Research Policy The internationalisation of technol - PDF document

D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266Anincreasingshareoftechnologyisownedbyrmsfromadifferentcountrythantheoneoftheinventors(whichmainlyreectsthefactthatcompanieshaveresearchfacilitiesabroad).Theimportanceofthisphenomenonisnotreallynew.ComparingUSpatentsgrantedin1969–1972andin1983–1986,Cantwell(1989)reportedanincreasingshareofpatentswiththeownerandinventorlocatedindifferentcountries.Datareportedbelowshowthatthistrendiscontinuing.Whydomultinationalrmshaveresearchfacilitiesabroad?Arstansweristhatthisisanaccident,resultingfromcross-bordersmergersandacquisi-tions.Whentwormsofdifferentnationalities,bothequippedwithresearchlabsmergetheyendupwithonermowningresearchfacilitiesabroad,evenifthereisnospecialtechnologicalstrategybehind.Thisisapparentlyacommoncase(Patel,1995).However,theremaybealsospecialpurposesforamultinationalrmtolocatesomeofitsresearchfacilitiesabroad(Niosi,1999).Firstisthewilltoadaptitsproductstolocalmarketsandtoprovidetechnologicalsupporttolocalsubsidiary.Suchtechnologytransferbenetsrsttherecipientcountry.Secondisthewilltomoni-tornewtechnologydevelopmentsoccurringinforeigncountries,to“tap”foreigntechnology.Itisnotenoughtoreadtechnicaljournalstokeeppacewithadvancingtechnology,itisalsonecessarytobepartofresearchersnetworks,whichmeanstobeactiveinresearchintheareasofinterest.Bygoingdirectlytotheplaceswithmoreexpertiseinagiventechnologicaleldthermisabletopenetrateatalowercostsuchnetworks.Thirdisanattempttodevelopspecialtechnologyinwhichtherecipientcountryhascomparativeadvantageandwhichcomplementstherm’scoretechnology.Thesecondandthirdmechanismsarealsosomekindoftechnologytransfer,fromtherecipientcountrytotheinvestingone.Whilethersttwomechanismshavefoundsomeempiricalconrmation,thethirdonehasnot(PatelandVega,1999;DunningandWymbs,1999;LichtenbergandvanPottelsberghe,2001).However,internationalisationoftechnologyisnotrestrictedtoownershipandtransfers:itfurtherextendstotheverygenerationofknowledge,asresearchersbasedindifferentcountriesco-operateonthesameprojectsandjointlyinventnewdevices.Itiswidely SeeArchibugiandIammarino(1999)forausefultaxonomyoftheglobalisationofinnovation.acceptedthat,asapublicgood,knowledgediffusesacrossborders:butitsgenerationisalsoincreas-inglyorganisedonaninternationalbasis.Thankstodecreasingcommunicationcoststhecostofnetwork-ingisreduced,andduetoincreasingspecialisationofresearchersthebenetsfromnetworkingareget-tingbigger.Internationalco-operationfacilitatestheconvergenceofvarioussourcesofknowledgethatisnecessarytogeneratenewtechnologyinmanyelds(e.g.biotech,whichrequiresbothbiologyandcomputerscience),asspecialisedexpertiseisoftenavailableindifferentcountries.Thisstudyexploresmeasurementandanalyticalissuesrelatingtotheinternationalisationoftechno-logy,basedonpatentsdata.Wepresentandanalysethreenewpatent-basedindicatorsofinternationalisa-tionoftechnology.Theseindicatorsarecalculatedatthecountrylevelforeachofthe29OECDmembercountries.Theyrelatetotheresearchactivityabroadofdomesticmultinationalrms,tothedomesticactiv-ityofforeignmultinationalrmsandtointernationalco-operationinresearch.Weareprimarilyconcernedwiththefollowingissues:Whataretheadvantagesanddrawbacksoftheseindicators?Aretheydifferentfromotherindicatorsofinternationalisationoftech-nology(e.g.R&Dabroadofmultinationalrms)ordotheyconveyconsistentmessages?Hastheinterna-tionalisationoftechnologyincreasedoverthepast15years?Doesitaffectallcountriestoasimilarextent?Thispaperalsoanalysesthefactorsthataffectthedegreeandthepatternofinternationalisationoftechnology:Doesthesize(GDP)andtechnologicalin-tensityofacountryaffectitsdegreeofinternationali-sation?Whichcountriesarerelativelyopen(closed)toforeigntechnologywhenotherfactorsarecontrolledfor?Doestechnologicalproximity,geographicaldis-tance,andothercountry-specicfactorsaffectthegeo-graphicaldistributionofinternationalisation(i.e.withwhichpartnerwillacountryhavemorelinkages?).Thecountrylevelapproachtakeninthisstudycom-plementsthecompanylevelapproachtakeninmostoftheliteratureinthiseld.Thecompanyapproachisabletocaptureaspectssuchastheindustrytowhichtheconcernedactorsbelong,itisnon-ambiguousonthenationalityofthepatentee(therm)atonepointintime.Hence,thecountrylevelapproachcannotaddressissuessuchashowdoesinternationalisationrelatetothecorporatestrategy(e.g.ismultinational D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266rms’researchabroadrelatedtotheircoreactivitiesortocomplementaryactivities?),whichistreatedindepthincertainrmlevelstudies(Zander,1995;CantwellandJanne,1999).Ontheotherhand,thecountryapproachisexhaustive,asallpatentsaretreated,whoeverthepatentee,insteadofaselectionoflargecompanies.Itfurtherallowstocovermorecoun-triesandtogiveforeachcountryamorecompletepic-tureofinternationalisation.Moreover,theboundariesofcountriesaregenerallystableovertime,contrarytothoseofrms,whichfacilitatestimeconsistencyinthetreatmentofthenationalityofpatentees.ThestudyhasbeenperformedonadatabaseofpatentsappliedtotheEuropeanPatentOfce(EPO)overtheperiods1985–1987and1993–1995.AfewtestsreportedbelowshowthatresultsaresimilarwhenbasedonpatentsgrantedbytheUSPTO.Thenextsectiondenesthevariousindicatorsofinterna-tionalisationandpresentstheirtrendsattheOECDwidelevel.Section3describesthecountrypatternsofinternationalisationoftechnologyandcomparesthepatent-basedindicatorswithotherindicatorsofinter-nationalisation.Sections4and5evaluateempiricallythedeterminantsoftheextentofinternationalisationandofthegeographicaldistributionofinternationali-sation,respectively.ConcludingremarksarereportedinSection6.2.ConceptsandglobaltrendsPatentsareincreasinglyrecognisedasarichsourceofinformationregardingtechnologicalperformance.Amongtheinformationavailablefrompatentlesaretheinventorandtheapplicant(theownerofthepatentatthetimeofapplication)addresses,hencetheircoun-triesofresidence.Whenstatisticallyelaborated,thisinformationallowstomapsomeaspectsoftheinterna-tionalisationprocess.Theadvantagesofpatentsinthisareaaretheirbroadavailability(availableforallcoun-triesintheworld),internationalcomparability(whenafewsourcesofbiasaredealtwith),andpossibilitiestobematchedwithothertypesofdata(e.g.rmlevel, Forthestrengthsandweaknessesofpatentsasindicatorofinnovativeoutput,seeGriliches(1990)andfortheuseofpatentdataasindicatoroftechnologicalspecialisation,seeArchibugiandPianta(1992).countryorindustrylevel).Theirmajordrawbackisadifcultytointerpret,insomecases,themeaningoftheindicators,i.e.whichunderlyingactivityisactu-allyreectedinthepatent.Patentsthatareofinterestformeasuringinter-nationalisationoftechnologyarethosewithseveralapplicantsfromdifferentcountries,orseveralin-ventorsfromdifferentcountries,oranapplicantandinventorfromdifferentcountries.Mappingthesepopulationsofpatentsandcomparingthemtootherpatentsisthepurposeoftheindicatorswhicharepresentedinthenextparagraphs.Cross-borderownershipofinventionshappenswhenatleastoneinventorandtheapplicantresideindifferentcountries.ItisdeemedtoreectthelocationofR&Dactivitiesofmultinationalrms.FormostEPOpatents(ashareusuallyestimatedtobehigherthan90%),theapplicantisaninstitution(arm,auniversity,apubliclaboratory).Theinventorisalwaysanindividual,usuallyaresearcheremployedbytheapplicant.Mostoften,theaddressoftheinventoristheaddressofthelaboratoryhe/sheworksin.Then,whentheinventorandtheapplicantofapatentdonotresideinthesamecountry,thisreectsinahugemajorityofcasesthefactthatthepatentprotectsaninventionperformedinaresearchfacilityabroadofamultinationalrm.Twoindicatorshavebeencalcu-latedbasedonthisdata,whichmirroreachother.SHIAistheshareforagivencountryofpatentswithadomesticinventorandaforeignapplicantinthecountry’stotaldomesticinventions.Itreectstheextenttowhichforeignrmscontrol(own)domes-ticinventions.Algebraically:PFisthenumberofinventedbytheresidentsofcountryatleastpartlyownedbytheresidentsofcountryisthetotalfractionalnumberofpatentsinventedbytheresidentsofcountrycontrolledbyforeignresidents.SHIAistheshareofpatentscontrolledbyforeignresidentsinthetotalfractionalnumberofpatentsinventedbyresidents(PFI). Patentsarecountedfractionally:whenapatenthasseveralin-ventorsresidingindifferentcountries,itis“shared”betweenthesecountries,eachofwhichisattributedafractionthatcorrespondstoitsshareinthenumberofinventors.Forinstance,ifapatenthastwoinventorsfromcountryAandthreeinventorsfromcountryB,thencountryA’snumberwillbe0.4andcountryB’swillbe0.6. D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266SHAIistheshareforagivencountryofpatentswithaforeigninventorandadomesticapplicantinthecountry’stotaldomesticapplications.Itreectstheextenttowhichdomesticrmscontrolforeigninventions.Algebraically:PFisthenumberofpatentsownedbytheresidentsofcountryandatleastpartlyinventedbytheresidentsofcountryisthetotalnumberofpatentscontrolledbytheresidentsofcountryandinventedbyforeignresidents.SHAIPFAistheshareofpatentsinventedbyforeignresidentsinthetotalfractionalnumberofpatentscontrolledbyresidents(PFA).Internationalcollaborationinscienceandtechnologytakesplacewhenapatenthasseveralinventorsresidingindifferentcountries.Thiskindofin-ternationalcollaborationbetweenresearcherscantakeplaceeitherwithinamultinationalcorporation(researchfacilitiesinseveralcountries),orthrougharesearchjointventurebetweenseveralrms.Itisreectedbythefollowingindicator.SHIIistheshareforagivencountryofpatentswithaforeignresidentasco-inventorinthepopulationofpatentswithadomesticinventor.Algebraically:isthenumberofpatentsco-inventedbyresi-dentsofcountryandresidentsofcountryisthetotalnumberofpatentsinventedbytheresidentsofcountryincollaborationwithforeignresearchers.SHIIistheshareofpatentsresultingfrominternationalresearchco-operationinthetotalnumberofpatentsinventedbyresidentsofagivencountry(PI).Table1presentssomeillustrativeexamplesofpat-entsconcernedwiththethreeindicators.ThefourTable1Examplesofcross-borderownershipandinternationalco-operation ApplicationnumberApplicantInventorSHAI/SHIASHII NameResident 