http://www.diva-portal.org
This is the published version of a paper published in .
Citation for the original published paper (version of record):
Silveira, F F., Sbragia, R., Lopez-Vega, H., Tell, F. (2017)
Determinants of reverse knowledge transfer for emerging market multinationals: The
role of complexity, autonomy and embeddedness.
Revista de Administração, 52(2): 176-188
https://doi.org/10.1016/j.rausp.2016.12.007
Access to the published version may require subscription.
N.B. When citing this work, cite the original published paper.
Open Access
Permanent link to this version:
Availableonlineatwww.sciencedirect.com
Revista
de
Administração
http://rausp.usp.br/ RevistadeAdministração52(2017)176–188
Technology
Management
Determinants
of
reverse
knowledge
transfer
for
emerging
market
multinationals:
the
role
of
complexity,
autonomy
and
embeddedness
Determinantes
da
transferência
reversa
de
conhecimento
em
multinacionais
de
mercados
emergentes:
o
papel
da
complexidade,
da
autonomia
e
da
integra¸cão
Factores
determinantes
de
la
transferencia
inversa
de
conocimiento
en
multinacionales
de
mercados
emergentes:
el
papel
de
la
complejidad,
la
autonomía
y
la
integración
Franciane
Freitas
Silveira
a,∗,
Roberto
Sbragia
b,
Henry
Lopez-Vega
c,
Fredrik
Tell
daUniversidadeFederaldoABC,SãoBernardodoCampo,SP,Brazil bUniversidadedeSãoPaulo,SãoPaulo,SP,Brazil
cJönköpingUniversity,JönköpingInternationalBusinessSchool,Jönköping,Sweden dUppsalaUniversity,Uppsala,Sweden
Received29June2016;accepted19August2016 Availableonline2January2017 ScientificEditor:MariaSylviaMacchioneSaes
Abstract
Subsidiariesconductinnovationactivitiesinforeignmarketseithertocapturevaluableknowledgethatisnecessarytoadapttheirproductstolocal
marketsortocreatevaluableknowledgeforheadquarters.Foremergingmarketmultinationals,moststudieshaveoverlookedthedeterminantsof
successfulreverseknowledgetransferfromsubsidiarieslocatedinemerginganddevelopedmarkets.Thispaperanalyzedtheresponsesofasurvey
administeredto78Brazilianmultinationalsthatownsubsidiariesindevelopedandemergingmarkets.Wefoundthatknowledgecomplexity
devel-opedatthesubsidiary,itsautonomyandembeddednessintheforeignmarketdeterminethesuccessfulreverseknowledgetransfertoheadquarters
ofemergingmarketmultinationals.Thispapercontributestopreviousstudiesofreverseknowledgetransferbyunderlyingthemaindriversfor
emergingmarketmultinationals.
©2016DepartamentodeAdministrac¸˜ao,FaculdadedeEconomia,Administrac¸˜aoeContabilidadedaUniversidadedeS˜aoPaulo–FEA/USP.
PublishedbyElsevierEditoraLtda.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).
Keywords: Reverseknowledgetransfer;Emergingmultinationals;Brazilianmultinationals
Resumo
Subsidiáriasrealizamatividadesdeinovac¸ãoemmercadosestrangeiros,querparacapturaroconhecimentovaliosoqueénecessárioparaadaptar
seusprodutosaosmercadoslocaisouparacriarconhecimentodealtovalorparaasede.Nocontextodemultinacionaisdemercadosemergentes,
amaioriadosestudostêmnegligenciadoosdeterminantesdatransferênciadeconhecimetnoprovenientesdesubsidiárias(transferênciareversa).
Foramanalisadasasrespostasdeumapesquisarealizadacom78multinacionaisbrasileirasquepossuemsubsidiáriasemmercadosdesenvolvidos
∗Correspondingauthorat:AlamedadaUniversidade,s/n◦–CEP09606-045,SãoBernardodoCampo,SP,Brazil.
E-mail:franciane.silveira@ufabc.edu.br(F.F.Silveira).
PeerReviewundertheresponsibilityofDepartamentodeAdministrac¸ão,FaculdadedeEconomia,Administrac¸ãoeContabilidadedaUniversidadedeSãoPaulo –FEA/USP.
http://dx.doi.org/10.1016/j.rausp.2016.12.007
0080-2107/©2016DepartamentodeAdministrac¸˜ao,FaculdadedeEconomia,Administrac¸˜aoeContabilidadedaUniversidadedeS˜aoPaulo–FEA/USP.Published byElsevierEditoraLtda.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).
eemergentes.Verificou-sequeacomplexidadedoconhecimentodesenvolvidonasubsidiária,bemcomoasuaautonomiaeinserc¸ãonomercado
externodeterminamofluxodetransferênciareversa deconhecimentonaempresamultinacionalemergente. Estetrabalhoenriqueceestudos
anterioressobretransferênciareversadeconhecimentodestacandoosprincipaisdriversparaasmultinacionaisdosmercadosemergentes.
©2016DepartamentodeAdministrac¸˜ao,FaculdadedeEconomia,Administrac¸˜aoeContabilidadedaUniversidadedeS˜aoPaulo–FEA/USP.
