The Incumbent - PDF document

The IncumbentÕs Curse / 1ol.64 (July 2000),1Ð17Rajesh K. Chandy & Gerard J. TellisThe IncumbentÕs Curse?Incumbency,Size,and RadicalProduct Innovationvations. Such firms tend to solidify their market positions with relatively incremental innovations. They may even Rajesh K.Chandy is Assistant Professor of Marketing, Carlson School ofManagement, University of Minnesota.Gerard J.Tellis is Jerry and NancyNeely Professor of Marketing, Marshall School of Business, University ofSouthern California.This study was supported by a grant from the Mar-eting Science Institute and a research initiation grant from the University One lesson from stories of corporate innovation is that itÕsrare for incumbent firms in an industry to reinvent thatindustry. Leadership in the typewriter industry, for exam-ple, changed hands from Remington to Underwood toIBM (with the Ògolf-ballÓ typewriter) to Wang (with theadvent of word-processing) and now to Microsoft. Neveronce did the leader at a particular stage pioneer the nextBusiness Times Journal of Marketing,July 2000argues that incumbents in a particular product generation arereaucracy that they fail to introduce the next generation ofradically new products.xample,Henderson(1993,p.248)arguesthatincumbentsÕeffortswithrespecttoradicallynewtech-nologiesarecharacterizedbyÒincompetenceÓandÒunder-investment.ÓGhemawat(1991,p.161)suggeststhatincumbentsarepronetoÒtechnologicalinertiaÓbecauseoftheirmanyinvestmentsintheexistingmarket.Scherer(1980,p.438)citescasestoarguethatÒnewentrantscon-tributeadisproportionatelyhighshareofallreallyrevolu-tionarynewindustrialproductsandprocesses.ÓSomeauthorsclaimthatestablishedfirmsareslowtointroducenotonlyradicalproductinnovationsbutalsoÒseeminglyminorÓchanges(HendersonandClark1990,p.9).Everychangeinconfigurationinthecomputerdiskdriveindus-try(e.g.,from8.5-inchdisksto5.25-inchdisksto3.5-inchdisks)wasinitiatedbyanonincumbentandledtothedownfallofthepreviouslydominantfirm(Christensen1993).Indeed,RosenbloomandChristensen(1994,p.655)notethattheinabilityorunwillingnessofincum-bentstointroduceradicallynewproductsisoneoftheprominentÒstylizedfactsÓintheliteratureoninnovationHowever,thepreviousconclusionsarebasedoncasestudiesofindividualproducts,notoncross-sectionalstudiesthatuselargesamplesofproducts(e.g.,Ghemawat1991;HendersonandClark1990;Mitchell1991).Thecasesstud-iedhavebeenquitespecialized,includingphotolithographicaligners(Henderson1993;HendersonandClark1990),medicaldiagnosticimagers(Mitchell1991),andprivatebranchexchanges(Ghemawat1991).Thefewmultiproductstudiesthatexistuseconvenienceratherthanformalsam-pling(CooperandSmith1992;RosenbloomandChristensen1994;TushmanandAnderson1986;Utterback1994).More-mostofthesestudiesuse data from theUnitedStates.Thisstateofknowledgeraisesthefollowingquestions:How pervasive is the phenomenon of the incumbentÕs curse?In particular, what percentage of radical innovations followthe script outlined previously?How consistent is the phenomenon? Has the increasing sizeand complexity of firms over time accentuated it? Does itIs the phenomenon a curse that invariably afflicts largeincumbents in current industries? Is it driven by incumbencyThecurrentstudyexploresthesequestions.Researchonthistopicisimportantforseveralreasons.First,radicalinno-ationisanengineofeconomicgrowthandasourceofbet-terproducts.Anunderstandingoftheseissuesisimportantforconsumersandpublicpolicymakers.Second,radicalinnovationchangestheentireshapeofindustriesandmakesthedifferencebetweenthelifeanddeathofmanyfirms(see,e.g.,CooperandSchendel1976;Schumpeter1942).Indeed,thehistoryofbusinessislitteredwiththegraveyardsofentireindustriesthatweredestroyedbyradicalproductinno-ations.Thetelegraph,gaslighting,andtypewriterindus-triesarecasesinpoint.Thus,managersneedtoknowhowtoinitiateandmanageradicalproductinnovation.Third,manylargeincumbents,especiallyintechnologicallyintenseindustries,spendhugeamountsofresourcesonresearchanddevelopment(R&D).AconfirmationoftheincumbentÕscursewouldsuggesttheneedforareconfigurationofeithertheirresourcesortheirdepartmentsresponsibleforinnova-tion.Forexample,afterlisteningtospeakerafterspeakeremphasizetheincumbentÕscurseattheMarketingScienceInstituteÕs1997conferenceonreallynewproducts,oneseniormanagerfromalargemultinationalfirmcommentedthathisoverarchingfeelingwasÒoneofhopelessness.ÓMoreresearchontheextentandcausesoftheseproblemsmayengenderinsightinsteadofhopelessness.Insightisthegoalofthecurrentstudy.Weaimtobuildonagrowingstreamofresearchinthemarketinglitera-tureonthesourcesofradicallynewproductsandthecharacteristicsofradicallyinnovativefirms(e.g.,ChandyandTellis1998;GatignonandXuereb1997;Ghemawat1991;MoormanandMiner1997;Olson,Walker,andRuekert1995).Inthenextthreesections,wedescribethetheory,method,andresultsofthisstudy.Thelasttwosec-tionsdiscusstheimplicationsandlimitationsoftheTheoryradicalproductinnovationisanewproductthatincor-poratesasubstantiallydifferentcoretechnologyandpro-videssubstantiallyhighercustomerbenefitsrelativetopre-viousproductsintheindustry(ChandyandTellis1998).Aradicalproductinnovatoristhefirmthatfirstcommercial-izesaradicalproductinnovation(EttlieandRubenstein1987).Inaninnovationcontext,incumbencyreflectswhetherafirmparticipatedinthepreviousgenerationofproducts.Thus,anincumbentisafirmthatmanufacturedandsoldproductsbelongingtotheproductgenerationthatprecededtheradicalproductinnovation(Henderson1993;Mitchell1991;MitchellandSingh1993).Firmsizereferstothescaleofoperationsofanorganization(PriceandMueller1986).provide a theoretical background to understand thebehavior and performance of firms in the realm of radicalinnovations, we next discuss the theory of S-curves.Theory of S-CurvesThetheoryofS-curvescomesfromthetechnologymanage-mentliteratureandexplainstheoriginandevolutionofradicalinnovations(Foster1986;Sahal1985;Utterback1994;Utter-backandAbernathy1975).Thistheorysuggeststhattechnolo-giesevolvealongaseriesofsuccessiveS-curvesthatdrivevar-iousnewproductintroductions(ChandyandTellis1998).TheS-curveemergesbecauseanewtechnologyoffersfewcon-sumerbenefitswhenfirstintroduced,offersrapidlyincreasingconsumerbenefitsasitdevelops,andoffersslowlyincreasingconsumerbenefitsasthetechnologymatures(seeFigure1).understand this phenomenon better, consider an), which is in its maturity. Assumethat at some point during the maturity of this technology, anew technology, T, emerges, which leads to a new product, The IncumbentÕs Curse / 3 S-Curves Notes: Figure adapted from Chandy and Tellis (1998).Initially, because of problems implementing the technology,benefits are inferior to those of T, so the new productÕssales are below those of existing products and occur mostlyto highly innovative, price-insensitive consumers. However,with research, Tbegins to improve rapidly in consumerbenefits and ascends its own S-curve. A point may come(point b in Figure 1) when Tpasses the existing technology,, in benefits. At this point, the market considers the newproduct based on Ta radical product innovation. Sales ofthe new product then take off as consumers increasinglyshift to the new product to satisfy their needs. Sales of thenology, T, may make a renewed effort to be competitive byredoubling their efforts in the old technology. This effortmay result in some short-term improvement in Tmarket breakthrough (point c in Figure 1). However, invest-ments in the new technology, Tgreater benefits to consumers than do those in Tucts based on Tcontinue to outsell those based on Told product dies out when the new product completelyAstimepasses,Tceasestoimprovesubstantially,andsalesbasedonthenewproductslowdown.Ifanotherradi-calproductinnovationemerges,thecyclemayrepeatitself.Ifthisnewinnovationneversurpassestheoldoneincon-sumerbenefits,itdiesoutwithoutevertriggeringatakeoff.Incumbents and Radical InnovationThetheoryofS-curvessuggeststhreereasonsincumbentsmaybereluctanttointroduceradicalinnovations:perceivedincentives,organizationalfilters,andorganizationalrou-. Incumbents may perceive smallerincentives to introduce radical product innovations thannuanced analytical view, see Tirole 1988). The reason is thatthey derive a significant stream of rents from existing prod-ucts based on the current technology, whereas nonincum-bents derive no such rents. Note in Figure 1 that radicallynew products hold the potential to make existing productsobsolete. Introducing a radically new product could there-fore jeopardize the rents from existing products. Incumbentstherefore have a lower marginal incentive than nonincum-bents to develop or commercialize radical innovations in thecal models of innovation indicate that incumbents wouldmaximize overall profits by not introducing a radical prod-uct innovation even when (1) the radical innovation itself ismore profitable than the existing product (Reinganum 1983)and (2) the fixed cost of offering the radical innovation iszero (Ghemawat 1991). What these models do not fully takeinto account is the dynamic nature of the radical innovationspletely replace the existing market in the long