstitutionsseemstobeanempiricalissue.Addingonemorelayerofcomplexity,ChinaissubjecttounderdevelopedcorporatePleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx5

governanceandtheseparationofgovernmentandbusinessisfarfromcomplete.Thereforeverticalintegrationdecisionsofthesefirmscouldbeaffectedbynon-valuepursuingmanagerswhichthenextsectionturnsto.

2.2.4.Politicalconnection

OnerousregulationslimitingtherighttooperateabusinessinChinaleadtoseveralveryimportantadditionalconsiderations.Bureaucrats'powerstoallocatetheserights,andtointerfereinbusinesses'operations,fosteraspecializedclassofrent-seekingfirms,whichgainbusinessopportunitiesbytradingfavorswithbureaucratsandsustaintheircompetitiveadvantagethroughcorporateinsiderswhoonceservedasbureaucrats.

Theserent-seekingfirmsoftenusetheirpoliticalinfluencetoobtainlocalizedstate-enforcedmonopolies.Theywouldthenuseintegrationtoextendthescopeofmonopolypowerbyexpandingthemonopolist'smarketpowerupstreamanddownstream.Thatis,averticallyintegratedfirmmightmagnifythereturnstoitsinvestmentinpoliticalcloutbyforeclosingcompetitioninthoseindustriesaswell.

Suchlocalmonopolistswouldalsooptforverticalintegrationtoavoid“doublemarginalization”(Spengler,1950).Doublemarginalizationariseswhereverticallyrelatedmonopolistsfailtointernalizetheimplicationofone'shighpriceontheother'sprofits,andhencecollectlessmonopolyrentsthaniftheycoordinatetheiractiontomaximizetheirjointprofits.Contractualarrangementsmightachievesuchcoordination,sointegrationisparticularlyappealingwherecontractingoptionsarelimited(Cabral,2000).

StateandPartyofficialsmightwelcomesuchexpandedmonopolies.Corruptofficialsmightcooperateintheirestablishmentforashareofthemonopolyrentscreated,buthonestofficialsmightalsofindthatdealingwithasinglefirmsimplifiessocialengineeringnegotiationswiththebusinesssector.

Theseconsiderationsmotivateourfourthhypothesis:themoredeeplyafirmisconnectedwithofficialbureaucrats,themoreverticallyintegratedisthefirm'sbusinessstructure.

2.2.5.Goodgovernment

Atamoregenerallevel,regulationpersemightwellmatterlessthantheoverallqualityofgovernment.Agovernmentmorerifewithbureaucrats'intentonextractingrents,collectingbribes,orevenexplicitlyexpropriatingprivatepropertywouldmotivatefirmstoinvestmoreheavilyinofficialconnectionsmerelytocope.Moreoversuchanenvironmentraisesthecostsofmarkettransactions,andthusreinforcesfirms'incentivestointegratevertically.Notonlywouldfirmswithextensiveinvestmentsinofficialconnectionsverticallyintegratetoenhancetheirrent-extraction,butpoliticallyunconnectedfirmswoulddothesametoavoidarm'slengthtransactionsinmarketsrendereddysfunctionalbypredatorygovernmentofficials.

Hence,ourfifthhypothesisisthatfirmsaremorelikelytointegrateverticallyinregionswithlowerqualitygovernment.2.2.6.Summary

WethusexpectafirmtobemoreverticallyintegratedifitsmanagersarepoliticallyconnectedandifitliesinaChinaprovinceorspecialdistrictcharacterizedbypoorqualitygovernment,weaklegalsystems,and/orill-developedmarkets.82.3.Verticalintegrationandeconomyperformance

Overall,thisreasoningsuggeststhatcoordinationand(expost)bargainingissuesdeterundiversifiedfirmsfromforminginverticallyrelatedindustries.Whereweakmarketdisciplinaryforcesanddysfunctionallegalinstitutionsmakearm's-lengthcontractsofuncertainvalue,verticalintegrationcanbeanefficientsecondbestsolution.Indeed,thisinsightunderliesRosenstein-Rodan's(1943)viewthatastate-coordinatedBigPushisanecessaryprecursortotherapidindustrializationofadevelopingeconomy.Eveninadevelopedeconomy,verticalintegrationcanbeaneconomicallyrationalresponsetoextensivegovernmentintervention—forexample,Stigler(1983)attributesAmerica'slate-1940smergerwavetofirmscircumventingthecountry'spolicyofcontinuingwartimepricecontrols.TheseconsiderationssuggestthatverticalintegrationmightalsobegrowthenhancinginChina.

However,itisalsopossiblethatverticalintegrationinChinacouldreflectpoliticallyconnectedfirms'successfulrent-seekingaimedataccumulatingmonopolypower—withunclearimplicationsforeconomicefficiencyandgrowth.Ontheonehand,theverticalintegrationofmultiplemonopolists,eachalreadyexistingbydintofpoliticalconnections,couldenhanceefficiency(ShleiferandVishny,1993).9Ontheotherhand,verticalintegrationaimedatbuildingnewmonopoliesorextendingexistingmonopoliesfurtheralongproductchainsworsensinefficiencyand,bysignifyinghighreturnstopoliticalrent-seeking,coulddiscourageeconomicactivityandretardgrowth(Krueger,1974;Murphy,ShleiferandVishny,1991).

OnepossiblesynthesisoftheseconsiderationsmightpositmoreverticalintegrationbeingcorrelatedwithelevatedpercapitaGDPlevelsandgrowthratesifverticalintegrationislargelyatacticusedbypoliticallyunconnectedfirmsseekingtocopewithweakinstitutions,butwiththecorrelationbeingofuncertainsignifverticalintegrationismorecommonlyassociatedwith

KhannaandOberholzer-Gee(2005)findthattheextentofgovernmentinterferenceineachprovincecorrelateswiththeexistenceoflargefirms.Theyarguethateitherpoliticalinterferenceprotectsincumbentsorincumbentsmustattainacertainsizetocombatinterference.9Theargumentisasfollows:ahighwaywithonlyonetollboothgeneratesmoreefficienttrafficpatternsthanasimilarhighwayfeaturingmultipleindependentlyruntollbooths,witheachtollboothownermaximizinghertollrevenues.Acoordinationprobleminthelattercaseleadseachoperatortochargetoohighatoll,resultingininefficientlylowtraffic.

8Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

6J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

politicallyconnectedfirmsamassingmonopolypower.Ifso,wemightpositthatpercapitaGDPshouldbemorepositivelycorrelatedwithverticalintegrationwherefirmsarelesspoliticallyconnected.3.Data,measurement,andbasicstatistics

WeexaminethevalidityoftheabovehypothesesusingChinaasourempiricalsetting.Becauseofsubstantialvariationinboththestrengthofmarketforcesandthequalityofinstitutionsacrossitsregions,Chinaprovidesanaturaltestinggroundfortheseissues.Thisvariation,againstalessheterogeneousculturalbackground,providesauniqueopportunitytoexaminetherolesofmarketandlegalinstitutionsinfirmverticalintegrationdecisions.AnotheradvantageisthatpubliclyavailableinformationletsusidentifywhichChinesefirmsreceivepreferentialtreatmentorare‘connected’withgovernmentofficials.

Thefollowingfirstdescribesoursampleandourmeasureofverticalintegration,andthenexplorespatternsoffirmverticalintegration.3.1.Sample

OursampleincludesmostcompanieslistedontheShanghaiandShenzhenstockexchangesfrom2002to2010.TheChinaSecuritiesRegulatoryCommissionrequiresthatlistedcompaniesdisclosesegmentinformationforallbusinesssegmentscomprisingmorethan10%ofconsolidatedsales,assetsorprofits.Disclosedinformationaboutagivensegmenttypicallyincludesanindustryname,adescriptionofproductsorservices,andsegmentsales.Wemanuallycollectthesedatafromannualreportsstartingin2002,sincefromthatyearoncoverageandreportingqualitysubstantiallyimprove.

Companiesreportingnon-positivesalesorincompletesegmentsalesandindustrysectorinformationaredropped.Financialfirmsareexcludedbecausetheirfinancialstatementsarenotcomparabletothoseofotherfirms.Ourfinalsampleconsistsof1912firmsand11,563firm-yearobservations.Thesamplefirmsaccountforalmost80%oftotallistedcompanies.

Wedefineafirm'sprimaryindustryastheindustryinwhichithasthelargestsales.Ourfirms'primaryindustriesspanthewholeeconomy,withthemostcommonbeingthemachinery,equipment,andinstrumentsector,followedbytheglass,mineralandmetalsector,thepetroleumandchemicalsector,andthecommerce(trade)sector.Table1describesthesamplebyyearandindustry.

