ContentslistsavailableatSciVerseScienceDirect
Research Policy
jo u rn a l h om epa ge :w w w . e l s e v i e r . c o m / l o c a t e / r e s p o l
What hampers innovation? Revealed barriers versus deterring barriers
Pablo D’Este
a,c,∗, Simona Iammarino
b,c, Maria Savona
c,d, Nick von Tunzelmann
caINGENIO(CSIC-UPV),SpanishCouncilforScientificResearch(CSIC)-PolytechnicUniversityofValencia(UPV),Spain
bLondonSchoolofEconomicsandPoliticalScience,DepartmentofGeography&Environment,UK
cScienceandTechnologyPolicyResearch(SPRU),UniversityofSussex,UK
dFacultyofEconomicsandSocialSciences,UniversityofLille1,France
a r t i c l e i n f o
Articlehistory:
Received14September2011 Accepted16September2011 Available online 26 October 2011
JELclassification:
O31 O32 O33 Keywords:
Barrierstoinnovation
Engagementininnovationactivities Innovativefirms
Non-innovators
a b s t r a c t
Innovatingfirmsarelikelytofaceseveralchallengesandexperiencedifferenttypesofbarriers.Inthis paperwearguethatitisnecessarytodistinguishbetweentwokindsofbarrierstoinnovation.Thefirst correspondstowhatwedescribeasrevealedbarriersandreflectsthedegreeofdifficultyoftheinnovation processandthelearningexperienceconsequentonthefirmengagingininnovationactivity.Thesecond typeofimpediment,whichwelabeldeterringbarriers,encompassestheobstaclesthatpreventfirms fromcommittingtoinnovation.Weusedatafromthe4thUKCommunityInnovationSurvey(CIS4)to investigatetherelationshipbetweenfirms’engagementininnovationandtheirassessmentofthebarriers toinnovation.Weshowthattherelationshipiscurvilinearinthecaseofcostsandmarketbarriers.These resultshaveimportantimplicationsforinnovationpolicyandinnovationmanagement.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
Successfulinnovationdependsonthefirmcombiningarange ofcapabilities,includingcapacitytoaccessfinance,understanding marketneeds,recruitinghigh-skilledstaff,andestablishingeffec- tiveinteractionswithotheractors.Innovatingfirmsnecessarilyare forcedtocopewithmost,ifnotallofthesechallenges.Somefirms, however,aredeterredfromengagementininnovationbecauseof thedifficultiesinvolved,andremainlockedintoestablishedrou- tines.Otherfirmsdotrytoinnovateandinvestinformalorinformal researchanddevelopment,butmayfailtobringnewproductsor processestomarketbecausetheyareunabletoovercomethese barriers.Itisimportant,fortworeasons,todistinguishbetween thedifferenttypesofinnovationbarriers.
First,adistinctioniscrucialfromaninnovationpolicyperspec- tive;inorderforpolicymakerstobeabletodesignappropriate measurestotacklesystemicfailuresthatpreventfirmsfromengag- ingininnovationactivities,theyneedtoidentifywhyandtowhat extentfirmsareexcludedfromtheinnovationcontest(Woolthuis, 2005;ChaminadeandEdquist,2006).Second,fromtheperspective
∗ Corresponding authorat: INGENIO(CSIC-UPV), Universidad Politécnicade Valencia,CiudadPolitécnicadelaInnovación,Edif8E4Planta,CaminodeVeras/n, 46022Valencia,Spain.Tel.:+34963877007x77048;fax:+34963877991.
E-mailaddress:pdeste@ingenio.upv.es(P.D’Este).
ofinnovationmanagement itis importanttoidentifythebarri- ers commonly faced byfirms engagingin innovation activities, andespeciallythosethatresultinfailuretointroducenewprod- ucts/processesinthemarket;thisshouldprovidecrucialinsights formanagers,toinformcorporatestrategiesorientedtoovercom- ingtheobstaclestoinnovation.
Thispaper aims toexamine two arguments. First, we claim thattoanalysetherelationshipbetweenengagementininnova- tionandthechallengesfacedbyfirms,itisnecessarytodistinguish betweentwomain barrierstoinnovation: revealedbarriersand deterringbarriers.Whiletheformerreferstothefirm’sawareness ofthedifficultiesinvolvedasaresultofengagementininnovation activities—pointingtoa“disclosing”or“learning”outcomebased ondirectexperience,thelatterreferstoabarrierthatisseenby firmsasbeinginsurmountable.Second,drawingonthisdistinction, wearguethatdifferenttypesofobstaclesarelikelytohavedistinct effectsaseitherdeterringorrevealedbarriers.Inotherwords,some typesofbarriersmightdetersomefirmsfromanyengagementin innovationactivities,andsometypesmaybeexperiencedmostly byfirmsheavilyengagedininnovation—irrespectiveoftheirsuc- cessatintroducingnewproductsorprocessesintothemarket.The objectiveofthispaperistoprovidenewinsightsregardingthese claims.