run.Organizationalfiltersanizationalfiltersarecogni-tivestructuresthatscreenoutinformationunrelatedtotheorganizationÕsimportanttaskstofocusitsattentiononthesetasks.Organizationaltheoristsarguethattheorganizationalltersofincumbentsmakethemlesseffectiveatradicalinnovation(e.g.,HannanandFreeman1977;HendersonandClark1990;NelsonandWinter1982).IncumbentsÕsuccessinthecurrentproductcategoryispartlyduetoorganizationalltersthatenablethemtofocusefficientlyontheircurrentchallenges.Inparticular,thesefiltershelpthemprocesscon-sumerrequestsorcomplaints,channelthemtothemanufac-turingdepartmentordistributors,andensurethatthecurrentproductsmeetconsumerexpectationsaseffectivelyaspossi-ble.Thus,thefiltersservetodirectmanagersÕattentiontomaximizetheutilityofthecurrenttechnologyforcurrentcustomers.However,radicalproductinnovationsinvolveasubstantiallynewtechnology.Asaresult,theseveryorgani-zationalfiltersmaycauseincumbentstobelesseffectivethannonincumbentsatspotting,developing,andmarketingradicalproductinnovations(Henderson1993).Organizationalroutinesanizationaltheoristsalsostatethatincumbentsdeveloporganizationalroutinesorprocedurestocarryouttherepetitivetasksofmanufacturinganddistribut-inglargevolumesofthecurrentproductefficiently(HannanandFreeman1977;HendersonandClark1990;Nelsonandinter1982).WithintheR&Ddepartment,theroutinesaregearedtowardefficientlydevelopingincrementalinnovationsbasedonthecurrenttechnology.However,theroutinesareineffectiveatdevelopingradicalproductinnovations,becausethelatterarebasedonasubstantiallydifferenttechnology(Henderson1993).Moreover,adoptionofradicalinnovationsouldobsoletemanyoftheseroutinesandrequirethedevel-opmentofnewroutines,whichisdifficult,costly,andrisky(HannanandFreeman1977;NelsonandWinter1982).Man-agersmayreasonthattheseveryroutineshavehelpedthefirmridethroughthesuccessofthecurrenttechnology.Thus,theyhaveavestedinterestinthecurrentorganizationalroutines(Staw1981)andarereluctanttoembraceradicalinnovations.. Although the previous the-ory suggests that incumbents are less likely to innovate rad- Journal of Marketing,July 2000tunities of incumbents in terms of market capabilities (e.g.,Srivastava, Shervani, and Fahey 1998). Indeed, if incum-bents can muster the willingness to cannibalize their owninvestments (Chandy and Tellis 1998), they can exploit theirmany resources to lead with radical innovations. We high-light three market capabilitiesÑcustomer knowledge, cus-tomer franchise, and market powerÑand note the opportu-nities they create for radical innovation.First,incumbentshavegreaterknowledgeaboutcustomersinthemarket.Theycommunicatewiththemandunderstandtheirneedsandmayhavedetailedrecordsoftheirpurchasesandbehavior.Thus,incumbentsmaybebetterabletogaugethevalueofradicalinnovationsandmarketthemwhentheyareintroducedthannonincumbents.Second,incumbentsenjoyacustomerfranchisethatcanbebeneficialinthecontextofradicalinnovation.Consumersperceiveradicalproductinnovationsasriskypurchases(seeBauer1960;Folkes1988;Gregan-PaxtonandJohn1997).Totheextentthatconsumersinaparticularmarketarefamiliarwithanincumbent,theyouldbelessapprehensiveaboutpurchasingaradicalproductinnovationfromthisfirm.Third,incumbentsalsopossessgreatermarketpower,whichgivesthempreferentialaccesstodistributionchannelsrelativetononincumbents(Mitchell1989).Itmayalsoenablethemtosustaintheirmarketpresenceduringthelongandunprofitableperioduntiltheproducttakesoff.Recentresearchindicatesthatthisperiodtilltakeoffaver-aged17.7yearsforproductsintroducedbeforeWorldWarIIand5.8yearsforproductsintroducedafterWorldWarII(GolderandTellis1997;seealsoUrbanandHauser1993).. The literature and extant theory strongly sug-gest that incumbents are less likely than nonincumbents tointroduce radical innovations. At the same time, incumbentshave many market capabilities to do so. Although the focusof empirical studies so far has been mostly on noninnovativedeveloping and introducing radical innovations. This con-large cross-section of innovations, that can identify theactual pattern of radical innovation by incumbents.Firm Size and Radical Product InnovationIncumbents that have successfully survived in a market tendalso to be large. Thus, size and incumbency are likely to bepositively correlated. Our concern is the role of size in radi-cal innovation after we control for the influence of incum-bency. Here again, the literature strongly suggests that largeirms may not be radical innovators, primarily because ofthe theory of inertia (Acs and Audretsch 1991; Cohen 1995;Cohen and Levin 1989; Scherer 1991). We review this the-of large firms.. As firms grow large, they are prone tothe forces of bureaucratic inertia(Tornatzky and Fleischer1990). The key factor that contributes to such inertia is thenumber of employees that work in large firms. The numbersof employees make it difficult to manage large firms, sothese firms develop layers of administrative staff and formalSchoenherr 1971; Kasarda 1974; Terrien and Mills 1955).Although the staff and rules may cause large firms to func-tion more effectively in serving current customers with cur-rent technology, they can also render them slow to react toradically new products (e.g., Kimberly 1976).NoteinFigure1thatradicalinnovationsnotonlyareanoutgrowthofthecurrenttechnologybutalsoresultfromtheapplicationofasubstantiallydifferenttechnologytothesameproblem.Inalargefirm,newideasthatgivebirthtoradicalinnovationsmustmovethroughmorelayersofadministration.Innovativeemployeesoftenmustlaborthroughlayersofureaucraticresistancetogetapprovalfortheirideas.Theprocessincreasesthelikelihoodofscreeningortemperingofradicalideas.Alternatively,thetimeandtroubleittakestogainapprovaldistractsscientistsfromconcentratingoncreativeork(AcsandAudretsch1991;Scherer1980).Intheextremecase,theseimpedimentscanfrustrateinnovatorsandleadthemtolookelsewhereforsupportortostarttheirowncompanies.relatedproblemisthatofrewardsforinnovatorsinlargeorganizations.Asorganizationsgrowinsize,theyinvolvemorelevelsofscreeningandgroupdecisionmaking.Thus,thecontributionsofindividualsgetincreasinglydiluted,soinnovatorsarelessabletocapturethebenefitsoftheirefforts(Cohen1995;Schumpeter1942)andhavefewerincentivestodevelopradicallynewproducts(CohenandLevin1989).Insummary,largefirmsarelesslikelythansmallfirmstoprovidetheresponsive,risk-takingatmosphereneededforthedevelopmentofradicalproductinnovations.Opportunities of large firms. Large firms have greatopportunities to develop and introduce radically new prod-ucts. In particular, large firms have enormous financial andtechnical capabilities, which they can harness for radicalinnovation. Large firms may have the economies of scope tospread the risks of new ventures widely (Arrow 1962; Gal-braith 1968). Similarly, large firms have a large volume ofsales over which to spread the fixed costs of R&D(economies of scale; Comanor 1965). As a result, large firmsare less vulnerable to the failure of a particular developmentproject, because it would entail a smaller proportion of theirresources than it would for a small firm.The deeper pockets of large firms also enable them tomaintain state-of-the-art scientific facilities and hire qualityscientific personnel. Moreover, some researchers argue thatcapital market imperfections provide an advantage to largeirms, because these firms have greater access to internaland external funds to finance risky R&D projects (Cohenand Levin 1989). Thus, large firms may be more capable ofpursuing radical innovations than small firms.. The theory of inertia strongly suggests thatlarge firms would be unlikely to introduce radical innova-tions. At the same time, large firms have many financial andtechnical capabilities to do so. Thus, there is a conflict aboutthe role of size on radical innovation. This conflict beganwith the work of Schumpeter (1942) and continues today(Chandy and Tellis 1998). Empirical research that helpsenlighten the debate would be useful.Country,Incumbency,and Radical InnovationTheproportionofinnovationsfromincumbentscomparedwithnonincumbentsislikelytobelowerintheUnitedStatesthaninotherindustrializednations,suchasJapanandthoseinWesternEurope(AcsandAudretsch1991;Imai1990; The IncumbentÕs Curse / 5Scherer1991).Thereasonforthisdiscrepancymaybedif-ferencesininstitutionsandpopularculturebetweentheUnitedStatesandtheseothernations(PatelandPavitt1995).TheUnitedStateshashistoricallyenjoyedanactivemarketforventurecapital,whichmakesfinanc-ingforlessestablishedfirmseasierintheUnitedStatesthaninmanyothercountries(Saxenian1994).Thus,finan-cialinstitutionsmayplayaroleinencouragingradicalinnovationamonglessestablishedfirmsintheUnitedStatescomparedwithJapanandEurope.Atthesametime,governmentpoliciesinthelattercountrieshavetradition-allyfavoredtechnologyventuresbylarge,establishedirmsoverthosebysmallfirms(FitzroyandKraft1991;Scherer1991;Urabe,Child,andKagono1988).Largeincumbentsinthesecountriesenjoyfinancialandtechno-logicalsupportthatisunavailabletosimilarfirmsintheUnitedStates(MagazinerandPatinkin1989).Therefore,incumbentsandlargefirmsinJapanandEuropearemorelikelytoindulgeinradicalinnovationsthanthoseintheUnitedStates.The entrepreneur in popular culturethe United States celebrates risk takers. Indeed, Americanentrepreneurs are celebrated figuresÑthe pride of Ameri-cans, the ideal of would-be entrepreneurs, and the envy ofStates, failure is stigmatized less than in some other coun-tries (see Patel and Pavitt 1995). On the contrary, succeed-ing after a string of failures enhances the glory of the entre-preneur. Therefore, people are more motivated to engage inentrepreneurial ventures to commercialize new technologyin the United States than in other countries. This factorcould increase the pool of small, nonincumbent firms thatare likely to become radical innovators in the United States.. On the basis of the previous arguments, wexpect the profile of U.S. radical innovators