472,807MicrosoftCorporationUSFrance(2)XXUS(1)859,431ColgatePalmoliveCompanyUSBelgium(2)X828,191AlcatelAlsthomCGEFranceGermany(2)X463,418GeneralHospitalCorporationUSBelgium(2)XXRijksuniversiteitBelgiumUS(1) Source:USPTOWebsitehttp://www.uspto.gov/.patentslistedhavebeengrantedbytheUSPTO.Theyallrepresentasituationofcross-borderownership.ThepatentsownedbyColgatePalmolive(US)andAlcatelAlsthomCGE(France)witnessacross-borderowner-shipofinventionmadebyBelgianandGermaninven-tors,respectively.Oneexampleofbothinternationalco-operationbetweeninventorsandcross-borderown-ershipisprovidedbytheMicrosoftpatent,whichwasinventedbyresearcherswithresidenceindifferentcountries(Franco-UScollaboration).Thefourthpatentillustratesthethreetypesofinternationalisationalto-gether.Itisco-ownedbyaUSHospitalandaBelgianUniversity(i.e.aninternationalco-application)andhasbeeninventedbytworesearcherswithaBelgianresidenceandoneresearcherwithresidenceintheUS.Thesevariousindicatorsarenotindependentfromeachother.Forinstance,allpatentswithco-inventorsresidingindifferentcountrieswillalsohave,foratleastoneoftheconcernedcountries,aforeignapplicant.Asaconsequence,anypatentcountedinSHIIforacountrywillalsobecountedeitherinSHIAorinSHAIforthiscountry.Analysingtheshareofco-inventionsinthepopulationofforeignownedpatentsallowstoanalysetheextenttowhichcross-borderownershipfavoursinternationalcircula-tionofknowledge.AnotherfeatureisthatworldwideSHIAisequaltoSHAI.Thisfactdoesnotprecludelargedifferencesbetweenthetwoindicatorsatthecountrylevel.AnindicatorsimilartoSHAIhasbeenproposedbytheUniversityofReading,relyingonpatentcountdataobtainedfromtheSciencePolicyResearchUnit(SPRU;seeespeciallyPatelandPavitt(1991,2000),Dunning(1994)andDunningandWymbs(1999)).Thisindicatorhasbeencomputedwithdata D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266 Fig.1.Globaltrendsintheinternationalisationoftechnology.onthepatentsregisteredintheUSby727oftheworld’slargestrms.ItreectstheshareofUSpatentsledbythesermsattributabletoresearchinforeignlocations(outsidethehomecountryoftheparentcom-pany).ContrarytoSPRU,weuseallpatents(includingthoseledbysmallrms,largedomesticrms,pub-licinstitutions,universities,etc.),whichmeansthatthedegreeofinternationalisationofacountryisrep-resentedmorecompletely.However,identifyingtheownerrmasSPRUdoesand,aswedonot,allowstoidentifytheactualownerofthepatent,beyondthedirectowner:hence,foreignafliatesofmultinationalrmscanbeidentiedandtreatedassuch,whichisnotthecaseinourapproach.Theindicatorsproposedhereshouldbeconsideredaslower-boundsindicatorsofinternationalisationoftechnology.Amajorfactorofunder-estimationofactualinternationalisationinourdataisrelatedtotheincreasingtendency,sincetheearly1980satleast,ofcross-bordermergersandacquisitions(M&A).AsunderlinedbyDunning(1994),animportantraisond’êtreforthegrowingshareoftheseM&A’sisto Itisworthmentioningthatthecross-countryandcross-industriesdifferencesobservedwiththeSPRUindicatoraresimi-lartothethoseobservedwithSHAI.acquiretheinnovativecapacityofthetargetedrms.Patentdatabasesdonotregistersuchchangesinownershipofpatents.Whatdotheseindicatorstellus?ThetimetrendsoftheindicatorsofinternationalisationoftechnologyattheOECD-widelevelarepicturedinFig.1.Overthe1980–1995period,thedegreeofinternationalR&Dcollaboration(SHII,theshareofpatentsinvolvingatleasttwoinventorsfromdifferentcountries)hasmorethandoubled,from2.1%in1980to4.7%in1995.Thatis,almost5outof100patentedinventionsintheOECDareaarethefruitofinternationalresearchcollaboration.However,thecross-countryaverageshareofpatentsco-inventedwithforeignresidentsinOECDcountriesisalmost10%(thecross-countryaverageshareishigherthantheOECD-widesharesinceinthelattercasethereissomeclearing:eachpatentco-inventedinseveralcountriesiscountedonlyoncewhereasintheformercaseitiscountedineachTheshareofcross-borderownership(attheOECD-widelevelSHIAisequaltoSHAIbyconstruction)hasbeenquitestableinthe1980s,uctuatingaround8.5%oftheOECDpatents.From1990onwards,cross-borderownershipofinventionshasgrownsteadilytoreachashareof12%ofOECDpatents.In D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266otherwords,morethanoneoutof10patentsappliedtotheEPOissubjecttocross-borderownership.Thatislessthantheshareofforeignafliatesofmultina-tionalrmsinR&Dexpenditure,whichaveragesabout15%OECD-wide:butitstillhigh,andmore-overitisincreasingrapidly.3.CountrypatternsofinternationalisationofBeyondtheseglobaltrends,countriesdisplayvariouspatternsofinternationalisation,asreportedinTable2.FourobservationscanbedrawnfromthisFirst,thereisastrikingheterogeneityacrosscoun-tries.VeryfewinventionsmadeinJapanarecontrolledbyforeignrms,inventedinco-operationwithforeignresearchers.Similarly,onlyfewinventionscontrolledbyJapaneseresidentsareinventedabroad.WiththeexceptionofJapan,itisclearthatasignicantpropor-tionofpatentsaresubjecttocross-borderownershipandinternationalcollaboration.Amongstthelargestcountries,UKischaracterisedbyarelativelyhighdegreeofinternationalisationofitstechnology,withratiosrangingfrom11to20%.Smallercountries(suchasBelgium,AustriaandIreland)and/orlessdevelopedcountries(Turkey,Mexico,Poland)arehighlyinternationalised.Itisalreadywellknownthatsmalleconomiesaremoreinternationalisedthanlargeoneintermsoftradeaswellasforeigndirectinvest-ment.Itturnsoutthatthisappliestotechnologyaswell.The“technologicalsize”ofacountry(measuredforinstancebythenumberofpatentswithadomesticinventororapplicant)seemstohaveacloserelation-shipwithitsdegreeofinternationalisation.Concern-ingtheshareofresearchco-operation,thismaypartlybeexplainedbythefactthateachresearcherfromasmallcountryhasfewerlocalcolleaguesintheeldandmustthereforelookabroadforcollaboration.Second,inmostcountriestheshareofdomesticinventionsownedbyforeignrms(SHIA)issub-stantiallyhigherthantheshareofforeigninventionsintotaldomesticapplications(SHAI).Thereverseistrueforonlyeightcountries.Thisisduetoconcentra-tionofownershipofcross-borderpatentsinthehandsofafewcountries.Actually,threecountriesarethelargestownersofpatentscoveringforeigninventions:Table2Threeindicatorsofinternationalisationoftechnology,bycountry,prioritydate1993–1995 