Table1

Samplebyyearandindustry.Oursampleconsistsofnon-financiallistedcompaniesinShanghaiandShenzhenStockExchangesfrom2002to2010.Allsuchfirmsareincludedexceptcompaniesreportingnon-positivesalesorincompletesegmentsalesandindustrysectorinformation.PanelA:SamplebyyearYear200220032004200520062007200820092010Total

PanelB:SamplebyindustryIndustry

Agriculture,forestryandfishingMining

Foodandbeverages

Textile,apparelandleather

Lumber,furniture,paperandprinting

Petroleum,chemicals,rubberandplasticproductsGlass,mineralsandmetals

Machinery,equipmentandinstrumentMedicineandbiologicalproductsUtility

ConstructionTransportationCommerceRealestateServicesTotal

Obs.156180042033712551336255968552426347212768511,563

Percentage1.3%1.6%4.7%3.6%2.9%10.9%11.6%22.1%5.9%4.5%2.3%4.1%10.8%7.8%5.9%100%

Obs.9651048110912231208131414871457175211,563

Percentage8.3%9.1%9.6%10.6%10.4%11.4%12.9%12.6%15.2%100%

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx7

3.2.Verticalintegrationproxy

Giventhatdirectfirm-levelverticalintegrationdataareunavailable,wemodifythemethodologyofFanandLang(2000)touseChina'sinput–outputmatrixandourfirm-leveldata,describedabove,toconstructafirm-levelproxyforverticalintegrationintensity.Theconstructionofthisverticalintegrationmeasureinvolvestwosteps.

Inthefirststep,wecreateinter-industryverticalrelatednesscoefficients.Thisinvolvescomputingthecoefficientsbetweeneachpairof122industriesdefinedinthe2002Chineseinput–outputtable.Thetablereports,foreachpairofindustriesiandj,thevalueofinputsfromindustryiusedbyindustryjasafractionofthevalueofindustryj'stotaloutput.Wedenotethisfractionasvij.Eitherahighvijorahighvji,orboth,suggestsopportunitiestointegrateactivitiesiniandjinthesamefirm.Wethereforedefinetheverticalrelatednesscoefficientbetweenindustriesiandjas

󰀁󰀃

Vij¼maxvij;vji:

½1󰀂

Inthesecondstep,weconstructafirm-levelverticalintegrationmeasurebycomputingtheweightedaverageverticalrelatednesscoefficientsofeachpairofitssegments,excludingsame-segmentpairs.Todothis,weconstructamatrixofindicatorvariables[δi,j]foreachfirm,settingδi,jtooneifthefirmoperatesinbothindustriesiandj,andtozerootherwise.Wethendefinethefirm'sverticalintegrationmeasureas:

nX1X

¼wiVijδi;j

n−1i¼1

j≠i

Vmax½2󰀂

withVi,jdefinedasinEq.(1)andwithwiaweightequaltothesalesofthefirm'ssegmentinindustryi.WedividetheweightedsuminEq.(2)byn−1toaccountforitsmechanicalincreasewiththenumberofsegmentsthefirmcontains.10Foraone-industryfirm(n=1),wesetVmaxtozero.

Anexampleillustrates.HuangshanTourismDevelopmentCo.,Ltd.hasthreesegmentsin2002:touroperation,hotels,andtransportation,accountingfor59%,20%,and21%offirmsales,respectively.Usingtheinput–outputtable,weestimatethattouroperationemploys0.0700ofhotelservicestoproduceoneyuanofoutput,andconverselyhotelsconsume0.0020oftouroperationproductsforeveryyuanofoutputgenerated.Theinter-industryrelatednesscoefficientoftouroperationwithhotels,Vij,isthus0.0700,themaximumofthetwoinputrequirementratios.Similarly,therelatednesscoefficientofhotelswithtransportationis0.0056,andthatoftouroperationwithtransportationis0.0465.Theequationabovethenletsusestimatetheoverallextentofverticalintegrationofthefirm,definedas1/2×[59%×(0.0700+0.0465)+20%×(0.0056+0.0700)+21%×(0.0465+0.0056)],or0.0474.

Inotherwords,afirm'slevelofverticalintegrationisthesales-weightedaverageoftheverticalrelatednesscoefficientsofallpairsofindustriesinwhichthefirmreportssales.

WecalculateVmaxforeachfirmeachyear.Table2summarizestheverticalintegrationmeasure.PanelAshowstheaveragesofVmaxtobe0.0258.Thatis,thefirmspotentiallycouldtransact2.58fen(theChineseunitfor“cent”)ofitsbusinessdealingswithin-houseaffiliates,foreveryyuanofoutputitproduces.Intermsoftimeseries,theaveragedegreeofverticalintegrationisstableovertime.Acrossthe15broadindustries,ourproxyindicatesverticalintegrationtobemostpronouncedinmining,agriculture,forestryandfishing,butislowestinmedicineandbiologicalproducts.

WeareinterestedincomparingtheextentofverticalintegrationbetweenChineseandU.S.firms.Inordertomakethecomparisonwefirstexaminewhetherindustriesinthetwocountriesexhibitsimilardegreesofverticalrelatedness.WefindthatChineseinter-industryverticalrelatednesscoefficientsarehighlycorrelatedwiththosecalculatedfromaU.S.input–outputtable.ThePearsoncoefficientarrivesat0.2andissignificantattheonepercentlevel.Then,wecomparetheverticalintegrationintensityofoursamplefirmswiththatoftheU.S.firmsasreportedinFanandLang(2000).TobeconsistentwiththeFanandLang'sstudy,weredefinetheinter-industryverticalrelatednesscoefficientasVij=1/2(vij+vji)andfocusonprivatelyownedmulti-segmentfirms.ThemeanverticalintegrationintensityfortheChinesefirmsbetween2002and2010is0.0235.FanandLang(2000)reportthattheaverageverticalintegrationintensityofU.S.firmsbetween1979and1997is0.0195.ItappearsthatChinesefirmsinthe2000'saremoreverticallyintegratedthantheU.S.firmsinthe1990s.

Oursisoneoftheseveralpossiblemethodsforconstructingaverticalintegrationindex.Onealternativewoulddistinguishinputintegrationfromoutputintegration.Inputintegration,vij,measuresthefractionofindustryj'sinputsourcedfromindustryi;thatisscalingthe(i,j)entryintheI/Otablebyindustryj'stotalinput.Outputintegrationisthenthefractionofoutputfromindustryiallocatedtoindustryj;thatisscalingthe(i,j)entryintheI/Otablebyindustryi'stotaloutput.Onemightlegitimatelydefineverticalrelatednessineitherway,orastheaverage,minimum,ormaximumoftheinputandoutputmeasures,dependingonthequestionathand.Allthesealternativesgenerateresultsqualitativelysimilartothoseshowninthetables.

ForstudiesusingChineseinput–outputdata,acriticalissueisnoiseduetoendconsumptionandimports.Largeamountofendconsumptionand/orimportswouldinducenoiseintotaloutputwhichisusedasthescalingfactor.Uponconsultingexpertson

Forexample,considerathree-segmentfirmwithone-thirdofitssalesineachindustry.Iftheverticalrelatednesscoefficientsforpairsofthesegmentswereeachone,theweightedsumofeachpairofverticalcoefficientsistwo.Dividingthissumbytwo(threeminusone)rescalestheverticalmeasurebacktoone.Unadjusted,thesumwouldrisewiththenumberofsegments.

10Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

8J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

Table2

Patternsofverticalintegration.Summarystatisticsbyyearandindustryforthefirm-levelverticalintegrationmeasure.FirmsaremoreverticallyintegratediftheirsegmentslieinindustriesthatsourceorsellmoreextensivelytoeachotherinChina'snationalinput–outputtable.PanelA:VerticalintegrationintensitybyyearYear200220032004200520062007200820092010Total

Obs.965104811091223120813141487145717521,1563

Mean0.02440.02510.02630.020.02670.02580.02670.02660.02440.0258

Median0.01250.01310.01310.01310.01310.01310.01210.01160.00970.0124

Std.dev.0.04880.05040.05170.05190.04940.04980.060.05270.04830.0509

Min.0.00000.00000.00000.00000.00000.00000.00000.00000.00000.0000

Max.0.52990.52990.52990.52990.52990.58810.68370.68370.68370.6837

PanelB:VerticalintegrationintensitybyindustryIndustry

Agriculture,forestryandfishingMining

Foodandbeverages

Textile,apparelandleather

Lumber,furniture,paperandprinting

Petroleum,chemicals,rubberandplasticproductsGlass,mineralsandmetals

Machinery,equipmentandinstrumentMedicineandbiologicalproductsUtility

ConstructionTransportationCommerceRealestateServices

Obs.1561800420337125513362560685524263472127684

Mean0.08430.08600.03910.03860.01930.02620.03260.02320.01780.02460.02370.01320.02500.00780.0209

Median0.03460.02990.01430.01660.00780.01620.00650.00720.00150.01750.02060.01020.02580.00410.0126

China'sofficialstatistics,weareconfidentthatourverticalintegrationindicesarelikelytobelesscontaminatedbytheseproblemsandthusreflectdomesticindustrysourcingfromeachother,scaledbydomesticindustrytotaloutput.113.3.Datacaveats

Ourdataareobservationsofindividualfirms.However,verticalintegrationisaccomplishedbyplacingtwoormoredistinctone-industryfirmsundercommoncontrol.Commoncontrolismostreadilyeffectivebycombiningthesefirmsintoone,butcanalsoariseifagroupofseeminglyindependentfirmscomeunderthedirectorindirectcontrolofasingledominantshareholder(Morck,WolfenzonandYeung,2005).12BecausebusinessgroupsarepresentinChina(Fanetal.,2013)andwecannotidentifyverticalintegrationviabusinessgroupsduetodatalimitations,wemaywellunderestimatetheoverallincidenceofverticalintegration.

Ifintegrationviabusinessgroupisthenorm,weshouldfindnodiscernablerelationshipbetweenindividualfirms'verticalintegrationandourvariousothervariables.Ifintegrationoccursthroughbothmechanisms,andifbusinessgroupssimplyaddnoisetoourvariables,anysignificantfindingsbecomeallthemorecredible.

Ofcourse,ifbusinessgroupformationisgovernedbyunknownfactorsthatinteractinauspiciouslywithourvariables,cleareconometricproblemsarise.DealingwiththeseeffectivelywouldrequirethecompletecategorizationofallgrouptiesamongalllistedChinesefirms.Theconstructionofsuchadatasetlieswellbeyondthescopeofthisstudy;however,wehopetoexplorethisissueinsubsequentresearch.3.4.Focalindependentvariables

Thissectiondescribestheindependentvariablesusedtoproxyforpossibleeconomicdeterminantsofverticalintegration.Someoftheseareprovince-levelvariables.ThePeople'sRepublicofChinaisafederalstatecomposedoftwenty-twoprovinces(shěng,or省),fourprovince-levelmunicipalities(zhíxiáshì,or直辖市),fiveautonomousregions(zīzhìqù,or自治区),andtwo

WethankShuchangQi,deputydirectorofcensuscenterinNationalBureauofStatistics,andShouzhongGe,professorofnationaleconomicstatisticsinShanghaiUniversityofFinanceandEconomics,foransweringourinquiry.12Otheralternatives,suchascross-holdings,interlockingboards,directorappointmentclausesincorporatecharters,andthelikearealsopossible.SeeKhannaandYafeh(2007).

11Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx9

specialadministrativeregions(tèbiéxíngzhèngqū,or特别行政区).OurlistofprovincesexcludesTaiwan,whichthePeople'sRepublicconsidersabreakawaytwenty-thirdprovince.ThemunicipalgovernmentsofBeijing,Tianjin,Shanghai,andChongqingexerciseprovince-levelpowers.Thelegislaturesofthefiveautonomousregions–,Guangxi,Xinjiang,InnerMongolia,andNingxia–exerciseslightlybroaderpowersthanthoseoftheprovinces,andeachcontainssubstantialnon-Hanpopulations.Thegovernmentsofthetwospecialadministrativeregions,andMacau,exercisemanyofthepowersofnationalgovernment,thoughnotforeignaffairsordefense.Thetwospecialadministrativeregionsarenotincludedinoursample.Forsimplicity,weusethetermprovincetorefertoalloftheseincludedregions,whichthePeople'sRepublicclassifiesasprovince-leveladministrativedivisions.

3.4.1.Assetspecificityanduncertainty

Verticalintegrationisaresponsetoassetspecificityanduncertaintyassociatedwitharm'slengthtransactions.Wecaptureassetspecificitywithakeyinstitutionalfeature—thequalityofeachprovince'stransportationinfrastructure,asreflectedinthetotallength,inkilometers,ofrailways,waterways,andhighwaysintheprovincedividedbyitstotalgeographicarea,insquarekilometers.Inadequateregionaltransportationinfrastructurerestrictsparties'abilitytofindalternativebusinesspartners,andsoheightensassetspecificityproblemsbyaggravatingthepotentialforhold-upproblems.Thiselevatesthecostsofmarkettransactions,renderingverticalintegrationmoredesirable.

Wegaugeperformanceuncertaintybyinputpriceuncertainty,asinLieberman(1991)andFan(2000).Wedefinetheinputpriceuncertaintyinayearasthestandarderroroftheresidualsofaregressionofthelogofanindustry'sannualinflation-adjustedprimaryinputpriceindexonatimetrendusingtheprevioustenyeardata.Weemploytheinput–outputtabletoidentifytheprimaryinputofanindustry.Priceuncertaintyofafirmisthefirm'sprimaryindustry'sinputpriceuncertaintyasdefinedabove.Higherpriceuncertaintyshouldraisearm's-lengthtransactioncostsandthereforerenderverticalintegrationmorelikely.

3.4.2.Institutionalfactors

Section2hypothesizesthatverticalintegrationshouldcorrelatewithvariousinstitutionaldeficiencies,suchasweaklegalinstitutions,lowqualitygovernment,andlaggingmarketdevelopment.

OurindexoflegaldevelopmentispublishedbyTheAnnualReportofUrbanCompetitivenessinChina(Li,2002–2010).ResearchersattheChineseAcademyofSocialSciencesurveyasampleofordinarycitizens,entrepreneurs,andscholarsinmultiplecitiestogaugethe“competitivenessofcities”inChina.Thequestionsrelatedtolegaldevelopmentmeasure(i)thelikelihoodthatpropertyrightsareprotected;(ii)theprotectionofintellectualrights;and(iii)thedegreeofcontractenforcementbycourts.Applyingaprincipalcomponentanalysistothesesurveydatacreatesacitylevelindexoflegaldevelopment.Weaggregatethecitylevelindextotheprovincelevelbytakingitsaverageacrossallcitiesineachprovince.

ThemeasureofgovernmentdevelopmentalsocomesfromTheAnnualReportofUrbanCompetitivenessinChina(Li,2002–2010),whichisbasedonthesurveyofmultiplecitiesacrossChina.Therelatedsurveyquestionsinclude(i)thedegreeofbureaucratization;(ii)thefrequencyofgovernmentexpropriation,and(iii)thelevelofcitizensatisfactionwithgovernments.Similarly,aprincipalcomponentanalysisisappliedtocreateacitylevelindexofgovernmentdevelopment.Weaggregatethecitylevelindextotheprovincelevelbytakingitsaverageacrossallcitiesineachprovince.

Finally,werequireageneralmeasurecapturingtheextenttowhichintermediategoodsareallocatedbymarketsratherthanbybureaucrats.Unfortunately,suchameasureisnotavailable.Wethereforesubstituteaproxy:theimportanceofemploymentintheprivatesectorasopposedtothestatesector.Specifically,ourmarketdevelopmentproxyisthepercentageofworkersintheprovinceofficiallyregisteredasemployedintheprivatesector–thatis,registeredasemployedinprivateenterprisesorasself-employedindividuals(gètíhùor个体户)–ratherthanasemployeesofstate-ownedenterprisesorstateandPartyorgans.ThesedataarefromTheIndexofMarketizationofChina'sProvinces,ascompiledbyFan,WangandZhu(2002–2010).Theindexmeasuresmarketdevelopmentineachprovincebyyear.Weexpectthatahighdegreeofprivatesectoremploymentisassociatedwithavibrantgeneralmarketactivity.

3.4.3.Rent-seekingpotentialandpoliticalconnections

Section2alsopointsoutthatverticalintegrationmightbemotivatedbyrent-seeking.Wethereforeconstructasetofpoliticalconnectionvariablestocapturefirms'potentialreturnstopoliticalrent-seeking.

FollowingFanetal.(2007),ourfirstpoliticalconnectionvariableisapparatchikCEO,anindicatorvariablesettooneifafirm'sCEOis,orhasbeen,astateorCommunistPartyofficialofthecentralgovernment,theregionalgovernment,oranindustrybureau.