to be differentfrom that of non-U.S. radical innovators. Specifically, wexpect a greater proportion of U.S. innovators than innova-tors from other countries (especially Japan and WesternEurope) to be small firms or nonincumbents. Stated differ-ently, the incumbentÕs curse is more likely to apply in theUnited States than in non-U.S. contexts. Table 1summarizesThe next section describes our research approach, samplingResearch Approachcollected our own data to address the questions posedpreviously, because we could not find any satisfactory data-base of radical innovations. We use the historical approachto data collection (Golder 2000; Savitt 1980; Smith and Lux1993) for three reasons. First, we study events from the past,many from the distant past. Second, the easier alternativeapproach of surveying current managers can suffer fromsevere memory or self-report biases. Third, the historicalapproach enables us to study the effects of time on theincumbentÕs curse. Specifically, this approach can providean answer to the question, Are todayÕs incumbents more orless innovative than the incumbents of the distant past? Antime order of events that is best obtained by the historicaling, but well worth the effort because of the insight and nov-elty of findings it provides. Thus, in scope and design, ourstudy is similar to Golder and Tellis (1993), though in con-tent it is similar to Foster (1986) and Utterback (1994).use the following five criteria to include data in ourConfirmation: At least two published sources cite the sameNeutrality: The sources have no overt interest to bias their: The sources are based on independent obser-ation (i.e., they did not come from a single source, such as: The sources are well respected or have a historyContemporaneity: The sources report as close to the time ofthe event as possible.Overall,weuseinformationfrommorethan250booksand500articlesinperiodicals.Theinformationsearchanddatacollectiontasksaretimeandeffortintensive,involvingoneauthorandninetrainedassistantsoveraperiodoffouryears.oid sampling biases, we use a relatively formal sam-pling frame to choose the product categories and innova-tions for our study. This approach contrasts with prior Summary of Theoretical Discussion ype of FirmTheory Against Radical InnovationTheory for Radical InnovationIncumbentIncentives, filters, and routines enhance Market capabilities facilitate radical innovation.Large firmBureaucratic inertia dampens radical innovation.Financial and technical capabilities facilitate U.S. nonincumbentsÑInstitutions and culture foster radical innovation Journal of Marketing,July 2000research, which relies on convenience samples. We choosethe sample to satisfy three objectives:First,intheinterestsofeffortandcomparability,werestrictourstudytotwobroadproductclasses:consumerdurablesandofficeproducts.Thesetwoproductclasseshavebeenstudiedinpreviousresearch,especiallyintheliteratureoninnovationdiffusionandmarketpioneering(GolderandTellis1993,1997;GortandKlepper1982;Sultan,Farley,andLehmann1990).Thecurrentresearchonthesecategoriesaddstothecumulativeknowledgeinthearea.Thesetwoproductclassesarealsoattractivebecauseinnovationsintheseclasseshavewidelyvaryingdatesofintroduction.Thus,theyenableustoidentifygen-eralizationsovertimeorhistoricaltrendsinthepatternofradicalproductinnovation.Second, we seek product categories with high unit sales. Inpractice, we restrict the sample to categories with more thandata were available when this study began. This cutoff valuefor sales is a conservative figure; it helps ensure that the cat-gories in our sample include innovations that have truly hada large enough impact on consumers to form huge markets.The 1994 volume of Predicastsprovides the list of categoriesand their annual sales. This goal leads to a list of 49 productcategories.innovation in the category varies substantially from the tech-nology used in the previous product generation. Prior theory(Cooper and Smith 1992; Utterback 1994). To determine aradically new technology in each category, we use a two-stepprocedure. First, we identify the most significant product inno-ations in each product category. We obtain information on theinnovations from books on the history of the respective cate-gories, as well as from past issues of business and technologyperiodicals. To ensure the eligibility of information andincrease the relevance to current products, we restrict the sam-ple to innovations introduced after 1850. The procedure leadsto 93 significant innovations. Second, each significant innova-tion is assigned a rating for radicalness relative to the previousproduct generation by a team of three experts. The MeasuresOur sample size compares favorably with those in otherhensive meta-analysis on diffusion models of new products,Sultan, Farley, and Lehmann (1990) show that prior studieson average have a sample size of 14 product categories.Golder and TellisÕs (1993, 1997) studies on market pioneer-ing and sales takeoff have sample sizes of 35 and 31 prod-uct categories, respectively.The sample derived from this procedure enjoys severalstrengths. First, the sample does not suffer from survivalbias. The reason is that we include any incumbent or entrant,large or small, surviving or dead, so long as it was the firstto introduce the radical innovation. We are able to do so byreferencing articles about the radical innovation writtenclose to the time the innovation occurred, not by relying onself-reports by current participants in the industry. Second,try of origin of the firm or the innovation is not a criterionfor selection. Third, our sample covers an extensive timeperiod: Introduction dates span close to 150 years. Fourth,all our data come from publicly available sources that areaccessible to any interested party.Radical Innovation. At present, the literature does notcontain a measure of the radicalness of innovations, yet thisis a critical variable in the field of innovation and new prod-ucts. To initiate research in scale development and providegreater objectivity to the classification and rating of innova-tions, we develop a new index of radicalness.Recall that our definition of a radical product innovationinvolves two dimensions: whether a new product (1) incor-porates a substantially different core technology and (2) pro-vides substantially higher customer benefits relative to theprevious product generation in the category. We had threexperts rate the innovations on these two dimensions of rad-icalness. Each dimension involves a nine-point scale. Fordifferences in core technology relative to the previous prod-uct generation, they rated each innovation on a scale rangingfrom 1 (not at all different) to 9 (substantially different). Forsuperiority in user benefits relative to the previous productgeneration, they rated each innovation on a scale rangingAllthreeratersareknowledgeableaboutthehistoryofinnovationintheproductclassesstudied.Allthreeratershavepublishedarticlesoninnovationornewproductsinleadingmarketingjournals.Nevertheless,becausewearestudyingawidevarietyof94innovations,forwhichevenexpertsmaynotrememberallthedetails,andbecausewewantedtoensurethatallraterswereexposedtoconsistentinformation,wegavetheraterskeyinformationoneachofthetwodimen-sions(consumerbenefitsandtechnology)foreachofthecur-rentinnovationsanditspreviousproductgeneration.irmsizeirmÕssizehasmanymeasures,themostcommonbeingnumberofemployees,salesvolume,orvalueofassets.Empiricalresearchindicatesthatinthecontextofradicalinnovation,thesealternativedefinitionsoffirmsizeprovidesimilarresults(Agarwal1979a,b;ChandyandTellis1998;Child1973).ThemostcommonmeasureofsizeintheinnovationliteratureisthenumberofafirmÕsemployees(CohenandLevin1989;Pavitt,Robson,andTownsend1987).Thismeasureistheoreticallyappealing,becausemanyoftheproblemsoflargefirms(suchasincreasedbureaucracyandinertia)areduetotheincreasedneedforcoordinationasafirmemploysmorepeople(Kimberly1976).Therefore,weopera-tionalizefirmsizeasthenumberofemployeesinthefirm.measure firm size, we determine the number of full-time employees in the firm at the time the radical productinnovation was commercialized (e.g., Cohen and Levin1989; Pavitt, Robson and Townsend 1987). We define a firmas small if it employed fewer than 100 employees, mediumif it employed between 100 and 2500 employees, and largeif it employed more than 2500 employees (see Pavitt, Rob-son, and Townsend 1987). In our analysis, we use both acontinuous measure of firm size and the previous categori-cal measure. Our results are robust to the cutoff betweensmall and medium firms in the categorical measure.or publicly traded firms, we obtain size informationfrom publications such as MoodyÕs Industrial Manual(which extends back to the turn of the century) and the dard & PoorÕsmanual. For privately held firms, we obtaininformation from historical records such as company direc- The IncumbentÕs Curse / 7 This definition thus excludes all firms that commercialized theinnovation after the first firm. As we note in the Additional Analy-sis section, relaxing this definition to include later entrants makesour results even stronger.Industrial Laboratories Directory),contem-(e.g., employee time sheets for 1880 from the Edison Elec-tric Light Co.). For some small start-up firms, we are unableto nail down the precise number of full-time employees atthe time the firm commercialized the radical product inno-ation. However, in every case we are able to determine