PublicadoporElsevierEditoraLtda.Este ´eumartigoOpenAccesssobumalicenc¸aCCBY(http://creativecommons.org/licenses/by/4.0/).
Palavras-chave:Trasnferênciareversadeconhecimento;Multinacionaisemergentes;Multinacionaisbrasileiras
Resumen
Filialesrealizanactividadesdeinnovaciónenlosmercadosextranjeros,yaseaparacapturarelconocimientovaliosoqueesnecesarioparaadaptar
susproductosalosmercadoslocales,oconelfindecrearconocimientodealtovalorparasusede.Respectoalasmultinacionalesdemercados
emergentes,enlamayorpartedelosestudiosnosehadadoladebidaatenciónalosfactoresdeterminantesdelatransferenciadeconocimientoa
partirdefiliales(transferenciainversa).Enesteestudioseanalizanlasrespuestasdeunaencuestarealizadaa78multinacionalesbrasile˜nasque
poseenfilialesenmercadosdesarrolladosyemergentes.Losresultadosindicanquelacomplejidaddelconocimientodesarrolladoenlafilial,así
comosuautonomíaeinserciónenelmercadoexternodeterminanelflujodetransferenciainversadeconocimientoenlaempresamultinacional
emergente.Conestetrabajo,secolaboraaldesarrollodelosestudiosacerca delatransferenciainversa deconocimiento,conénfasisenlos
principalesdriversparalasmultinacionalesdemercadosemergentes.
©2016DepartamentodeAdministrac¸˜ao,FaculdadedeEconomia,Administrac¸˜aoeContabilidadedaUniversidadedeS˜aoPaulo–FEA/USP.
PublicadoporElsevierEditoraLtda.Esteesunart´ıculoOpenAccessbajolalicenciaCCBY(http://creativecommons.org/licenses/by/4.0/).
Palabrasclave: Transferenciainversadeconocimiento;Multinacionalesemergentes;Multinacionalesbrasile˜nas
Introduction
Themultinationalenterprise(MNE)isadifferentiated net-workinwhichitscontrolledsubsidiariesvarywidelyinterms of dutiesand responsibilities(Nohria &Ghoshal,1994). For example,whilesomesubsidiariesevolvethroughthe headquar-ters’mandatesothersfocusontheirowninitiatives(Mudambi, Piscitello,&Rabbiosi,2014).Sincethelate1990s,the recog-nitionthat headquarters operate as knowledgereceivers from their internationallydispersed subsidiarieshas gained signifi-cance ininternational business research (Ambos, 2015). The strategicimportanceoftheMNE’subsidiarieshascontinuedto grow,inthat itisanaccesspathway toknowledgeandtothe technologysituated atthesubsidiaries’localmarkets(Borini, Oliveira,Silveira,&Concer,2012;Criscuolo&Narula,2007; Frost& Zhou, 2005),which canactively contribute tovalue creationandsubsequentgainofcompetitiveadvantageforthe entireMNE(Bartlett&Ghoshal,1989;Cantwell&Mudambi, 2005;Yang,Mudambi,&Meyer,2008).
An underlying idea is that MNE make use of knowledge generatedbyforeignsubsidiaries.From thisperspective, sub-sidiaries upgrade their competence enhancing role such as marketexpansion, costreduction andsupplieradaptation and begintoplayamoreactiverolethrough knowledge develop-ment. For example, foreign subsidiaries might develop new products, new technologies, create new practices, new skills thatwilllatershapetheirowncompetencecreatingpathwaysas wellasaccumulatedifferentdegreesoftechnologicalcapability (Birkinshaw,1997;Borinietal.,2012;Borini,Costa,Bezerra,& Oliveira,2014;Cantwell&Mudambi,2005;Figueiredo&Brito, 2011;Frost,Birkinshaw,&Ensign,2002;Ghoshal&Bartlett, 1988; Govindarajan & Trimble, 2012; Mudambi, Mudambi, &Navarra,2007;Nohria &Ghoshal,1997).Moreover, com-petence creating subsidiaries could enhance their innovation
outcomes which enables them to compete domestically and internationally (Bell & Pavitt, 1995; Cantwell & Mudambi, 2005;Figueiredo&Brito,2011).Fromasubsidiaryperspective, reverseknowledgetransfer(RKT)givesvisibilitytosubsidiaries thatcouldleveragetheirstrategicpositioninthemultinational network(Borinietal.,2012;Holm&Pedersen,2000).