Thepaperisorganisedasfollows.Section2reviewsthemain literaturestreamsfocusingonthenatureandrelevanceofbarriers thatpreventorslowinnovationactivity.Section3discusseshow 0048-7333/$–seefrontmatter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.respol.2011.09.008
theinformationfrom4thUKCommunityInnovationSurvey(CIS4) isusedtoexaminetherelationshipbetweenfirms’engagementin innovationactivitiesandtheirassessmentsoftheimportanceofthe barrierstoinnovation.Sections4and5examinetheserelationships indetailandreporttheresultsoftheeconometricexercise.Section 6concludeswithadiscussionofthemainfindingsandprovides someimplicationsofourresearch.
2. Literaturebackground:revealedversusdeterring barrierstoinnovation
2.1. Barrierstoinnovation
Theempiricalliteraturebasedoninnovationsurveys,suchas theEuropeanCIS,whichexploresthenatureandimpactofinnova- tionacrossfirmsandsectors,islargeandconsolidated.Regarding theroleofbarrierstoinnovation,empiricalworklargelyfocuses eitheron(i)thefactorsaffectingperceptionsoftheimportanceof barriers(MohnenandRosa,2000;BaldwinandLin,2002;Baldwin andHanel,2003;GaliaandLegros,2004;Iammarinoetal.,2009) oron(ii)theimpactof(mainlyfinancial)obstaclesonthepropen- sitytoinnovateand/ortheintensityofinnovation(Arundel,1997;
TourignyandLe,2004;MohnenandRöller,2001,2005;Savignac, 2006,2008;Tiwarietal.,2007;MancusiandVezzulli,2010).We arguethatthesetwocategoriesofcontributions(whosemainfind- ingssubstantially converge),failtoidentifythedifferentnature of thebarriers to innovation in terms of theirrevealed versus deterringeffects,andthecontextsinwhichtheymightco-exist.
Also,mostsurvey-basedcontributionstendtofocusontheeffects of financial obstacles where there is a range of non-financial barriers—market,knowledgeandregulation—thatarecrucialinthe contextofinnovationpolicyandmanagement.
2.2. Revealedbarrierstoinnovation
Moststudiesthatfocusonthefactorsaffectingfirms’percep- tionsoftheimportanceofbarriers,showthatthegreaterthefirm’s involvementinR&Dand otherinnovationactivities,thegreater willbetheimportanceattachedtotheimpedimentstoinnovation.
Forinstance,BaldwinandLin(2002),inarepresentativesample ofCanadianmanufacturingfirms,examinewhetherthepropor- tionsoffirmsthatexperienceobstaclesdifferbetweeninnovators andnon-innovators(andbetweenadoptersandnon-adoptersof advancedtechnologies).Theyfindthatalargerproportionofinno- vatorsandadoptersofadvancedtechnologiesreportimpediments to technology adoption compared to non-innovators and non- adopters of advanced technologies. Mohnen and Rosa’s (2000) resultsfromanempiricalanalysisofCanadianservicesoverthe period1996–1998,basedonlyoninnovatorsandusingR&Dinten- sityasaproxyforinnovationintensity,aresimilarinsofarasthey findthatthemostinnovation-intensivefirmsarealsothosereport- ingmorefrequentobstaclestoinnovation.Alongthesamelines, Iammarinoetal.(2009),usingdatafromtheItalianCIS3andfocus- ingonfirmownership(i.e.foreignmultinationalsversusnationally ownedgroupsandsingledomesticfirms)andregionallocation,find supportforthehypothesisofapositiveassociationbetweenfirms’
perceptionofobstaclesandtheirinnovationpropensity.
Thesestudiestendtoexplain thissomewhatsurprisingfind- ingasduetoinnovatorsbeingmorelikelytohaveexperiencedthe barrierstoinnovationand,therefore,beingmorelikelytorecog- nisetheirimportance.AsGaliaandLegros(2004)(p.1189)suggest
“itisplausiblethatcertainproblemsarenoteffectivelyencoun- tereduntilfirmsfacethem.[...]innovativefirmsfaceproblems andmoreinnovativefirmshavemoreproblems”.Thiswouldimply thattheperceptionofobstaclesbyinnovativefirmsmayslow,but
notpreventfirms’engagementininnovationactivity.Amorecon- troversialinterpretationofthepositivelinkbetweeninnovation propensity/intensityandthelikelihoodofrecognisingthebarriers toinnovationasimportant,isofferedbyBaldwinandLin(2002)and TourignyandLe(2004).Theysuggestthattheobstaclestoinnova- tion,atleastasmeasuredininnovationsurveyssuchastheCIS, shouldnotbeinterpretedaspreventinginnovationortechnology adoption,butratherasanindicationofhowsuccessfulthefirmis atovercomingthem.
Thisfirstgroupofstudiesthusoffersarevealedbarriersinterpre- tationoftherelationshipbetweeninnovationeffortsandobstacles.
