Tropes

Open innovation in SMEs Trends, motives and management challenges.pdf

Technovation29 (2009) 423 437 Open innovationinsmes: Trends, motivesandmanagementchallenges Vareska vandevrandea,,Jeroenp. J. dejongb, Wimvanhaverbekec, Maurice derochemontd acollege ofmanagementoftechnology, Ecolepolytechniquefe'de'rale delausanne (EPFL), Odyssea1. 19, Station5, 1015lausanne, Switzerland beim Businessandpolicyresearch, Thenetherlands cfaculty ofbusinessstudies, Hasseltuniversity, Belgium deindhoven Universityoftechnology, Thenetherlands Abstract Open innovationhassofarbeenstudiedmainlyinhigh-tech, multinationalenterprises. Thisexploratorypaperinvestigatesifopen innovation practicesarealsoappliedbysmall-andmedium-sizedenterprises (SMES. Drawingonadatabasecollectedfrom605 innovative SMESINTHENETHERLANDS, weexploretheincidenceofandapparenttrendtowardsopeninnovation. Thesurveyfurthermore focuses onthemotivesandperceivedchallengeswhensmesadoptopeninnovationpractices. Withinthesurvey, openinnovationis measured witheightinnovationpracticesreflectingtechnologyexplorationandexploitationinsme. Wefindthattherespondingsmes engage inmanyopeninnovationpracticesandhaveincreasinglyadoptedsuchpracticesduringthepa t7years. Inaddition, wefindno major differencesbetweenmanufacturingandservicesindustries, butmedium-sizedfirmsareonaveragemoreheavilyinvolvedinopen innovation thantheirsmallercounterparts. Wefurthermorefindthatsmespursueopeninnovationprimarilyformarket-related motives suchasmeetingcustomerdemands, orkeepingupwithcompetitors. Theirmostimportantchallengesrelatetoorganizational and culturalissuesasaconsequenceofdealingwithincreasedexternalcontacts. r 2008 Elsevierltd. Allrightsreserved. Keywords: Open innovation; SMES; Technologymarkets; Incidence; Perceivedtrend; Motives; Managerialchallenges 1. Introduction Open innovationhasbeenproposedasanewparadigm for themanagementofinnovation (Chesbrough, 2003; Gassmann, 2006. Itisdefinedas‘theuseofpurposive inflowsandoutflowsofknowledgetoaccelerateinternal innovation, andtoexpandthemarketsforexternaluseof innovation, respectively.'('Chesbroughetal.,, 2006, p. 1). It thus comprisesbothoutside-inandinside-outmovements of technologiesandideas, alsoreferredtoas‘technology acquisition'and‘technologyexploitation'(Lichtenthaler, 2008. Open innovationhasreceivedincreasinglyattentionin scientificresearch, butsofarithasmainlybeenanalyzedin large, high-techmultinationalenterprises (MNES) drawing on in-depthinterviewsandcasestudies (e g. Chesbrough, 2003; Kirschbaum, 2005. Fewstudieshavedemonstrated that openinnovationalsoexistsinsmallerorganizations. Moreover, allofthemfocusonveryspecificindustries, for exampleopensourcesoftware (Henkel, 2006) ortabletop role-playinggames (Lecocqanddemil, 2006. Whenever large samplesofenterprisesareexplored, thefocusison specific issuesratherthanthefullopeninnovationmodel (e g. Laursenandsalter, 2006; Chesbrough, 2002. Toour knowledge, only Lichtenthaler (2008) hassofarattempted to empiricallystudytheincidenceofopeninnovationina broader sampleofenterprises. Hefocusedonmedium-sized andlargemanufacturersingermany, Switzerland and Austria, assmallenterprisesandserviceindustries were notsurveyed. Thisstudyaddressesthisgapbyfocusingonsmall-and medium-sizedenterprises (SMES. Itisafirst, explorative study measuringtowhichextentsmesapplyopen ARTICLEINPRESS www. elsevier. com/locate/technovation 0166-4972/$-seefrontmatter r 2008 Elsevierltd. Allrightsreserved. doi: 10.1016/j. technovation. 2008.10.001 Correspondingauthor. Tel.:++41216930048; fax:++41216930020. E-mail addresses: vareska. vandevrande@epfl. ch (V. vandevrande), jjo@eim. nl (J. P. J. dejong), wim. vanhaverbeke@uhasselt. be (W. Vanhaverbeke), m d. rochemont @tm. tue. nl (M. derochemont. innovationpracticesandwhetherthereisatrendtowards increasedadoptionoftheopeninnovationmodelover time. Indoingso, wedevelopandtestpropositionsonthe differences betweenmanufacturingandservicesfirms, and between medium-sizedandsmallenterprises. Furthermore, we explorethemotivesofsmestoengageinopen innovationandperceivedmanagementchallengesin implementingopeninnovation. Toourknowledgethis study isthefirsttoinvestigatetheincidenceofopen innovationinabroadsampleofsmes. Indoingso, it assesseswhetheropeninnovationisatrendthatisnotonly relevantforhigh-techmnesbutalsoforabroaderrange of firmsandbusinesses. Aswedrawonasurveydatabase of 605smesinthenetherlands, thepaperalsoaccounts for thepotentialcriticismthatopeninnovationhassofar been studiedmainlyinamericanenterprises (e g. Ches-brough, 2003; Chesbroughandcrowther, 2006; Lecocq and Demil, 2006) andnotinotherspartsoftheworld (a notableexceptionis Lichtenthaler, 2008. The remainderofthepaperisstructuredasfollows. Section2discussesopeninnovationandthedimensionsof technologyexploitationandexplorationthatcanbeused to classifyopeninnovationpractices. Next, wedevelop some tentativepropositionsontheadoptionofopen innovationpracticesinmanufacturingandservicefirms, and betweendifferentsizecategoriesofsmes. Section4 describesourdata, whilesection5analysestheincidence and trendtowardsopeninnovation, andmotivesand hamperingfactorsofsmes. Finally, Section6concludes and discussesthelimitationsandimplicationsofourwork. 2. Openinnovation Traditionally, largefirmsreliedoninternalr&dto createnewproducts. Inmanyindustries, largeinternal R&d labswereastrategicassetandrepresenteda considerableentrybarrierforpotentialrivals. Asaresult, large firmswithextendedr&dcapabilitiesandcomple-mentaryassetscouldoutperformsmallerrivals (Teece, 1986. Thisprocessinwhichlargefirmsdiscover, develop and commercializetechnologiesinternallyhasbeenlabeled the closedinnovationmodel (Chesbrough, 2003. Although this modelworkedwellforquitesometime, thecurrent innovationlandscapehaschanged. Duetolabormobility, abundantventurecapitalandwidelydispersedknowledge across multiplepublicandprivateorganizations, enter-prisescannolongeraffordtoinnovateontheirown, but rather needtoengageinalternativeinnovationpractices. As aresult, agrowingnumberofmneshasmovedtoan open innovationmodelinwhichtheyemploybothinternal and externalpathwaystoexploittechnologiesand, concurrently, toacquireknowledgefromexternalsources (Chesbrough, 2003. Open innovationisabroadconceptencompassing different dimensions. Followingthedefinitionmentioned earlier, moststudiesdistinguishbetweenpurposiveout-flows andinflowsofknowledgetoaccelerateinternal innovationprocessesandtobetterbenefitfrominno-vative efforts, respectively (e g. Chesbroughetal. 2006; Chesbroughandcrowther, 2006. Purposiveoutflowsof knowledge, or technologyexploitation, impliesinnovation activities toleverageexistingtechnologicalcapabilities outside theboundariesoftheorganization. Purposive inflows, whichwewillrefertoas technologyexploration, relates toinnovationactivitiestocaptureandbenefitfrom external sourcesofknowledgetoenhancecurrenttechno-logical developments. Inafullyopensetting, firmscombine both technologyexploitationandtechnologyexploration in ordertocreatemaximumvaluefromtheirtechnological capabilitiesorothercompetencies (Chesbroughand Crowther, 2006; Lichtenthaler, 2008). 2. 1. Technologyexploitation In ordertobetterprofitfrominternalknowledge, enterprises mayengageinvariouspractices. Inthispaper, three activitiesrelatedtotechnologyexploitationwillbe distinguished: venturing, outwardlicensingofintellectual property (IP), andtheinvolvementofnon-R&dworkersin innovationinitiatives. Venturingisdefinedhereasstartingupneworganiza-tions drawingoninternalknowledge, i e. itimpliesspin-off and spin-outprocesses. Supportfromtheparentorganiza-tion mayalsoincludefinance, humancapital, legaladvice, administrativeservices, etc. Previousopeninnovation studies haveprimarilyfocusedonventuringactivitiesin large enterprises (e g. Chesbrough, 2003; Lord etal. 2002). ) The potentialofventuringactivitiesisregardedtobe enormous, e g. Chesbrough (2003) illustrated thatthetotal market valueof11projectswhichturnedintonewventures exceeded thatoftheirparentcompany, Xerox, byafactor of two. IP playsacrucialroleinopeninnovationasaresult of thein-andoutflowsofknowledge (Arora, 2002; Chesbrough, 2003,2006; Lichtenthaler, 2007. Enterprises have opportunitiestoout-licensetheiriptoobtainmore value fromit (Gassmann, 2006. Out-licensingallowsthem to profitfromtheiripwhenotherfirmswithdifferent businessmodelsfindprofitable, externalpathstothe market. Thedecisionoffirmstolicenseoutdependson anticipatedrevenuesandprofit-dissipationeffects (Arora et al. 2001), i e. outwardlicensinggeneratesrevenuesin the formoflicensingpayments, butcurrentprofits might decreasewhenlicenseesusetheirtechnologyto competeinthesamemarket. Priorresearchhasshownthe importance ofestablishingareputationasaknowledge provider inordertoincreasethemonetaryandstrategic benefits oftechnologyout-licensing (Lichtenthalerand Ernst, 2007. A thirdpracticetobenefitfrominternalknowledgeisto capitalizeontheinitiativesandknowledgeofcurrent employees, includingthosewhoarenotemployedatthe internal R&ddepartment. Severalcasestudiesillustrate that informaltiesofemployeeswithemployeesofother ARTICLEINPRESS V. vandevrandeetal.