Horizontal and Vertical R&D Cooperation

HorizontalandVerticalR&DCooperationJoachimInkmann*DepartmentofEconomicsandCenterofFinanceandEconometrics(CoFE)UniversityofKonstanz,BoxD124,78457Konstanz,GermanyAugust1999Abstract:Thispaperintroducesasecond,verticallyrelatedindustryintotheusualone-industryoligopolyframeworkofcooperativeR&Dinvestmentbetweenfirmsoperatingonthesameproductmarket.R&Deffortsareaffectedbyintra-andinter-industryR&Dspillovers.HorizontalandverticalR&DcooperationscenariosarecomparedtoR&Dcompetition.ItturnsoutthatverticalR&DcooperationisusuallytheonlystableequilibriuminthesensethatnofirmhasanincentivetochoseanyotherR&Dscenario.EmpiricalimplicationsconcerningtherelationshipbetweenR&DintensitiesandR&Dspilloversarederivedandempiricalevi-denceisgivenusingdataofGermanmanufacturingfirms.Keywords:R&Dcooperation;R&Dspillovers;verticalrelationsJELClass.:O31,L13,C21*E-mail:[e-mail]perwaspresentedatthe1998NorthAmericanWinterMeetingoftheEconometricSocietyinChicago.FinancialsupportbytheDeutscheForschungsgemeinschaft(DFG)isgratefullyacknowledged.IwishtothankGautamGowrisankaran,GeorgLicht,andWinfriedPohlmeierforhelpfulcommentsandtheCentreforEuropeanEconomicResearch(ZEW)inMannheimformakingavailablethedatasource.21.IntroductionItisgenerallyacceptedthattheincentivesoffirmstoinvestinresearchanddevelopment(R&D)aredistortedbecauseofthepublicgoodcharacteristicofnewinformation.Inparticu-lar,theappropriabilityproblemhasbeenwidelydiscussedintheliterature(cf.Spence,1984,CohenandLevinthal,1989),whichcausesfirmstounderinvestinR&DbecausetheycannotcompletelyinternalizethesocialreturnsoftheirprivateeffortsinthepresenceofR&Dspill-overs.Threeinstrumentsareusuallyconsideredtorestorethefirms’incentivestoengageinR&D:Taxpoliciesanddirectsubsidies,ex-postR&Dcooperationthroughpatentsandli-censing,andex-anteR&Dcooperation(cf.KatzandOrdover,1990).Whilethefirsttwoin-strumentsrequiregovernmentinterventiontodeterminetaxesandsubsidiesortostrengthenpropertyrights,thethirdinstrumentisassumedtoworkthroughprivateincentivesbecauseofthepossibilitytointernalizeR&Dspilloversbetweencooperatingfirms1.OtheradvantagesofR&DcooperationincludetheeliminationofwastefulduplicationofR&Deffortsandthedis-tributionofriskandfixedcostsamongparticipants(cf.Jacquemin,1988).StartingwiththeworkbyKatz(1986)andD’AspremontandJacquemin(1988)alargebodyoftheoreticalliteraturehasemergedoverthepastdecadetryingtoformalizeafirms’privateincentivestoengageinR&Dcooperationbyusingoligopolymodelswhichallowforstrategicinteractionsbetweenfirms.2Usually,two-stagegamesareanalyzedinwhichfirmschooseeithernoncooperativelyorcooperativelytheiramountofR&Dinvestmentinthefirststageandcompeteontheproductmarketinthesecondstage.IfR&Dspilloversaresuffi-cientlyhighthisframeworkexplainstheprivateincentivesforR&DcooperationbetweenrivalswhichleadstoincreasedR&Dinvestment,outputandsocialwelfare.Whilethesemod-elsdifferinmanydetails(e.g.,Bertrandvs.Cournotcompetitioninthesecondstage),theyhaveincommonthattheyfocusonasingleindustrybyanalyzingintra-industrycooperationbetweenrivalsontheproductmarket.Thisstandsincontrasttothelargeamountofempiricalliteraturesearchingforbothintra-andinter-industryR&Dspilloversandclaimingtheparticularimportanceofthelatterforproductivitygrowth(e.g.GrilichesandLichtenberg,1984,andreferencescitedinCapronetal.,1996).DescriptiveevidenceofR&Dcooperationbetweensectorsandwithinsectorsre-1Nevertheless,inmanycountiesR&Dcooperationisnowexplicitlysupportedbyantitrustpolicy(fortheEC,cf.Jacquemin,1988,andKatsoulacosandUlph,1997).2ExamplesincludeDeBondtandVeugelers(1991),Kamienetal.(1992),Suzumura(1992),Vonortas(1994),Ziss(1994),SalantandShaffer(1998).3vealsthatinter-industryagreementsaremuchmorefrequentthanintra-industryagreements.AccordingtoChesnais(1988),80%ofJapaneseinter-firmR&Dcooperationinvolvefirmsfromdifferentsectors.Licht(1994)presentssimilarevidenceforsixEuropeancountries.Inthesecountries,themostprevalentformofR&Dcooperationincludeseithercustomersorsuppliers.UsingGermandatafrom1994,Harabi(1997)showsthat84%ofallinnovatingfirmsareengagedinR&Dcooperationwithcustomersorsuppliers.Thisformofcooperationisusuallylabeledverticalanddistinguishedfromhorizontalcooperationwithfirmsoperatinginthesameindustry(cf.Geroski,1992).VonHippel(1986)andVanderWerf(1992)presentcasestudiesfortheUSAforinnovationswhichareinitiatedbycustomersandsuppliers,re-spectively.VonHippelexplainscustomerdriveninnovationsby‘leadusers’whohavebettercapabilitiestoforecasttheirfutureneedsortofilltheircurrentneedsthanproducers.VanderWerfarguesthatsuppliersofintermediategoodshaveanincentivetoencouragedownstreaminnovationinordertoincreasetheirowndemand.Inbothcasesverticalcoopera-tioninR&Dmayincreasetheinnovativesuccessofparticipatingfirms.Geroski(1992,1995)pointsoutthatverticalR&DcooperationmaybesuperiortohorizontalagreementsbecausethelattermayleadtocollusivepricingfortheproductsembodyingthejointR&Defforts.ProbablythemostprominentexamplefortheimportanceofverticalR&DcooperationbetweenmanufacturersandsuppliersistheJapaneseauto