ifthe firm had fewer than 100 employees at that time.Informationonthenumberofemployeesismorecom-monlyavailablethanthatonthealternativemeasures.Never-theless,tovalidatethismeasureoffirmsize,wealsocollectinformationonfirmsalesandassetsintheyearofintroductionoftheinnovationfor27oftheinnovatingfirmsinoursampleforwhichdataareavailable.Weconvertthesalesandassetinformationinto1980U.S.dollarsbymultiplyingtherawsalesandassetsfiguresbytheappropriateexchangerateandinflationrateindices.Wethencorrelatethesestandardizedsalesandassetsvariableswiththenumberofemployeesintherelevantfirms.Thecorrelationbetweenthenumberofemployeesinafirmanditsstandardizedsalesis.75(andthatbetweenemployeesandstandardizedassetsis.73(.001).Thus,inadditiontotheoreticalsupport,ourmeasureofsizeisalsocloselyrelatedtothealternativemeasures.InnovatorandincumbentCooperandSchendelÕs(1976)andUtterbackÕs(1994)studies,wedefinetheradicalinnovatorasthefirmthatfirstcommercializedtheradicalinnovation.ollowingGolderandTellis(1993,1997),GortandKlepper(1982),andGortandWall(1986),wedefinetheintroductiondateasthedateoffirstcommercializationoftheradicalinnovation.Tomeasureincumbency,wefirstidentifytheproductgenerationthatprecededtheradicalproductinno-ation.FollowingHenderson(1993),Mitchell(1991),andMitchellandSingh(1993),wethendefineafirmasanincum-bentifitmanufacturedorsoldproductsthatbelongedtothepreviousproductgenerationontheintroductiondate.Wedefineitasanewentrantifitdidnot.Sixoftheinnovationsinoursamplefulfilledneedsthatwerenotmetbyanyprevi-ousproduct.Forexample,nospecificproductsfulfilledtheneedslatermetbytelephoneansweringmachines.Inthesecases,thepreviousproductgenerationisdefinedastheser-viceortechnologybywhichtheparticularneedwasfulfilledbeforetheintroductionoftheinnovation(e.g.,humananswer-ingserviceinthecaseoftelephoneansweringmachines).. To study trends in the incumbentÕs curse overtime, we compare the profile of radical innovators beforeand after World War II. However, to detect continuous trendsranges from 1 for the first year in our sample (1851) to 148When presenting the categorical analyses, we focus onthe preÐ and postÐWorld War II period in particular, becausethat period saw the birth of many fundamental new tech-puting fields (Sakudo and Shiba 1994; Teitelman 1994).These technologies were first applied to civilian uses in the significant innovations. The key results from this analysis are con-period after the war. World War II was also a major eco-nomic event that significantly altered the business environ-ment of many countries. Many incumbents found their for-tunes dramatically changed by the war, and a newgies developed during this period. Finally, separatingpreÐWorld War II innovations from postÐWorld War II inno-sample generally represent breakthroughs in electrification,mechanization, and chemistry. Products in the later sampleare largely based on breakthroughs in electronics and com-. We measure nationality as the countrywhere the firm was headquartered at the time it introducedthe radical innovation. In all but one of the innovations inour sample, the firmsÕdevelopment efforts leading to theradical innovation were based in the country where the firmsFrom the original set of 93 innovations, we could collectreliable information on each of the key variables of interestfor 64 innovations, which constitute the sample for ouranalysis. The Appendix presents the list of innovations in thesample. This section first presents bivariate categoricalanalyses of the dependent variable and then presents multi-ariate analyses of the continuous variables.Categorical Analysisget a better feel for the phenomenon, we first presentbivariate categorical analyses of the key relationships. To doso, we categorize three of our continuous variables: radicalinnovation, time, and size. First, we classify the sample of64 innovations as radical if the average rating from all threenine-point scale. Fifty-three innovations meet this criterion.Second, we label firms with fewer than 100 employees assmall, 100Ð2500 employees as medium, and more than 2500employees as large. Unfortunately, we have only sixmedium firms by this criterion, so we collapse the class ofsmall and medium firms into one group. Third, we catego-rize time as preÐ and postÐWorld War II, for reasons statedpreviously. This section reports the characteristics of radicalinnovations on each of the key variables of interest.. Recall that the incumbentÕscurse suggests that incumbents are much less likely thannonincumbents to introduce radical innovations. Of theinnovations in the sample, 53% are from nonincumbents,whereas 47% are from incumbents. These proportions arenot significantly different from each other (This result implies that incumbents may be as likely to intro-duce radical innovations as nonincumbents. Thus, the over-all results do not support the incumbentÕs curse. Journal of Marketing,July 2000Incumbency Status of Radical Innovators over Time FIGURE 3 Size of Radical Innovators over Time Ourtheoreticaldiscussionsuggeststhatlargefirmsarelesslikelytointroduceradicalinnovationsthansmallfirmsbecauseofinertiaandbureaucracy.Totestthisproposition,weclassifyradicalinnovationsbyfirmsize.Overall,58%oftheinnovationsinthesamplearefromsmallandmediumirms,whereas42%oftheinnovationsarefromlargefirms.Thedifferencebetweenthesetwoproportionsisstatisticallysignificant(.001).Thisresultseemstosupportthetheoryoftheinertiaandbureaucracyoflargefirms.alidquestionatthispointishowthecombinationofsizeandincumbencyaffectsradicalinnovations.Tableshowsadramaticinteractioneffectofthesetwovari-ables.Smallernonincumbentsarealmostfourtimesaslikelytoberadicalinnovatorsthanlargenonincumbents.Incontrast,largeincumbentsarealmosttwiceaslikelytobetheradicalinnovatorsthanaresmallandmediumincumbents.Thus,sizeseemstofavorincumbentsanddis-ends over time. The previous results indicate thatwhen analyzed separately, incumbents tend to introduce rad-ical innovations in roughly the same proportion as nonin-cumbents. Has this pattern remained steady over time? Weuse the term ÒolderÓfor preÐWorld War II innovations andÒrecentÓ for postÐWorld War II innovations.Figure2showsthattheolderinnovationsinthesamplearemostlyfromnonincumbents(73%);relativelyfewarefromincumbents(.0001).Butaverydifferentpictureemergeswhenwefocusonrecentinnovations.Incumbentssignificantlyoutnumbernonincumbentsforrecentinnovationsinthesample(74%to26%,.0001).TheseresultsindicatethatwhereastheincumbentÕscursemayhavebeenaprobleminthelatenineteenthandearlytwentiethcenturies,itisnotcommoninrecenttimes.Aswithincumbency,thereversalovertimealsooccursforthesizeofinnovators.Figure3showsthatsmallandmediumfirmsaccountforamajority(83%)ofolderinno-ationsinthesamplerelativetolargefirms(17%)(.0001).Thepatternchangesdramaticallyinrecenttimes,whensmallerfirmsaccountforonly26%oftheinnovationscomparedwith74%forlargefirms(These results suggest that small firms or nonincumbentswere more radically innovative previously, probablybecause of their nimbleness or lack of inertia relative to thecomplexity of prevailing technology. However, in recenttimes, large firms and incumbents account for more radicalinnovations, probably because their large financial, techni-cal, or market capabilities enable them to master the com-plex technologies better. A review of the historical evidencebrings to light an additional explanation for the innovative-ness of incumbents and large firms in recent years. In thepostÐWorld War II period, large firms and incumbents insti-tuted organizational features that better support radical inno-ation (Chandler 1956; Williamson 1975). These organiza- Radical Innovators by Incumbency and Size Small and MediumLargeNonincumbent42%11%Incumbent17%30% The IncumbentÕs Curse / 9Radical Innovators by Incumbency Status and Country of Origin tional features may make them willing to cannibalize theirwn past investments (Chandy and Tellis 1998). We coverInternationalcomparisonshatproportionofradicalinnovationscomefromU.S.firms?Oftheinnovationsinthesample,62%arebyU.S.firms(001).How-er,unlikethepreviousresults,thisproportionremainssteadyacrosstimeperiods(Figure4).Oftheolderinnova-tionsinthesample,67%arefromU.S.firms,comparedwith57%oftherecentinnovations,butthisdifferenceisnotsignificantlydifferentfromzero(Themajorchangeacrossthesetwoperiodsistheemer-genceofJapanasasourceofinnovation.Whereasallbutoneoftheoldernon-U.S.innovationsarefromEuropeanfirms,bothJapaneseandEuropeanfirmsareresponsibleforrecentnon-U.S.innovations.Theseresultssuggestthatintheprod-uctclasseswestudy,theUnitedStatesdominatesbutdoesnotexclusivelycontrolthefieldofradicalinnovation.ItalsosuggestsnostrongtemporalpatternsinU.S.innovativenessinthesecategories.WesternEuropeannationsseemtohavelostsomegroundinrecentyearstoJapanesefirms.hadexpectedthatbecauseofinstitutional(e.g.,governmentpolicies,availabilityofventurecapital)andcultural(e.g.,attitudestowardentrepreneurship)factors,U.S.innovationswouldbemorelikelytocomefromnon-incumbent,smallfirmsthannon-U.S.innovations.FigurepresentstheresultsonincumbencyamongU.S.andnon-U.S.innovators.OftheU.S.innovationsinthesample,55%arebynonincumbents,comparedwith45%byincumbents,thoughthisdifferenceisnotstatisticallysig-nificant(.29).Non-U.S.innovationscomefromincumbentsandnonincumbentsinequalproportionsAn analysis of the size distribution of U.S. versus non-U.S. innovators reveals similar patterns. Figure 6shows thata majority of the U.S. innovations in the sample are