Thatis,engagementininnovationactivityincreasesfirms’aware- nessoftheassociateddifficulties(i.e.increasesconsciousnessand knowledgeofthefactorsconstraininginnovationthroughthe“dis- closing”or“learning”outcomeofdirectexperience),althoughit doesnotpreventthemfromengagingininnovationactivitiesor beingsuccessfulinnovators. Thisinterpretationisconfirmedby theevidencefromanotherstreamofliteratureinwhichanum- berofworksexaminefirms’experientiallearningfromtheirown (Mineretal.,1999;HaunschildandSullivan,2002;Denrell,2003) andotherorganisations’innovationfailures(KimandMiner,2007;
BaumandDahlin,2007).Theinnovationpathisinvariablypunc- tuatedbysetbacksandfailures(Ferrianietal.,2008),whichmight bemorevaluableforlearningthanaccomplishmentsandsuccesses (Mineretal.,1999;BaumandDahlin,2007).
Amongthestudiesinthesecondgroupofsurvey-basedcontri- butions,whichfocusontheimpactofobstaclesonthepropensity toinnovate,someaccountforpossibleestimationbiasduetothe endogeneityoftheregressors,thatis,reversecausalitybetween theperceptionofobstaclesandfirms’innovationefforts(Mohnen andRöller,2001,2005;Savignac,2006,2008;Tiwarietal.,2007;
MancusiandVezzulli,2010).Thesestudiesshowthatfirms’engage- mentininnovationactivitiesissignificantlyreducedordiscouraged bythepresenceofobstacles.Forexample,Savignac(2006),using dataon Frenchmanufacturing firms,shows that the likelihood that a firm will implement innovative projects is significantly reduced by the presence of financial constraints. Tiwari et al.
(2007),onthebasisoftheDutchCIS,estimatetheeffectofper- ceivedfinancialobstaclesonR&Dinvestment,1 andfindastrong and significantdeterrent effectoftheformer onthelatter.The authorstestthereverserelationship(i.e.theimpactoffirms’char- acteristicsontheprobabilitythatfinancialfactorsareperceived asimportant) and showthat, correcting for endogeneity,inno- vativenesshasapositive—althoughnotsignificant—effectonthe probabilitythatfinancialobstacleswillbeseenasimportant.This result is in line withthe findings of thefirst group of studies consideredabove.2Someofthemostrecentcontributionsinthis second lineof research,which focuses exclusively onfinancial obstacles,pointoutthatthecounter-intuitive,positivecorrelation betweentheperception ofbarriersand engagementin innova- tion, can be attributed to a combination of severalsources of bias (Savignac, 2008; Mancusi and Vezzulli, 2010).First, it can beattributedtothepresenceofheterogeneousunobservedfirm- specificfactors(suchasentrepreneurialbehaviouroridentification ofmarketopportunities)thatmayimpactonbothaspectsofthe relationship.Second,itcanbeattributedtothesimultaneousdeter- minationofthedecisiontoinnovateandthedecisiontofinance R&D projects.In addition tothese two sourcesof endogeneity, theremaybeselectionbiasifthefirmsampleincludesfirmsthat, for variousreasons, arenot interested in innovating.Thepres- enceofthesefirmsmaywellinduceapositivespuriouscorrelation
1Tiwarietal.(2007)includeonedummyforallotherobstaclesgroupedtogether, andfocusontheroleoffinancialbarriers.
2SeealsoMohnenetal.(2008).
Table1
Proportionoffirmsreportingbarriersasimportantbydegreeofengagementininnovativeactivities(numberofobservations:12,024).
TypeofBarriers Degreeofengagementininnovativeactivities
Zero 1–2 3–4 5–7 Chi-square(2)
Costfactors 30.7 29.0 36.6 42.8 136.69a
Knowledgefactors 12.1 10.8 13.1 15.2 25.26a
Marketfactors 19.0 15.3 17.4 19.7 23.95a
Regulationfactors 16.8 14.5 15.4 18.5 18.03a
aStatisticallysignificantat1%.
betweenperceptionofbarriersandfirminnovativeness(Savignac, 2008; Mancusiand Vezzulli, 2010), and may over-estimatethe roleofrevealedbarrierswhileunderestimatingorignoringdeter- rents.
2.3. Deterringbarrierstoinnovation
Supportforbarriersinterpretedasa realimpedimenttothe firm’sinnovationactivities—thatis,theviewofbarriersasinducing deterringeffects—comesfromdifferentstreamsoftheliteraturein thefieldsofinnovationmanagementandindustrialorganisation, whichcomplementsurvey-basedstudiesbydrawingonin-depth casestudyevidenceoranalysesofentrybarriersinparticularindus- tries.Theseresearchstrandsinvestigate,forinstance,whydifferent typesofcompaniesarelikelytoconfrontdifferenttypesofbarriers toinnovation.Distinctionsaremade,forexample,betweenlarge establishedfirmsandsmallnewfirmsandthereisabroadconsen- susthat,althoughtheformeraresuitedtodevelopingincremental innovations,smallnewfirmsarebetteratdevelopingradicalinno- vation(HamiltonandSingh,1992;Henderson,1993;Christensen andBower,1996).Theinnovationprofilesofthesetwogroupsof firmsdifferdue,amongotherfactors,tothetypesofobstaclesto innovationthattheyface.