//Technovation29 (2009) 423 437 424 organizationsarecrucialtounderstandhownewproducts are createdandcommercialized (e g. Chesbroughetal. 2006). ) Manypractitionersandscientists, alsooutsidethe field ofopeninnovation, endorsetheviewthatinnovation by individualemployeesisameanstofosterorganizational success (e g. Van deven, 1986. Workhasbecomemore knowledge-basedandlessrigidlydefined. Inthiscontext, employeescanbeinvolvedininnovationprocessesin multiple ways, forexamplebytakinguptheirsuggestions, exemptingthemtotakeinitiativesbeyondorganizational boundaries, orintroducingsuggestionschemessuchasidea boxes andinternalcompetitions (e g. Van Dijkandvan den Ende, 2002). 2. 2. Technologyexploration Technologyexplorationreferstothoseactivitieswhich enable enterprisestoacquirenewknowledgeandtechno-logiesfromtheoutside. Inthesurvey, fivepracticeswere distinguishedrelatedtotechnologyexploration: customer involvement, externalnetworking, externalparticipation, outsourcingr&dandinwardlicensingofip. Open innovationtheoristsrecognizethatcustomer involvementisoneimportantalternativetoinforminternal innovationprocesses (Gassmann, 2006. Drawingonthe work of Von Hippel (2005) users areincreasinglyregarded not asjustpassiveadoptersofinnovations, buttheymay rather developtheirowninnovationswhichproducerscan imitate. Usersforexampleregularlymodifytheircurrent machines, equipmentandsoftwaretobettersatisfyprocess needs, andbecauseproducersfailtoprovideanadequate supply (Von Hippel, 2005. Firmsmaybenefitfromtheir customers'ideasandinnovationsbyproactivemarket research, providingtoolstoexperimentwithand/or develop productssimilartotheonesthatarecurrently offered, orbyproducingproductsbasedonthedesignsof customersandevaluatingwhatmaybelearnedfrom general productdevelopment. External networkingisanotherimportantdimension which isconsistentlyassociatedwithopeninnovation (Chesbroughetal. 2006). ) Itincludesallactivitiesto acquire andmaintainconnectionswithexternalsourcesof social capital, includingindividualsandorganizations. As such, itcomprisesbothformalcollaborativeprojectsand more generalandinformalnetworkingactivities. Networks allow enterprisestorapidlyfillinspecificknowledgeneeds withouthavingtospendenormousamountsoftimeand money todevelopthatknowledgeinternallyoracquireit through verticalintegration. Networksmayalsoevolve into formalcollaborativeeffortssuchasr&dalliances. Such alliancesbetweennon-competingfirmshavebecomea popularvehicleforacquiringtechnologicalcapabilities (Gomes-Casseres, 1997. External participationsenabletherecoveryofinnova-tions thatwereinitiallyabandonedorthatdidnotseem promising. Enterprisesmayinvestinstart-upsandother businessestokeepaneyeonpotentialopportunities (Chesbrough, 2006; Keil, 2002. Suchequityinvestments provide opportunitiestofurtherincreaseexternalcolla-borationincasetheirtechnologiesprovetobevaluable (Van devrandeetal. 2006). ) Enterprisesmayalso outsourcer&dactivitiestoacquireexternalknowledge. At theheartoftheopenparadigmistheassumptionthat enterprisescannotconductallr&dactivitiesbythem-selves, butinsteadhavetocapitalizeonexternalknowledge which canbelicensedorbought (Gassmann, 2006. Technicalserviceproviderssuchasengineeringfirmsand high-tech institutionshavealsobecomemoreimportantin the innovationprocess. Intheopenmodelitisconsidered fully legitimatetobringkeyknowledgedevelopment outsidetheorganizationalboundary (e g. Prencipe, 2000. Finally, enterprisescanexternallyacquireintellectual property, includingthelicensingofpatents, copyrightsor trade marks, tobenefitfromexternalinnovationopportu-nities (Chesbrough, 2006. Thismaybeanecessitytofuel one's businessmodelandtospeedupandnurtureinternal research engines. To conclude, incomparisonwiththeclosedmodel, the open innovationmodelimpliesthatthemanagementand organizationofinnovationprocessesbecomesmorecom-plex, i e. openinnovationincludesmanymoreactivities than justthosethatwereassignedtoatraditionalr&d department. 3. Innovationinsmes Havingdiscussedhowopeninnovationcanbedefined and operationalized, thecurrentsectiondevelopssome tentativepropositionsontheincidenceofopeninnovation in SMES, whatdifferencescanbeanticipatedbetween industriesandsizeclasses, andwhatmotivesandmanage-ment challengesmaybeencountered. 3. 1. Incidenceandtrends In theclosedinnovationmodelenterprisesmustgenerate their ownideasandthendevelop, build, market, distribute, and supportthemontheirown. Thismodelcounsels enterprisestobestronglyself-reliant, implicitlyrecom-mending organizinginnovationininternalr&ddepart-ments. Incontrast, theopenmodelprescribesenterprisesto draw onbothexternalandinternalideasandpathstothe market, whenenterpriseslooktodiscoveranddevelop innovativeopportunities (Chesbrough, 2003. Indoingso, the openinnovationmodelrecognizesthatsmallerfirms take anincreasinglyprominentroleinthecontemporary innovationlandscape. Somefirsttentativeevidenceis found in Chesbrough (2003) as hecitedstatisticsofhow small enterprisescontributetototalindustrialr&d expenses intheus. Theyaccountedforaround24%of all R&dspendingin2005, comparedtoonly4%in1981 (Nationalsciencefoundation, 2006. Likewise, inan interview-basedstudyof12enterprisesinmainlylow-tech industries, Chesbroughandcrowther (2006) foundthat ARTICLEINPRESS V. vandevrandeetal.//Technovation29 (2009) 423 437 425 basicallyallrespondentshadtosomeextentpickedup open innovationpractices, withaclearfocusontechnology explorationactivities. Anotherexampleis Lichtenthaler (2008) who conductedasurveyamongmedium-sizedand large manufacturersingermany, Austriaandswitzerland. He foundthat32. 5%oftherespondentswassomehow engaged inopeninnovation. Besides, therehavebeenmultiplestudiesonthestrengths and weaknessesofsmesintheirorganizationofinnova-tion processes (e g. Vossen, 1998; Acs andaudretsch, 1990. Thisworkconcludesthatinnovationinsmesis hampered bylackoffinancialresources, scantopportu-nities torecruitspecializedworkers, andsmallinnovation portfoliossothatrisksassociatedwithinnovationcannot be spread. SMESNEEDTOHEAVILYDRAWONTHEIRNETWORKSTO find missinginnovationresources, andduetotheir smallness, theywillbeconfrontedwiththeboundariesof their organizationsrathersoonerthanlater. Intoday's increasinglycomplexandknowledge-intensiveworld with shortenedproductlifecycles, suchnetworking behavior hasbecomeprobablyevenmoreimportant than before. Giventheseconsiderations, weanticipatethat open innovationpracticesarenotexclusivelyappliedby MNES, butwillalsobepresentinsmes, andwillbe increasinglyadopted. 3. 2. Industriesandsizeclasses Prior researchgivestheimpressionofindustrialdiffer-ences regardingtheincidenceofandtrendtowardsopen innovation. Inthecurrentpaper, weexplorethedifferences between manufacturingandservicesindustries. Services differ fromphysicalgoodsintermsofintangibility, inseparability, heterogeneityandperishability (Atuahene-Gima, 1996. Giventhedistinctnatureoftheofferingsof manufacturingandservicesfirms, differencesinthe adoptionofopeninnovationmaybeveryplausible. As physicalgoodsaremoreseparableandhomogenous, itis much easiertooutsourcepartsofther&dprocessorto in-sourcenewideasandtechnologiesthatfitwithcurrent businesslines. Gassmann (2006) proposes thatindustries are morepronetoengageinopeninnovationiftheyare characterizedbyglobalization, technologyintensity, technology fusion, newbusinessmodelsandknowledge leveraging. Wearguethatespeciallythefirstthree characteristicsasdefinedby Gassmann (2006) are more applicabletomanufacturersthantoservicesenterprises, i e. manufacturingenterprisesgenerallytendtooperatein largergeographicalregionsandthenatureoftheir processesdemandshigherinvestmentsincapitaland technologies. Forservices duetotheirrelativelyintangi-ble, simultaneousandheterogeneousnature theopposite applies. Indeed, descriptivestatisticsofdutchenterprises offered by Statisticsnetherlands (2006) demonstratethat manufacturersareonaveragemoretechnology-intensive, invest moreinr&d, andoperateinlargerregions. We thereforeanticipatethattheincidenceandadoptionof open innovationwillbestrongerinmanufacturing industries. Besidesindustrydifferences, thesizeofenterprisesmay also influencetheadoptionofopeninnovation. Oursurvey results containinformationonbothsmallenterprises (defined as10 99employees) andmedium-sizedones (100 499employees. Pastworkhasshownthatthereisa great dealofdifferenceintheinnovationstrategiesofsmall and largefirms (e g. Vossen, 1998; Acs andaudretsch, 1990. Innovationprocessesoflargerfirmsaretypically more structuredandprofessionalized. Assmesgrowthey increasinglydevelopandapplyformalstructures, also marked byrecruitingspecializedworkers, andintroducing manageriallayers, rulesandprocedures (Greiner, 1972. Once acriticalsizeisreached, theymaybebetterableto formalizetheirinnovationpracticesandtodevelop structures forlicensingip, venturingactivitiesandexternal participations. Theirlargersizealsoenablesthemto maintain largeanddiversifiedinnovationportfolios (to spreadrisks) andtoreservestructuralfundstofinance innovation. Thismayhaveimportantimplicationsfor the applicationofopeninnovationinthesefirms. The extent towhichtheycanengageintechnologyexploitation and explorationactivitiesislikelytobecontingenton their size. Asaresult, weproposethatopeninnovation is morecommonlyappliedbymedium-sizedenterprises and thatanytrendtowardsopeninnovationisstrongerin this group. 