Thesefactorshavehighlightedthatreverseknowledge trans-ferisakeyvariableinthestudyofcross-borderknowledgeflows inMNEs(Ambos,2015).Asaresult,theknowledgetransferin thereversedirection,thatis,fromsubsidiariestoMNE headquar-ters,hasemergedasaprominentthemeininternationalbusiness studies(Ambos,2015;Ambos,Ambos,&Schlegelmilch,2006; Criscuolo,2005;Frost&Zhou,2005;Gupta&Govindarajan, 2000; Håkanson & Nobel, 2001; Rabiosi, 2008; Rabiosi & Santangelo, 2011;Rabiosi, 2011;Yangetal., 2008).Whilea numberofarticlesexploretheantecedents,successamountand successkey-factorsindifferentfunctionalconFigurationsatthe multinationalcorporation(Ambos,2015),additionalresearchis needed (Michailova&Mustaffa,2012).First,sincethe trans-ferof knowledge inMNEs hasgrownconsiderably inrecent years,becomingthereforemorepronetovariousdefinitionsand measurementsofthesameconstructsresultinginconclusions, oftencontradictoryandambiguous.Second,whilerecognizing theimportanceofinvestigatingtherelationshipofthesubsidiary withexternalcompanieslocatedinthehostcountries,the liter-ature oftenfocuses onlyon the research of knowledge flows withinthe MNE.Thisnarrowattention considerssubsidiaries areprimarilyrecipientsofknowledge(Michailova&Mustaffa, 2012).
Fromemergingmultinationalsenterprises(EMNEs)’s view-point, the ability to transfer knowledge in reverse direction seemstobe evenmorecrucial. Forexample,authorssaythat theEMNEsstrategic modelsareguidedbythepursuitof for-eigncapabilities,suchastechnologicalknowledge,whichcan
becombinedwiththeexistingresources (Bartlett&Ghoshal, 2000).Thatissobecause,insteadofinternationalizingtoutilize existingadvantages,emergingmarketmultinationalswill inter-nationalizeaimingatacquiringnewadvantagesandcapabilities (Guillén&García-Canal,2009;Mathews,2006;Ramamurti& Singh,2009)andshouldtodoitquickerthantraditional multi-nationalsdidintheirexpansionpaths(Mathews,2006).
InthecontextofBrazilianmultinationals,recentstudieshave soughttounderstandtheprimaryfactorsthatinfluencetheRKT. ThestudyofBorinietal.(2012)arguesthatthereverse knowl-edge transfer is a function of the strategic guidance of the: (1)subsidiaries’R&D laboratories,(2)integration (communi-cation) between headquarters andsubsidiaries, (3) subsidiary entrepreneurialorientation,(4)subsidiarylifetimeand(5)entry viagreenfieldinvestments.Moreover,thestudyofBezerraand Borini(2015)teststheimpactthatanationdevelopmentexerts onthereverseinnovationtransferinproductsandprocesses.In thisstudy,theysoughttounderstandwhichdeterminantsofRKT arepresentinBrazilianmultinationals.Inourstudy,weshowthat RKTisrelatedtothedegreeof:(1)knowledgecomplexitythat isbeingtransferred,(2) subsidiaryautonomyand(3) external embeddednes.Ashighlightedbynumerousauthors(Minbaeva, 2007;VanWijk,Jansen,&Lyles,2008),suchaspectsare iden-tified as key ones for understanding the RKT phenomenon. Althoughtherearemanykindsofknowledgetobetransferred throughconventionaland/orreversedirection,thisstudyfocuses specificallyonthetechnologicaltypeofknowledge(ofproduct andprocess).Ourfindingsarebasedinananalysisofthesurvey responsesadministeredto78Brazilianmultinationalsthatown subsidiariesindevelopedandemergingmarkets.
Asacontribution,itisexpectedthatourstudyadds knowl-edgetotheinternationalbusinessestheory,sincetheknowledge transferhasbeentreatedasakeyfactorofcompetitiveadvantage ofMNEs(Borinietal.,2012,2014;Govindarajan&Ramamurti, 2011)and,specifically,oftheemergingmultinationals compa-nies(Cuervo-Cazurra,2012;Immelt,Govindarajan,&Trimble, 2009;Ramamurti,2008).Sincemostresearchthatexplainsthis phenomenonisbasedonMNEswithsubsidiariesand headquar-tersindevelopedcountries,lessaffectedbyinstitutionaldistance (Rabiosi,2011;Yangetal.,2008),onecannotassumethatthe factorsthatinfluenceRKTfromforMNEsarethesameasthose for EMNEs (Borini,Costa,& Oliveira,2016).A practitioner contributionofthisstudyseekstoinformEMNEmanagersabout thestrategicdriversofRKT.
Thispaperisstructuredasfollows:thenextsection, “Con-ceptualframework”sectionpresentstheproposeddeterminants ofreversetechnologytransfer.“Methodology”sectionoutlines our research strategy and field procedures. “Findings” sec-tionpresentsourresultsanddiscussesthe implicationsof our findingsfor firmsinemergingmarkets.Finally,“Conclusion” sectionpresentsourmainconclusions,somelimitationsofthe study,andavenuesforfurtherresearch.
Conceptualframework
Theliteraturearguesthatknowledgetransfer,whether aris-ingfrominternalorexternalsources,hasanimportantimpact
onorganizationalperformanceandinnovationcapacity(Lyles &Salk,1996;Powell,Koput,&Smith-Doerr,1996;Tsai,2001; VanWijketal.,2008).Theunderlyingideaisthatthetransferred knowledge contributes to the development of organizational capabilities that are difficult to imitate and can laterlead to betterperformance(Szulanski,1996).Knowledgetransfer sti-mulates the combinationof the existing knowledge with the newly acquired oneandincreasesthe capabilityof aunit for carrying out new combinations (Jansen, Van Den Bisch, & Volberda,2005).