Largeestablishedfirmsexperiencebarrierstoinnovationdueto pathdependenceandlockin,whichresultinaresistancetoadjust competenciesand previouslysuccessfulorganisationalpractices (Ferrianietal.,2008).Forinstance,organisationalinertiaandstruc- turedroutinesmaylimittheabilityofincumbentfirmstoidentify newopportunitiesand adapttoenvironmentalchanges(Nelson andWinter,1982;HannanandFreeman,1984;Dougherty,1992), strengthentheresistancetoengageinradicalinnovationtoavoid cannibalisingexistingproductsordestabilisingcorecompetencies (TushmanandAnderson,1986;Henderson,1993),orfosteranar- rowcommitmenttoafewmaincustomers(Christensen,1997).The obstaclesfacedbynewfirms,ontheotherhand,mayberelated, principally,tolackofresourcesandmarketstructure.Theformer includesknowledgeandorganisationalskills,suchasthelackof expertiseinthetechnologiesusedinmanufacturing-intensivesec- tors(GortandKlepper,1982;KatilaandShane,2005),andlackof finance(Schoonhovenetal.,1990;KatilaandShane,2005),while marketstructure,inthetraditionsofbothSchumpeter(1942)and Arrow(1962),mayimposesevereconstraintsintheformofcom- petition,firmsizeandappropriabilityconditions.3Newfirmsare likelytofacehigherbarrierstoinnovationinlargerandlesscom- petitivemarketswhereincumbentsarebetterplacedtocapitalise onthecapabilitiesforcoordinatingcomplementaryassetswhich newfirmsoftendonotpossess(Schoonhovenetal.,1990;Tripsas, 1997;Deanetal.,1998).
3ResearchontechnologicalregimesintheSchumpeter(1942)traditionempha- sizesthepositiverelationshipbetweenfirmsize,marketpowerandinnovativeness.
However,Arrow(1962)showsthat,underconditionsofincompleteappropriabil- ityofthereturnstoinnovation,thisrelationshipdoesnotnecessarilyhold.Within thisvastliteratureseeNelsonandWinter(1982),Freeman(1982),Winter(1984), MalerbaandOrsenigo(1993,1995).
In short, the innovation survey-based literature pays scant attentiontonon-innovativefirms,eitherfocusingonthebarriers faced byinnovativefirms(e.g.Galiaand Legros,2004; Mohnen andRosa,2000)ortreatingallnon-innovatorsasasingle,undif- ferentiatedgroup(e.g.BaldwinandLin,2002;Iammarinoetal., 2009).Inaddition,thesestudiesgenerallyfocusonlyonfinancial obstaclesandlargelyignoreotherconstraints(Tiwarietal.,2007;
Savignac,2008;MancusiandVezzulli,2010).Ourempiricalstudy triestoovercomesomeoftheselimitationsby:(a)explicitlycon- sideringtwotypesofinnovationbarriers:revealedanddeterring;
and(b)examiningwhethercost,knowledge,marketandregulation barriersplaydistinctrolesasdeterrentsbydiscouragingengage- mentininnovationtoutcourt,orasrevealedobstaclesbydisclosing thedifficultiesinherentintheinnovationprocessesofsuccessful innovators.
3. Dataandconstructs 3.1. Datasources
Thispaperusesdatafromthe2005UKInnovationSurvey(part ofthefourthiterationoftheCommunityInnovationSurvey-CIS4 -coveringEUcountries),whichrefertotheperiod2002–2004.The surveysampledmorethan28,000UKenterpriseswithtenormore employees,andhadwidesectoralcoverageincludingbothmanu- facturingandservicesectors.ThedataarestratifiedbyGovernment OfficeRegionsinEngland,Scotland,WalesandNorthernIreland.
Therawdataconsistofarepresentativesampleof16,445firms.
InordertoinvestigatetheissuesraisedinSection2,wedrawon theresponsestoquestionsintwosectionsofCIS4askingrespec- tively about firms’ assessment of the barriers related to their innovationactivities,andtherangeofinnovationactivitiesthey engagein.Sincetheresponsestothesetwoquestionsarecrucial fortheanalysisinthisstudy,wediscusstheirexactframingbelow.
3.2. Barrierstoinnovation
Thequestionasksfirmstoreportwhethertheyhaveexperi- encedbarriers,andifso,toassesstheirimportance.Itincludes11 itemsthatcapturefactorsthathamperinnovationeffortsorinflu- encethedecisionnottoinnovate(TableA.1intheAppendix).In TableA.1,the11itemsaregroupedinto4categories:costfactors, knowledgefactors,marketfactors,others.Othersincluderegula- tionissues and, inwhat follows,we refertoregulation factors.