3. 3. Motivesandchallenges Manyfirmsstartedtoimplementopeninnovationasa necessaryorganizationaladaptationtochangesinthe environment (Chesbrough, 2003. Inaworldofmobile workers, abundantventurecapital, widelydistributed knowledgeandreducedproductlifecycles, mostenter-prises cannolongeraffordtoinnovateontheirown. A further explorationofmotiveswasdoneby Chesbrough and Crowther (2006. Inaninterview-basedstudythey found thatthemostcommonreasonforexternaltechnol-ogy acquisitionwasacommonbeliefthatitiscriticalto maintain growth. Itisanticipatedthatbasicentrepreneur-ial valuessuchasgrowthandrevenueswillbeamongthe key motivesofenterprisestopracticeopeninnovation. Previous workonmotivesforopeninnovationfocusedon MNES andusuallycoveredonlyfewopeninnovation practices. EIRMA (2003) for exampleshowedthatr&d managersoflargecorporationsengageinventuringfor market-relatedmotivessuchasmeetingcustomerdemands, but alsotoacquirenewknowledge. Likewise, Jacobsand Waalkens (2001) foundthatorganizations‘innovatetheir innovationprocesses'toreducetime-to-marketandto better utilizeinternalcreativity. Hence, weexpectmarket considerationsandknowledgecreationtobekeymotives for openinnovation. Otherpotentialmotivescanbederivedfrominno-vation collaborationstudies. Thisliteraturesuggeststhat ARTICLEINPRESS V. vandevrandeetal.//Technovation29 (2009) 423 437 426 enterprisesmayengageincollaborationtoacquiremissing knowledge, complementaryresourcesorfinance, tospread risks, toenlargeitssocialnetworks, ortoreducecosts (Hoffman andschlosser, 2001; Mohrandspekman, 1994. Such motivesmainlyreflectoutside-inconsiderationswhile motivestoconductoutboundactivitiesseemtobemissing. Koruna (2004) howeveridentifiedvariousobjectivesfor firms toexternallyexploittheirknowledge, including revenues andaccesstoknowledge, butalsotosetindustry standards, toprofitfrominfringements, torealizelearning effects, andtoguaranteefreedomtooperatebyestablish-ing cross-licensingagreementswithotherorganizations. As forthechallengesofopeninnovationinsmes, our data setcontainsinformationonperceivedbarriersto adoptopeninnovationpractices. Theopeninnovation literaturehassofarwitnessedfewattemptstoexplorethis subject. Chesbroughandcrowther (2006) for example identifiedthenot-invented-here (NIH) syndromeandlack of internalcommitmentasmainhamperingfactors. The NIH syndromehasbeenpreviouslyfoundtobea prominentbarrierforexternalknowledgeacquisition (e g. Katz andallen, 1982. Althoughfocusedontheexternal acquisitionofknowledge, itsunderlyingantecedentsare also applicabletotechnologyexploitation, leadingtothe‘only-used-here'(OUH) syndrome (Lichtenthalerand Ernst, 2006. Morepotentialbarrierscanagainbefound in therelatedliteratureoncollaborativeinnovation. Boschma (2005) for exampleidentifiedvariousformsof‘proximity'whichareessentialforeffectivecollaboration. These includecognitive, organizational, culturaland institutionaldifferencesbetweencollaborationpartners, implyingthatpotentialproblemsmayarisedueto insufficientknowledge, culturesormodesoforganization, or bureaucraticelements. Tomentiononlyafew, other potentialbarriersincludelackingresources, free-riding behavior, andproblemswithcontracts (Hoffmanand Schlosser, 2001; Mohr andspekman, 1994). 4. Methods 4. 1. Sample To analyzethetrends, motivesandmanagement challengesofsmeswithregardstoopeninnovation, we use asurveydatabasethatwascollectedbyeim, adutch instituteforbusinessandpolicyresearch. Thesurveywas commissionedbythedutchadvisorycouncilonscience and Technologytosupportapolicyadviceonopen innovation (see AWT, 2006. Thesurveytargetedsmes, definedasenterpriseswithnomorethan500employees, and wasimplementedbymeansofcomputer-assisted telephoneinterviewing. Datacollectionwasdoneovera 3-week periodindecember2005. Toreliablyidentify trends onlyrespondentswithlongtenureandrepresenting enterprisesthatsystematicallyinnovate, wereselected. The survey thereforestartedwithscreeningquestions. Respondents firstindicatediftheircompanyhaddevelopedatleast one innovationinthepast3years. Thiscouldeitherbea product-,process-,organizational-ormarketing-related innovationasdefinedbytheoslomanual (asetofintegral guidelinesforthecollectionofinnovationdata, see OECD, 2005. Secondly, thesurveyaskedifrespondents'enterprises hadformulatedaninnovationstrategy. Thirdly, respondentshadtobeemployedintheircurrentjobsforat least 7years. Inthisway, thescreeningensuredthat respondentsallrepresentedsmeswithsystematicinnova-tion efforts, andtheywereinapositiontoadequatelyjudge if andhowinnovationprocesseshaddevelopedoverthe past 7years. Thesamplewasdisproportionallystratifiedacross manufacturingandserviceindustriesandtwosizeclasses (10 99employeesand100 499employees. Enterpriseswith lessthan10employees (micro-enterprises) wereexcluded sincetheygenerallyhavenoorlimitedidentifiableinnova-tionactivities, andthispopulationusuallycontainsmany start-ups. Itwasanticipatedthatveryfewmicro-enterprises wouldpassthescreening. Thesamplewasdrawnfromthe Dutchchambersofcommercedatabase. Interviewers explicitlyaskedforthosewhowereresponsiblefor innovation, i e. smallbusinessowners, generalmanagers, R&d managersorstaffmanagingnewbusinessdevelop-mentactivities. Intotal2230respondentswerecontacted, of whom1206persons (54%)werewillingtoparticipate. To checkfornon-responsebias, respondentsandnon-respondentswerecomparedacrossindustriesandsize classes. Contrastingbothgroupswith w2-testsrevealedno significantdifferencesatthe5%level (p 0. 23fortypeof industryand p 0. 55forsizeclasses. Atotalof605 respondentspassedthescreeningphase, correspondingwith a finalsamplingrateof27%.%Table1 showshowtheseres-pondentsaredistributedacrosssizeclassesandindustries. 4. 2. Variables The surveyproceededwithquestionsonthenatureof firms'innovationprocesses. Morespecifically, eightopen innovationpracticesweredistinguished, whicharedefined in Table 2. Afterthescreeningquestions, therespondentswere asked iftheirenterprisehadengagedinanyventuring activitiesinthepast3years. Throughoutthesurveyatime space of3yearswasused, resemblingthecriterionusedby Statisticsnetherlandstoidentifyinnovativeenterprises (Statisticsnetherlands, 2006; OECD, 2005. Secondly, respondentswereaskedifventuringactivitiesintheir enterprisehadincreased, remainedstable, orhaddecreased in thepast7years (ifventuringactivitiesweremissingthe questionwasrephrasedasifventuringhadbeenstableor decreased. Thirdly, incasethefirmswereinvolvedinopen innovationactivities, theintervieweraskedtoprovidethe motives todoso. Theiranswerswererecordedinanopen-ended format. Finally, respondentswereaskediftheyhad perceivedanybarrierstoimplementopeninnovation practices, andifso, todescribethem. ARTICLEINPRESS V. vandevrandeetal.//Technovation29 (2009) 423 437 427 The otherinnovationpracticesweresurveyedwithan identical sequenceofquestions. Theonlyexceptionwasthe innovationpracticeofoutwardiplicensing. Here, the sequence wasprecededbyascreeningquestionchecking whetherthefirmactuallypossessedanyip. During thethirdandfourthpartofthesurvey, respondentswereaskedtoclarifytheirmotiveswhenthey get involvedinthedifferent‘openinnovation'practices. The variousanswersoftherespondentstothequestion what drivesthemtogetinvolvedinopeninnovation practices werecoded, resultinginthecategoriesdescribed in Table 7. Asimilarapproachwasadoptedforthe perceivedbarrierstoadaptopeninnovation. Thisresulted in elftypesofbarriersdescribedin Table 8. Thecoding process wasorganizedwithtworesearchers. Theyfirstread all open-endedanswersandtogetheridentifiedanumberof preliminarycategories. Next, theycarefullystudiedall answers andclassifiedthemintothescheme. New categories couldbeproposedwhenevertheyfeltthatthe categories wereinsufficientorshouldberefined. Finally, all classificationswerecomparedanddifferentopinions discussed andresolved. Becauseonlyfewsmespossess and tradeip (see Table 3), thedatadidnotcontainenough records toprovidereliableinsightsaboutrespondents'motives andchallengesonthistopic. 