However, transferringknowledge betweenunitsof asame organization isnoteasierthanconductingexternalknowledge transfers(Kogut&Zander,1992).Thisisparticularlythecase whenitcomestoRKT.Thisprocesscanbeevenmore challeng-ing,sincewhile“[...]theconventionaltransferisaprocessof teaching,thereversetransferisaprocessofpersuading(Yang et al.,2008)”. In thiscase,the effortis muchhigher because its effectiveness dependson convincing headquarters. There-fore,thetransferdependsonheadquarter’sassessmentthatthe featuresandrelevanceofthesubsidiary’sknowledgeiscrucial sothat thereversetransferdoesoccur.TheRKTisdefinedas “an intra-organizationalexchange of information, technology or know-how from internationalsubsidiaries (located inhost countries)tocorporateheadquarters(homecountries).Theterm ‘reverse’ is used todistinguishthesetransfers from themore conventional form of‘forward’ transfers –from headquarters tosubsidiaries–,and‘lateral’transfers betweensubsidiaries” (Ambos,2015).
Somestudieshavehighlightedthatsubsidiariescreate com-petitive advantages for MNEs when valuable knowledge is transferred to the headquarters (e.g. Gupta & Govindarajan, 2000; Håkanson & Nobel, 2001; Rabiosi, 2011; Yang et al., 2008).ForMNEs, someof thedeterminants of RKTinclude the:(1)knowledgefeaturesbeingtransferred(Minbaeva,2007), (2) organizational characteristics(size, age,autonomy)(Frost et al., 2002; Gupta& Govindarajan,2000),(3) role of orga-nizational mechanisms (Håkanson & Nobel, 2001; Rabiosi, 2011),(4)thesubsidiaries’roles(Ambosetal.,2006;Rabiosi, 2011; Yang et al., 2008), (5) the host country economic development(Cantwell&Mudambi,2005;Frostetal.,2002; Gupta & Govindarajan, 2000), (6) the absorptive capacity (Ambosetal.,2006),(7)theknowledgerelevance(Yangetal., 2008),(8)theinternalembeddedness(subsidiary/headquarters) and (9) the external embeddedness (subsidiary/partners) (Figueiredo,2011;Meyer,Mudambi,&Narula,2011).
Following, it is explained how knowledge characteristics, suchascomplexity,autonomy,andexternalembeddedness influ-enceRKTfromsubsidiariestoheadqueartersofEMNEs. Subsidiary’sautonomy
Subsidiary’s autonomy could be defined as the extent to which a subsidiary is allowed to make decisions on its key strategic issues(Mudambi&Navarra,2004),withouta head-quartersdirectintervention(Roth&Morrison,1992).Ahigher level of autonomyisoften relatedtoknowledge creationand development at the MNE (Ghoshal & Nohria, 1989; Gupta
&Govindarajan,1991;Nohria&Ghoshal,1994),since inde-pendentsubsidiaries,(1)havestrategicmandates(Birkinshaw, Hood,&Jonsson,1998),(2) makequick decisions(Cantwell &Piscitello,1999),(3)recognizeandtakeadvantageoflocal opportunities(Frostetal.,2002),(4)developnewknowledgeas oflocalknowledgebases(Andersson,Forsgren,&Holm,2002), (5)generateintrinsicmotivationonindividuals(Mudambietal., 2007),(6)haveinitiativeandwillingnesstosharetheknowledge acquired(Gupta&Govindarajan,2000;Tsai,2002).Onthe con-trary,alowlevelofautonomy,maylimitthesubsidiaryfreedom, hindering its knowledge creation anddevelopmentcapability (Ghoshal & Bartlett, 1988). Foss and Pedersen (2002) also explainthathighlevelsofsubsidiaryautonomy–associatedwith lossofcontrol–couldbeovercomedbytheincreasein knowl-edge exchange amongst subsidiaries. While opposite results havealsobeenreported(Frostetal.,2002;Gammelgaard,Holm, &Pedersen,2004),mostresearches havesuggestedmainlya positiverelationshipbetweenknowledge decentralization and transfer(Cantwell&Mudambi,2005;Foss&Pedersen,2002; VanWijketal.,2008).
Recently,Rabiosi(2008)argued thatRKTiscoupled with subsidiary autonomy, i.e. mechanisms of personal communi-cation between subsidiary and headquarters. Yet, regarding subsidiariesof EMNEs, it is argued that, due totheir recent progressintheinternationalmarketand,therefore,duetotheir earlyage, theyare stronglydependent onheadquarters’ deci-sionmakingpower(Dunning,1993).Thismightnotbedifferent inBrazilianMNEs,thattendtobemorecentralizing,limiting therefore their subsidiaries’ knowledge creation possibilities (Chu&Wood,2008).Thisisanunfavorablesituation forthe developmentofexistingandnewknowledgeattheheadquarter. However,inthesamewayastraditionalMNEs,the international-izationprocessofEMNEsrequiresthecapabilitytoacquireand developknowledge(Mathews,2006).Hence,subsidiariesplay acentralroleinthepursuitofnewknowledge(Borini&Fleury, 2011). Different authors state that EMNEs survival depends evenmoreheavilyonresourcesthathavebeendevelopedabroad whencomparedtothemultinationalsfromdevelopedcountries (Guillén&García-Canal,2009;Mathews,2006).Therefore,this studyadvocatesthatsubsidiaryautonomyiscriticalforRKTin EMNEs,whichallowsustohypothesizethat:
H1. The greater the subsidiary autonomy, the greater the reverseknowledgetransfer.
