Innovation and Learning

InnovationandLearning:TheTwoFacesofR&DAuthor(s):WesleyM.CohenandDanielA.LevinthalSource:TheEconomicJournal,Vol.99,No.397(Sep.,1989),pp.569-596Publishedby:BlackwellPublishingfortheRoyalEconomicSocietyStableURL::15/08/201022:54YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat://=black.EachcopyofanypartofaJSTORtransmissionmustcontainthesamecopyrightnoticethatappearsonthescreenorprintedpageofsuchtransmission.JSTORisanot-for-profitservicethathelpsscholars,researchers,andstudentsdiscover,use,andbuilduponawiderangeofcontentinatrusteddigitalarchive.Weuseinformationtechnologyandtoolstoincreaseproductivityandfacilitatenewformsofscholarship.FormoreinformationaboutJSTOR,[e-mail]lackwellPublishingarecollaboratingwithJSTORtodigitize,preserveandextendaccesstoTheEconomicJournal.(SeptemberI989),569-596PrintedinGreatBritainINNOVATIONANDLEARNING:THETWOFACESOFR&D*WesleyM.CohenandDanielA.LevinthalEconomistsconventionallythinkofR&Dasgeneratingoneproduct:newinformation.WesuggestthatR&Dnotonlygeneratesnewinformation,butalsoenhancesthefirm'sabilitytoassimilateandexploitexistinginformation.InthispaperweconsidertheimplicationsofthisdualroleofR&Dforthefirm'sincentivetoinvestinR&D.WearguethatrecognitionofthissecondroleofR&DsuggeststhattheeaseandcharacteroflearningwithinanindustrywillbothaffectR&DspendingandconditiontheinfluenceofappropriabilityandtechnologicalopportunityconditionsonR&D.Forexample,weshowthat,contrarytothetraditionalresult,intra-industryspilloversmayencourageequilibriumindustryR&Dinvestment.ScholarsoftechnologicalchangehaveobservedthatfirmsinvestinownR&Dtobeabletoutiliseinformationwhichisavailableexternally(e.g.,Tilton,I97I;Allen,I977;Mowery,i983).1Tilton,forexample,statesthatoneofthemainreasonsfirmsinvestedinR&Dinthesemiconductorindustrywasthat,anR&Deffortprovidedanin-housetechnicalcapabilitythatcouldkeepthesefirmsabreastofthelatestsemi-conductordevelopmentsandfacilitatetheassimilationofnewtechnologydevelopedelsewhere'(I97I,p.7I).Accordingly,wearguethatwhileR&Dobviouslygeneratesinnovations,italsodevelopsthefirm'sabilitytoidentify,assimilate,andexploitknowledgefromtheenvironment-whatwecallafirm's'learning'or'absorptive'capacity.Whileencompassingafirm'sabilitytoimitatenewprocessorproductinnovations,absorptivecapacityalsoincludesthefirm'sabilitytoexploitoutsideknowledgeofamoreintermediatesort,suchasbasicresearchfindings*Cohen'sresearchforthispaperwassupportedbytheDivisionofPolicyResearchandAnalysisoftheNationalScienceFoundation(PRA83-I0664).Fortheircomments,wethankMarkKamlet,BruceKogut,JonathanLeland,RichardLevin,RichardNelson,andSidneyWinter,andtheparticipantsintheI988biennielCongressoftheInternationalJosephA.SchumpeterSociety,especiallyF.M.SchererandJianyeWang.WearealsoparticularlygratefultoStevenKlepperandtwoanonymousrefereesfortheirthoughtfulandextensivecomments.WethankMichaelDodmanandNamPhamforcomputingassistance,andMariaViskochilfortyping.Therepresentationsandconclusionspresentedhereinarethoseoftheauthors.TheyhavenotbeenadoptedinwholeorinpartbytheFederalTradeCommission,itsBureauofEconomics,oranyotherentitywithintheCommission.TheFTC'sDisclosureAvoidanceOfficerhascertifiedthatthedataincludedinthispaperdonotidentifyindividualcompanylineofbusinessdata.1Inadifferentcontext,EvensonandKislev(I973)makeasimilarpointwhentheyobservethattheinternationaltransferofagriculturaltechnologydepends,inpart,upontherecipients'ownresearchefforts.21[569]ECS99570THEECONOMICJOURNAL[SEPTEMBERthatprovidethebasisforsubsequentappliedresearchanddevelopment.Also,inlightofthedependenceofindustrialinnovationuponextramuralknowledge,2absorptivecapacityrepresentsanimportantpartofafirm'sabilitytocreatenewknowledge.Inthisregard,theexerciseofabsorptivecapacityrepresentsasortoflearningthatdiffersfromlearning-by-doing,thefocusofindustrialeconomists'workonfirmlearninginrecentyears(e.g.,Spence,I98I;Lieberman,I984).Learning-by-doingtypicallyreferstotheautomaticprocessbywhichthefirmbecomesmorepracticed,and,hence,moreefficientatdoingwhatitisalreadydoing.Incontrast,withabsorptivecapacityafirmmayacquireoutsideknowledgethatwillpermitittodosomethingquitedifferent.TherolethatR&Dplaysinlearninghasreceivedlittleattentioninthepastbecause,followingArrow(I962)andNelson(I959),economistshaveassumedthattechnologicalknowledgewhichisinthepublicdomainisapublicgood.Likearadiosignalorsmokepollution,itseffectsarethoughttobecostlesslyrealisedbyallfirmsloc