5. Results 5. 1. Incidenceandtrends Table3 shows theincidenceofopeninnovationpractices in oursampleofinnovativesmes. Thethreelastcolumns also giveanoverviewoftheevolutionoftheuseofthese practices indutchinnovativesmes. Thetableshowsthe shares ofrespondentsconductingvariousaspectsof technologyexploitationandtechnologyexploration, and the extenttowhichtheyperceivedanincrease, stabilization or decreaseintheapplicationofthesepracticesinthe past 7years. Table3 shows thatcustomerinvolvement external networking andemployeeinvolvementarefairlycommon innovationpractices. Outwardandinwardlicensingofip, venturingandexternalparticipationsinotherenterprises are conductedbyonlybyaminorityoftherespondents, while R&doutsourcingisdonebyhalfofthesample. ARTICLEINPRESS Table 1 Distributionofrespondentsacrossindustriesandsizeclasses Type ofindustry Size class 10 99 employees 100 499 employees Total Manufacturing Food andbeverages (NACECODES15 16 ) 40 21 Chemicals, rubberandplastics (NACECODES23 25) 54 22 Machinery andequipment (NACECODES29 34) 19 32 Other manufacturers (NACECODES17 22; 26 28; 35 37) 47 53 160 128 288 Services IT (NACECODE72) 53 17 Business services (NACECODES73 74) 59 24 Other services (NACECODES50 71; 93) 104 60 216 101 317 Total 376 229 605 Table 2 Surveyed openinnovationpractices Practice Definition Technologyexploitation Venturing Startingupneworganizationsdrawingoninternal knowledge, andpossiblyalsowithfinance, human capital andothersupportservicesfromyourenterprise. Outward IP licensing Selling orofferinglicensesorroyaltyagreementsto other organizationstobetterprofitfromyour intellectual property, suchaspatents, copyrightsor trade marks. Employee involvement Leveraging theknowledgeandinitiativesofemployees who arenotinvolvedinr&d, forexamplebytakingup suggestions, exemptingthemtoimplementideas, or creating autonomousteamstorealizeinnovations. Technologyexploration Customer involvement Directly involvingcustomersinyourinnovation processes, forexamplebyactivemarketresearchto check theirneeds, orbydevelopingproductsbasedon customers'specificationsormodificationsofproducts similar likeyours. External networking Drawing onorcollaboratingwithexternalnetwork partners tosupportinnovationprocesses, forexample for externalknowledgeorhumancapital. External participation Equity investmentsinneworestablishedenterprisesin order togainaccesstotheirknowledgeortoobtain others synergies. Outsourcing R&d Buying R&dservicesfromotherorganizations, suchas universities, publicresearchorganizations, commercial engineers orsuppliers. Inward IP licensing Buying orusingintellectualproperty, suchaspatents copyrightsortrademarks, ofotherorganizationsto benefit fromexternalknowledge. V. vandevrandeetal.//Technovation29 (2009) 423 437 428 The tablealsoshowsthatforeverysurveyedpractice, the share ofrespondentsperceivinganincreaseoverthepast7 years issubstantiallylargerthanthesharewithadecrease. These resultssuggestthatopeninnovationisnotjust conductedbymnes, butratheralsoappliestoabroad sampleofsmes, andmoreover, openinnovationison average increasinglyadopted. 5. 2. Industriesandsizeclasses Table 4 compares theincidenceandtrendtowardsopen innovationbetweenmanufacturingandservicesenter-prises. Foreaseofpresentation, trendscoreshavebeen averaged. Weappliedvariousteststoanalyzesignificant differences. As t-testprocedureswerelesssuitablebecause most dependentvariablesviolatedtherequirednormal distribution, Table 4 reportsnon-parametricmann Whitneytestsonsignificantmediandifferences. Wedid routinelycheckifourresultswererobustforthechosen test. Itappearedthat w2-andindependentsamples t-tests produced nearlyidenticalresults. Besides, asdutch manufacturerstendtoberelativelylargeorganizations (Bangma, 2005), wealsoranmultivariateanalysisof variancemodelsinwhichsizeclasseswereenteredas control variables. Again, significancesofthedifferences between manufacturingandserviceswerenearlyidentical (output availableonrequest. The left-handsideof Table 4 showsonlyfewsignificant differences betweenmanufacturingandservicesenter-prises. Employeeinvolvement, customerinvolvementand externalnetworkingappeartobemaintypesofopen innovationconductedbybothmanufacturersandservices enterprises. Wedoremarkthatthesepracticesweredefined very broadly (Table 2) andhencemaybluranysignificant difference (alsoseediscussionsection. Nevertheless, the other indicatorsrevealnosystematicpatternofdifferences between industries. Inmanufacturingthereseemstobe somewhat moreattentionfortechnologyexploration, i e. manufacturersrelativelyoftenengageinr&doutsourcing and inwardiplicensing. Incontrast, servicesenterprisesdo better onventuringactivities (33%versus24%,po0. 05. The right-handsideof Table 4 reveals thatthetrend towards openinnovationisobservedinbothindustries, i e. average trendscoresareconsistentlypositive. Weagain find onlyfewsignificantdifferences. Manufacturershave adopted R&doutsourcingmoreoften (0. 23versus0.13, po0. 01) whiletheoppositeappliestoventuringactivities. In arecentsurveyofmanufacturers, Lichtenthaler (2008) analyzedindustrydifferencesinmoredetailandalsofound no significantdifferences. Inall, wedonotfindmajor differences betweenthemanufacturingandservicesin-dustries withregardstotheincidenceandtrendtowards open innovationpractices. Table5 provides similaroutputforthedifferences between small-andmedium-sizedenterprises. Again, significanceswereanalyzedwithdifferenttests (including multivariateanalysisofvariancewithindustrycontrols) and provedtoberobust. ARTICLEINPRESS Table 3 Incidence andperceivedtrendsinopeninnovationpractices (n 605) Incidence(%)Perceived trend Increase(%)Stable(%)Decrease(%)Technology exploitation Venturing 29 14842 Outward IPLICENSING10 4 951 Employee involvement9342571 Technology exploration Customer involvement9738611 External networking9429674 External participation3216831 Outsourcingr&d5022735 Inward IPLICENSING20 5 932 Table 4 Incidence ofandperceivedtrendsinopeninnovationpracticesbetweenindustries Incidence Perceived trenda Manufacturing (n 288)(%Services (n 317)(%Mann Whitney Z (U) Manufacturing (n 288) Services (n 317) Mann Whitney Z (U) Technology exploitation Venturing 24 33 2. 44 0 09 0. 15 2. 14 Outward IPLICENSING11 8 1. 2 0. 02 0. 02 0. 1 Employee involvement94 93 0. 7 0. 41 0. 41 0. 2 Technology exploration Customer involvement98 97 0. 8 0. 34 0. 40 1. 3 External networking95 94 0. 6 0. 24 0. 26 0 . 4 External participation29 34 1. 2 0. 14 0. 15 0. 3 Outsourcingr&d59 43 4. 0**0. 23 0. 13 2. 6 *Inward IPLICENSING25 15 3. 2*0. 04 0. 03 0. 6**po0. 001,*po0. 01, 4po0. 05. aaverage scorewithincreasecoded1, stablecoded0anddecreasecoded 1. V. vandevrandeetal.//Technovation29 (2009) 423 437 429 Table 5 shows thatmedium-sizedenterprises (100 499 employees) aremorelikelytoengageinopeninnovation. On alltechnologyexploitationandexplorationpractices they aredoingslightlyorsubstantiallybetter. Bearingin mind thatemployeeinvolvement, customerinvolvement and externalnetworkingwerebroadlydefined, the differences betweenbothsizeclassesarenotsignificant. As forperceivedtrends, theright-handsideof Table 5 shows substantialdifferences. Allvaluesinthecolumnof respondentswith100 499employeesare (much) larger. Especiallyforthetechnologyexplorationactivitiesmed-ium-sizedenterprisesaremuchmoreinvolvedintheseopen innovationactivities. Thisresultcontraststhefindingsby Lichtenthaler (2008), whoconcludedthatfirmsizedidnot have amajorimpactonthedegreeoftechnology exploration, butitdidinfluencetechnologyexploitation. In sum, wefindthatmedium-sizedenterprisesapplyand adoptopeninnovationmoreoftenthantheirsmaller counterparts, asexpected. 5. 3. Clusteranalysis To exploretheincidenceofopeninnovationinmore detail, wedecidedtoclustertherespondentsingroupsof SMES thatarehomogenousintheiropeninnovation strategy andorganizationofinnovationpractices (seealso Lichtenthaler (2008) for asimilarapproach. Theanalysis was basedontheeightdichotomousvariablesmeasuring the incidenceoftechnologyexploitationandexploration practices. Westartedtheanalysiswithaprincipal componentanalysis (PCA) toreducethenumberof dimensionsinourdataandappliedclusteranalytic techniquestofindhomogeneousgroupsofenterprises. Finally, thedifferencesbetweenclusterswereexploredwith non-parametrictests. PCA summarizesthevarianceofasetofvariablesina limited numberofcomponents. Thisprovidesuncorrelated componentscoresattheintervallevelwhicharemore suitable forclusterprocedures, andpreventsthatsingle variables dominateaclustersolution (Hair etal. 1998). ) A first exploratoryrundemonstratedthatourdatawere suitable forpca (i e. MSAVALUESALL 40.57, KMO measure 0. 61 and p (Bartlett) o0. 