fromsmall and medium firms rather than large firms ()wever, non-U.S. innovations come from smalland medium firms rather than large firms in roughly equalalso conducted a further analysis of the trends overtime within the sample of radical innovations introduced by Nationality of Radical Innovators over Time Radical Innovators by Size and Country of Origin 10/ Journal of Marketing,July 2000U.S. firms (n= 33). This analysis indicates that the sourcesof U.S. radical innovations also change dramatically in thepostÐWorld War II period. The older U.S. innovations tendto come from smaller firms and nonincumbents, whereas themore recent U.S. innovations tend to come from large firmsand incumbents, thus paralleling the results for the overallsample. However, because of the limited sample size in thisanalysis, we do not present quantitative results or signifi-cance levels (see Frank and Althoen 1994).The results so far are based on a bivariate analysis ofradical innovations by categorized independent variables.The previous results do not distinguish between highly rad-ical and less radical innovations, nor do they explain theariation of the relationship over the range of time and firmsize. Therefore, we next present multivariate results fromour full sample of 64 innovations, thus using the full rangeof information available in our sample.Multivariate Results understand the extent to which our key variables (incum-bency, size, time, and nationality) explain the level of radi-calness of the innovations, we estimate the following regres-(1) Radicalness of Innovation= Incumbent+ Largeime+ (Time (Time Large)(Time US)+ where Radicalness of Innovation is a variable= i= 1 (Technology Rating)i= 1 (Benefits Rating), (Tech-is rater iÕs rating of the extent to which thecore technology in the innovation is substantially differentfrom that used in the previous product generation, and (Ben-efits Rating)is rater iÕs rating of the extent to which theproduct provides substantially superior benefits relative tothe previous product generation. Incumbent, Large, and USare dummy variables defined as previously; Time is a con-tinuous variable ranging from 1 (for the year 1851) to 148s are coefficients to be estimated, andis a vector of errors assumed to be i.i.d. normal.Thechoiceandspecificationofindependentvariablesfol-lowsfromourtheoreticaldiscussion.Totesttheextenttowhichourresultsarerobusttodifferingoperationalizationsofirmsize,wealsoestimatethefollowingregressionequation:(2)Radicalness of Innovation= Employees+ ime+ (Time (Time Employees)(Time US)+ whereEmployeesisthenumberoffull-timeemployeesinthecompanyintheyearitintroducedtheinnovation,arecoef-icientstobeestimated,isavectoroferrorsassumedtobei.i.d.normal,andallothervariablesaredefinedaspreviously.able 3presents the stepwise results of estimating Equa-tion 1. The regression results, as well as the tests of bivari-the categorical analyses. In particular, the regression analy-sis indicates that the main effect of incumbents (Model 1)and large firms (Model 3) on radical innovations is negative.However, in the regression with all independent vari-ables (Model 4), neither the main effect of incumbency norits interaction with time is significantly different from zero.This result indicates that innovations introduced by incum-bent firms are no less radical than those introduced by non-incumbents. (Incumbency and size are positively related, sothe difference in these results from the bivariate case alsomay be due to the collinearity between incumbency andsize.) The results also indicate a strong interaction betweenirm size and time, similar to that in the categorical analysis.The results indicate that though larger firms introduce lessradical innovations, in recent years the trend is just the oppo-site. Indeed, the size and sign of the standardized coefficientnegative effect of size on radicalness. Thus, in recent years,innovations introduced by large firms are more radical thanthose introduced by smaller firms. Similarly, there is aUS. The coefficient indicatesthat in recent years, innovations introduced by U.S. firms aremore radical than those introduced by non-U.S. firms.The previous analysis uses a truncated measure of size.ous measure of sizeÑnumber of employees. Note that thepattern of results is similar across the two measures, despitedifferent sample sizes. These results indicate that the effects Regression Results Dependent Variable:Radicalness of InnovationModel 1Model 2Model 3Model 4Model 5 t-Valuet-Valuet-Valuet-Valuet-ValueIncumbentÐ.21 (1.7)Ð.41 (1.3)Ð.19(.6)Ð.31 (1.0)Ð.27 (.4)LargeÑÑÐ.85(1.6)Ð1.05(1.9)ÑÐ4.06(2.1)USÑÑÑÐ.51 (1.7)Ð1.28(1.0)earÑÐ.47 (2.7)Ð.56(3.2)Ð.92 (3.6)Ð.96(1.5)IncumbentÑ.43 (1.1).00(.0).14 (.3).29(.3)LargeÑÑ1.17(2.0)1.45 (2.3)ÑUSÑÑÑ.59 (1.9)1.42(1.3)4.37(2.2) 3%12%16%19%5% The IncumbentÕs Curse / 11of firm size are strong and robust. We observe them in thecategorical analyses, the bivariate regression, and the multi-ariate analysis with two alternative measures.Additional Analysesfurther discussion: relevant population, definition of firmsize, and definition of radical innovator.Relevant PopulationWhenanalyzingthesourcesofradicalinnovations,wecon-trastinnovativelargefirmsandincumbentstoinnovativesmallirmsandoutsiders,respectively.However,tointerprettheseiguresfully,wemustalsocontrasttheproportionofsmall,medium,andlargefirmsinthesamplewiththeproportionofsmall,medium,andlargefirmsintheoverallmanufacturingeconomy.Similarly,wemustcontrasttheproportionofincumbentsandoutsidersinthesamplewiththerelevantpro-portionsintheoverallmanufacturingeconomy.Forexample,ourcategoricalanalysesindicatethatsmall,medium,andlargefirmsaccountfor47%,11%,and42%,respectively,oftheradicalinnovationsinthesample.Butifsmall,medium,andlargefirmsalsoaccountfor47%,11%,and42%ofthemanufacturingeconomy,ourresultswouldonlyindicatethatirmsineachsizeclasscontributeradicalinnovationsinanumberproportionatetotheirnumberinthepopulation.on the number of small, medium, and large firms in the man-ufacturing economy. The U.S. Census Bureau has reportedCensus of Manufacturesing in 1909. Thus, we are able to include the information forall U.S. innovations introduced after 1909. We also includesimilar information on seven of the eight Japanese innova-tions and one each of the Dutch and German innovations(through their respective Census of Manufacturesperiod of analysis, the proportion of large firms (i.e., thosewith 2500 or more employees) never exceeds .25% of thetotal number of firms in the U.S. manufacturing economy.The proportion of large firms in the U.S. manufacturingeconomy ranges from a high of .22% (in 1967) to a low of.04% (in 1909). The proportion of medium (i.e., 100Ð2499employees) firms in the U.S. manufacturing economyranges from a high of 10.82% (in 1972) to a low of 5.27%(in 1909), whereas the proportion of small firms ranges froma high of 94.69% (in 1909) to a low of 88.99% (in 1972).small firms would be expected to contribute a much largernumber of innovations than large firms. But this is not thecase. Large firms account for a substantially larger propor-tion of radical innovations relative to their number in theeconomy. The data from the non-U.S. sources also providesimilarly large contrasts.similarargumentcanbemadefortheproportionofincumbentfirmsintheeconomy,relativetotheproportionofinnovationsaccountedforbythem.Thenumberofincum-bentsinanyparticularproductclassislikelytobemanytimessmallerthanthenumberofnonincumbentsintheecon-omy,simplybecausetheeconomyconsistsofmanysuchproductclasses.Yetincumbentsaccountforalmosthalfthenumberofradicalinnovationsoverallintheproductclassesstudiedhereandaboutthree-fourthsoftheradicalinnova-tionsintroducedafterWorldWarII.Asdolargefirms,incum-bentsaccountforadisproportionatelylargernumberofradi-calinnovationsrelativetotheirnumberintheeconomy.Definition of Firm SizeInthecategoricalanalyses,wedefineafirmassmallifithadfewerthan100employees,mediumifithadbetween100and2499employees,andlargeifithad2500ormoreemployees.Whathappensifweselectdifferentcutoffpointsforfirmsize?Table4presentstheresultsofasensitivityanalysisreflectingdifferentcutoffsforsmallandmediumfirms.ThecutoffpointsarebasedonthecategoriesusedbytheU.S.CensusBureauinclassifyingthesizeoffirms.Becausethe2500ormoreemployeescategoryisalreadytheU.S.CensusBureauÕscategoryoflargestfirmsinthemanufacturingecon-omy,wedonotvaryourdefinitionoflargefirms.As Table 4 indicates, there is a sharp decrease in the pro-portion of innovations from medium firms as the lower cut-off for medium firms increases from 20 employees to 100employees. Beyond this point, the decreases are much lesssteep. Thus the cutoff point of 100 employees captures mostof the innovative small firms in our sample.Definition of Radical InnovatorRecall that we define a radical innovator as the firm that firstcommercializes a radical innovation. This definitionxcludes other early entrants in the product category, evenafter the first entrant. What would the size and incumbencyprofile of such early entrants be?Severalresearchersarguethatincumbentswouldbethemostlikelytoentermarketsearly(e.g.,Ali1994;Conner1988;seealsoNaultandVandenbosch1996).Theysuggestthatbecauseoffearsofcannibalization,manyestablishedirmsrefrainfromcommercializingtheirinnovationsaslongaspossibleÑperhapsuntiloutsidersintroducetheradicallyinnovativeproductstothemarket.Thus,suchfirmsdonotseektobethefirsttocommercializetheradicallynewtech-nology.Butwhenproductsbasedonthenewtechnologyenterthemarket,incumbentsrushinwiththeirownequivalentproducts.Theycapitalizeontheirmarketingandtechnologyresourcestotakefulladvantageoftheirslightlylaterentry.If