EPO(%)USPTO SHIAcSHAIdSHIIeSHIAcSHAIdSHIIe Australia11.75.313.823.812.616.5Austria18.47.814.933.613.727.1Belgium31.711.821.351.717.630.4Canada24.015.319.920.818.115.5CzechRepublic31.25.726.861.5046.2Denmark10.89.316.128.321.527.0Finland5.27.68.08.716.311.3France8.35.37.919.59.313.0Germany6.34.56.615.210.211.6Greece13.14.222.652.066.752.0Hungary11.72.317.850.810.040.0Iceland92.316.737.142.90.014.3Ireland26.537.221.153.147.536.7Italy9.42.35.817.67.411.8Japan2.61.52.43.03.02.3Korea4.53.48.43.13.32.6Luxembourg42.780.331.7Mexico48.19.240.441.220.035.3TheNetherlands12.230.912.661.944.725.1NewZealand18.18.821.124.810.014.0Norway12.614.215.321.229.625.7Poland35.57.834.764.1064.1Portugal18.510.325.058.366.741.7Spain15.64.414.732.214.224.9Sweden8.49.49.918.419.514.8Switzerland12.426.918.034.741.730.1Turkey61.527.851.2100–88.9UK20.010.711.938.818.922.6US5.08.76.82.68.63.7EuropeanUnion6.53.75.118.93.59.4OECD11.711.74.77.28.46.2 Source:OECDSecretariat.TheguresfortheUSPTOcorrespondtothedateofgrant1998(approximately1993–1996forthedateofpriority).ShareofpatentapplicationsatEPO(USPTO)ownedbyforeignresidentsintotalpatentsinventeddomestically(%).ShareofpatentapplicationsatEPO(USPTO)inventedabroadintotalpatentsownedbycountryresidents(%).ShareofpatentapplicationsatEPO(USPTO)withatleastoneforeignco-inventorintotalpatentsinventeddomestically(%).theUS(although,becauseofitssize,theshareoffor-eigninventionsisjustabovetheOECDaverage:butlevelishigh),theNetherlands,andSwitzerland.Thesearealsocountrieswithwell-known,strongmultinationalrms.AretheseobservationsspecictoEPOpatents,ordotheyapplytoothertypesofpatents?Atest D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266onpatentsgrantedbytheUSPTO(seethelastthreecolumnsofTable2)showsthatoverallthecountrypatternsaresimilar.Cross-countrycorrelationbetweenUSPTOandEPOindicatorsis0.65(SHIA),0.95(SHAI)and0.74(SHII).However,thereseemstobeasystematicbiasbetweenthetwosources.PatentsledbyresidentsfromEuropeancountriesaremuchmoreinternationalisedattheUSPTOthanattheEPO.Forinstance,almost20%ofpatentsinventedbyresidentsoftheEuropeanUnionandgrantedbytheUSPTOarecontrolledbynon-EUresidents,againstlessthan7%withEPOapplications.ThereverseholdsforresidentsfromtheUS,Canada,KoreaandMexico.Theremaybetworeasonsforthis:thehigher“proximity”oftheEPO(USPTO)toEuropean(non-European)countriesleadsthemtopatenttheretheirpurelydomesticin-ventionstoalargerextentthaninotherpatentofces(“homeadvantage”).Itmayalsohappenthatpatentsresultingfromco-operationbetweenUSresidentsandnon-US/non-EuropeanresidentsaremoreoftenpatentedattheUSPTOthanattheEPO.Cross-industriesdifferencesarereportedinTable3.Thevarianceissignicantlyloweracrossindustriesthanacrosscountries,suggestingthattheinternationalisationoftechnologyismorerelatedtocountrypeculiaritiesthantotechnologicalones.Foursectorsareneverthelesshighlyinternationalised:chemicals,oilrening,drugs,andfoodandbeverages.Shipbuildingandaerospace,twosectorsgenerallyconsideredtobesubjecttospecialgovernmentatten-tionsaretheleastinternationalised.ThesedifferencesacrossindustriesconrmthendingsofDunningandWymbs(1999).Theirsurveyoftheworldlargestmultinationalsshowsthatpharmaceuticalrmsobtainmoreoftheircompetitiveadvantagefromforeignsourcesthanothersectors,whereasrmsfromtheaerospacesectorrelythemostondomesticsources.Towhichextentdoescross-borderownershipofpatentsreectthecirculationofknowledgeor,instead,“purelynancial”relationship?Onewayofaddressingthisquestionistoanalysethepopulationofinternationallyownedpatentsfromthepointofviewofco-inventions.Theunderlyingassumptionsarethatco-inventionsreectcirculationofknowledge, WeusedtheOST/MERITconcordancetablebetweenIPC,theinternationalpatentsclassication,andISIC/rev2,theinternationalstandardindustryclassication.Table3Threeindicatorsofinternationalisationoftechnology,bysector, SHIA/SHAISHII Electricmachine(exceptelectronics)7.25.0Electronics7.35.1Chemistry(exceptpharmacy)10.712.6Pharmacy8.814.7Oilrening11.610.6Shipbuilding3.01.4Motorvehicles6.44.6Aerospace2.01.2Othertransport5.14.4Ferrousbasicmetals6.48.5Non-ferrousbasicmetals3.76.0Metalproducts(exceptmachines)6.35.1Instruments6.56.4Computersandofcemachines5.63.5Othermachinery6.95.9Food,beverages,tobacco11.915.1Textiles,clothes,etc.7.25.9Rubberandplasticproducts7.55.9Stone,clayandglassproducts8.08.0Paper,printingandpublishing7.57.7Woodandfurniture4.22.8Otherindustrialproducts6.85.5 