Weusethesefourcategoriestopresenttheinformationfromthe firms’responsestothissectionofthequestionnaire,andtobuild ourbarrierconstructs.
3.3. Engagementininnovationactivities
Thequestiononengagementininnovationactivitiesasksfirms toreportwhethertheyhaveengagedinanyofseventypesofactiv- itiesandallowsustoidentifywhetherthefirmdidnotinnovate orwasengagedinoneorseveral(upto7)innovationactivities.
Table2
Listofvariables:descriptivestatistics(Numberofobservations:12,024).a
Variables Mean St.Dev. Min Max
Dependentvariables
Costbarriers 0.34 0.47 0 1
Knowledgebarriers 0.12 0.33 0 1
Marketbarriers 0.17 0.38 0 1
Regulationbarriers 0.16 0.37 0 1
Independentvariables
Zeroinnov.activities 0.23 0.42 0 1
1–2innov.activities 0.32 0.47 0 1
3–4innov.activities 0.27 0.45 0 1
5–7innov.activities 0.17 0.38 0 1
LnEmployees 4.11 1.51 2.20 11.01
Partofalargercompany 0.39 0.49 0 1
Startup 0.15 0.36 0 1
Internationalmarket 2.34 1.15 1 4
aDuetomissingvalues,thenumberofobservationsislowerthan12,024forsomeofthesevariables.
TableA.2(intheAppendix)reportshowthequestionwasformu- latedanddescribeseachoftheitems.
3.4. Identifying“potentialinnovators”
Inlinewithsomeof thecontributionsmentioned inSection 2 (i.e. Savignac, 2008; Mancusi and Vezzulli, 2010), firms that attempttoundertakeorhavealreadyundertakeninnovationactiv- itiesneedtobedistinguishedfromfirmswithnoaspirationsor intentionstoinnovate.Followingasimilarproceduretothatpro- posedinSavignac(2008),weexcludefromtheanalysisallthose firmsthatreportedthemselvestobenoninnovation-active(i.e.that respondedpositivelytoQuestion20inthesurvey,seeTableA.3 intheAppendix)anddidnotexperienceanybarrierstoinnova- tion(i.e.hadnotexperiencedanyofthe11obstaclesincludedin thequestiononbarriers,seeTableA.1).Thisexclusionis based ontherationalethatthesefirmsareunlikelytohaveanyaspira- tionstoinnovate(atleast,intheperiodconsideredinthesurvey).
About60%ofthesecompaniesindicated“noneedduetomarket conditions”amongthereasonswhytheyconsideredinnovation notnecessaryor notpossible.We classifiedthese3126as“not innovation-oriented”, and excluded them from the subsequent analysisbecausetheirinclusioncouldleadtoselectionbias(D’Este etal.,2008;Savignac,2008;MancusiandVezzulli,2010).
Theremaining12,024firmsareincluded,andareconsidered
“potentialinnovators”inthesensethattheyengagedininnovation activities(regardlessofwhethertheyhavemanagedtointroduce aneworsignificantlyimprovedproductorprocess)ordidnotdo sobecauseofthebarrierstoinnovation.4
3.5. Engagementininnovationactivitiesandbarriersto innovation
We provide a preliminary description of the relationship betweentheextenttowhichfirmsengageininnovationactivities andtheirassessmentofthebarriersinvolved.Todothis,wecon- siderthefourcategoriesofbarrierslistedabove(i.e.costs,market, knowledge,regulationfactors),settingfourintervalsfortheextent ofengagementininnovationactivity:noengagementatall(zero), engagementin1or2innovationactivities(1–2),engagementin3 or4innovationactivities(3–4)andengagementinmorethanfour
4 Thereasonwhythesefigures(12,024and3126)donotaddupto16,445isdueto missingvalueswithrespecttoinformationonproductandprocessinnovation,and toinconsistentresponsepatternsthatwereremovedfromtheanalysis(e.g.firms thatrespondedthat“factorsconstraininginnovation”wereamongthefactorsmak- inginnovationnotnecessaryorpossible,butindicatedthattheydidnotexperience anyofthebarrierslisted).
innovationactivities(5–7).Weusetheseintervalstoidentifyfirms thatdidnotengageininnovationactivity,fromthosefirmswith variousdegreesofengagement.
Table1considerseachofthefourbarriertypesseparatelyand showstheproportionoffirmsthatreportatleastoneitem(barrier) ashighlyimportantfordifferentlevelsofengagementininnovation activities.Table1providespreliminarysupportfortheproposi- tionthatassessmentofbarriersisnotindependentoftheextentof firmengagementininnovationactivity.Itshowsthat,regardlessof thetypeofbarrier,assessmentofbarriersasimportantincreases betweenlow(i.e.1–2)andhigh(5–7)levelsofinnovationactiv- ity.However,firmsthatreportednoinnovationactivitiestended toassessthebarrierstoinnovationascomparativelymoreimpor- tantthanfirmswithlowlevelsofengagement—andparticularlyin thecaseofmarketbarriers.