this reasoning holds, our results would underrepresentthe innovative performance of large firms and incumbents.After a new entrant commercializes a radically new product,centage of Innovations by Small and Medium- Sized Firms,Based on Definition of Small Firm Definition of Small Firm% Small% Medium� 20 employees36%23%� 100 employees47%11%� 500 employees49%9%� 1000 employees55%4%� 2500 employees58%0% 12/ Journal of Marketing,July 2000many large firms and incumbents may also enter with simi-lar products. Thus, large firms and incumbents may be evenmore innovative than our results indicate.Thissampleofconsumerdurablesandofficeproductsshowsthatsmallfirmsandnonincumbentsareslightlymorelikelytointroduceradicalproductinnovationsthanlargefirmsandincumbents.Yetinrecentyearsthepatternhaschangeddra-matically.Recently,largefirmsandincumbentsaresignifi-cantlymorelikelytointroduceradicalinnovationsthansmallfirmsandnonincumbents.Furthermore,theinnova-tionsintroducedbyrecentlargefirmsandincumbentsarenolessradicalthanthoseintroducedbysmallfirmsandnonin-cumbents.Thus,ourresultsindicatethattheincumbentÕscursemayapply,buttoanoldereconomicperiod.ThecursemayapplyevenlesstocountriesoutsidetheUnitedStates.Inourlimitedsample,althoughtheUnitedStatesaccountsforsimplemajorityofradicalinnovations,itdoesnotaccountforanoverwhelmingproportionofthem.Also,itsshareofradicalinnovationshasnotchangedmuchovertimeperiods.In the hope of motivating further research on some of thecounterintuitive findings from this study, we highlight twoimportant issues raised by this study: lessons from large orincumbent innovators and opportunity for nonincumbentsand small firms. The factors we highlight could also providedirections to current managers in similar innovation con-texts. Because we do not have precise measures for theseactors, this discussion is exploratory in nature.Lessons from Large or Incumbent InnovatorsContrarytoconventionalwisdom,ourresearchindicatesthattodayÕsincumbentsandlargefirmsaccountformanyradicalinnovations,especiallysinceWorldWarII.Thisfindingdove-tailswithrecentresearchthatsuggeststhataconsiderablepro-portionofdominantfirmsintodayÕshigh-techindustriesarewillingtocannibalizetheirownpastinvestmentstointroduceradicalproductinnovations(ChandyandTellis1998).Yetnotallincumbentsandlargefirmsareableorwillingtomakethetransitiontothenewtechnologythatisembodiedinaradicalproductinnovation.Forexample,inthemarketforwatches,SocieteSuissepourIÕIndustrieHorlogere,adominantmechanicalwatchproducer,didnotintroducequartzwatchesuntillateintheirlifecycle.Bythattime,itsmarketpositionhadbeenconsiderablyweakened(Glasmeier1991).However,Hattori-Seiko,anotherdominantproducerofmechanicalatches,wasthefirstfirmtointroducetheanalogquartzatch.Stillanotherincumbent,theHamiltonCompany,wasthefirsttointroducethedigitalquartzwatch.Whydosomedominantfirmsmaintaintheirinnovativevigordespitethesupposedliabilitiesofsizeandincumbency?Thisquestionhasstrongimplicationsforlargeandincumbentfirmsthatarecurrentlycontemplatingradicalinnovations.Althoughwedonothaveconclusiveevidenceonthisquestion,ourhistoricalresearchsuggeststwopossiblecauses:dynamicorganiza-tionalclimatesandstrongtechnologicalcapability.DynamicorganizationalclimatesOnereasonfortheinnovationperformanceofsomelarge,incumbentfirmsmaybethatsuchfirmshaveorganizationalclimatesthatresem-blethoseofsmallfirms.AfterWorldWarII,fundamentalchangesoccurredinthestructureofmanylarge,incumbentirms(Chandler1990).Manyofthesefirmscreatedautonomousbusinessunitswithsignificantauthorityovertheirlinesofbusinessandseparateprofitandlossresponsi-bilities(Chandler1956;Williamson1975).Thebenefitsofthisorganizationalstructurequicklybecamepopular,sothatbythemid-1950s,thepracticeofdecentralizationbecamewidespreadamonglargeU.S.corporations(Chandler1956,p.111).Asdiscussedpreviously,growthinsizeandcom-plexitycanleadtobureaucraticinertiathatdampenstheinnovativenessoffirms.Incontrast,decentralizationleadstosmaller,autonomousorganizationalunitsthatenablethelargefirmtorespondtoandcreatetechnologicalinnovationswhilemaintainingitsresourceadvantages.Decentralizationmayalsohavefosteredinternalcompe-tition:incumbentandnonincumbentbusinessunitswithinalargeincumbentthatcompeteformarkets(ChandyandTel-lis1998;Forrester1965).Asaresult,eventhoughaparticu-larbusinessunitmayhaveastrongstakeintheexistingproductcategory,otherbusinessunitswithintheincumbentirm,whichdonotderivemanyrentsfromtheexistingprod-ucts,maynotbeascommittedtotheseproducts.Theselat-terbusinessunitsalsodonothavetheestablishedroutinesthatmayconstraintheactionsoftheincumbentunits.Theyarethereforelikelytosupportradicalproductinnovations,becausetheseproductsrepresentconsiderableopportunityrelativelylittlethreattotheirexistinglinesofbusiness.Thefirmasawholeisthussupportiveofradicalproductinnovationsdespiteitsincumbencyandevenintheabsenceofexternalcompetitors.Thehistoryofthequartzwatchprovidesagoodillustra-tionoftheeffectsofautonomyandinternalcompetitiononradicalinnovation.Thefirstquartzwatchwascommercial-izedbyHattori-Seikoin1969.ThisproductresultedfromaÒtechnologycontestÓbetweenthecompanyÕsSuwaSeikoshaandDainaSeikoshadivisions(BusinessWeek1978).Theorganizationwasstructuredsuchthatthesetwodivisionsmaintainedseparateresearch,design,andmanu-acturingfacilities.HattoriÕscentralofficeinformedbothdivisionsofitsanticipatedproductneeds.ThesedivisionsthenindependentlydevelopedproductprototypesfromwhichHattorichosemodelstomass-produceforthemarket(Hoff1985).nological capabilitying radical innovation by large firms. Incumbent firms withstrong technological capability are likely to become awareof scientific breakthroughs at an early stage and are in ainnovations. The General Electric CompanyÕs historicalemphasis on basic research provides a rich illustration of therole of technological capability in radical innovation.The General Electric Company established its Researchthe laboratory published actively in leading scientific jour-nals. The lab was insulated from immediate businessdemands and staffed by people with advanced scientific The IncumbentÕs Curse / 13training (Reich 1985). Willis Whitney, its founding director,as elected to serve as president of the American ChemicalSociety in 1909, and Irving Langmuir, a scientist at GeneralElectric, won the 1932 Nobel Prize for chemistry for workconducted at the firm between 1912 and 1915 (Brown andeeks 1952; Wise 1985). The research conducted at the lab-oratory played an important role in the companyÕs ability todevelop and commercialize fluorescent lamps even while itas the dominant player in incandescent lamps. Few othersmall firms at that time had the technological capability tointroduce these new products.meager payoffs if firms do not also have an organization thatis suited to developing and marketing commercially viableinnovations. Although basic research may provide a sourceof ideas in the early stages of radical product development,a dynamic organizational climate is critical in the commer-cialization of radically new products. XeroxÕs failure tocommercialize the many innovations emerging from its Palo(Smith and Alexander 1988).OpportunityforNonincumbentsandSmallerFirmsSuccessful development of a radical innovation today canrequire huge expenditures in R&D. The reason is thatadvances in technology make new products far more com-plex than they were a century ago, or even a few decadesago. At the same time, the noise from competing advertisingand promotion for myriad old and new brands raises a for-midable barrier for any new entrant. Thus, the image oftinkererÐinnovators fashioning radical innovations in theirsmall garages may not always be true. However, ourresearch indicates that even in these noisy markets, smallirms or nonincumbents introduce radical innovation.How can a firm with relatively few resources succeed ina process that presumably requires large outlays in R&D?The case histories of innovative small and medium firms inour sample provide some clues.Oneoptionistomakeuseofspilloversfromresearchconductedatother,resource-richfirms.SomesmallfirmsfocusonthedevelopmentpartoftheR&Dprocess,relyingonoff-the-shelfcomponentsfromotherindustriestointro-ducetechnologiesthatareradicallynewinadifferentindus-try.Forexample,thefirstpersonalcomputerwasdevelopedandintroducedbyEdRoberts,apracticingphysicianandfounderofMicroInstrumentationandTelemetrySystems,whoincorporatedmanycomponents(suchasintegratedcir-cuits)thatweretheresultofbasicresearchbyothercompa-nies.However,suchabusinessmodelleavesthefirmvul-nerabletoimitators.Successmaybehardtosustainunlessthereisasteadystreamofeverimprovingoff-the-shelfcom-ponentsintheindustryortheproducttakesoffquicklyenoughtoprovidethefirmwiththeresourcestobuilditstechnologicalbase.A different (and perhaps more sustainable) innovationmodel is that which the Haloid Corporation followed in the1949; Jewekes, Sawers, andStillerman 1969). The idea for electrostatic copying camefrom Chester Carlson, an inventor, who also developed aprimitive prototype of the product. After failing to interestany of the large incumbents in the photographic copier busi-ness (the previous product generation), Carlson finally suc-Institute, then the worldÕs largest nonprofit research organi-zation. The managers at the Haloid Corporation, a medium-sized incumbent