DenitionsprovidedinTable2.Source:OECDSecretariat.whereasco-applicationsaremoreonthenancialside.Thebreakdownofinternationallyownedpatentsshowsthatmorethan40%involvedinternationalresearchco-operationin1995,against30%in1985.Thatis,internationalownershipincreasinglyinvolvesinternationalcirculationofknowledge.Howsimilartoeachotherarethethreepatent-basedindicatorsofinternationalisationoftechnology?Andaretheysimilartootherindicatorsofinternationa-lisation?Inotherwords,dothedifferentdimensionsofinternationalisationgotogether,sothatcertaincountriesaremoreopenedthanothersinallrespectsorconversely,arethedifferentdimensionsofinterna-tionalisationsubstitutetoeachother,allowingdiffer-entpatternsofinternationalisationtoexist?Table4reportscross-countrycorrelationbetweentheseindi-cators.Allpairsofindicatorsrelatedtotheinterna-tionalisationofinventionsaresignicantlycorrelatedwitheachother.Thehighestcorrelation(0.83)isforthepairSHII/SHIA:thehigheracountry’sshareofdomesticinventionscontrolledbyforeigncompanies,themoreitcollaborateswithforeigncountries. D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266Table4Cross-countrycorrelationsbetweenindicatorsofinternationalisation SHIASHAISHII PatentdataSHIA200.47SHAI200.47SHII200.83ActivitiesofforeignafliatesShareofforeignoutput(SHFP)120.86ShareofforeignR&D(SHFR)110.900.560.89ForeigndirectinvestmentsShareofinwardFDI(SHINF)200.250.130.32ShareofoutwardFDI(SHOUF)200.010.410.07TechnologybalanceofpaymentsShareofTBPReceipts(SHINTBP)170.420.330.41ShareofTBPpayments(SHOTBP)170.55InternationaltradeShareofimportsinGDP(MGDP)200.76ShareofexportsinGDP(XGDP)200.69Numberofdomesticinventions20 Correlationacross20OECDcountrieswithatleast200patentsinventedovertheperiod1993–1995.SHIA:shareofdomesticinventionswithforeignapplicants;SHAI:shareofdomesticapplicationswithforeigninventors;SHII:shareofdomesticinventionswithatleastoneforeigninventor;SHFP:shareofdomesticoutputproducedbyforeignrms;SHFR:shareofdomesticR&Dinforeignrms;SHINF:shareofinwardFDIingrossxedcapitalformation;SHOUF:shareofoutwardFDIingrossxedcapitalformation;SHINTBP:shareoftechnologyreceiptsinGDP;SHOTBP:shareoftechnologypaymentsinGDP;MGDP:shareofimportsinGDP;XGDP:shareofexportsinGDP;:numberofavailableobservationsforeachvariable.Thecoefcientsthataresignicantata5%probabilitythreshold.Thecoefcientsthataresignicantata1%probabilitythreshold.Thethreeindicators(especiallySHIA)arehighlycorrelatedwiththeshareofproductionbyafliatesofforeignownedrmsintotaldomesticproduction.OnlySHIAandSHIIhaveasignicantcorrelationwiththeshareofforeignafliatesintotaldomesticR&Dexpenditure.Thatis,internationalcollaborationandforeignownershipofdomesticinventionsarecloselyrelatedtotheinnovativeactivitiesofforeignafliates.TheshareoftechnologypaymentsinGDPisalsosignicantlycorrelatedwiththethreepatentsindica-tors,andespeciallywiththeshareofforeigninventionsintotalapplicationsbyresidents(SHAI).Thehighertheshareofforeigninventionsinapplicationsownedbyresidentsfromagivencountry,thelargerthetech-nologypaymentsofthiscountry(whichprobablycor-respondtovarioustypesoffundingofinventors).Allpatent-relatedindicatorsarepositivelycorrelatedwithopennesstoexternaltrade(importsandexportsrela-tivetoGDP).TheonlyindicatorsthatdonotseemtoprovidesimilarpicturesofinternationalisationareFDI(bothinwardandoutward)andtechnologyreceipts.Finally,thelastrowofTable5showsthatSHIIandSHIAarenegativelycorrelatedwiththepatentsinventedbyresidents(thenumberofpatentsreectstheleveloftechnologicalactivity).Thesmallerthe“technologicalbasis”ofacountry,thehighertheshareofthisbasiswhichiscontrolledbyforeignapplicants,andthemorethiscountry’sresidentsarelikelytocollaboratewithforeignresearchers.This“sizeeffect”probablyincreasesthecorrelationofpatentsindicatorswithopennesstotrade:smallercountriesaregenerallymoreinternationalisedonallaspectsoftheireconomy.Asmallsizeinvolvesarelativelynarrowrangeofactivities(higherspecial-isation),thereforeadeeperinsertionininternationaldivisionoflabour.Section4partlytacklesthisissuewithinaneconometricanalysis. D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266 D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–12664.ThedeterminantsoftechnologicalinternationalisationofcountriesWhatarethecommondeterminantsoftheopennessofcountriestoforeigntechnology?Thefewstudiesonthedeterminantsofinternationalisationoftechnol-ogyhaveessentiallyfocusedontheshareofforeignR&D,atthermlevel(seethesurveybyGranstrandetal.(1993)).Themajordeterminantsaretheageoftherm,itssize,itsstageofcorporatedevelopmentanditsinternationalpatternofmanufacturing.Weshallfocusontwomainfactorsforeachcountry:therelativeleveloftechnologicalendowment,proxiedbytheresearchintensity(GERD/GDPratio),andthesizeofthecountry(GDP).Bothvariablesaretakeninlogarithmicform,inordertocapturenon-lineareffects.Theempiricalmodelhasthefollowingform:whereSHXX(XXIA,AI,II)istheindicatorofisanintercept,IRDisresearchistheerrorterm,aretheparam-eterstobeestimated,isthecountryindex.Table5reportstheeconometricestimatesforeachofthethreeindicators(withEPOpatents).Therstpartofthetableshowstheresultswith27OECDcountriesandthesecondpartwiththe21countrieswithatleast150patentsinventedovertheperiod1993–1995.ThethirdpartofTable5presentssimilarestimatesfortheperiod1985–1987.Inallcases,thewithdrawalofsixcountriesfromthesampleandtheconsiderationofanearlierperioddonotaffectsubstantiallythesignandsignicanceoftheparameters,whichwitnessestheTable6Cross-countryanalysisoftheextentofinternationalisationoftechnology(USPTO) DependentvariablesUSPTO—allcountries(NOBSUSPTO—largerinnovators(NOBS SHIASHAISHIISHIASHAISHII Constant1.00(2.02)0.65(1.28)0.80(2.39)0.81(1.69)1.29(3.10)0.650.06(2.93)0.02(0.46)-test22.811.9120.80.460.070.620.210.200.44 