Knowledgecomplexity
The increasing specialization and sophistication in R&D requires companies to integrate distinct knowledge areas to developnewproducts.Asaresultknwoledgeturnstobehighly complexanddifficulttoconductintra-knowledgetransfers.A paradox emerges:the greater the number of functional areas andscientific disciplines necessary todevelop newproducts, themorecomplexitistotransfertheknowledge(Ciabuschi& Martín,2012).Knowledgecomplexityisassociatedtothe ampli-tudewhichistheextentofspecializationfields(Grant,1996)and theambiguityofthereferencedknowledge(Reed&DeFillippi,
1990). The greater the number of techniques, organizational routines,peopleandresourcesinvolvedthatareconnectedtoa particularknowledge,themorecomplexitbecomes.These con-ditionsmoderatetheinformationamountthatmustbeprocessed fortheunderstandingofcomponentsinvolved(Simonin,1999). Thus, management scholars tend to agree on the idea that complexity hinders knowledge transfersinceit decreases the receiver’sabilitytoidentify,understandandintegratethe knowl-edgetobeacquired(Simonin,1999).Yet,oppositeresultshave beenfoundintheliterature(Minbaeva,2007).Since,complex knowledgeisthemostvaluabletothecompany’s competitive-ness.Studieshaveshown,forexample,thatglobalteamsareable tosharecomplexknowledgethroughrulesandcodescommon totheexchangingarea(Reddy,2011).
IntheEMNEperspective,additionaleffortstosharethiskind of knowledgecanbeadvantageous since,as itsimitationand substitution ishampered, it maybe useful tothe buildingof strategic capabilities (Nair,Demirbag, &Mellahi, 2015) due to the prevailing need to use the available foreign resources (Mathews, 2006). A study conducted in Indian multination-als, for example, found that RKT happens regardless of the knowledgecomplexity.InBrazilianMNEs,itissuspectedthat onlylesscomplexknowledgefromsubsidiariesistransferredin reversedirection,consideringthattheforeignsubsidiariesrole isdeterminedbytheBrazilianheadquarters(Galina&Moura, 2013)whichstill holdsgreater centralization inthedecisions andinnovations.Accordinglyitwasformulatedthefollowing hypothesis:
H2. Thelowerthecomplexityofthesubsidiary’sR&D knowl-edge,thegreaterthedegreeofreverseknowledgetransfer. Localembeddedness
EmbeddednesisrelatedtothenotionthatMNE’scompetitive performancecanbefacilitatedthroughthesocialrelationships they create withseveral business players such as customers, universities and local research institutions (Grabher, 1993; Granovetter, 1985; Uzzi, 1996). More specifically, embed-dedness refers tothe mutual adaptation of activitiesbetween two companies as much as a common understanding of the collective targets and appropriate ways to work in a social system (Tsai & Ghoshal, 1998). Therefore, it is considered as astrategic resourcefor MNEs.It provides easyaccess to the resources and capabilities that are outside the company (Anderssonetal.,2002;Uzzi&Gillespie,2002)thatareable to generate a large knowledge transfer among the partners (Figueiredo,2011;Uzzi&Gillespie,2002).
Thedegreeofembeddednessbyforeignsubsidiaries, mea-sured by the proximitytolocal partners, reflectssubsidiary’s abilitytoabsorbknowledgefromitslocalnetwork,which some-timesmightresultinnewknowledgecreation(Anderssonetal., 2002).Thisscenariotendstodirectlyfostersubsidiary’s innova-tivecapacity,i.e.improvementofexistingproductsandservices ornewproduct,service,technologydevelopment(Andersson, Björkman,&Forsgren,2005;Cantwell&Mudambi,2005;Frost etal.,2002;Håkanson&Nobel,2001;Yamin&Otto,2004).
Indirectly,localsubsidiaryembeddednesscanfosterknowledge transfertootherMNE’sunits(Powelletal.,1996;Yamin&Otto, 2004),constructing,inturn,thesubsidiary’spowerrelationships withintheMNE(Andersson,Forsgren,&Holm,2007).
Higher levels of subsidiary embeddedness are related to an understanding of the context in which the local knowl-edge resides. Frequently,subsidiariesinteractwithits closest networkof local companiesandinstitutions inorder tolearn aboutcustomersandtechnologiesand, therefore‘capture’the localknowledge(Figueiredo,2011).Subsequently,itmustuse theconnectivityalreadyestablished withintheMNE network for transferring the knowledge in reverse direction (Meyer et al., 2011; Najafi-Tavani, Giroud, & Andersson, 2013). Regarding MNEsof emerging markets,Child andRodriguez (2005),Mathews(2006)andLuoandTung(2007)emphasize the importance of relationships andknowledge opportunities availableatsubsidiarieshostingmarkets.Forexample, provid-ing easy access to technologies found in developed markets (Figueiredo,2005).RamamurtiandSingh(2009,pp.126–127) showthatEMNEscanpursueseveraldifferentstrategies,such as“low-costpartners”,“globalconsolidators”and“globalfirst movers”.Basedonthesearguments,thefollowinghypothesisis suggested:
H3. Thegreatertheembeddednessofaforeignsubsidiary,the higherthereverseknowledgetransfer.