4. Assessmentofbarriersandengagementininnovation:
empiricalanalysis
Followingthesurvey-based literatureoninnovationbarriers discussedinSection2,weanalysehowtheperceptionofthedif- ferentobstaclesisaffectedbythefirm’sengagementininnovation activity,controllingforvariousfirmandenvironmentalcharacter- istics.
Fourdependentvariables,oneforeachsetofbarriers,arecon- structed as dichotomous variables: for each set of barriersthe variabletakesthevalue1ifthefirmassessesatleastoneitemas highlyimportant,and0otherwise.Sinceperceivedobstaclesare mostlikelytoberelated,possiblecorrelationbetweenthediffer- entdependentvariablespresentsanestimationproblem.Following GaliaandLegros(2004),totakeaccountofnon-independenceof categoriesofbarriersandtheneedtocontrolforpotentialcorrela- tionintheerrorterms,weranaMultivariateProbitModel(MPM) forthefourcategoriesofobstacles.5
Thefirm’sdegreeofengagementininnovationactivityismea- suredby a setof three dummyvariables thattakethe value1 ifthefirm hasengagedin1–2,3–4or5–7innovationactivities (thereferencecategoryisfirmsthatengagedinzeroinnovation activities).Firmcharacteristicsinclude:(a)firmsize,measuredby numberofemployees(logvalues);(b)whetherthefirmispartof anenterprisegroup;(c)whetherthefirmwasestablishedafter1 January2000;and(d)degreeofmarketinternationalisation.6We
5TheMPMallowstheerrortermstobefreelycorrelatedacrossequations,simi- lartoseeminglyunrelatedleastsquareregressions(SURmodels).TheuseofMPM allowsustoaccount(andtocontrol)forthefactthatthebarrierratingsmaybe correlated(seeGreene,2000,andforMaximumLikelihoodEstimationoftheMPM mvprobitprograminSTATA,seeCappellariandJenkins,2003).
Table3
Multivariateprobitresults.Dependentvariables:whetherthefirmassessesatleast1barrier-itemashighlyimportant,foreachsetofbarrierfactors.
Explanatoryvariables Costbarriers Knowledgebarriers Marketbarriers Regulationbarriers
Coefficient S.E. Coefficient S.E. Coefficient S.E. Coefficient S.E.
Noinnovationactivity Reference Reference Reference Reference
1–2InnovationActive −0.077** 0.034 −0.065 0.042 −0.188*** 0.038 −0.061 0.039
3–4InnovationActive 0.140*** 0.035 0.089** 0.044 −0.101** 0.040 0.039 0.041
5–7InnovationActive 0.299*** 0.0399 0.219*** 0.049 −0.030 0.045 0.241*** 0.046
LnEmployees −0.049*** 0.009 −0.070*** 0.012 −0.037*** 0.010 −0.074*** 0.011
Partofalargercompany 0.027 0.027 −0.038 0.034 0.011 0.031 −0.079** 0.032
Startup 0.111*** 0.033 0.047 0.041 0.072** 0.038 −0.076** 0.040
Internationalmarket 0.004 0.012 −0.035** 0.015 0.041*** 0.014 −0.096*** 0.015
Constant −0.311*** 0.072 −0.831*** 0.088 −0.722*** 0.080 −0.776*** 0.086
RegionalDummies Included Included Included Included
SectorDummies Included Included Included Included
Rho1 Rho2 Rho3 Rho4
Rho1 1.000
Rho2 0.431*** 1.000
Rho3 0.372*** 0.399*** 1.000
Rho4 0.359*** 0.337*** 0.297*** 1.000
No.totalobservations 11747
LogLikelihood −21049.7
Wald2(96) 723.0***
Twotailedttest.
*p<0.10
**p<0.05.
***p<0.01.
alsoincluderegionalandsectoraldummies.7Summarystatistics forthevariablesusedintheestimationarereportedinTable2.
5. Results
The resultsof the MPM are reportedin Table 38 and show thattherelationshipbetweenbeinganinnovation-activefirmand theimportanceattachedtobarriersisrathernuanced.Firmsthat engageheavilyininnovativeactivitiesaremorelikelytoassess barriersas importantcompared tofirms thatdo not engagein innovation activities(the reference category), withthe notable exception of “market related”barriers. Table 3 shows that the relationship between engagement in innovation activities and assessment of barriersappears todiffer acrossthe foursets of barriers.Forinstance,inthecaseofcostand,especially,market barriers,thedeterringeffectemergesasparticularlystrong,while itisweakerforknowledgeandregulationbarriers(weobserveno statisticalsignificanceatstandardlevels).