OLSestimates,withHeteroskedastic-consistentstandarderrors;-statisticvaluesbetweenparentheses.Theparametersthataresignicantlydifferentfromzeroata10%probabilitythreshold.Thereare26observationscorrespondingtotheOECDcountries;LuxembourgismissingandTurkeyandIcelandweredroppedfromthesample,duetoatoosmallnumberofpatents(8and16patentsinventedduringthe1993–1995period,respectively).robustnessoftheestimates.SimilarregressionshavebeenrunonUSPTOpatents,withalmostidenticalresults(Table6).Theseresultsleadtothefollowingobservations(focusingmainlyontherstpartofTable5).Thethreeestimationsareofunevenstatisticalqual-ity:thevarioustypesofinternationalisationdonotseemtorespondtothesamedeterminants.However,inallcasestheeffectofGDPisnegative(althoughnotalwayssignicant),showingthatsmallercountriesaremoreinternationalisedthanlargeronesceterisparibus,asexpected.Similarly,R&DintensityhasanegativeimpactonSHIIandSHIA,apositiveoneonSHAI.Theextentofcollaborationwithforeignresearchers(SHII)isverywellexplained(adjustedbythemodel.Thehigher(lower)theR&Dinten-sityofacountry,theless(more)itsresearchersenterintocollaborationwithforeigncolleagues.LowR&Dintensityandsmallcountriesrelymoreonexternalco-operationduetotheirweakerowncapabilities,thusbenetingfromknowledgeowsfromabroad.SHIA(whichisalsoquitewellexplainedbythemodel,withanadjustedequalto0.36)tendstodecreasewithR&Dintensity:thehigher(lower)therelativeR&Dspendingofacountry,thelower(higher)istheshareofitsresidents’inventionthatiscontrolledbymultinationalrms.Inotherwords,nationalcontroloverdomesticinventionsincreaseswithdomesticinventiveeffort.Thereisonlyasmall(statisticallyinsignicant)negativeeffectofGDP.SHAIislesswellexplainedbythemodel.Thesizeofacountryisanegativeandsignicantdeterminantoftheextenttowhichitcontrolsforeigninventions,whereasitstechnologicalintensityisapositivebut D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266notsignicantdeterminant.Thesmalleracountryis,thehigheristheshareofinventionsitownsthatareinventedabroad.ThisasymmetrybetweenSHIAandSHAIwasevenmorepronouncedinthemideighties,whenahighertechnologicallevelwassyn-onymouswithalargeramountofforeigninventionscontrolleddomestically.Inanutshell,themoreacountryisintensiveinresearch,thelessitsownin-ventionsarecontrolledbyforeignrmsandthelessitentersintointernationalresearchco-operation.Thelargerthecountry,thelowerisitsshareofpatentapplicationsthathavebeeninventedabroadandthelowerisitspropensitytoenterintointernationalTheasymmetrybetweenSHIAandSHAItendstoconrmtheideathatthermsbasedinleadingedgecountriesexploittheirtechnologicaladvantagemorethroughforeignacquisition(cross-borderownership),i.e.throughSHAI.Thenegativesignoftherela-tionshipbetweenSHIAandR&Dintensitydoesnotsupporttheargumentthatleadingedgecountriesarebeing“techno-sourced”,atleastnotthroughforeignownershipoftheirowninventionfacilities.Whensizeandtechnologicalintensityarecon-trolledfor,differencesremainacrosscountriesintheirdegreeofinternationalisation.Somearemoreinterna-tionalisedthanpredictedbythemodel,othersareless.Othercountry-specicfactorsseemtobeatwork.Table7reportsforSHIIandSHIAthecountrieswhoseactualdegreeofinternationalisationiswellabove(orunder)thepredictedvalues.ResultsarefragileforcountrieswithfewpatentsandarenotAmonglargerpatentingcountries,theUK,Belgium,CanadaandtheUShaveahigherthanpre-dictedshareoftheirdomesticinventionsownedbyforeigners,thereversebeingtrueforFinland,Korea,ItalyandJapan.Forinternationalco-operationinresearch,theUS,SwitzerlandandCanadarankmuchhigherthantheirsizeandR&Dintensitywouldhaveledtopredict,whereasItaly,FinlandandJapanarebelowtheirpredictedvalue.TheestimatedparametershavebeenappliedtotheaggregateR&DintensityandGDPoftheEuropeanUnion.ThepredictedvalueofSHIIiswellaboveitsactualvalue,whichmeans Predictedvaluesresultfromsimulatingthemodelwitheachcountry’sactualvariables(GDP,R&Dintensity).Table7Countrieswithdegreeofinternationalisationaboveorunderthepredictedlevel SHIASHII AbovepredictionUK[0.96]US[1.57]Belgium[0.91]Switzerland[0.50]Luxembourg[0.83]Canada[0.49]Canada[0.73]Mexico[0.53]US[0.43]UnderpredictionFinland[0.66]Italy[Korea[0.57]Finland[Greece[0.57]Japan[Hungary[Italy[0.43](EU[Japan[ Theratiooftheerrorofpredictiontothepredictedvalueofthedependentvariableisshownbetweenbrackets,computedfromtheresultspresentedinTable5,with27countries.Thecountrieslistedarethoseforwhichtheratioisthehighestorthelowest(0.4wasselectedasathresholdvalue).thatEuropeasawhole(netofintra-zonecollabo-ration)islessopenedthatitssizeandtechnologicalintensitywouldleadonetopredict.Finally,Japan,althoughstillshownaslittleinternationalisedinthisanalysis,doesnotappearasthemostinsulatedOECDcountryasitiswhensizeandR&Dintensityarenotcontrolledfor:beingabigandR&Dintensivecoun-trycontributestoJapan’slowdegreeofopenness,althoughitdoesnotexplainitentirely.5.Whomateswithwhom?Thissectionaddressesthegeographicaldistribu-tionoftheinternationalisationoftechnology:withwhichpartnersdoeachcountrytendtoco-operatemore,andless?Forinstance,regardingthepatentsinventedbyresidentsandownedbyforeignappli-cants(SHIA),theshareofUSresidentsishigherinnon-Europeancountries,exceptIrelandandtheUK.TheshareofJapanishigherinKorea,AustraliaandtheUS(butthisispartlyastatisticalartefactsincetheUSisnotpartnertotheUS,whichtendstoin-atetheshareofallcountriesintheUSascomparedtotheirshareinothercountries).InEuropeancoun- D