Oursecondmeasure,businessprivilege,alsoanindicatorvariable,issettooneifanyofthefirm'slinesofbusinessisaheavilyregulatedsector(utilities,postalservices,communication,railroadtransportation,airtransportation,mining,metals,orfinance)andtozerootherwise.InChina,therighttooperateinthesesectorsishighlyrestricted,andmustbegrantedbytheState.

WesurmisethatcorporateinsiderswithPartyorgovernmentbackgrounds,orwhoobtainedtherighttooperateinaheavilyregulatedindustry,arelikelytobepoliticallyconnected.Wethereforetakethesetwovariablesasreflectingpoliticalrent-seekingability.Weassumethatsuchabilityopensfurtherpoliticalrent-seekingopportunities,andhenceprovidessuperioraccesstobusinessopportunities.

Ourthirdproxyforpoliticalconnectionislong-termleverage,measuredbyeachfirm'slong-termdebtoveritstotalassets.Sapienza(2004),KhawajaandMian(2005),andAllenetal.(2005)arguethatpoliticallyconnectedfirmshavebetteraccesstoPleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

10J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

bankfinancingingeneral.BecauseChina'sbanksareallstate-controlled,thisbiasisquitelikelyevenstrongerthere.Wethereforesetanindicatorvariablehighleveragetooneifafirm'slong-termdebtovertotalassetsisabovethemedianforallfirmsinthatprovince,andtozerootherwise.Acaveatisinorderhere,however,forthisvariablehasapotentialalternativeinterpretation—betteraccesstocapitalfacilitatesacquisitions,includingacquisitionsofverticallyrelatedfirms.

3.4.4.Othercontrols

Weconsiderseveralmorevariablesascontrols.First,largerfirmsmightbemoreverticallyintegratedthansmalleronessimplybecausetheyarelarger.Forexample,economiesofscaleinacorelineofbusinessmighttranslateintosimilareconomiesinsecondaryandtertiarylinesofbusiness,renderingverticalintegrationlesscostly.Wegaugefirmsizebythenaturallogofthefirm'stotalassets.

YoungfirmsmightbemoreverticallyintegratedthanoldfirmsbecauseittakestimetoestablishexchangerelationshipsinChina'semergingmarketplace.Oryoungfirmsmightbemoreverticallyintegratedifformerlyfullystate-ownedenterprisesarehighlyverticallyintegratedattheirIPOs,andacquiremorefocussubsequently.Tocontrolforeithereventuality,wemeasurefirmagebythenumberofyearsthefirmhasbeenlisted,anddenotethisasyearslisted.

Wealsoincludeprovincialpercapitagrossdomesticproduct(GDP)toaccountfortheeffectsofregionaleconomicconditionsonfirmorganizationalstructure.Thiscontrolalsomitigatesthepossibilitythatourtransportationinfrastructureandinstitutionalvariablesmightproxyforgeneralprovince-leveldevelopment.

Table3summarizesthedefinitionsanddatasourcesforthesevariables.3.5.Descriptivestatistics

Table4reportsbasicdescriptivestatisticsacrosspooledobservationsforeachvariable.Thenumberofobservationsdiffersacrossvariablesbecausesomeareprovince-levelwhileothersareindustry-orfirm-level.SubstantialvariationininstitutionalqualityacrossChina'sregionsisconfirmedbythesubstantialstandarddeviationsoftheprovince-levelinstitutionalvariables.

Table5reportsPearsoncorrelationcoefficientsbetweenpairsoftime-serialmeanvariablesofinterest.Verticalintegrationcorrelatespositivelywithpriceuncertainty,andnegativelywithtransportationinfrastructure.Verticalintegrationalsocorrelatesnegativelywithinstitutionalstrength,asmeasuredbyourprovince-levellegal,government,andmarketdevelopmentindexes.Verticalintegrationsignificantlycorrelatespositivelywithallthreemeasuresofpoliticalconnection.Thesefindingsleavescopeforallthedriversofverticalintegrationconsideredabove.

Finally,thetablewarnsofpotentialmulticollinearitybetweenthetransportationinfrastructurevariableandlegaldevelopment(ρ=0.727),governmentdevelopment(ρ=0.607),marketdevelopment(ρ=0.655)andalsotheprovinciallevelGDPpercapita(ρ=0.787).Likewise,percapitaGDPalsocorrelatessignificantlypositivelywithlegaldevelopment,

Table3

Definitionofvariables.VariableLegal

developmentGovernmentdevelopmentMarket

developmentApparatchikCEO

Definition

Datasource

TheannualreportonurbancompetitivenessinChinaTheannualreportonurbancompetitivenessinChina

TheindexofmarketizationofChina'sprovincesFanetal.(2007)Corporateannualreports

Anindexmeasuring(i)thelikelihoodthatpropertyrightsareprotected;(ii)theprotectionofintellectualrights;and(iii)thedegreeofcontractenforcementbycourts.

Anindexmeasuring(i)thedegreeofbureaucratization;(ii)thefrequencyofgovernmentexpropriation,and(iii)thelevelofcitizensatisfactionwithgovernments.

Anindexmeasuringthepercentageofemploymentinprivateenterprisesorself-employedindividuals.

Anindicatorsettooneifthefirm'sCEOisnoworhaseverbeenabureaucratofthecentralgovernment,alocalgovernmentoranindustrybureau;andtozerootherwise.

BusinessprivilegeAnindicatorsettooneifthefirmoperatesinaheavilyregulatedindustry(utilities,postalservices,

communication,railroadtransportation,transportationbyair,mining,metals,orfinance);andtozerootherwise.

HighleverageAnindicatorsettooneifthefirm'stotaloutstandinglong-termloansexceedthemedianlevelforall

firmsintheprovince,andtozerootherwise.

PriceuncertaintyThestandarderroroftheresidualsofaregressionoflogofthefirm'slargestsegment'sprimaryinput

priceonatimetrendovertheprevioustenyears.

TransportationTotallengthofallhighway,railway,andwaterwayintheprovinceinkilometersdividedbyitstotalinfrastructuregeographicareainsquarekilometers.FirmsizeThenaturallogarithmoffirmassets.YearslistedThenumberofyearssincethefirm'sIPO.PercapitaGDPTheprovince'spercapitaGDPintensofthousandsofyuan.PercapitaGDPTheannualgrowthrateofpercapitaGDP.growthCapitalThenaturallogarithmofcumulativeannualcapitalinvestment,estimatedfromaperpetualinventory

modelwith7%depreciationrate,intheprovincefrom1984.

EducationThefractionoftheprovince'spopulationabletoreadandwrite.

aCSMARaChinaPriceYearbookChinaStatisticalAlmanacCSMAR

CSMAR

ChinaStatisticalAlmanacChinaStatisticalAlmanacChinaStatisticalAlmanacChinaStatisticalAlmanac

CSMARisChinaStockMarket&AccountingResearchDatabase,developedbyShenzhenGTAInformationTechnologyCo.,Ltd.

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx11

Table4

Descriptivestatisticsofvariables.VariablesaredefinedinTable3.Samplesforfirm-levelvariablesare11,563firm-yearobservationsfor1912firmsfrom2002to2010.Province-levelvariableshaveupto279province-yearobservationsover31provincesfrom2002to2010.Priceuncertaintyisanindustry-levelvariablewith848industry-yearobservationson111industriesfrom2002to2010.Variable

Assetspecificity&uncertaintyPriceuncertainty

TransportationinfrastructureInstitutionaldevelopmentLegaldevelopment

GovernmentdevelopmentMarketdevelopmentPoliticalconnectionApparatchikCEOBusinessprivilegeHighleverage

OthervariablesFirmsizeYearslistedPercapitaGDP

PercapitaGDPgrowthCapitalEducation

Obs.848279

Mean0.16070.6532

Median0.17280.5337

Std.dev.0.06270.4746

Min.0.00750.0326

Max.0.25032.3585

2792792790.560.557.42220.700.62307.45000.11100.23063.24300.42400.09800.34001.00001.000014.6500

11,56311,56311,5630.12520.20380.50160.00000.00001.00000.33100.40280.50000.00000.00000.00001.00001.00001.0000

11,56311,56327927927927921.328.78011.97580.14949.25040.1221.28299.00001.49330.14809.39020.91361.19234.31441.50930.09231.03320.078610.84221.00000.3153−0.20665.63070.451428.135721.00009.49200.451411.33510.9725

governmentdevelopment,andmarketdevelopment.Werevisitthesecorrelationsbelowwhenwediscussourmultipleregressions.