Also,therevealedorlearningeffectfrommoreintensiveinnova- tionactivityismorepronouncedinthecaseofcostandknowledge barriers,showingthat innovationexperiencegenerallyhelpsto reduce uncertainty - especially in relation to cost and knowl- edge,but alsoinrelationtoregulationinthecase ofvery high engagementininnovationactivity.Thelearningeffectisweakin thecaseofmarketbarriers,confirmingthatentrybarriersdueto
6Thisvariabletakesthevalues1–4dependingonwhetherthemostdistantmar- ketservedbythefirmisthelocalmarket(“1”),theUK(“2”),Europe(“3”)ora non-Europeancountry(“4”).
7AshighlightedinSection2,perceptionofobstaclesmaydependonwhether thefirmistryingtointroduceaninnovationorisengagingininnovationrelated activities,whileinnovation-activestatusmaydependonhavingexperiencedobsta- cles.Thiswouldrequirecorrectingforthepresenceofpotentialendogeneitybiasin theestimation.However,duetothelackofappropriateinstruments(Mohnenand Röller,2005;Iammarinoetal.,2009),wedonottrytocorrectforendogeneity.In ourview,thepresenceofendogeneityisnotlikelytoaffectthenatureordirection ofourfindings.
8Wealsoransomerobustnesschecksusingdifferentspecificationssuchas orderedprobitregressions,takinganorderedcategoricaldependentvariablefor thenumberofbarrieritemsrankedasimportant:theresultswereconsistentwith thoseinTable3andareavailablefromtheauthorsonrequest.
marketconcentrationortheriskofnotmeetingdemandexpecta- tions,dopreventfirmsfromengagingininnovation.
Table3showsthatforcostandmarketbarriersthereisevi- denceofanon-linearrelationship,becausefirmsneedtoprogress beyondacertainthresholdofengagementininnovationactivity beforeapositiverelationshipemerges.Belowthisthreshold,the relationshipisnegative;thatis,firmsthatdonotengageatallin innovationactivityaremorelikelytoassessbarriersasimportant, comparedtofirmsthatengagetoasmallextent(1–2activities)in thecaseofcostbarriers,oreventoamoderateextent(3–4activ- ities)inthecaseofmarketbarriers.Theexistenceofanon-linear relationshipbetweenengagementininnovationandassessment ofbarriersrepresentssomereconciliationbetweenthetwoappar- entlyconflictinginterpretationsofinnovationbarriersdiscussed inSection2.TheU-shapedrelationshipshowsthatcostsandmar- ketbarriershindercommitmenttoinnovationactivity(deterring effects)forsomefirmsandreflectlearningfromdirectexperience ofengagementininnovation(revealedeffects)forothers.
Regardingtheotherfirm-specific variables,asexpected,size significantlyaffectsperceptionsoftheobstaclestoinnovation(i.e.
largerfirmsperceivethemaslessrelevantthansmallerfirms),and beinga newfirmincreasestheprobabilityofassessingcostand marketbarriersasimportant,whichisinlinewiththeliterature reviewedinSection2andwiththestrongdeterrenteffectrelated tomarket obstacles.Internationalisation of thefirm’s customer baseseemstopromotelearningeffectswithregardtoovercom- inginnovation-relatedbarriers,strengtheningthefindingsofthe literatureon“learningbyexporting”(Sofronisetal.,1998,among others), but increasesperceptions of market concentrationand demanduncertainty—suggestingthatthelearningeffectislimited toknowledgeandregulationbarriers.
6. Summaryandconclusions
Thispaperhastriedtohighlighttheimportanceofdistinguish- ingbetweentwodifferenteffectsoftheobstaclesthatfirmsfacein undertakinginnovationactivity.Thefirsteffectisrelatedtoincreas- ing awareness of the difficultiesinvolved in innovating, or the
“disclosing”and“learning”contentofdirectexperience,described
hereastherevealedeffectofbarriers.Thesecondeffectisrelated totheperceptionoftheimpedimentstoinnovationbyfirmsthat otherwisewouldbekeentoengageinthisactivity:describedhere asthedeterringeffectofbarriers.
Wehaveshownthat,inthecasesofcostandmarketbarriers,the relationshipbetweenassessmentofthebarriersandengagementin innovationactivitiesischaracterisedbyanon-linearrelationship, indicatingthepresenceofbothdeterringandrevealedeffects.That is,theassessmentofbarriersasimportantishigherattheextremes:
whenfirmsdonotengageininnovationactivity,andwhenfirms arehighlyinnovative.
Firmsthatengageininnovationactivitiesgenerallyfacerevealed barriers,assessing barriers as highly important alongsidetheir engagementininnovation.Policymeasurestoremoveorreduce theobstaclestoinnovationforthesefirmsshouldbedirectedtothe micro-level,andinvolveencouragementofbettermanagementof innovationactivityinordertominimisetheimpactoftheseobsta- cles.Itislikelythat,forinnovators,theirrelativefailuresmightbe asvaluableastheirsuccesses,producingapositivelearningcycle.