Methodology
Sampleanddatacollection
Thesample ofthisstudy consistsof BrazilianMNEswith manufacturing,salesorR&Dsubsidiariesabroad.Weexpected thatsubsidiarieswithmorestrategicactivitieswouldhavemore opportunitiesof transferringknowledgetotheheadquartersin reversedirection.Thedatawascollectedusinganeletronic sur-veywithBrazilianMNEssubsidiariesestablishedabroad(see Appendix1).Duetothenon-existenceofanofficialnumberof Brazilianmultinationalsowning subsidiarieswitheither man-ufacturingorR&Dcentersinstalledabroad.Thefirststepwas toidentifyBrazilianmultinationalspresentingthese character-isticsfromsecondarydatasources,suchasGINEBRAProject (ManagementSystemfor theInternationalizationofBrazilian Enterprises) that resulted in the publication ‘Business Man-agementfor the Internationalization of BrazilianCompanies’ (coordinatedbyFleury,2010),anannualsurveyoftheFundac¸ão DomCabral(DomCabralFoundation),ValorEconômico (Eco-nomicValue),andSOBEET(BrazilianSocietyofTransnational Corporations)surveysaswellasdatafromtheBrazilian Multi-nationalsObservatory (CenterofBrazilian Multinationals)of theESPM(SchoolofHigherEducationinAdvertisingand Mar-keting).
In this secondary sources, 63 multinational companies were listed, being possible to identify 240 subsidiaries with foreign manufacturing operations and/or R&D centers. Of thispopulation,39Brazilianmultinationalsparticipatedinthe survey(61.9%),with78responses,correspondingto32.5%of
allsubsidiaries.Thismeansthatinsomecasesresponseswere receivedfrommorethanonesubsidiaryperheadquarter.
Intheattempttoidentifypossibleshortcomingsor misunder-standingsinthesurvey,apre-testwasconductedtogetherwith specialists from academia andindustry(Cooper &Schindler, 2003)whichhelptogeneratenewinsightsandadjustmentsin thequestionnaire.Following,theelectronicsurveywassentto participants,withafollow-upphone-calltoclarifyanyquestions fromrespondents.Thetotalperiodof datacollectionwasfive months, fromOctober 2013 uptoFebruary 2014.Responses were collectedfromR&Doffices andtherespondentsranged fromsubsidiarydirector,internationalbusinessandR&D direc-tor,andengineeringmanagers.
Measures
Dependentvariable
Thedependentvariable(reverseknowledgetransfer–RKT) represents, overthe lastthreeyears,therate of RKTof tech-nology andmarket knowledge that the subsidiary transferred back totheheadquarters. Inordertodetailthe typesof tech-nologicalcontent,itwasappliedtheIammarino,Padilla-Pérez, andVonTunzelmann(2008)scale,whichwasvalidated previ-ouslybyotherauthors(Lall,1992;Bell&Pavitt,1995;Ariffin andFigueiredo(2003)),whorankthetechnologicalknowledge transferintermsofproductandprocess.Onafive-pointscale (rangingfrom1“notatall”to5“toaverygreatextent”).For ensuring therobustnessresults,it wasalsoinserted adummy variablewhichallowedtherespondenttoindicatethecasesin whichthe subsidiary hadneverdone or haddonethe reverse transferofaspecificproductorprocessknowledge(0or1). Independentvariables
The knowledge complexity construct measures the num-ber of interdependent technologies, routines,individuals,and resources linkedtoaparticularknowledge orasset (Simonin, 1999).Moreover,thecomplexityconstructwasmeasuredusing a six-item Likert scale based on responses (1=strongly dis-agree;5=stronglyagree)(adaptedfromSimonin,2004;Zander &Kogut,1995).Thesubsidiaryautonomymeasureindicatesthe extenttowhichasubsidiaryisallowedtomakedecisionsabout its key strategic issues (Rabiosi, 2011).The measureof sub-sidiaryautonomywasbasedonascaleoriginallydevelopedby GhoshalandNohria(1989)andlaterusedbyBirkinshawetal. (1998)andRabiosi(2011).Afive-itemLikertscaleassessedit. Thesubsidiaryembeddednessindicatesthecollaborationdegree withthelocalnetworks.Inparticular,thisstudyfocusesonthe subsidiary embeddedness withlocalcustomersandsuppliers. ThisconstructwasdevelopedbasedonAnderssonetal.(2002, 2005).Afive-itemLikertscaleassessedit.
Controlsvariables
The MNE literaturesuggestsseveral factorsthat mightbe correlatedtoRKT.Inparticular,itisexpectedthatsubsidiaries locatedindevelopedcountriesandmoreancientsubsidiariesare morelikelytotransferreverseknowledge.