001, see Hair etal.,1998). ) Todeterminethenumberofcomponentsweapplied the latentrootcriterion (eigenvalues 41.0). ) Asaresultwe obtainedathree-dimensionalsolutionexplaining57%of the variance. Intheappendixofthispaper, thematrixof componentloadingsisshown. Thefirstcomponentreflects the practicesofemployeeinvolvement, externalinvolve-ment andexternalnetworking. Thesecondcomponent containsr&doutsourcingandoutwardandinwardip licensing. Thethirdonerelatestoventuringandexternal participation. Sincethepcawasdonetoreducethe number ofdimensions, wedidnotattempttolabelthese components, butinsteadusedthethreefactorscoresasa basis forourclusterexercise. In theclusteranalysiswecombinedhierarchicaland non-hierarchicaltechniques. Thishelpstoobtainmore stable androbusttaxonomies (Milligan andsokol, 1980; Punj andstewart, 1983. Thehierarchicalanalysiswas done withward'smethodbasedonsquaredeuclidian distances. Next, non-hierarchicalclusteranalyseswere done todetermineafinalsolution. Weconsideredarange of initialsolutionsfromthehierarchicalanalysiswitheither ARTICLEINPRESS Table 5 Incidence ofandperceivedtrendsinopeninnovationpracticesbetweensizeclasses Incidence Perceived trenda 10 99 employees (n 376)(%100 499 employees (n 229 )(%Mann Whitney Z (U) 10 99 employees (n 376) 100 499 employees (n 229) Mann Whitney Z (U) Technologyexploitation Venturing 27 32 1. 4 0. 11 0. 14 1. 2 Outward IP licensing 6 16 4. 3**0. 01 0. 04 1 . 5 Employee involvement 92 96 1. 7 0. 37 0. 48 2. 8*Technologyexploration Customer involvement 97 98 1. 1 0. 30 0. 50 4. 6**External networking 94 95 0. 4 0. 20 0. 33 3. 2*External participation 24 44 5. 2**0. 13 0. 18 2. 04 Outsourcing R&d 42 645.1**0. 140.242.54 Inward IP licensing 14 294.7**0. 020.072.24**po0. 001,*po0. 01, 4po0. 05. aaverage scorewithincreasecoded1, stablecoded0anddecreasecoded 1. V. vandevrandeetal.//Technovation29 (2009) 423 437 430 two, three, fourorfivegroups (assuggestedbythe dendogram. Foreachnumberofgroups (k), weperformed a k-meansnon-hierarchicalanalysis, inwhichsmeswere iterativelydividedtothegroupsbasedontheirdistanceto the centroidsofourinitialhierarchicalsolutionsfor (following Milligan andsokol, 1980; Punj andstewart, 1983. Toassesswhichsolutionwasmoststablewe computedkappa, thechancecorrectedcoefficientof agreement (Singh, 1990), betweeneachinitialandfinal solution. Thethree-clustersolutionappearedtobeoptimal (k 0. 95, while ko0. 94 fortheothersolutions). A basicvalidityrequirementisthatoneshouldfind significantdifferencesbetweenthevariablesusedto develop theclusters (Hair etal. 1998). ) Kruskal wallis tests confirmedthisforallvariables (Table 6). Again, all significancesreportedherearerobust, i e. eitherparametric or non-parametrictestsgiveidenticalresults. Firms incluster1aremoststronglyinvolvedinopen innovation. Theyuseabroadsetofinnovationpracticesto improvetheirinnovationperformanceandareonaverage largerandarerelativelymorebasedinmanufacturing industriescomparedtotheothertwoclusters. Cluster2is the largestgroupoffirms; theseenterprisesnearlyalways rely ontheinvolvementofemployeesandcustomers, and externalnetworking, featureswhicharesharedwithcluster1. Cluster 3includesinnovativefirmsthatrelyheavilyon customerinvolvementbutmostofthemarenotinvolvedin relativelycomplexandformalizedtransactionformsof open innovationactivitiessuchasventuring, IP-trading, outsourcingofr&dandparticipationinotherfirms. The clusters provideasimilarviewonhowsmesapplyopen innovationpracticesaswasearlieridentifiedby Lichtenthaler (2008) for medium-sizedandlargemanufac-turers. Mostenterpriseshaveadoptedeitheropenorclosed strategiesonbothtechnologyexplorationandexploitation activities, i e. onlyfewrespondentsarefoundwith decidedlyhighscoresononedimensionandlowscores on theother, andtherearenotsufficientofthemtoform separate clusters. To furtherexplorethedifferencesbetweenclusters, Table 7 comparesaveragetrendscoresfortheapplication of innovationpracticesinthepast7years. Respondents in cluster1, whicharestronglyembracingopeninnovation, also intensifiedtheadoptionoftheopenmodelthe most. Theoppositeappliestothethirdcluster. Inother words, thedifferencesbetweenthethreeclustersare growing overtime. Nevertheless, thereisatrendtowards increasedadoptionofopeninnovationinallclusters; only inwardiplicensingisbecominglesspopularinthe third cluster. ARTICLEINPRESS Table 6 Incidence ofopeninnovationpracticesacrossthreeclusters Cluster1 (n 133)(%Cluster2 (n 411)(%Cluster3 (n 61)(%Kruskal-wallis w2 (df 2) Technology exploitation Venturing 40 27 15 14.5*Outward IPLICENSING44 1 0 227.3**Employee involvement98 99 38 340.5**Technology exploration Customer involvement98 99 77 109.3**External networking99 100 44 310.2**External participation44 31 11 20.4**Outsourcingr&d 70 48 21 41.5**Inward IPLICENSING 86 0 5 486.9***po0. 001,*po0. 01, 4po0. 05. Table 7 Perceived trenda in openinnovationpracticesacrossthreeclusters Cluster1 (n 133) Cluster2 (n 411) Cluster3 (n 61) Kruskal-wallis w2 (df 2) Technology exploitation Venturing 0. 17 0. 11 0. 05 5. 2 Outward IPLICENSING 0. 11 0. 00 0. 00 26.0**Employee involvement 0 . 53 0. 43 0. 07 36.1**Technology exploration Customer involvement 0. 52 0. 38 0. 05 36.3**External networking 0. 29 0. 27 0. 05 11.5*External participation 0. 23 0. 14 0. 02 14.6*Outsourcingr&d 0. 21 0. 18 0 . 07 4. 9 Inward IPLICENSING 0. 17 0. 00 0. 03 47.4***po0. 001,*po0. 01, 4po0. 05. aaverage scorewithincreasecoded1 , stablecoded0anddecreasecoded 1. V. vandevrandeetal.//Technovation29 (2009) 423 437 431 We alsoinvestigatedifenterprisesinthethreeclusters are evenlydistributedacrossindustriesandsizeclasses (see Table 6). Asforindustries, 58%oftherespondentsin cluster 1aremanufacturingcompanies. Inclusters2and3 these percentagesare55and43, respectively. AKRUSKAL Wallistestshowsthatthesedifferencesaresignificantat po0. 05 (Kruskal wallis w2 7. 3, df 2). Focusingon size classes, 55%oftherespondentsincluster1are medium-sizedenterprises. Inclusters2and3theseshares are 34%and25%,respectively. Again, thedifferencesare significant, nowat po0. 001 (Kruskal wallis w2 23.1, df 2). Itthusappearsthatenterprisesincluster1 (open innovators) tendtobelargerorganizations. Theseresults suggest asequenceintheadoptionofopeninnovation practices asorganizationsgrow. Cluster3containsmany small enterpriseswithmodestapplicationofopeninnova-tion, butevenhereamajorityoffirmsinvolvescustomers in theirinnovationprocesses. Themostdistinctivefeature of cluster2isthatthesesmesallengageinpracticeswhich can beorganizedinformallyandwhichdonotnecessarily requiresubstantialinvestments, includingemployeeinvol-vement andexternalnetworking. Medium-sizedenterprises are clearlyover-representedandtheirinnovationactivities are alsomarkedbypracticeswhichusuallydemand substantialinvestments, includingventuring, external participations, IPLICENSINGANDR&DOUTSOURCING. 5. 4. Motivesandchallenges The resultsanalyzedintheprevioussectionshowthat SMES clearlyhavetakenupamoreopenapproach towards innovation. Animportantpartofthesurvey focused onthemotivesandchallengesofsmeswhen pursuing openinnovation. Table 8 shows thatforalmost all openinnovationpracticespursuedbysmes themost important motivesaremarket-relatedones. Forthe majorityofrespondents, usingnewinnovationmethodsis regardedasawaytokeepupwithmarketdevelopments and tomeetcustomerdemand, whicheventuallyshould result inincreasedgrowth, betterfinancialresults, or increased marketshare. Market-relatedmotivesarethe most importantdeterminantforcompaniestoengagein venturing (31), %toparticipateinotherfirms (36%)andto involve userintheinnovationprocess (61%).%Manysmes believe itisnecessarytouseabroadsetofmethodstomeet the ever-changingcustomerdemandandtopreventthe firm frombeingoutperformedbycompetitorsornew entrants. Motivesrelatedtocontrol, focus, costsand capacity arementionedlessfrequently. An importantfindingisthatthedifferentinnovation practices seemtohavethesameunderlyingmotives. This implies thatventuring, participationinotherfirms, inter-organizationalnetworksandcustomerinvolvementare ARTICLEINPRESS Table 8 Motives toadoptopeninnovationpractices Category Examples Technologyexploitation Technologyexploration Venturing (n 83)(%Employee involvement (n 256)(%Customer involvement (n 232)(%External networking (n 175)(%External participation (n 94)(%Outsourcing R&d (n 134)(%Control Increasedcontroloveractivities, better organizationofcomplexprocesses 1 9 1131 Focus Fitwithcorecompetencies, clearfocusof firm activities 8 113 Innovation process Improvedproductdevelopment, process-/market innovation, integrationofnew technologies 23 19 21 24 8 Knowledgegainknowledge, bringexpertisetothe firm 4 535644 Costs Costmanagement, profitability, efficiency13 2 2 11 9 Capacity Cannotdoitalone, counterbalancelackof capacity 1 37513 Market Keepupwithcurrentmarket developments, customers, increasegrowth and/or marketshare 3 1361223614 Utilizationoptimaluseoftalents, knowledge, qualities, andinitiativesofemployees 30 Policy Organizationprinciples, management convictionthatinvolvementofemployees is desirable 15 Motivation Involvementofemployeesinthe innovationprocessincreasestheir motivationandcommitment 22 Other 19 11 9 11 14 8 Total 100 100 100 100 100 100 V. vandevrandeetal.