Firmsthatdonotengageininnovation,conformtoapattern characterisedbydeterringbarriers.For firmstryingtoenterthe innovationcontest,obstacles—particularly thoserelatedtomar- ketstructureanddemand,andthecostofinnovationactivity—do representreasonsforwithdrawaland“failurewithoutlearning”.
Revealedanddeterringeffectsmaybothbepresent,depending onthespecificphaseintheinnovationtrajectory:whendeciding whethertoentertheinnovationcontest,someobstaclesmayactas realimpediments;aftercommittingtoinnovationandinthecourse oftheinnovationactivities,overcomingobstaclesmayresultina learningprocesswhichinturnproducesmoresuccessfulperfor- mance.
Ourfindingthatdeterringeffectsareparticularlystronginthe caseofmarketbarriersisimportantsincemuchoftheempirical literatureoninnovationbarriersdealsonlywithfinancialandcost relatedhamperingfactors.AsdiscussedinSection2,theimportant deterringroleofmarketfactorsmightbeinterpretedinlightofthe presenceofmarketsdominatedbyestablishedincumbentswhere
itisnotfeasiblefornew,smallerfirmstoengageininnovation- basedcompetition(e.g.Schoonhovenetal.,1990;Tripsas,1997;
Deanetal.,1998).Acomplementaryexplanationmightberelated totheuncertaintyinthedemandforinnovativegoodsandservices.
Previousstudiesfindthatthemarket’sresponsetotheintroduction ofanewproduct/servicecanbeseenasabarrierforfirmsdeciding whetherornottoinnovate(Iammarinoetal.,2009).Theriskofnot meetingdemandand,therefore,offailingtoincreasemarketshare, maypreventfirmsfrominnovating.Thisresultcouldbeinterpreted atthemacro-levelofanalysiswithina“Schmooklerian”framework, accordingtowhichthedecisiontoinvestininnovationismainly
“demand-led”.Itisimportantthatthisisaccountedforinpolicy aimedatenlargingthebaseofinnovators.Inthiscase,policymea- suresshouldbemacro-orientedinordertostimulateconsumption, increasingmarket’sresponsetonewproductstomatchdemand andsupplysideofinnovation.Thiswouldreducethestructuralrisk ofnotmeetingmarket’sdemandorfacinghighmarketconcentra- tionandthereforeencourageinitialinnovativeefforts.Adetailed analysisoftheissueofdemand-sideinnovationpolicyinrelation toobstaclesisbeyondthescopeofthispaper,butisanitemonour futureresearchagenda(D’Esteetal.,2008).
Acknowledgments
Theauthorsgratefullyacknowledgefinancialsupportfromthe (former)UK Department of Trade and Industry (DTI). Theyare extremelygratefulforsuggestionsandcommentsfromparticipants in theDTI Innovation Economics Conference, held 1 December 2006inLondon,theDIME InternationalWorkshopon“Agglom- erationandGrowthinKnowledge-basedSocieties”,KielInstitute, 20–21April2007,theInnovationandGrowthWorkshopatCRENoS, Cagliari,on11June2007,theDRUIDSummerConference,Copen- hagen,18–20June2007,andtheKiTES-CESPRISeminarSeries,28 February2009,Milan.Thefindingsandconclusionsofthepaper arethesoleresponsibilityoftheauthors.Allerrorsandomissions remainourown.
AppendixA.
TableA.1
Barrierstoinnovation:duringthethree-yearperiod2002–2004,howimportantwerethefollowingfactorsasconstraintstoyourinnovationactivitiesorinfluencinga decisionnottoinnovate?
Barrierfactors Barrieritems Factornotexperienced Degreeofimportance
Low Medium High
Costfactors Excessiveperceivedeconomicrisks
Directinnovationcoststoohigh
Costoffinance
Availabilityoffinance
Knowledgefactors Lackofqualifiedpersonnel
Lackofinformationontechnology
Lackofinformationonmarkets
Marketfactors Marketdominatedbyestablishedenterprises
Uncertaindemandforinnovativegoods/services
Regulationfactors NeedtomeetUKGovernmentregulations
NeedtomeetEUregulations
TableA.2
Engagementininnovativeactivities:duringthethree-yearperiod2002–2004,didyourenterpriseengageinthefollowinginnovationactivities?
Yes No
Intramural(in-house)R&D
Creativeworkundertakenwithinyourenterpriseonanoccasionalorregularbasistoincreasethestockofknowledgeanditsusetodevise newandimprovedgoods,servicesorprocesses
AcquisitionofR&D(extramuralR&D)
Sameactivitiesasabove,butpurchasedbyyourenterpriseandperformedbyothercompanies(includingotherenterpriseswithinyour group)orbypublicorprivateresearchorganisations
Acquisitionofmachinery,equipmentorsoftware
Acquisitionofadvancedmachinery,equipmentandcomputerhardwareorsoftwaretoproduceneworsignificantlyimprovedgoods, services,productionprocesses,ordeliverymethods