4.Empiricalresults

4.1.Verticalintegrationanditsdeterminants

Wenowturntomultipleregressionsthatcontrolforotherfactorsthatmightaffectverticalintegration.Weruntheseonpooledfirm-yeardata.Ourprovincelevelvariablesexhibitnovariationacrossfirmswithineachprovince.Likewise,wehaveindustry-levelvariablesthatexhibitlittleornovariationacrossfirmsineachindustry.Finally,someofourfirmlevelvariablesmaywellexhibitlittlevariationthroughtime.Toavoidupwardbiasedt-statistics,weemploystandarderrorsclusteredbyfirmandyearthroughout,asrecommendedbyPetersen(2009).

ThevaluesofourverticalintegrationmeasuresVmaxareboundedbelowbyspikesofobservationsatzero—presumablybecausesomefirmsdecidethenetcostsofoperatinginmultipleindustriesalongthevaluechainoutweighthenetbenefitsofspecialization.Toresolvethisdatacensorproblem,weemployTobitregression.Table6reportstheregressionresults.

ThefirstcolumninTable6,denotedasregression6.1,revealspriceuncertaintysignificantlypositivelycorrelatedwithverticalintegrationintensity.Further,thecoefficientoftransportationinfrastructureissignificantlynegative,suggestingthataweaktransportationsystemcorrelatessignificantlywithgreaterverticalintegration.Weinterpretthispatternasconsistentwithgreaterverticalintegrationoccurringwhereassetspecificityanduncertaintyproblemsposegreaterthreatstooverlyspecializedfirms.

Thesecondpatternisevidentinregressions6.2through6.5,whichallshowmoreextensiveverticalintegrationinprovinceswithworselegalprotectionforprivatepropertyrights,worsegovernment,andlessdevelopedmarketeconomies.Theseinstitutionalvariablesareclearlyeconomicallysignificanttoo.Forexample,basedonregression6.2,settingalltheindependentvariablesattheirmeans,a10%improvementinthepropertyrightsreducesverticalintegrationby7.73%,almost1.52timeslargerthanthestandarddeviationoftheverticalintegrationmeasure.Similarly,a10%improvementin“governmentquality”reducesverticalintegrationby4.77%.Marketdevelopmenthasasimilareconomicsignificance,anda10%improvementreducesverticalintegrationby5.08%.

Wetakethispatternasconsistentwithatransactioncostexplanationofverticalintegration:itletsfirmscircumventweakinstitutionsthatrenderprivatetransactionsproblematic.

ThethirdpatternevidentinTable6pertainstoourmeasuresofpoliticalconnections.BecauseTable5showsthethreeproxiesfor“connections”significantlycorrelatedwitheachother,Table6presentsseparateregressions–6.3,6.4,and6.5–includingeachpoliticalconnectionproxyinturn,ratherthanasingleregressionincludingallthree.Thepoliticalconnectionvariables,apparatchikCEO,businessprivilegeandhighleverage,areallsignificantlypositivelycorrelatedwithverticalintegration.Politicalconnectionshaveveryhigheconomicimpact.Forexample,switchingafirm'sCEOfromnotbeingapriorgovernmentbureaucrattobeingoneincreasesverticalintegrationby40.45%.Weinterprettheseresultsasconsistentwithpoliticallyconnectedfirmsbeingmoreverticallyintegratedduetorentseeking.

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

12J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

Table5

Pearsoncorrelationofcoefficients.ThevariablesaredescribedinTable3.Numbersinparenthesesarep-levelsforrejectingzerocorrelation.Allcorrelationsofvariablesareoftime-seriesmeans.

Vmax1

2

3

4

5

6

7

8

9

10

11

12

13

Assetspecificityanduncertainty1Priceuncertainty0.0518

(0.02)

2Transportation−−

infrastructure0.06080.0056

(0.01)(0.80)Institutionaldevelopment3Legaldevelopment−

0.1007(0.00)

4Government−

development0.1044

(0.00)

5Market−

development0.0980

(0.00)Politicalconnection6ApparatchikCEO

0.0117(0.61)−

0.0384(0.09)−

0.0321(0.16)

0.7274(0.00)

0.60710.7025(0.00)(0.00)

0.65530.84900.6653(0.00)(0.00)(0.00)

7Businessprivilege

8Highleverage

0.04690.0462−

(0.04)(0.04)0.0909

(0.00)

0.12090.2797−

(0.00)(0.00)0.1278

(0.00)

0.080.10740.0003(0.00)(0.00)(0.99)

0.0923(0.00)−

0.1433(0.00)−

0.0072(0.75)

0.0928(0.00)−

0.1084(0.00)0.0087(0.70)

0.1010(0.00)−

0.1393(0.00)−

0.0047(0.84)

0.0508(0.03)

0.00270.2187(0.90)(0.00)

Othervariables9Firmsize10Yearslisted

11PercapitaGDP

12PercapitaGDP

growth13Education

14Capital

0.4320(0.00)−

0.0214(0.35)

−−0.00140.10940.1247(0.95)(0.00)(0.00)

−−−−−−0.03820.08720.04410.26230.25060.06680.20430.0376(0.09)(0.00)(0.05)(0.00)(0.00)(0.00)(0.00)(0.10)

−0.00200.33810.32350.29700.3011−−−

0.0232(0.93)(0.00)(0.00)(0.00)(0.00)0.02700.03360.0010(0.31)(0.23)(0.14)(0.96)−−0.58220.78530.68760.7466−−0.00680.07220.0348(0.00)(0.00)(0.00)(0.00)0.12450.1063(0.76)(0.00)(0.13)(0.00)(0.00)

0.1063(0.00)−

0.0183(0.42)−

0.0405(0.07)0.0027(0.90)

0.1453(0.00)0.0962(0.00)

0.0990(0.00)−

0.1330(0.00)

0.00080.7874(0.97)(0.00)0.0294(0.20)−

0.2271(0.00)0.7087(0.00)0.0286(0.21)−

0.3133(0.00)0.56(0.00)0.0298(0.19)−

0.2800(0.00)0.7300(0.00)

0.06(0.00)0.1663(0.00)0.3001(0.00)0.0282(0.21)

0.0232(0.31)

0.1408−

(0.00)0.1913

(0.00)

−−

0.01990.2277(0.38)(0.00)0.1260−

(0.00)0.0617

(0.01)

0.0340−

(0.14)0.3293

(0.00)

0.3834(0.00)0.4914(0.00)0.5012(0.00)

−0.2421(0.00)−0.0623(0.01)

0.4149(0.00)

Table6showsthatthecoefficientontransportationinfrastructurebecomesinsignificantafterincludinglegal,governmentandmarketdevelopmentmeasures.Thismightbecausedbyhighcorrelationbetweentransportationinfrastructureandinstitutionvariables.Totheextentthattransportationinfrastructureispoorinunderdevelopedregions,thevariablemaycapturetheregionalinstitutionalenvironmentinsteadoftransactiondifficultiesfromassetspecificity.

AllthespecificationsinTable6controlforthepercapitaGDPoftheprovinceinwhichthefirmisbased.Thiscontrolissignificantlypositive,indicatingthatfirmsaremoreverticalintegratedintheregionsofhigherlivingstandards.Further,theresultsshowthatlargerfirmsaremoreverticallyintegratedthansmallerones,butolderfirmsarelessverticallyintegratedthanyoungerones.

5.Robustness

Asrobustnesschecks,weapplyaninverselogistictransformationtoourverticalintegrationmeasure,mappingVmaxintoVmax′=ln[1/(1−Vmax)].WethenregressthetransformedmeasureonthevariablesintheTable6,usingeitheraTobitorleastsquaresregression,againwithstandarderrorsclusteredbyfirmandyear,andagaingetsimilarresultstothoseshown.