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266tries(excepttheUK,IrelandandLuxembourg),thehighestshareofforeignownedpatentsgoestootherEuropeancountries.SimilarpatternsappearforSHAIandSHII.Inordertocontrolforthesizeeffect(i.e.theUSisthelargestpartnertoallcountries,simplybecauseitisthebiggestpatentingcountryofall)anindexof“revealedgeographicaldistribution”(RGD)hasbeencomputedforeachofthethreeindicatorsofinter-nationalisationoftechnology.Thisindexissimilar,inspirit,to“revealedcomparativeadvantages”thatinternationaleconomistsarefamiliarwith.Basically,itiscountry’sshareincountry’sforeignrelation-shipsrelativetoitsshareinOECD.Forinstance,itistheshareofGermanresidentsinFrenchinventors’patentsownedbyforeigners,dividedbyGerman’sresidentsshareintotalOECDforeignownedpatents.Algebraically,wehaveTheRGDofforeignownershipofdomesticinven-tionsisequaltotheshareofcountryincountry’spatentsownedbyforeignresidentsdividedbytheshareofcountryintheworldwidepatentssubjecttocross-borderownership:RGD IAijPFIA][PFIAPFIA].TheRGDofdomesticownershipofforeigninven-tionsisequaltotheshareofcountryincountry’spatentsownedbyforeignresidentdividedbytheshareofcountryintheworldwidepatentssub-jecttocross-borderownership:RGD AIijPFAI][PFAIPFAI].Byconstruction,RGD AIisthetransposedmatrixRGD TheRGDofinternationalco-inventionsisequaltotheshareofcountryinacountry’spatentsownedbyforeigndividedbytheshareofcoun-intheworldwidepatentssubjecttocross-borderownership:RGD IIij IIIIij ][ IIIIij 
IIIIII ].Theseindicatorshavebeencalculatedforeachpairofcountries,resultinginthree2929matrices.Weidentiedandtestedvefactorsthatmayexplaintherevealedgeographicaldistributionofthedifferentindicatorsofinternationalisationoftechnol-ogy.Therstoneisthetechnologicalproximity(TP)betweenpairsofcountries.Tomeasuretheproximityofcountries,weusetheuncenteredcorrelationofthetwocountries’distributionvectorsofpatentsacross30technologicalclassesin1992–1995(),asfollows:ws:  ji ]12Thisindicatorisequaltooneforthepairsofcoun-trieswhosetechnologicalspecialisationsareidentical,itisequaltozeroforpairsofcountrieswhosevectorsareorthogonal,anditisboundedbetween0and1forallotherpairsofcountries.Thesecondexplanatoryfactoristhegeographicaldistance(DGD),proxiedbyadummyvariabletakingthevalueoneifcountrieshaveacommonborder,and0otherwise.ThencomethreeparticulardummyvariablesreectingthefactthatcountriesaremembersoftheEuropeanUnionornot(DEU,included12countriesforthetwosub-periods),whethertheyarebothNordiccountries(DNORD)andwhethertheyshareacommonlanguage(DLANG),beitEnglish,SpanishorGerman.ThesedummiesaimattestingwhetherthecommonmembershiptotheEuropeanUnionorcommonlanguages(andhenceculturalandhistoricalsimilarities)increasethepropensityofrmsfromtwodifferentcountriestocollaboratewitheachother.Theestimatedmodelhasthefollowingform whereXXrepresentseitherIA(cross-borderowner-ship)orII(internationalco-inventions).EstimatesoftheveparametersofinterestofEq.(2)arereportedinTable8fortwosub-periods(1993–1995and1985–1987).Theyprovidefairlygoodresultsthatleadtothefollowingobservations.First,technologicalproximitymatters.Theclosertwocountriesareintheirtechnologicalspecialisation,themoretheyco-operateinresearchandownpatentsinventedbyresearchersoftheothercountry.Second, ThisindicatorissimilartoJaffe’s(1986)indicatoroftechno-logicalproximitybetweenUSrms.The2727countriesmatrixoftherevealedgeographicaldistri-butionindexesofinternationalco-inventions(II)issymmetricandthereforeonlyhalfoftheobservationshavebeenused.Regardingcross-borderownership,thematrixforIA(theshareofdomesticinventionscontrolledbynon-residents)isthetransposedofthematrixforAI(theshareofpatentcontrolledbyresidentsthatareinventedabroad)andthereforeallobservationsareused. D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266Table8Thedeterminantsoftherevealedgeographicaldistributionofinternationalisation 1993/19951985/1987 (RGD—IA)(RGD—II)(RGD—IA)(RGD—II) Technologicalproximity(TP)12.07(4.62)11.58(4.33)7.33(3.34)6.86(3.14)3.74(5.06)7.08Commonborders(DGD)3.40(3.28)2.11(1.87)4.05(4.52)2.83(2.99)2.19(6.65)2.85EuropeanUnion(DEU)1.47(1.68)0.06(0.08)0.10(0.39)0.09(Nordiccountries(DNOR)5.78(2.31)7.65(3.58)3.72(5.40)4.60Commonlanguage(DLANG)2.52(1.72)2.12(2.48)1.77(4.30)2.13(4)42.30Observationnumber676676351351600325Numberleftcensored337337136136328157 Left-censoredTobitestimates(lowerboundatzero);allregressionsincludeanintercept.Include27OECDcountries,TurkeyandIcelandnotincludedbecauselessthan20patentsinventedduringtheperiod1993–95.Indicatestheparametersthataresignicantata10%probabilitythreshold.thegeographicaldistancehasalsoasignicanteffect,forthetwotypesofinternationalisationoftechnol-ogy.Countrieswithacommonborderenterintocross-borderownershipandco-operatemore.Intheageofglobalisation,geographystillmatters(alongwithhistory,whichismoresharedbetweenclosercountries).ThecommonhistoryandrelativelyweakculturaldifferencesthatcharacterisetheNordiccoun-triesisprobablythemainexplanationofthepositiveandsignicantparameterassociatedwithDNORD.Third,countriesthatshareacommonlanguageco-operatemorewitheachotherandhaveahigherpropensitytoenterintocross-borderownership.Fi-nally,pairsofcountriesthatarebothmembersoftheEuropeanUnionhaveslightlymorecross-borderown-ershipofpatents,butnotmoreresearchco-operation.Thispositiverelationshipbetweencross-borderown-ershipandEuropeanUnionmembershipappearedbetweenthemid-1980sandthemid-1990s.Onein-terpretationwouldbethatthepoliciesfosteringtheEuropeanintegrationhavestimulatedaprocessofin-dustrialandnancialconcentration(throughM&A’s)betweenEuropeanrms.Suchcross-borderconsol-idationtranslatesintomorecross-borderownershipofpatents,asthenewrmownsresearchfacilitiesindifferentcountries.Europeancountriesco-operatemorewitheachotherthanwithnon-Europeanones,butnotmorethantheirlanguagesimilarityandtheirgeographicalandtechnologicalproximitywouldHence,industrialandnancialconcentra-tionhasnotyetresultedincloserresearchlinks:thereisnotyetrealintegrationofEuropeancountriesintheeldofbusinessR&D.6.ConcludingremarksThispaperpresentedandanalysedthreeindicatorsofinternationalisationoftechnologyderivedfrominformationavailableinpatentdata.Inaccordancewiththeexistingliterature,theseindicatorswitnessanincreasingtrendtowardstheinternationalisationoftechnologyofOECDcountries.However,therearelargedifferencesintheextentofinternationalisationacrosscountries.Internationalisationofacountry’stechnologicalactivitiesdecreaseswiththelevelofitsGDPandwithitsR&Dintensity.Researchersinlargercountriesndmoreeasilycolleaguesforpart-neringintheirowncountry,andcountrieswithhighertechnologicalleveldonotneedasmuchasothersco-operationwithforeignresearcherssincetheirownknowledgebaseislarge.Thispartlyexplainstherel-ativeinsulationofJapanforinstance.Ontheother Otherestimates,excludingthesmallerpatentingcountries,ledtosimilarresults.FortheEuropeanUnionmembercountries,theproximitydummyvariableaccountsfor25%oftheEuropeanUniondummy,i.e.about25%ofthepairsofEUMembercountriesshareacommonborder. D.Guellec,B.vanPottelsberghedelaPotterie/ResearchPolicy30(2001)1253–1266end,theUKandtheUSseemtobemoreopenthantheirsizeandresearchintensitywouldleadtopre-dict.Languagemaybepartoftheexplanation,assuggestedinresultsconcerninggeographicalpatternsofinternationalco-operation.Anotherinsightfromthestudyisthatthemajoraimofmultinationalrmswhenestablishingresearchfacilitiesabroadistoadapttheirproductstolocalconditionsratherthanto“tap”foreigntechnology:actually,themajorstreamofR&DinvestmentabroadcomefromhighlyR&Dintensivecountriestolowintensiveonesratherthantheopposite(asitwouldbethecaseiftechnologysourcingwasthedominantobjective).Whoco-operateswithwhomislargelyexplainedbygeographicalproximityandtechnologicalproximity(similarspecialisation)ofthepartneringcountries.Inaddition,sharingacommonlanguagefostersbilaterallinksintechnology.PairsofcountriesthatarebothmembersoftheEuropeanUnionhaveslightlymorecross-borderownershipthantheaverage,butnotmoreresearchco-operationthanitisimpliedbytheirgeographicalandtechnologicalproximity.AcknowledgementsWethanktwoanonymousrefereesforveryhelpfulcommentsandsuggestionsonapreviousversionofthisstudy.AllopinionsexpressedinthisarticlearethoseoftheauthorsanddonotreectnecessarilytheviewsoftheOECDorUniversitéLibredeBruxelles.ReferencesArchibugi,D.,Iammarino,S.,1999.ThePolicyImplicationsoftheGlobalisationofInnovation.Chapter12,In:Archibugietal.(Eds.),Innovationpolicyinaglobaleconomy,CambridgeUniversityPress,Cambridge,pp.242–271.Archibugi,D.,Pianta,M.,1992.TheTechnologicalSpecialisationofAdvancedCountries.AreporttotheEEConInternationalScienceandTechnologyActivities.KluwerAcademicPubli-shers,Boston.Cantwell,J.,1989.TechnologicalInnovationandMultinationalCorporations.BlackwellPublishers,NewYork.Cantwell,J.,Janne,O.,1999.Technologicalglobalisationandinnovativecentres:theroleofcorporatetechnologicalleadershipandlocationalhierarchy.ResearchPolicy28,119–144.Dunning,J.H.,1994.Multinationalenterprisesandtheglobaliza-tionofinnovatorycapacity.ResearchPolicy23,67–88.Dunning,J.H.,Wymbs,C.,1999.TheGeographicalSourcingofTechnology-BasedAssetsbyMultinationalEnterprises.Chapter10,In:Archibugietal.(Eds.),Innovationpolicyinaglobaleconomy,CambridgeUniversityPress,Cambridge,pp.185–224.Granstrand,O.,Häkanson,L.,Sjölander,S.,1993.Internationa-lizationofR&D—asurveyofsomerecentresearch.ResearchPolicy22,413–430.Griliches,Z.,1990.Patentstatisticsaseconomicindicators:asurvey.JournalofEconomicLiterature28,1661–1707.Jaffe,A.B.,1986.TechnologicalopportunityandspilloversofR&D:evidencefromrm’spatent,protsandmarketvalue.TheAmericanEconomicReview76(5),984–1001.Lichtenberg,F.,vanPottelsberghe,B.,2001.Doesforeigndirectinvestmenttransfertechnologyacrossborders?TheReviewofEconomicsandStatistics.Niosi,J.,1999.TheinternationalisationofindustrialR&D:fromtechnologytransfertothelearningorganisation.ResearchPolicy28,107–117.Patel,P.,1995.Localisedproductionoftechnologyforglobalmarkets.CambridgeJournalofEconomics19(1),141–154.Patel,P.,Pavitt,K.,1991.Largermsintheproductionoftheworld’stechnology:animportantcaseofnon-globalization.JournalofInternationalBusinessStudies1–20.Patel,P.,Pavitt,K.,2000.Nationalsystemsofinnovationunderstrains:theinternationalizationofcorporateR&D.Miméo,SPRU,UniversityofSussex.Patel,P.,Vega,M.,1999.Patternsofinternationalisationofcorporatetechnology:locationvs.homecountryadvantages.ResearchPolicy28,145–155.Zander,I.,1995.Technologicaldiversicationinthemultinationalcorporation—historicalevolutionandfutureprospect.In:Schiatarella,R.(Ed.),NewchallengesforEuropeandInternationalBusiness.Condustria,Rome.