//Technovation29 (2009) 423 437 432 complementaryinnovationactivitiesinimprovingproduct development, integratingnewtechnologiesandkeepingup with currentmarketdevelopments. Employee involvementistheonlyitemwheremotives are differentthanfortheotheritems. SMESCAPITALIZEON the knowledgeandinitiativesoftheir (non-R&d) employees foroptimaluseofhumancapitalandformarket considerations. However, employeeinvolvementisalsothe outcome ofan‘internalorganizationalpolicy'oritis stimulatedtoimprovemotivationandcommitmentof employees. Thesetwomotivesarenotnecessarilydictated by innovationobjectives. Table 9 identifies themainmanagerialandorganiza-tional challengesthatsmesperceivewhentheyadoptopen innovationpractices. Weremindthatinterviewersfirst asked ifrespondentshadexperiencedanybarrierstoopen innovation. Ifrespondentsansweredpositively, theinter-viewer exploredthenatureofthesebarriersbyopen-ended questions. Themainbarrierstoinnovationmentionedby the respondentsarerelatedtoventuring (mentionedby 48%oftherespondents), externalparticipation (48%),and outsourcingofr&d (43%.%Table 9 shows theextenttowhichthebarriersmentioned abovematterforeachofthedifferenttypesofopeninnova-tionactivities. Organizationand corporateculture-related issues thattypicallyemergewhentwoormorecompanies are workingtogetherareclearlythemostimportant barriers/thatfirmsfacewhentheyengageinventuring (35%),participationinotherfirms (75%),andthe involvementofexternalpartiesandusers (resp. 48%and 30%.%Thesetypesofopeninnovationrequirecooperation among differentorganizations, or, inthecaseofventur-ing, employeeswholeavetheorganization. Theseinter-organizationalrelationshipsfrequentlyleadtoproblems concerning thedivisionoftasksandresponsibility, the balance betweeninnovationandday-to-daymanagement tasks, andcommunicationproblemswithinandbetween organizations. The availabilityoftimeandresourcesisanotherbarrier. This isabarrierforalmostalltypesofopeninnovation practices buttherelativelylowscoresin Table 9 indicate that timeandresourcesarenotthemostimportantbarriers to implementopeninnovationpractices. Administration-related problemsoccurmuchmorefrequently, typicallyin the contextofventuring (28%),participationinotherfirms (13%)andtheinvolvementofexternalparties (10%),more specificallywhencooperatingwithgovernmentalorother not-for-profitinstitutions. Administrativeburdensarealso prominentwhenthecompanyreceivesgovernmental subsidies andgrants. Governmentalsupportisexperienced ARTICLEINPRESS Table 9 Hampering factorswhenadoptingopeninnovationpractices Category Examples Technology exploitation Technology exploration Venturing (n 40)(%Employee involvement (n 88)(%Customer involvement (n 68)(%External networking (n 53)(%External participation (n 45)(%Outsourcing R&d (n 57)(%Administrationbureaucracy, administrativeburdens, conflicting rules 28 10 13 19 Finance Obtainingfinancialresources 10 5 4 Knowledge Lackoftechnologicalknowledge competent personnel, orlegal/administrativeknowledge 5 5 Marketing Insufficientmarketintelligence, market affinity, marketingproblemsofproducts 10 5 Organization/culture Balancing innovationanddailytasks, communicationproblems, aligningpartners, organizationofinnovation 35 30 48 75 36 Resources Costsofinnovation, timeneeded 5 17 10 7 10 IPR Ownership ofdevelopedinnovations, user rights whendifferentpartiescooperate 10 5 Quality of partners Partner doesnotmeetexpectations, deadlines arenotmet 24 28 Adoption Adoptionproblems, customerrequirements misjudged 14 Demand Customerdemandtoospecific, innovation appears nottofitthemarket 28 Competencesemployeeslackknowledge/competences, not enough laborflexibility 24 Commitmentlackofemployeecommitment, resistanceto change 51 Idea management Employeeshavetoomanyideas, no managementsupport 8 Other 7 8 1 2 3 Total 100 100 100 100 100 100 V. vandevrandeetal.//Technovation29 (2009) 423 437 433 as beinghighlyinflexible, alsobecauseitisnotallowedto change partnersandsuchprogramscannotbeended prematurely. In addition, everysingleopeninnovationpracticecreates its ownspecificproblems. Forinstance, whencompanies involveexternalpartiesintheinnovationprocess, they frequentlyreportthatthesepartnerscannotmeetthe expectationsordelivertherequiredqualityofaproductor a service. Userinvolvementgoestogetherwithproblems related topropertyrights, adoptionandtoospecific customerdemands. Whenrelyingonemployeestoimple-ment openinnovation, itoftenturnsoutthattheydonot have therequiredcapabilitiesorskillstomakeavaluable contributiontoinnovation, ortheylackmotivationtodo so. Italsohappensthatintheend, managementdecidesnot to takeupanyoftheideasprovidedbyemployeesorthat the numberofideascomingfromindividualemployeesjust gets toolargetohandleinanefficientway. This, inturn, poses newchallengestomanagerswhentheywanttoget the mostoutofthecreativityoflargenumbersof individuals. Eventuallytheycangetassistancefroma growingnumberofspecializedservicesfirmstoexecute this job. Overall, wecanconcludethatmanybarriersforopen innovationinsmesarerelatedtocorporateorganization and culture, nomatterwhichtypeofopeninnovationis pursued. Ontopofthat, differenttypesofopeninnovation also havetheirownspecifictypesofproblemsandbarriers to overcome. Remarkalsothatthenumberofobservations in Table 9 is quitesmallerthanin Table 8. Therearethree possibleexplanationsforthisobservation: first, itcan indicatethatmanyrespondentsdidnotexperienceany barrierstoimplementopeninnovationpractices; next, respondentsmaynotbeawareofanybarriersbecausethey cannotcomparethemwithbestpractices; finally, respondents wereawareofsomeproblemsbutcouldnot articulatethem. 6. Discussion 6. 1. Conclusions Open innovationresearchhassofarfocusedonlarge and multinationalenterprises (MNES. Openinnovation practices ininnovatingsmeshavebeenneglected. This study addressesthisgapbyexploringtheincidenceofand trends towardsopeninnovationinsmes. Drawingona survey databaseof605innovativesmesinthenether-lands, weconcludethatsmesarepracticingextensively open innovationactivities, and, moreimportantly, that they areincreasinglydoingso. Inall, openinnovationis relevantandpresentinbusinesslife, i e. itappliesnotjust to MNESBUTALSOTOAMUCHBROADERGROUPOFSMALL-and medium-sizedenterprises. Ourresultsareinlinewiththe recent surveystudyof Lichtenthaler (2008) who demonstrated thatmedium-sizedandlargemanufacturersem-brace openinnovationpractices. Drawingonanexistingdatabase, openinnovationwas operationalizedalongtwodimensions, i e. technology exploitation (reflectinginnovationpracticestoorganize purposiveoutflowsofknowledge) andtechnologyexplora-tion (purposiveinflowsofknowledge). Fortechnology exploitation, ourdatasuggeststhatmanysmesattempt to benefitfromtheinitiativesandknowledgeoftheir (non-R&d) workers. Fortechnologyexploration, byfar most SMESSOMEHOWTRYTOINVOLVETHEIRCUSTOMERSIN innovationprocessesbytrackingtheirmodificationsin products, proactivelyinvolvingtheminmarketresearch, etc. Thisresultconfirmstheimportanceofuserinnovation (Von Hippel, 2005) formanysmes: reducingthefocusof open innovationinsmestoscience-driveninnovations would seriouslybiasourunderstandingofopeninnovation for thiscategoryoffirms. Furthermore, externalnetwork-ing toacquirenewormissingknowledgeisanimportant open innovationactivityamongsmes. Incontrast, outward andinwardiplicensing, venturingactivitiesand external participationsareonlypracticedbyaminorityof the respondents. Themorepopularpracticeslikecustomer involvementandexternalnetworkingareinformal, unstructured practiceswhichdonotnecessarilyrequire substantialinvestments. IPLICENSING, venturingandex-ternal participationonthecontrary, requirefinancial investments, formalizedcontractsandastructuredinnova-tion portfolioapproachtomanagetherisks. Thisfindingis in linewithformerstudiesaboutinnovationinsmes (e g. Vossen, 1998. One ofthemajorobjectivesofthesurveywastoknow whether openinnovationisincreasinglypracticedby SMES duringthelast7years. Respondentsunequivocally perceive atrendtowardsincreasedpopularityanddis-seminationofopeninnovation. Ourfindingssuggestthat innovationinsmesisbecomingmoreopen. Thisisnot surprising, consideringtheincreasinglyimportantrole small-andmedium-sizedfirmsplayininnovation. After all, smallfirmsoftenlackresourcestodevelopand commercializenewproductsin-houseand, asaresult, are moreofteninclinedorforcedtocollaboratewith other organizations. Drawingonpreviousworkweexpectedthatthe incidenceandtrendtowardsopeninnovationwouldbe stronger formanufacturingcompaniesandmedium-sized enterprises (asopposedtoservicescompaniesandsmall enterprises, respectively. Manufacturingfirmsareon average moreactiveintheoutsourcingofr&dandthe out-licensingofip, aresultthatisnotsurprisinggiventhe technologicalcommitmentofthesefirms, buttheydonot differ fromservicefirmsonotheropeninnovation activities. Thisisanimportantfinding; openinnovation is asrelevantforservicefirmsasitisformanufacturing firms, andresearchaboutopeninnovationshouldnotbe limited tothosesmesthathaveformalr&dactivities. This resultisinlinewith Lichtenthaler'sfindings (2008. He investigateddifferencesbetweenindustriesinmore detail, andfoundnosignificantdifferenceseither. ARTICLEINPRESS V. vandevrandeetal.