in the photographic copier industry, saw adescription of the electrostatic technology in the April 1945odakÕs Monthly Abstract Bulletinother large incumbents (including Kodak) had turned downits development, Haloid signed a contract with Battelle topartly fund further development of the technology. AlthoughHaloidÕs $25,000 investment represented a significant pro-subsequent contributions by Haloid were insufficient to fundthe development of a commercially viable copier fully. Itas through BattelleÕs strength in basic research and, intechnological capability, relied on the resources of a strongresearch organization and funding from the federal govern-ment to develop its radically new product. Over time, Haloid(which changed its name to Xerox after the new productbecame an important revenue source) bought many of they Xerography patents from Battelle. This patent protec-tion, together with further research on its own part, enabledthe firm to protect its market position against later entrants.Insummary,smallnonincumbentshaveatleasttwooptionsindevelopingandintroducingradicalproductinno-ations:(1)useresearchspilloversfrommoreresource-richirmsand(2)activelypartnerwithorganizationswithtech-nologicalcapabilitiesandfinancialresourcestheydonothavethemselves.Ofthesetwooptions,thesecondmaybelessimitableandmoresustainableoptioninmanyindus-Althoughthehistoricalmethodallowsforuniqueandfairlyobjectiveinsightsonradicalinnovation,italsoimposescon-straintsonthescopeofourstudy.Theselimitationshighlighttheneedforadditionalresearchonthetopic.First,thelabor-intensivenatureofthestudylimitedoursampletoasmallnumberofinnovations(64)inalimitednumberofcategories(49)inonlytwoclassesofgoodsÑofficeproductsandcon-sumerdurables.Furthermore,ourlistofinnovationsdoesnotexhaustivelycoverallthesignificantinnovationsinthetwoproductclasseswestudy.Therefore,generalizations,especiallytootherclasses,mustbemadewithcaution.Sec-ond,ourstudyfocusesonrelativelysuccessfulinnovations.Includingfailedorlesssuccessfulinnovationsinthesamplemayleadtoadditionalinsights.Third,wehaveinformationononlyexternalcharacteristicsofthefirm,suchasincum- 14/ Journal of Marketing,July 2000 also checked if innovations introduced by incumbents andlarge firms were the result of their acquisitions of innovative smallirms and nonincumbents. We found that such acquisition-basedinnovations are rare in our sample. We further checked if, con-ersely, innovations introduced by nonincumbents and small firmswere the result of technology developed within incumbents andlarge firms, but commercialized by nonincumbents and smallirms. Such innovations are also rare in our sample.bency,size,andcountryofoperation.Furtherresearchcouldgatherinformationoninternalcharacteristicsofsuchirmstotestsomeoftheorganizationalpropositionssug-gestedpreviously.ConclusionInactive. Incompetent. Arrogant. These are some of theterms researchers use to describe how incumbent and largeirms have fared in radical product innovation (Ghemawat1991; Henderson 1993; Utterback 1994). Many academicstors of these firms. Radical innovation is likened to a gameBut do these terms reflect reality? Events in which themighty are humbled and the little guy finishes first are likelyto be more eye-catching than are those in which the mightyremain mighty. Unless a large, nonconvenience sample isused, observers may miss the latter, less salient events. Forthis reason, we research a relatively large cross-section ofradical innovations selected on certain explicit criteria.Our research of innovations in the consumer durablesand office product categories suggests that incumbents orlarge firms are not necessarily doomed to obsolescence bynimble outsiders. In particular, our research leads to the fol-lowing main conclusions regarding radical innovations inOver a 150-year period, small firms and nonincumbentsintroduce slightly more radical product innovations than largeHowever, the sources of radical product innovations changesubstantially after World War II. Large firms and incumbentsintroduce a majority of radical product innovations over thistime period. Thus, the incumbentÕs curse is less prevalent inThe United States accounts for almost two-thirds of radicalproduct innovations in the sample, and Western Europefew innovations, but those have been entirely in recent years.The distribution of radical innovations between the UnitedStates and other nations remains steady over time.Small firms and outsiders account for many more innovationsin the United States than they do in other countries. Thus, theincumbentÕs curse is less prevalent in Western Europe andDynamic organizational structures and strong techno-logical capability may keep large, incumbent organizationsnimble and innovative, but many managers and academicshave tended to focus on inertia-prone incumbents (e.g., Ghe-mawat 1991; Henderson 1993; Henderson and Clark 1990;Scherer 1980; Utterback 1994; for recent exceptions, seeChristensen 1997; Tushman and OÕReilly 1997). In focusingon the Remingtons and Underwoods of the world, let us notforget the examples of General Electric (in fluorescentlamps), Philips (in compact disc players), and Seiko (in ana-log quartz watches). Despite their large size and incum-bency in the incandescent lamp, tape recorder, and mechan-ical watch industries, these firms were the first to introduceradical innovations that changed the landscapes of theseindustries. Perhaps the incumbentÕs curse is not as inevitable Full List of Significant Innovations in Sample Radical Product InnovationFirst Commercialized byCommercializationAir conditionerBuffalo Forge Company1902AM radioWireless Telegraph and Signal Co.1897Analog answering machineAmerican Telegraphone Co. 1903Analog quartz watchSeiko1969Autofocus color celluloid roll cameraKonishiroku Photo Industry1977Black-and-white celluloid roll cameraEastman Dry Plate & Film Co.1889Ballpoint penEterpen Co.1943CamcorderSony1983Cassette tape playerPhillips1964Compact disc playerPhillips and Sony1979Cellular telephoneMotorola1983Color celluloid roll cameraLumiere Brothers1907Desktop computerMITS1975Digital answering machineSharp1988Digital cameraSony1983Digital high-definition televisionPanasonic1998Digital quartz watchHamilton Co.1972Digital video disc (DVD) playerToshiba1997Disposable shaverBic Corp.1975Dot-matrix printerRemington Rand1953 The IncumbentÕs Curse / 15 Continued Radical Product InnovationFirst Commercialized byCommercialization Acs, Zoltan J. and David B. Audretsch (1991), Innovation andhnological Change: An International ComparisonArbor, MI: University of Michigan Press.Agarwal, N.C. (1979a), ÒNature of Size-Structure Relationship:Some Further Evidence,Ó ÑÑÑ (1979b), ÒOn the Interchangeability of Size Measures,ÓAcademy of Management JournalAli, A. (1994), ÒPioneering Versus Incremental Innovation: Reviewand Research Propositions,Ó ournal of Product InnovationManagementArrow,Kenneth(1962),ÒEconomicWelfareandtheAllocationofResourcesforInvention,ÓinTheRateandDirectionofEco-nomicActivityPrinceton:PrincetonUniversityPress,609Ð25.Bauer, R.A. (1960), ÒConsumer Behavior as Risk Taking,Ó inDynamic Marketing for a Changing WorldChicago: American Marketing Association, 389Ð98.Birr, Kendall (1957), Pioneering in Industrial Research: The Storyof the General Electric Research Laboratory. Washington, DC:Public Affairs Press.Blau, Peter M. and Richard A. Schoenherr (1971), The Structure ofOrganizations. New York: Basic Books.Bright, Arthur A. (1949), The Electric-Lamp Industry: Technolog-ical Change and Economic Development from 1800 to 1947New York: Macmillan.Brown, Charles A. and Mary E. Weeks (1952), American Chemical Society. Washington, DC: AmericanChemical Society.Dry ink (electrostatic) copierHaloid Co.1951Electric blanketGeneral Electric1936Electric blenderWaring Blender Co.1937Electric can openerUdico Corporation1956Electric clothes washerHurley Machine Co.1908Electric dishwasherWillard and Forrest Walker1913Electric fanCrocker & Curtis Co.1882Electric garbage disposerGeneral Electric1934Electric percolatorLanders, Frary, & Clark1908Electric shaverSchick Inc.1931Electric toasterGeneral Electric1908Electric typewriterBlickensderfer Co.1902Electrochemical faxCaselli1865Electronic black-and-white televisionAllen B. Dumont Laboratories1939Electronic color televisionRCA1954Electronic desktop calculatorSharp1964Electronic pocket calculatorBowmar1971Electronic watchBulova1960Fluorescent lampGeneral Electric1938FM RadioGeneral Electric1937Incandescent vacuum lampEdison Electric Light Co.1880Instant cameraDubroni1864Internal combustion automobile (petrol)Carl Benz1888Laptop computerTandy Corp.1983Laser disc playerPhillips1978Laser printerIBM1976Magnetic tape player (reel-to-reel)AEG1934Mechanical black-and-white televisionTelevision Ltd.1930Mechanical color televisionCBSÐColumbia 1951Mechanical dishwasherJosephine Cochrane1889Mechanical refrigeratorJohn Gorrie1851Mechanical typewriterSholes/Densmore1872Mechanical vacuum cleanerVacuum Cleaning Co.1901Microwave ovenRaytheon1953Mini-disc playerSony1992Palm computerAmstrad1993PhonographEdison Speaking Phonograph Co.1878Portable computerOsborne Computer1981Safety shaver with disposable blades American Safety Razor Co.1903Single-player video gameNutting Associates1971Photoelectric scanning faxArthur Korn1907elephone set with cord Bell Telephone Co.1876ideocassette recorderAmpex Corp.1956oice mailECS1980 16/ Journal of Marketing,July 2000Business Times(1996), ÒWhen Being Efficient Is Not Enough,ÓBusinessWeek(1978), ÒK. HattoriÕs Blend of Internal Competi-tion,Ó (June 5), 90.Chandler,AlfredD.(1956),ÒManagementDecentralization:AnHistoricalAnalysis,ÓBusinessHistoryReviewÑÑÑ(1990),ScaleandScopeCambridge,MA:BelknapHarvard.Chandy, Rajesh and Gerard Tellis (1998), ÒOrganizing for RadicalProduct Innovation: The Overlooked Role of Willingness toCannibalize,Ó ournal of Marketing Research, 34 (November),Child, John (1973), ÒPredicting and Understanding OrganizationStructure,Ó Administrative Science QuarterlyChristensen, Clayton (1993), ÒThe Rigid Disk Drive Industry: AHistory of Commercial and Technological Turbulence,Ó ness History Review, 67 (Winter) 531Ð88.The InnovatorÕs Dilemma: When New Technolo-gies Cause Great Firms to Fail. Boston: Harvard BusinessCohen, Wesley (1995), ÒEmpirical Studies of Innovative Activity,ÓHandbook of the Economics of Innovation and TechnologicalChange, Paul Stoneman, ed. Cambridge, MA: Blackwell,ÑÑÑ and Richard C. Levin (1989), ÒEmpirical Studies of Inno-ation and Market Structure,Ó in Handbook of Industrial Orga-, Vol. 2, R. Schmalensee and R.D. Willig, eds. Nework: Elsevier Science Publishers, 1059Ð1107.Comanor, W.S. (1965), ÒMarket Structure, Product Differentiation,and Industrial Research,Ó Quarterly Journal of EconomicsConner, Kathleen (1988), ÒStrategies for Product Cannibalism,ÓStrategic Management JournalCooper,ArnoldandDanSchendel(1976),ÒStrategicResponsestoechnologicalThreats,ÓBusinessHorizonsÑÑÑ and Clayton Smith (1992), ÒHow Established FirmsRespond to Threatening Technologies,Ó Academy of Manage-Dessauer, John H. (1971), My Years with Xerox: The BillionsNobody Wanted. Garden City, NY: Doubleday and Company.Ettlie,J.andA.H.Rubenstein(1987),ÒFirmSizeandProductInno-ation,ÓournalofProductInnovationManagementFitzroy, Felix R. and Kornelius Kraft (1991), ÒFirm Size, Growth,and Innovation: Some Evidence from West Germany,Ó in vation and Technological Change: An International Compari-, Zoltan J. Acs and David B. Audretsch, eds. New York:Harvester Wheatsheaf, 152Ð59.olkes, Valerie S. (1988), ÒThe Availability Heuristic and Per-ceived Risk,Ó ournal of Consumer Researchorrester, Jay W. (1965), ÒA New Corporate Design,Ó ement Review, 7 (Fall), 5Ð17.(1949), ÒPrinting with Powders: A Process Named Xerog-raphy Puts Static Electricity to Work in the Graphic Arts,Óoster, R. (1986), Innovation: The AttackerÕs Advantage. NewFrank, Harry and Steven C. Althoen (1994), ork: Cambridge University Press.Galbraith,J.K.(1968),TheNewIndustrialStateHamishHamilton.Gatignon, Hubert and Jean-Marc Xuereb (1997), ÒStrategic Orien-tation of the Firm and New Product Performance,Ó ournal ofMarketing ResearchGhemawat, Pankaj (1991), ÒMarket Incumbency and Technologi-cal Inertia,Ó Marketing ScienceGlasmeier, Amy (1991), ÒTechnological Discontinuities andFlexi-ble Production Networks: The Case of Switzerland and theorld Watch Industry,Ó Research Policy, 20 (October), 469Ð85.Golder, Peter N. (2000), ÒThe Benefits of the Historical Method,Óournal of Marketing ResearchÑÑÑ and Gerard J. Tellis (1993), ÒPioneer Advantage: Market-ing Logic or Marketing Legend?Ó ournal of MarketingResearchÑÑÑ and ÑÑÑ (1997), ÒWill It Ever Fly? Modeling the Take-off of Really New Consumer Durables,Ó Marketing ScienceGort, Michael and Steven Klepper (1982), ÒTime Paths in the Dif-fusion of Product Innovations,Ó The Economic JournalÑÑÑ and Richard A. Wall (1986), ÒThe Evolution of Technolo-gies and Investment in Innovation,Ó The Economic JournalGregan-Paxton, Jennifer and Deborah Roedder John (1997), ÒCon-sumer Learning by Analogy: A Model of Internal Knowledgeransfer,Ó ournal of Consumer ResearchHannan,M.T.andJ.Freeman(1977),ÒThePopulationEcologyofOrganizations,ÓAmericanJournalofSociology(5),929Ð64.Henderson, Rebecca (1993), ÒUnderinvestment and Incompetenceas Responses to Radical Innovation: Evidence from the Pho-tolithographic Alignment Equipment Industry,Ó RAND JournalÑÑÑ and Kim B. Clark (1990), ÒArchitectural Innovation: TheReconfiguration of Existing Product Technologies and the Fail-ure of Established Firms,Ó Administrative Science QuarterlyDealers of Lightning: Xerox PARC andthe Dawn of the Computer Age. New York: Harper Collins.Hoff, Edward J. (1985), ÒHattori-Seiko and the World WatchIndustry in 1980,Ó Case # 9-385-300, Harvard Business School.Imai,Ken-ichi(1990),ÒPatternsofInnovationandEntrepreneurshipinJapan,ÓinEvolvingTechnologyandMarketStructure:StudiesinSchumpeterianEconomicsrnoldHeertjeandMarkPerlman,eds.AnnArbor,MI:UniversityofMichiganPress,187Ð202.Jewekes, John, David Sawers, and Richard Stillerman (1969), Sources of InventionSystem Size: A Three-Level Analysis,Ó American SociologicalReviewKimberly, John (1976), ÒOrganizational Size and the StructuralistPerspective: A Review, Critique, and Proposal,Ó AdministrativeMagaziner, Ira C. and Mark Patinkin (1989), The Silent War.Nework: Vintage Books.Mitchell, W. (1989), ÒWhether and When? Probability and Timingof IncumbentsÕEntry into Emerging Industrial Subfields,ÓAdministrative Science QuarterlyIncumbent and Newcomer Market Share and Survival WhenSpecialized Investments Retain Their Value,Ó Strategic Man-ement JournalÑÑÑ and K. Singh (1993), ÒDeath of the Lethargic: Effects ofExpansion into New Technical Subfields on Performance in aFirmÕs Base Business,Ó Organizational ScienceMoorman,ChristineandAnneMiner(1997),ÒTheImpactofOrga-nizationalMemoryonNewProductPerformanceandCreativ-ity,ÓournalofMarketingResearch(February),91Ð106.Nault, Barrie and Mark Vandenbosch (1996), ÒEating Your OwnLunch: Protection Through Preemption,Ó Organization ScienceNelson, Richard R. and Sidney G. Winter (1982), Theory of Economic Change The IncumbentÕs Curse / 17Olson, Eric M., Orville C. Walker, and Robert W. Ruekert (1995),ÒOrganizing for Effective New Product Development: TheModerating Role of Product Innovativeness,Ó atel, Pari and Keith Pavitt (1995), ÒPatterns of TechnologicalActivity: Their Measurement and Interpretation,Ó in of the Economics of Innovation and Technological Changevitt, K., M. Robson, and J. Townsend (1987), ÒThe Size Distrib-ution of Innovating Firms in the UK: 1945Ð1983,Ó ournal ofPredicasts(1994),PredicastsBasebookleveland,OH:PredicastsPrice,JamesL.andCharlesW.Mueller(1986),HandbookofOrga-nizationalMeasurementMarshfield,MA:PitmanPublishing.The Making of American IndustrialResearch: Science and Business at GE and Bell, 1876Ð1926New York: Cambridge University Press.Reinganum, Jennifer F. (1983), ÒUncertain Innovation and the Per-sistence of Monopoly,Ó The American Economic ReviewRosenbloom, Richard and Clayton Christensen (1994), ÒTechno-logical Discontinuities, Organizational Capabilities, and Strate-gic Commitments,Ó Industrial and Corporate ChangeSahal, Devendra (1985), ÒTechnological Guideposts and Innova-tion Avenues,Ó Research PolicySakudo, Jun and Takao Shiba (1994), orld War II and the Trans-. Tokyo: University of TokyoSavitt, Ronald (1980), ÒHistorical Research in Marketing,Ó of Marketing, 44 (Fall), 52Ð58.Saxenian, AnnaLee (1994), Regional Advantage: Culture andCompetition in Silicon Valley and Route 128Harvard University Press.Scherer, F.M. (1980), Industrial Market Structure and Economic. Chicago: Rand McNally.ÑÑÑ (1991), ÒChanging Perspectives on the Firm Size Prob-lem,Ó in Innovation and Technological Change, Zoltan Acs andDavid Audretsch, eds. New York: Harvester Wheatsheaf,Schumpeter, Joseph (1942), . New York: Harper.Smith, Douglas K. and Robert C. Alexander (1988), Future: How Xerox Invented and Then Ignored, the First Per-sonal Computer. New York: Morrow.Smith,RuthAnnandDavidS.Lux(1993),ÒHistoricalMethodinConsumerResearch:DevelopingCausalExplanationsofChange,ÓournalofConsumerResearchMarch),595Ð610.Srivastava,Rajendra,TasadduqShervani,andLiamFahey(1998),ÒMarket-BasedAssetsandShareholderValue:AFrameworkforAnalysis,ÓournalofMarketing(January),2Ð18.Staw, Barry M. (1981), ÒThe Escalation of Commitment to aCourse of Action,Ó Academy of Management ReviewSultan, Fareena, John U. Farley, and Donald R. Lehmann (1990),Meta-Analysis of Applications of Diffusion Models,Ó nal of Marketing Researcheitelman,Robert(1994),ProfitsofScience:TheAmericanMar-riageofBusinessandTechnologyork:BasicBooks.errien, Fred W. and Donald L. Mills (1955), ÒThe Effect ofChanging Size upon the Internal Structure of Organizations,ÓAmerican Sociological ReviewThe Theory of Industrial Organizationornatzky,LouisG.andMitchellFleischer(1990),TheProcessofhnologicalInnovationxington,MA:LexingtonBooks.shman, M.L. and P. Anderson (1986), ÒTechnological Disconti-nuities and Organizational Environments,Ó Administrative Sci-ÑÑÑ and Charles A. OÕReilly (1997), nning Through Innova-tion: A Practical Guide to Leading Organizational Change andRenewal. Boston: Harvard Business School Press.Urabe, K., John Child, and T. Kagono (1988), Innovation andManagement: International Comparisons. New York: Walter deGruyter.Design and Marketing ofNew Products, Englewood Cliffs, NJ: Prentice Hall.Mastering the Dynamics of Innova-. Boston: Harvard Business School Press.ÑÑÑ and W.J. Abernathy (1975), ÒA Dynamic Model of Processand Product Innovation,Ó OmegaashingtonPost(1997),ÒAndThenThereWere...3,Ó(July14),F5.illiamson,HaroldF.(1975),EvolutionofInternationalManage-mentStructuresark,DE:UniversityofDelawarePress.ise, George (1985), illis R. Whitney, General Electric, and theOrigins of US Industrial Research. New York: Columbia Uni-