Subsidiarylocation. Thehostcountryhasbeenrelatedto fac-tors that impact the subsidiary development and positioning (Birkinshaw & Hood, 1998; Gupta & Govindarajan, 2000; Cantwell&Mudambi,2005;Rabiosi,2011)aswellasthenature ofRKT(Yangetal.,2008).Particularly,thishappensbecausethe subsidiary’scapabilitiesandskillscouldreflectthecountry tech-nologicalandinstitutionalforces,suchaslegalandinstitutional factors(forexamplepatentprotectionandindustrialincentives) that ensure the proliferation of innovation. The assumption, therefore,isthatcompaniesinemergingmarketsgetinvolvedin lessinnovationthancompaniesindevelopedmarkets,duetothe lackofhightechnologyinemergingmarkets(Vernon-Wortzel& Wortzel,1998).Thus,thehighertheeconomicdevelopmentof thesubsidiary’shostcountry,thegreaterthebenefitsearnedby theheadquartersarisingfromthetransferredknowledge(Frost etal.,2002).
Foremerged market MNEs, subsidiarieslocated in devel-opedor high-incomecountriescanimpactthe rateandspeed ofRKT,sincetheresourcesavailableinthesemarketscanhelp increasetheheadquartersbreadthandnovelty(Mathews,2006). Onthecontrary,Aulakh(2007)andCuervo-CazurraandGenc (2008)arguethatemergingmarketMNEshavethesame knowl-edge resources than those operating in developed countries. In the specific case of Brazilian companies,Bezerra, Borini, and Maclennan(2015) have concluded that MNEs’ Brazilian subsidiarieslocatedindevelopedcountriestransfermore knowl-edgeinreversedirectionthansubsidiarieslocatedinemerging countries.Inordertocapturethesubsidiarylocationeffectson thelevels ofRKT,thedummyvariablelow-income countries (0)andhigh-incomecountries(1)wereaddedtothemodel. Subsidiary’sage. More ancientsubsidiariescould havesome advantagesovernewerones dueto(1) theincreased informa-tionandresources,(2)thehigherdevelopmentofR&Dskills, (3)acquiredexperienceandexpertise,and(4)increasedlearning curveeffects.Thereforetheymightbelessdependenton knowl-edge from headquarters (Foss &Pedersen, 2002; Yamin and Otto,2004).Previousstudiesshowbothpositiveandnegative effectsoforganizations’sageregardingthelearningand innova-tionoutcomes(Sørensen&Stuart,2000).Whilepositiveeffects are justified by the knowledge increase, accumulated experi-ence and possessionof stronger relationships with suppliers, andcustomersthatenabletheinnovationprocessimprovement (Cohen&Levinthal,1990).Negativeeffectsareassociatedwith upgrade difficultiesof moremature companies withexternal technologicaladvances,attheriskofbecominginertandlimited forlearningandadaptingtonewcircumstances.Inotherwords, thereisalossofinnovativecapacity(Sørensen&Stuart,2000; Tushman&Anderson,1986).
Inthisregard,intheBrazilianmultinationalscontext,Bezerra etal.(2015) foundthat the younger asubsidiary, the greater its extent of RKT.Thus,despite the inconclusive findings of subsidiaries’age,itisexpectedthatoldersubsidiariesaremore likelytodevelopandtransferbackknowledgetoheadquarters thanrecently establishedsubsidiaries. Particularly, duetothe periodofexistenceofBraziliansubsidiariesismuchlowerwhen comparedtoemergedmarketsubsidiaries.Inordertocapturethe
subsidiaryageeffectsonthelevelsofRKT,thedummyvariable young(0)andoldsubsidiary(1)wereaddedtothemodel.The detailsofeachvariable,includingindicatorsandauthors,used asbackgroundispresentedinAppendix1.
Dataanalysis
Adescriptiveanalysiswascarriedouttoidentifythe frequen-ciesofrespondents’answersforallconstructscomprisedinthe survey.ThePartialLeastSquare–StructuralEquationModeling (PLS-SEM)wasused toassessthedeterminants’influenceof RKT(Hair,Hult,Ringle,&Sarstedt,2014).Thestructuralmodel wasestimatedonSmartPLS3.0(Ringle,Sarstedt,&Schlittgen, 2014)usingthe‘path’weightingscheme.Thedecisiontouse thismethodtookintoaccount anumberof criteria,including (1)thefactthattheindicatorsdonothaveanormaldistribution, whichisoneofthe assumptionsfor theuseof themaximum likelihoodmethod(ML);(2)theuseofintervalscales(Joreskog &Wold,1982);(3)itsabilitytodealwithmorecomplexmodels ascomparedtoLISREL(Henseler,Ringle,&Sinkovics,2009); and(4)thesmallsamplesize.