//Technovation29 (2009) 423 437 434 In contrast, wefoundsignificantdifferencesinthe adaptionofopeninnovationpracticesbetweendifferent sizeclasses. Medium-sizedenterprisesengageinandadopt open innovationmoreoftenthansmallenterprises. These firmsdisposeoftherequiredscaleandresourcestoorganize a broaderrangeofinnovationactivities, andcomparedto smallenterprisestheymaybeconsideredaslarger repositoriesofknowledgethatcanbepurposivelyout-sourced. Thesurveyresultsfurthermorerevealthatopen innovationispresentinandincreasinglyadoptedbysmall enterprisesaswell, buttheadaptionrateforallexploration activitiesgrowsfasterformedium-sizedfirmsthanforsmall firms. Thisresultindicatesadivergentevolutionbetween medium-sizedfirmsandtheirsmallercounterparts. Clusteranalysisrevealedthreegroupsofsmes, clustering firms intogroupswithsimilaropeninnovationpractices. Theirfeaturesconfirm Lichtenthaler's (2008) conclusionthat companies seldomfocusoneithertechnologyexploitationor technologyexploration. Rather, openinnovatingcompanies tend tocombinethesetwoaspectsofopeninnovation. Besides, astheclusterofmost‘open'innovatorshas relativelymoremedium-sizedcompanies, theclustering implicitlysuggestsasequenceintheadoptionofopen innovation, startingwithcustomerinvolvement, following with employeeinvolvementandexternalnetworking, and ending withmore‘advanced'practiceslikeiplicensing, R&doutsourcing, venturingandexternalparticipations. The paperalsoexploredmotivesofsmestogetengaged into openinnovationandthebarriersmanagersexperience in implementingitintheorganization. Theresultsindicate that openinnovationinsmesismainlymotivatedby market-relatedtargets: SMESMAKEUSEOFSEVERALOPEN innovationpracticesatthesametimetoservecustomers effectivelyortoopenupnewmarkets, withhigher-order objectivestosecurerevenuesandtomaintaingrowth. This findingcorrespondswith Gans andstern (2003), who argued thatthemainproblemofsmallenterprisesisnotso much inventionbutcommercialization. Cooperationwith industryincumbentsmightbeonewaytoovercomethe difficultiesofcommercialization. Knowledgeacquisition and theeffectivenessofinnovationprocessesarealso frequentlymentioned, usuallyinthecontextoftechnology explorationpractices. Costandcontrolconsiderationswere mentionedmuchlessoften. The managerialandorganizationalbarrierstoopen innovationareverydiverse, butthemainbarriertoopen innovationinsmesisrelatedtotheorganizationaland culturalissueswhicharisewhensmesstarttointeractand collaboratewithexternalpartners. Theseissuesare encounteredinarangeofinnovationactivities, including venturing, customerinvolvement, externalnetworking, R&d outsourcingandexternalparticipations. 6. 2. Limitations The currentstudyisafirstexplorationoftheopen innovationpracticesinsmes. Consequently, ithasseveral limitations. Weidentifiedfourmajorlimitations. First, the measurementofsomeopeninnovationpracticeswasvery general assomepracticeswerebroadlydefined. This particularlyappliestoemployeeinvolvement, customer involvementandexternalnetworking. Theseinnovation practices wereintroducedtorespondentsinsuchawaythat most respondentsaffirmedtheywereapplyingthese practices. Althoughitisuncertainhowthedefinitions have influencedtheoutcomes, weprobablywouldgeta more preciseviewonopeninnovationinsmeswithmore narrowly definedpractices. Externalnetworkingwasfor exampledefinedas‘drawingonorcollaboratingwith externalnetworkpartnerstosupportinnovationprocesses, for exampleforexternalknowledgeorhumancapital'(Table2. Thispracticewouldincludeformalstrategic alliances withmultiplepartnerstoenableground-breaking research, butalsorelativelysimple, informalcontactswith supplierstodevelopprocessinnovations. Futureattempts to surveyopeninnovationinbroadsamplesofenterprises should delineatetheseveralpracticesinamoredetailed and accurateway. Next, thelistofopeninnovationindicatorsisprobably not acompletelist. Paststudieshaveproposedother practices thatwerenotincludedinthesurvey. Examples include theglobalizationofinnovationactivitiesandthe early involvedofsuppliersininnovationprocesses (see Gassmann, 2006. Onemayarguethatglobalizationofthe innovationprocessisnotrelevantforsmes. Nevertheless, we suggestthatglobalizationshouldbeincludedto complete thepicture. Asaconsequence, wecannotclaim that oursurveydatacapturethefulldomainofexternal technologyexploitationandexploration. Althoughoursampleofsmesisextensive, thereisstilla chance thatsometypesofenterpriseswerestilloverlooked. The screeningofrespondentsimpliedthatstart-upsand micro-enterprises (withlessthantenemployees) were excluded. Astheseenterpriseshavebeenrepeatedly identified assourcesofbreakthroughinnovations and challengersofincumbentinnovationactors (e g. Schumpeter, 1934), thisisanissuethatfutureresearchers should pickup. Moreover, thescreeningofrespondents based onthepresenceofinnovationactivitiesdistorts the‘representativeness'ofoursample, i e. resultscannot be generalizedtothepopulationofdutchenterprises with 10 499employees. Thisispartlyduetothescreening questions, butalsobecauseitwasdecidedthatmanu-facturers hadtobeover-sampledattheexpenseof services. Manufacturersareheavy-usersofinnovation policies, andfor‘political'reasonsthecommissionerof the surveyhadrequesteddetailedcoveringofthisgroup. Nevertheless, thesampledoesreflectabroadgroupof innovativesmesthatgoesbeyondthescopeofpastopen innovationstudies. Finally, motivesandperceivedchallengesweresurveyed only ifrespondentsreportedthattheyhadadoptedthe correspondingpractices. Duetolimitednumbersof respondentsourconclusionsareonlytentative, andfor ARTICLEINPRESS V. vandevrandeetal.//Technovation29 (2009) 423 437 435 outwardandinwardiplicensing, noresultscouldbe reported. Thisisregrettablebecauseiplicensingisan aspect ofopeninnovationthatisstillinitsnascentphase (Chesbroughetal. 2006) andprobablyinmostneedof detailedinvestigation (Lichtenthaler, 2007). 6. 3. Suggestionsforfurtherresearch Despite theselimitations, thefindingsofthecurrent study shouldencouragescholarstoanalyzeingreater depth openinnovationinsmes. Firstandforemost, our resultsindicatethatopeninnovationisrelevantformuch broader groupsofenterprisesthanjustlargeandmulti-nationalenterprisesorhigh-techmanufacturingfirms, i e. the openmodelispresentandincreasinglyapplied in thewholeeconomy. Futureresearchshouldbroaden the scopebystudyingopeninnovationinbroadersamples, also capturingsmallenterprisesandfirmsinservices industries. Open innovationstudieshavesofarbeendominated by qualitativeresearchapproaches, drawingheavilyon in-depthinterviewsandcasestudies. Suchmethodsare welcome tocharterrelativelynewphenomenaandto develop theories (Eisenhardt, 1989), butweanticipate that inthefurtherresearchonopeninnovation, quantitative researchmethodswillandshouldbeapplied more ofteninordertogeneralizeresearchoutcomes and totesthypotheses. Thisisalsorelevantforpolicy makers whowillfindithardtojustifyanddevelop policies foropeninnovationaslongasthereareno statistics demonstratingthatopeninnovationisrelevant for largebusinesspopulations. Weconsideritachallenge for statisticalofficestoadaptcurrentinnovation surveys tobetterreflectopeninnovation. Inthiscontext, we remarkthatcurrentinnovationsurveyssuchasthe CIS mainlyfocusonr&dandinnovationinvestments of enterprises, andexternalnetworkingactivities, but do notpayattentiontootheropeninnovationpractices (OECD, 2005). Especiallytechnologyexploitationactiv-ities areoverlooked. Thesurveypresentedhere might inspirestatisticalofficestomodifytheirsurveys, althoughtheabove-mentionedlimitationsshouldcertainly be accountedfor. The dynamicsofopeninnovationinsmesisanother researchareathatshouldbefurtherdeveloped. Our findings suggestthatsomeopeninnovationactivitiesare easy toimplementwhileothersmaybepickeduplaterin the growthcycleofthefirm. Clusteranalysisrevealedthree homogeneousgroupsofsmeswithsimilarapplicationof open innovationpractices. Theclustersimplicitlysuggesta sequence intheadoptionofopeninnovation, startingwith customerinvolvement, followingwithemployeeinvolve-ment andexternalnetworking, andendingwithmore advanced practiceswhichrequireformalbudgetsand greatersize, e g. IPLICENSING, R&doutsourcing, venturing and externalparticipations. Futureworkshouldfurther investigatehoworganizationsengageinopeninnovation during thesegrowthphases, andwhatmanagerialimplica-tions canbederived. In addition, thecurrentsurveydoesnotstudyhow large andsmallfirmsinteractinopeninnovation. Christensenetal. 