Theinput–outputstatisticsofthe“wholesale”and“retail”sectorsareaggregatedandlabeled“commerce”inTable2.Theseinputandoutputflowsmaybeexcessivelyaggregated,andthusqualitativelydifferentfromothersectors(Acemogluetal.,2009;Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx13

Table6

Tobitregressionsofdeterminantsofverticalintegration.Tobitregressionsexplainverticalintegrationwithpriceuncertaintyinthefirm'sprimaryindustry,thequalityofthetransportationinfrastructureoftheprovinceinwhichitislocated,province-levelindexesoflegaldevelopment,governmentdevelopment,andmarketdevelopment,ourapparatchikCEOdummyforapoliticallywell-connectedCEO,ourbusinessprivilegeindicatorthatthefirmoperatesinaheavilyregulatedsector,ahighleverageindicator,firmsize(logoftotalassets),yearslisted(yearssincethefirm'sIPO),andthepercapitaGDPofthefirm'sprovince.Afirmisassumedtobelocatedintheprovincecontainingitsheadoffice.Standarderrorsareclusteredatthefirmandyearlevel,withZstatisticsareinparentheses.One,two,orthreeasterisksdenotesignificanceat10%,5%and1%levels,respectively.Regression

Assetspecificity&uncertaintyPriceuncertainty

Transportationinfrastructure

6.1

0.0432***(4.20)

−0.0080***(4.07)

6.20.0458***(4.46)−0.0027(1.30)

6.30.0467***(4.)−0.0027(1.27)

6.40.0280***(2.70)−0.0017(0.83)

6.50.04***(4.42)−0.0026(1.24)

InstitutionaldevelopmentLegaldevelopmentGovernmentdevelopmentMarketdevelopment

−0.0181**(2.15)

−0.0122***(2.85)

−0.0010***(3.31)−0.0183**(2.18)

−0.0116***(2.70)

−0.0010***(3.30)−0.0183**(2.18)

−0.0138***(3.22)

−0.0009***(2.92)−0.0184**(2.19)

−0.0117***(2.73)

−0.0011***(3.38)

PoliticalconnectionApparatchikCEOBusinessprivilegeHighleverage

0.0058***(3.06)

0.0160***(9.91)

0.0047***(3.53)

OthercontrolsFirmsizeYearslistedPercapitaGDP

ClusterbyfirmandyearObservationsLoglikelihood

0.0043***(7.87)−0.0003*(1.71)0.0003(0.59)Yes11,56382690.0043***(7.78)−0.0002(1.32)0.0017***(2.85)Yes11,56382930.0042***(7.65)−0.0002(1.40)0.0018***(2.93)Yes11,56382980.0027***(4.80)−0.0002(1.14)0.0019***(3.11)Yes11,56383420.0035***(5.98)−0.0002(1.44)0.0018***(3.00)Yes11,5638299

FanandLang,2000).Asafurtherrobustnesscheck,wethereforereestimateourregressionsexcludingthesesectors.Thiscutsabout1250firm-yearobservationsfromthepanel,butyieldsqualitativelysimilarresultstothoseshown.

Tocontrolforpanelpersistence,weclusterresidualsbyfirmandtime.Anextremelyconservativerobustnesscheckcollapsesour2002through2010panelintoasinglecross-sectionoftime-seriesaverages.RunningregressionsanalogoustothoseinTable6onthiscross-section,Table7revealsqualitativelysimilarresultstothoseshown,exceptthatthe“legaldevelopment”variablebecomesinsignificant.

Overall,weinterprettheseresultsasconsistentwithverticalintegrationbeinglinkedtoassetspecificityandpriceuncertainty,andwiththesefactorsencouragingverticalintegrationasanorganizationalresponsetoweakpropertyrightsprotection,poorqualitygovernment,andill-developedmarketeconomies.Theseresultsarealsoconsistentwithalinkbetweenverticalintegrationandpoliticalrent-seeking.

5.1.Verticalintegrationandeconomyperformance

Wenowexaminetherelationshipbetweenverticalintegrationandregionaleconomydevelopment.Verticalintegrationasasecondbestresponsetoarm'slengthtransactiondifficultiesamidaweaklegalsystemandweakmarketforcescouldcorrelatewithelevatedpercapitaGDPlevelsandgrowthrates.Incontrast,verticalintegrationundertakentomagnifythereturnstopoliticalrent-seekinghasalesscertainimpact.Asafirstpassinvestigationofthisissue,weturntoasetofprovincelevelregressions.

Wefirstclassifyeachfirmaseitherpoliticallyconnectedornotaccordingtoeachofourthreepoliticalconnectionindicatorvariables—havinganapparatchikasaCEO,operatinginarestrictedindustry,andobtaininghigherthanmedianleveragefromstatebanks.Nextwecalculatetheaveragevalueofourverticalintegrationmeasureacrossallpoliticallyconnectedfirmsineachprovince.Thisyieldsthreemeasuresoftheintensityofverticalintegrationamongpoliticallyconnectedfirmsineachprovince,oneforeachofourpoliticalconnectionvariables.Wethenrepeatthisprocedureforpoliticallyunconnectedfirms,andobtainthreeanalogousmeasuresoftheintensityofverticalintegrationamongpoliticallyunconnectedfirmsineachprovince,oneforeachofourpoliticalconnectionvariables.Wedothisforall31provinceseachyearfrom2002to2010.

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

14J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

Table7

Cross-sectionregressionsofdeterminantsofverticalintegration.TobitregressionsinTable6arerepeatedonasinglecross-sectionofdata,constructedbytime-averagingthepanelofdatafrom2002to2010usedinTable6.Variablesareasinthattable,andaredefinedindetailinTable3.Zstatisticsareinparentheses.One,two,orthreeasterisksdenotesignificanceat10%,5%and1%levels,respectively.Regression

Assetspecificity&uncertaintyPriceuncertainty

Transportationinfrastructure

7.1

0.0637**(2.31)−0.0062(1.50)

7.2

0.0629**(2.30)0.0002(0.04)

7.3

0.0617**(2.26)0.0003(0.06)

7.40.0351(1.25)0.0005(0.11)

7.5

0.0601**(2.20)0.0003(0.07)

InstitutionaldevelopmentLegaldevelopmentGovernmentdevelopmentMarketdevelopment

−0.0125(0.44)−0.0267***(3.29)−0.0014*(1.71)−0.0131(0.46)−0.02***(3.26)−0.0014*(1.73)−0.0102(0.36)−0.0267***(3.30)−0.0013*(1.69)−0.0132(0.46)−0.0269***(3.32)−0.0014*(1.75)

PoliticalconnectionApparatchikCEOBusinessprivilegeHighleverage

0.0104*(1.87)

0.0147***(3.84)

0.0078*(1.95)

OthercontrolsFirmsizeYearslistedPercapitaGDPObservationsLoglikelihood

0.0056***(4.75)0.0008***(2.66)−0.0015(1.21)191219420.0052***(4.36)0.0002(0.71)0.0005(0.33)191219570.0050***(4.22)0.0002(0.46)0.0006(0.43)191219580.0038***(3.04)0.0003(0.88)0.0008(0.57)1912190.0041***(3.09)0.0002(0.75)0.0006(0.44)19121959

Ourleft-handsidevariablemeasureseconomyperformanceaseitherannualpercapitaGDPlevelorannualpercapitaGDPgrowth.Weregresstheseontheintegrationintensityoftheprovince'spoliticallyconnectedandunconnectedfirmsandonourinstitutionalvariables—legaldevelopment,governmentdevelopment,andmarketdevelopment.FollowingPetersen(2009)weclusterthestandarderrorsbyprovince.

Wesupplementthesewithafewothercontrols.Physicalandhumancapitalareknowntobeafirst-orderdeterminantsofeconomicgrowth(see,e.g.,BarroandSala-i-Martin,1995).Wemeasureaprovince'sphysicalcapital,denotedascapital.Toestimatecapital,weuseaperpetualinventorymodelthatsumstheannualprovincialcapitalinvestmentfrom1984withadepreciationrateof7%.13Humancapitalismeasuredasthefractionofeachprovince'spopulationclassifiedbytheNationalBureauofStatisticsofChinaas“literate”,whichisdenotedaseducation.Standardgrowththeoryalsoimpliesaconvergenceeffect:economieswithlowerinitiallevelsofpercapitaGDPcangrowfasterthanthosealreadysustaininghighincomes(BarroandSala-i-Martin,1995).WethereforeincludethelogarithmoflaggedpercapitaGDPasacontrolvariableinourgrowthregressions.

Table8displaystheresults.TheverticalintegrationamongpoliticallyunconnectedfirmsattractspositivecoefficientsintheregressionsofbothpercapitaGDPlevelandgrowth,buttheverticalintegrationamongpoliticallyconnectedfirmsgeneratesinsignificantcoefficients.

Theseresultsareeconomicallysignificanttoo.ConsidertheGDPgrowthregressionusinganapparatchikCEOtogaugepoliticalconnection.Allelseequal,aprovincewhoseapparatchik-freefirmshaveameanverticalintegrationintensityone-standarddeviationgreaterreaps0.extrapercentagepointsofadditionalpercapitaGDPgrowthannually.Theeffectisclearlylargeenoughtomatter,almostfourpercentofthemeanGDPgrowthrate.

Theseresultsarerobust.Weusemedianstogenerateourverticalintegrationintensitymeasures,andourfindingsarequalitativelyunchanged.Employingasset-weightedmeansgeneratesqualitativelysimilarresults.