2005) shows thatlarge, established companiesandsmallstart-upsmanageopeninno-vation differently, reflectingtheirdifferentialposition within theinnovationsystem. Hence, futureresearch should focusontherequirementsofopeninnovationon differences inculture, structureanddecisionmaking between partnersofdifferentsizesandfromdifferent industries. A finalrecommendationistostudythemotives and challengesrelatedtoopeninnovationinmore detail. Wefoundthatmarketconsiderationswerethe most importantreasonforsmestoengageinopen innovation. Thissuggeststhatsmesaremotivatedto capitalizeontheirinternalknowledgeandtofind alternativepathwaystomarkets. Itseemsthatfuture research shouldpaymoreattentiontothepurposive outflowsofknowledge, i e. technologyexploitationactiv-ities. Thisrecommendationisconsistentwith Lichtentha-ler's (2008) observationthattheinnovationprocessesof many enterprisesareincreasinglymarkedbyexternal technologyacquisition, butthatexternaltechnology exploitationtocommercializetechnologiesisofamore recent date (p. 148. Asforthemanagerialchallenges, we found thatorganizationalandculturalissuesarethe key barrierstoimplementopeninnovation. Thisiswell in linewithpastinterview-basedstudies (e g. Chesbrough and Crowther, 2006) andthecurrentliteratureon inter-organizationalcollaborationininnovation. However the questionremainshowsmescanbestdealwiththis major barrier. Annex. Principalcomponentanalysis Tablea1 shows componentloadingsofopeninnova-tion practicesonthreecomponentswitheigenvalues 41.0 (after varimaxrotation. Thissolutionexplains57%of the variance. ARTICLEINPRESS Table A1 Principalcomponentanalysisofopeninnovationpractices (n 605) Open innovationpracticecomponent1component2component3 Venturing 0. 02 0. 08 0. 83 Outward IPLICENSING 0. 04 0. 82 0. 04 Employeeinvolvement0. 72 0. 13 0. 01 Customerinvolvement0. 59 0. 08 0. 10 External networking0. 81 0. 07 0 . 01 External participation0. 11 0. 07 0. 81 Outsourcingr&d0. 21 0. 51 0. 13 Inward IPLICENSING0. 02 0. 80 0 . 06 Varianceexplained(%)25 17 15 V. vandevrandeetal.//Technovation29 (2009) 423 437 436 References Acs, Z. J.,Audretsch, D.,1990. Innovationandsmallfirms. MITPRESS, Cambridge, MA. Arora, A.,2002. Licensingtacitknowledge: intellectualpropertyrights and themarketforknow-how. Economicsofinnovationandnew Technology 4 (1), 41 59. Arora, A.,Fosfuri, A.,Gambardella, A.,2001. Marketsfortechnology and theirimplicationsforcorporatestrategy. Industrialandcorporate Change 10,419 450. Atuahene-Gima, K.,1996. Differentialpotencyoffactorsaffecting innovationperformanceinmanufacturingandservicesfirmsin Australia. Journalofproductinnovationmanagement13, 35 52. AWT, 2006. Openingvanzaken: Beleidvooropeninnovatie (Opening up: Policyforopeninnovation). ) Advisorycouncilforscienceand Technology, Denhaag, Thenetherlands. Bangma, K. L.,2005. Kleinschaligondernemen: Structuurenontwikkel-ing vanhetnederlandsemkb (Smallfirmsinthenetherlands: Structure anddevelopmentofthedutchbusinesssociety. EIM, Zoetermeer, Thenetherlands. Boschma, R. A.,2005. Proximityandinnovation: acriticalassessment. Regional Studies39 (1), 61 74. Chesbrough, H.,2002. Makingsenseofcorporateventurecapital. Harvard Businessreview, 4 11 march. Chesbrough, H.,2003. Openinnovation: Thenewimperativefor Creating andprofitingfromtechnology. Harvardbusinessschool Press, Boston, MA. Chesbrough, H.,2006. Openbusinessmodels: Howtothriveinanew Innovationlandscape. Harvardbusinessschoolpress, Boston, MA. Chesbrough, H.,Crowther, A k.,2006. Beyondhightech: earlyadopters of openinnovationinotherindustries. R&dmanagement36 (3), 229 236. Chesbrough, H.,Vanhaverbeke, W.,West, J.,2006. Openinnovation: Researchinganewparadigm. Oxforduniversitypress, London. Christensen, J.-F.,Oleson, M. H.,Kjaer, J. S.,2005. Theindustrial dynamicsofopeninnovation evidencefromthetransformationof consumer electronics. Researchpolicy34, 1533 1549. EIRMA, 2003. Innovationthroughspinninginandout. Workinggroup Report WG60. Eirma, Paris. Eisenhardt, K. M.,1989. Buildingtheoriesfromcasestudyresearch. Academy ofmanagementreview14 (4), 532 550. Gans, J. S.,Stern, S.,2003. Theproductmarketandthemarketfor‘‘ideas'':''commercializationstrategiesfortechnologyentrepreneurs. Research Policy32, 333 350. Gassmann, O.,2006. Openinguptheinnovationprocess: towardsan agenda. R&dmanagement36 (3), 223 228. Gomes-Casseres, B.,1997. Alliancestrategiesofsmallfirms. Small Business Economics9, 33 44. Greiner, L. E.,1972. Evolutionandrevolutionasorganizationsgrow. Harvard Businessreview50, 37 46. Hair, J. F.,Anderson, R. E.,Tatham, R. L.,Black, W c.,1998. Multi-variate Dataanalysis, fifthed. Prentice-hall, Englewoodcliffs, NJ. Henkel, J.,2006. Selectiverevealinginopeninnovationprocesses: thecase of embeddedlinux. Researchpolicy35, 953 969. Hoffman, W. H.,Schlosser, R.,2001. Successfactorsofstrategicalliances in smallandmedium-sizedenterprises: anempiricalsurvey. Long Range Planning34, 357 381. Jacobs, D.,Waalkens, J.,2001. Innovatiekwadraat: Vernieuwingeninde Innovatiefunctievanondernemingen (Innovationsquared: Innovation intheorganizationofinnovation. AWTACHTERGRONDSTUDIE, vol. 23. Kluwer, Deventer. Katz, R.,Allen, T. J.,1982. Investigatingthenot-invented-here (NIH)- syndrome: alookatperformance, tenureandcommunicationpatterns of 50r&dprojectgroups. R&dmanagement12, 7 19. Keil, T.,2002. Externalcorporateventuring: Strategicrenewalin Rapidlychangingindustries. Quorum, Westport, CT. Kirschbaum, R.,2005. Openinnovationinpractice. Researchon Technologymanagement48, 24 28. Koruna, S.,2004. Externaltechnologycommercialization-policyguide-lines. Internationaljournaloftechnologymanagement27 (2/3), 241 254. Laursen, K.,Salter, A.,2006. Openforinnovation: theroleofopennessin explaininginnovationperformanceamongukmanufacturingfirms. Strategic Managementjournal27, 131 150. Lecocq, X.,Demil, B.,2006. Strategizingindustrystructure: thecaseof open systemsinlow-techindustry. Strategicmanagementjournal27, 891 898. Lichtenthaler, U.,2007. Thedriversoftechnologylicensing: anindustry comparison. Californiamanagementreview49 (4), 67 89. Lichtenthaler, U.,2008. Openinnovationinpractice: ananalysisof strategic approachestotechnologytransactions. IEEETRANSACTIONS on Engineeringmanagement55 (1), 148 157. Lichtenthaler, U.,Ernst, H.,2006. Attitudestoexternallyorganis-ing knowledgemanagementtasks: areview, reconsideration and extensionofthenihsyndrome. R&dmanagement36 (4), 367 386. Lichtenthaler, U.,Ernst, H.,2007. Developingreputationtoovercomethe imperfectionsinthemarketsforknowledge. Researchpolicy36 (1), 37 55. Lord, M d.,Mandel, S w.,Wager, J. D.,2002. Spinningoutastar. Harvard Businessreview80, 115 121. Milligan, G. W.,Sokol, L. M.,1980. Atwo-stageclusteringalgorithmwith robust recoverycharacteristics. Educationalandpsychologicalmea-surement40, 755 759. Mohr, J.,Spekman, R.,1994. Characteristicsofpartnershipsuccess: partnershipattributes, communicationbehavior, andconflictresolu-tion techniques. Strategicmanagementjournal15, 135 152. National Sciencefoundation, 2006. Scienceresourcestudies, Surveyof Industrial Researchdevelopment. OECD, 2005. Oslomanual: Guidelinesforcollectingandinterpreting Innovationdata, thirded. OECD, Paris. Prencipe, A.,2000. Breadthanddepthoftechnologicalcapabilitiesin Cops: thecaseoftheaircraftenginecontrolsystem. Researchpolicy 29 (7 8), 895 911. Punj, G.,Stewart, D. W.,1983. Clusteranalysisinmarketingresearch: review andsuggestionsforapplication. Journalofmarketingresearch 20,134 148. Schumpeter, J. A.,1934. Theoryofeconomicdevelopment. Harvard Universitypress, Cambridge. Singh, J.,1990. Atypologyofconsumerdissatisfactionresponsestyles. Journal ofretailing661, 57 99. Statistics Netherlands, 2006. Kenniseneconomie (Knowledgeand Economy. Statisticsnetherlands, Voorburg, Thenetherlands. Teece, D.,1986. Profitingfromtechnologicalinnovation: Implicationsfor integration, collaboration, licensingandpublicpolicy. Researchpolicy 15,285 305. Van deven, A.,1986. Centralproblemsinthemanagementofinnovation. Managementscience32, 590 607. Van devrande, V.,Lemmens, C.,Vanhaverbeke, W.,2006. Choosing governancemodesforexternaltechnologysourcing. R&dmanage-ment 36,347 363. Van dijk, C.,Vandenende, J.,2002. Suggestionsystems: transferring employeecreativityintopracticableideas. R&dmanagement32, 387 395. Von Hippel, E.,2005. Democratizinginnovation. MITPRESS, Cambridge, MA. Vossen, R. W.,1998. Relativestrengthsandweaknessesofsmallfirmsin innovation. Internationalsmallbusinessjournal16 (3), 88 94. ARTICLEINPRESS V. vandevrandeetal.//Technovation29 (2009) 423 437 437

< Back - Next >