Althoughwefindthatthepositiveassociationbetweenverticalintegrationandeconomicvalueconcentratesinpoliticallyunconnectedfirms/regions,somecaveatsariseindrawingconclusions.First,ourdataareforlistedfirms,whichaccountforonlyafractionofprovincialeconomicoutput.Second,weleavethedirectionofcausalityunproveninthatbotheconomygrowthandverticalintegrationmightbeendogenouslydeterminedbylatentfactors.Thus,whileouranalysissuggeststhatverticalintegrationamongpoliticallyunconnectedfirmshasapositiveassociationwithlocaleconomydevelopment,theresultcouldreflectomittedfactorsandthereforebeopentootherinterpretations.

ThecalculationformulaisCapitali,t+1=Capitali,t+Ii,t−δCapitali,t,whereIi,tisannualcapitalinvestmentforprovinceiinyeart,andδisadepreciationrateof7%followingKingandLevine(1994)andFogel,MorckandYeung(2008).

13Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx15

Table8

Regressionsofeconomyperformanceonverticalintegration.DependentvariablesareeitherprovincialpercapitaGDPlevelorgrowthrate.Connectedfirmsaredefined,alternatively,byhavingpoliticalapparatchiksasCEOs,operatinginrestrictedindustries,orhavingabovemedianloansfrombanks.Verticalintegrationintensityamongpoliticallyconnectedfirmsisthemeanvalueofverticalintegrationofpoliticallyconnectedfirmsineachprovince.Verticalintegrationintensityamongpoliticallyunconnectedfirmsisdefinedanalogously.Legal,government,andmarketdevelopmentaredefinedasinTable3,capitaliscumulativeannualcapitalinvestmentfrom1984withadepreciationrateof7%,educationisliteratepopulationasafractionofthetotal,andpercapitaGDPdenotestheregionalpercapitaGDP.Standarderrorsareclusteredattheprovincelevel,withTstatisticsinparentheses.One,two,orthreeasterisksdenotesignificanceat10%,5%and1%levels,respectively.Regression

Connectiondefinedas:Performancedefinedas:

Verticalintegration

IntensityamongconnectedfirmsIntensityamongunconnectedfirms

8.1

ApparatchikCEOGDPlevel0.0662(0.10)5.2742***(3.09)

GDPgrowth−0.0022(0.03)0.3434**(2.45)8.2

8.3

BusinessprivilegeGDPlevel0.4207(0.36)3.6158*(1.86)

GDPgrowth0.0076(0.07)0.3778**(2.61)8.4

8.5

HighleverageGDPlevel0.0686(0.08)6.1610***(3.07)

GDPgrowth−0.0513(0.44)0.5952**(2.26)8.6

InstitutionaldevelopmentLegaldevelopmentGovernmentdevelopmentMarketdevelopment

0.4934(1.13)0.48**(2.07)0.1288***(5.39)−0.0232(0.39)0.0323(1.53)0.0052*(1.78)0.5003(1.13)0.5065**(2.14)0.1256***(5.25)−0.0188(0.31)0.0346(1.56)0.0050(1.69)0.4384(1.03)0.5041**(2.25)0.12***(5.34)−0.0274(0.46)0.0333(1.51)0.0051*(1.78)

ControlsCapitalEducationPercapitaGDPObservationsAdjustedR2−0.0483(0.57)1.7360(1.36)

2790.70

0.0018(0.27)0.0490(0.90)−0.0211***(3.74)2790.05

−0.0372(0.44)1.6682(1.30)

2790.69

0.0019(0.28)0.0452(0.81)−0.0210***(3.)2790.05

−0.0371(0.47)1.5687(1.30)

2790.70

0.0024(0.41)0.0351(0.70)−0.0213***(3.90)2790.06

6.Conclusions

PatternsofverticalintegrationinChinacorrespondwelltolikelyregionaltransactioncostdifferences.Firmsaremoreverticallyintegratedinprovinceswithweakerpropertyrightsprotection,worsegovernment,andlaggardmarketreforms.Theseobservationsareconsistentwithverticalintegrationbeingusedtoovercometransactiondifficultieswherepropertyrightsprotectionandmarketdisciplinaryforcesareweak.Theyarealsoconsistentwithverticalintegrationbeingusedtoovercomeimpedimentstodoingbusinessposedbyoverlybureaucraticgovernmentsthataredisinclinedtorespectprivatepropertyrights.Theseresultspersistaftercontrollingforlocaltransportationinfrastructure,inputpriceuncertainty,firmsize,ageasalistedfirm,andregionaleconomicdevelopment.

Firmswhosemanagershaveclosertiestobureaucratsarealsomoreverticallyintegrated.Provinceswheremarketinstitutionsareweakarelikelyalsoplaceswherepoliticalrent-seekingpaysahighreturn.Ifpoliticalrent-seekinghelpsfirmsgainstateenforcedmonopolypower,rent-seekersshoulddirecttheirlobbyingtowardsgainingverticallyintegratedmonopolies.Thisisbecauseaverticallyintegratedmonopolycanextractunambiguouslyhigherprofitsthanastringofverticallyrelatedsingle-marketmonopolies(Spengler,1950).Thissuggestsalesslaudableuseofverticalintegrationtomagnifythereturnstopoliticalrentseekingbyforeclosingcompetitionalongbroaderswathesofavaluechain.

OurfurtherfindingisthatmoreextensiveverticalintegrationinpoliticallyunconnectedfirmsisassociatedwithelevatedprovincialGDPlevelsandgrowthrates,whilemoreextensiveverticalintegrationinpoliticallyconnectedfirmsisnot.Thesefindingssupportthepropertyright(Alchian,1965)andtransactioncosts(Kleinetal.,1978)explanationsofverticalintegration.ButtheyalsopointtoaparallelpoliticaleconomyexplanationofverticalintegrationthatisrelevantinChina.Ourfindingofnoimprovementinprovincialeconomyperformancewhereverticallyintegratedfirmsarepoliticallyconnectedsuggeststhisalternativepossibilityatleastbetakenseriously.

Furtherworkisneededtotestthispossibilitymorethoroughlyanddiscountalternativeexplanationsofourfindings,andnumerouscaveatsapply.Certainly,China'sexperiencemay,butneednot,extendtoothereconomies.Nonetheless,theevidencehereprovidesabenchmarkforcomparisonwithotheremergingandtransitioneconomieswithsimilarinstitutionalasthenia.

InChina,successivereformsareprogressivelyseparatingbusinessfromgovernment,butatmarkedlydifferentpacesindifferentprovinces.Wherethatseparationisleastadvanced,stateandPartyofficialsarealsofreesttopreyuponprivatebusinesses.Perhapstheonlyviablebusinessesinsuchregionsarethoserunbythepotentialpredators,eitherdejureviadirectstatecontrolordefactoviatheircronies.

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

16J.P.H.Fanetal./JournalofCorporateFinancexxx(2014)xxx–xxx

Sincethemostpoliticallyconnectedmaynotbethemostcreativeentrepreneurs,growthintheseregionsmaywellbetriplyimpeded—byweakinstitutionsforcinggreaterverticalintegration;byentrustingcorporatedecision-makingtothepoliticallyconnected,ratherthanthemostableentrepreneurs;andbylargeverticallyintegratedmonopoliescapableofinflictingworseoverallwelfarelossesthanseparatesingle-marketmonopoliescouldmanage.Thepotentialcostsofsuchathreefoldself-reinforcingdragoneconomicgrowthshouldbeofconcerntoChinesepolicy-makersifincreasinginterregionalinequalityisviewedasapotentiallyserioussourceofsocialinstability.References

Acemoglu,D.,Aghion,P.,Griffith,R.,Zilibotti,F.,2010.Verticalintegrationandtechnology:theoryandevidence.J.Eur.Econ.Assoc.8(5),9–1033.

Acemoglu,D.,Johnson,S.,Mitton,T.,2009.Determinantsofverticalintegration:financialdevelopmentandcontractingcosts.J.Financ.(3),1251–1290.Alchian,A.,1965.Someeconomicsofpropertyrights.IIPolitico,30,pp.816–829.

Alchian,A.,Demsetz,H.,1972.Production,informationcosts,andeconomicorganization.Am.Econ.Rev.62(5),777–795.Allen,F.,Qian,J.,Qian,M.,2005.Law,finance,andeconomicgrowthinChina.J.Financ.Econ.77(1),57–116.Barro,R.,Sala-i-Martin,X.,1995.EconomicGrowth.McGrawHill,Columbus.Cabral,L.,2000.IntroductiontoIndustrialOrganization.MITPress,Cambridge.

Che,J.,Qian,Y.,1998.Insecurelegaldevelopmentandgovernmentownershipoffirms.Q.J.Econ.113(2),467–496.

Chipty,T.,2001.Verticalintegration,marketforeclosure,andconsumerwelfareinthecabletelevisionindustry.Am.Econ.Rev.91(3),428–453.Coase,R.,1937.Thenatureofthefirm.Economica4,386–405.

Cull,R.,Xu,L.,2005.Institutions,ownership,andfinance:thedeterminantsofprofitreinvestmentamongChinesefirms.J.Financ.Econ.77(1),117–146.DeSoto,H.,2000.TheMysteryofCapital:WhyCapitalismTriumphsintheWestandFailsEverywhereElse.BasicBook,NewYork.Fan,G.,Wang,X.,Zhu,H.,2002–2010.NERIIndexofMarketizationofChina'sProvinces.EconomicsSciencePress,Beijing.

Fan,J.P.H.,2000.Priceuncertaintyandverticalintegration:anexaminationofpetrochemicalfirms.J.Corp.Finan.6(4),345–376.Fan,J.P.H.,Goyal,V.,2006.Onthepatternsandwealtheffectsofverticalmergers.J.Bus.79(4),877–902.

Fan,J.P.H.,Lang,L.,2000.Themeasurementofrelatedness:anapplicationtocorporatediversification.J.Bus.73(4),629–660.

Fan,J.P.H.,Wong,T.,Zhang,T.,2007.PoliticallyconnectedCEOs,corporategovernanceandpost-IPOperformanceofChina'spartiallyprivatizedfirms.J.Financ.

Econ.84(2),330–357.

Fan,J.P.H.,Wong,T.,Zhang,T.,2013.Institutionsandorganizationalstructure:thecaseofstate-ownedcorporatepyramids.J.LawEcon.Org.29(6),1217–1252.Feenstra,R.,Hanson,G.,2005.OwnershipandcontrolinoutsourcinginChina:estimatingthelegaldevelopmenttheoryofthefirm.Q.J.Econ.120(2),729–761.Fogel,K.,Morck,R.,Yeung,B.,2008.Bigbusinessstabilityandeconomicgrowth:iswhat'sgoodforGeneralMotorsgoodforAmerica?J.Financ.Econ.(1),83–108.Henisz,W.,2000.Institutionalenvironmentsformulti-nationalinvestment.J.LawEcon.Org.16(2),334–3.

Jensen,M.C.,Meckling,W.H.,1976.Theoryofthefirm:managerialbehavior,agencycostsandownershipstructure.J.Financ.Econ.3(4),305–360.Joskow,P.,1987.Contractdurationandrelationship-specificinvestments:empiricalevidencefromcoalmarkets.Am.Econ.Rev.77(1),168–185.Khanna,T.,Oberholzer-Gee,F.,2005.Thepoliticaleconomyoffirm-sizedistributions:evidencefromChina.WorkingPaper,HBS.

Khanna,T.,Palepu,K.,1999.Policyshocks,marketintermediaries,andcorporatestrategy:evidencefromChileandIndia.J.Econ.Manag.Strateg.8(2),271–310.Khanna,T.,Yafeh,Y.,2007.Businessgroupsinemergingmarkets:paragonsorparasites?J.Econ.Lit.45(2),331–373.

Khawaja,A.,Mian,A.,2005.Dolendersfavorpoliticallyconnectedfirms?Rentprovisioninanemergingfinancialmarket.Q.J.Econ.120(4),1371–1411.King,R.,Levine,R.,1994.Capitalfundamentalism,economicdevelopment,andeconomicgrowth.Carn.-Roch.Conf.Ser.PublicPolicy40(1),259–292.Klein,B.,2014.Economictheoriesofthefirm.J.Corp.Financ.forthcoming.

Klein,B.,Crawford,R.,Alchian,A.,1978.Verticalintegration,appropriablerents,andthecompetitivecontractingprocess.J.LawEcon.21(2),297–326.Klein,B.,Leffler,K.,1981.Theroleofmarketforcesinassuringcontractualperformance.J.Polit.Econ.(4),615–1.Krueger,A.,1974.Thepoliticaleconomyoftherent-seekingsociety.Am.Econ.Rev.(3),291–303.

Lafontaine,F.,Slade,M.,2007.Verticalintegrationandfirmboundaries:theevidence.J.Econ.Lit.95(3),269–685.Lieberman,M.,1991.Determinantsofverticalintegration:anempiricaltest.J.Ind.Econ.39(5),451–466.Li,P.,2002–2010.TheAnnualReportofUrbanCompetitivenessinChina.SocialSciencesAcademy,Beijing.Lucas,R.,1978.Onthesizedistributionofbusinessfirms.BellJ.Econ.9(2),508–523.

Masten,S.,1984.Theorganizationofproduction:evidencefromtheaerospaceindustry.J.LawEcon.27(2),403–418.

Milgrom,P.,Roberts,J.,1988.Aneconomicapproachtoinfluenceactivitiesinorganizations.Am.J.Sociol.94(Supplement),1–179.

Monteverde,K.,Teece,D.,1982.Supplierswitchingcostsandverticalintegrationintheautomobileindustry.BellJ.Econ.13(1),206–213.

Montinola,G.,Qian,Y.,Weingast,B.,1995.Federalism,Chinesestyle:thepoliticalbasisforeconomicsuccessinChina.WorldPolit.48(1),50–81.Morck,R.,Wolfenzon,D.,Yeung,B.,2005.Corporategovernance,economicentrenchmentandgrowth.J.Econ.Lit.43(3),657–722.

Mulherin,J.,1986.Complexityinlong-termcontracts:ananalysisofnaturalgascontractualprovisions.J.LawEcon.Org.2(1),105–177.Murphy,K.,Shleifer,A.,Vishny,R.,1991.Theallocationoftalent:implicationsforgrowth.Q.J.Econ.106(2),503–530.Ohanian,N.,1994.VerticalintegrationintheU.S.pulpandpaperindustry1900–1940.Rev.Econ.Stat.76(1),202–207.

Petersen,M.,2009.Estimatingstandarderrorsinfinancialpaneldatasets:comparingapproaches.Rev.Financ.Stud.22(1),435–483.Qian,Y.,Weingast,B.,1996.China'stransitiontomarkets:market-preservingfederalism,Chinesestyle.J.PolicyReform.1(2),149–185.Qian,Y.,Weingast,B.,1997.Federalismasacommitmenttopreservingmarketincentives.J.Econ.Perspect.11(4),83–92.Rosenstein-Rodan,P.,1943.ProblemsofindustrializationofeasternandsoutheasternEurope.Econ.J.53(210/211),202–211.Sapienza,P.,2004.Theeffectsofgovernmentownershiponbanklending.J.Financ.Econ.72(2),357–384.Schliefer,A.,Vishny,R.,1993.Corruption.Q.J.Econ.108(3),599–617.

Shahrur,H.,2005.Industrystructureandhorizontaltakeovers:analysisofwealtheffectsonrivals,suppliers,andcorporatecustomers.J.Financ.Econ.76(1),61–98.Smith,A.,1776.TheWealthofNations.1904.Reprintedition1937In:Cannan,Edwin(Ed.),ModernLibrary,NewYork.Spengler,J.,1950.Verticalintegrationandantitrustpolicy.J.Polit.Econ.58(4),347–352.

Stigler,G.,1951.Thedivisionoflaborislimitedbytheextentofthemarket.J.Polit.Econ.59(1),185–193.Stigler,G.,1983.TheOrganizationofIndustry.UniversityofChicagoPress.

Teece,D.,1976.VerticalIntegrationandVerticalDivestitureintheU.S.OilIndustry,(WorkingPaper).Tullock,G.,1965.ThePoliticsofBureaucracy.PublicAffairsPress,WashingtonD.C.

Williamson,O.,1973.Marketsandhierarchies:someelementaryconsiderationsAm.Econ.Rev.63(2),316–325.Williamson,O.,1975.MarketsandHierarchies:AnalysisandAntitrustImplications.Free,NewYork.

Williamson,O.,1979.Transactioncosteconomies:thegovernanceofcontractualrelations.J.LawEcon.22(2),233–261.

Pleasecitethisarticleas:Fan,J.P.H.,etal.,InstitutionaldeterminantsofverticalintegrationinChina,J.Corp.Finance(2014),http://dx.doi.org/10.1016/j.jcorpfin.2014.05.013

因篇幅问题不能全部显示,请点此查看更多更全内容

Copyright © 2019- jqkq.cn 版权所有 赣ICP备2024042794号-4

违法及侵权请联系:TEL:199 1889 7713 E-MAIL:2724546146@qq.com

本站由北京市万商天勤律师事务所王兴未律师提供法律服务