Since the PLSalgorithmformulation(Hui &Wold, 1982; Lohmöller, 1989) is recognized that it is biased and is only “consistent at large”,whichmeans that the biasdecreases as the numberof indicatorsbylatent variableis increased.This issueoccursbecausetherelationshipsamongstlatentvariables (correlationsand pathcoefficients) are estimated as from the factorialscores,whichareobtainedasasumoraweighted aver-ageoftheirindicators,includingthemeasurementerrors.This factistreatedascorrelationattenuationinthemethodological referencesrelatedtopsychometrics,forexample(Nunnally& Bernstein,1994,p.212).However,despitethisbias,Hairetal. (2014,p.79)mentionsomesimulationswhereitisidentified thatthebiasissmallforpracticalpurposes.Forfourandeight indicatorsbylatentvariable,ChinandNewsted(1999,p.333) foundabiasequalto0.05.Tominimizethisbias(attenuation)the latentvariablesweremeasuredwithfivetosixindicatorseach, reachingreliabilityvalues(compositereliabilityandCronbach’s alpha)higherthan0.8(Table2).
Additionally,toassessthisbiassize,thedisattenuated cor-relations were calculated (or “correction” for attenuation as explained by Nunnally and Bernstein (1994, p. 241) of the dependentvariable(RKT)withtheotherindependentvariables (Table1).
Itisobservedthatthehighestbiaswasequalto0.053.Asthis isasmallbiasforpracticalpurposesandisintheconservative direction(underestimatingthepopulationparameter),theresults wereconsideredadequateforpurposesofresultsinterpretation fromthepointofviewofstatistical significanceandpractical importance.
Anotherwaytocheckthesamplesizeadequacyisthrough analyzing the statistical power sensitivity, performed with G*Power3software(Faul,Erdfelder,Buchner,&Lang,2009). Forasampleof78respondents,withasignificancelevelof5% andstatisticalpowerof0.80(Cohen,1998),thetest‘sensitivity analysis’ foundthatthe modelisable todetectan effectsize of0.1574,whichisconsideredamediumeffect(Cohen,1998).
Table1
DisattenuatedcorrelationsofthedependentvariableRKT.
Statistics 1.Age 2.Country 3.Complexity 4.Autonomy 5.Embeddedness 6.RKT
CorrelationwithRKT −0.150 0.350 0.270 0.300 0.410 1000
Compositereliability 1 1 0.890 0.880 0.900 0.870
DisattenuatedcorrelationwithRKT(usingCR) 0.161 0.375 0.307 0.343 0.463
Attenuation −0.011 0.025 0.307 0.343 0.053
Using the population effect formula f2=R2/(1−R2) (Cohen, 1998),itwasconcludedthattheresearchwilldetectaminimum R2of0.1507.
Findings
TherespondentsincludedalargevarietyofBrazilian multi-nationals ranging from natural resources (12%), consumer goods(21%), basicinputs(32%), manufacturing (19%), sys-temassembly(10%) andrawmaterialsforconstruction(6%). The responding subsidiaries locations were: Latin America (42%),NorthAmerica(24%),Asia(14%),Europe(14%)and Africa(5%).Atthe countrylevel,the largestnumberof sub-sidiaries are in the U.S. (15%), Argentina (15%), Colombia (10%)andMexico(9%).Moreover,China(6%)alreadyappears as an important destination for Brazilian subsidiaries. As to the size and numberof employees at the subsidiary, 56% of responding subsidiaries are in the range 100–1000 employ-ees, followed by 14% of subsidiaries employing more than 1000workers.Thisdescriptivestatistics showsthatarelative percentage of subsidiariesconsist of consolidated companies abroad.Asregardtothesubsidiaries’age,themajority(69%) is under ten years of age, 22% are between ten and nine-teenyears andonly9%are morethan 20yearsof activities. The entry mode of Brazilian subsidiaries abroad represents 77% acquisitions and 23% direct investment or greenfield investment.
Evaluationofthemeasurementmodel
In measuring the constructs, the model wasconducted by evaluatingtheconvergent,discriminantandreliabilityvalidity. AspresentedinTable1,theconstructs(alsocalledlatent vari-ables)weremeasured using reflectiveindicatorstoverify the adequatereliabilityoftheCronbach’salphavalues.Inaddition, all latent variablesachieved convergent validity, that is, they haveanaveragevarianceextracted(AVE)higherthan0.5,and compositereliabilityhigherthan0.7(Hairetal.,2014;Henseler etal.,2009;Tenenhaus,Vinzi,Chatelin,&Lauro,2005). How-ever,threeitemsofthescaleshadtoberemovedfromthemodel so that theAVE reached thereference value(3.7;4.2; 4.4in Appendix1).Thediscriminantvalidityisverifiedbythe For-nellLarckercriterionandwasevaluatedthroughthecross-loads analysis. Thisfacilitated todetermine whether a construct is trulydistinctfromotherconstructsthroughempiricalpatterns. Basedonthisresult,itwasnotedthatallcorrelationsamongstthe latentvariablesweresmallerthanthesquarerootoftheaverage
varianceextractedoftheirlatentvariables(Fornell&Larcker, 1981).Thus,itcanbesaidthatthemodelpresentedconvergent, discriminantandreliabilityvalidity.Themeans,standard devi-ations,reliabilityestimatesandfactorcorrelationsarereported inTable2.
Assessmentofthestructuralmodel
Thestructuralmodelisabletospecifytherelationship pat-ternsamongsttheconstructs.Themodelwasassessedusingfive criteria:(i)pathcoefficients(β);(ii)pathsignificant(p-value); (iii) varianceexplain(R2);(iv)effectsize(f2)and(v) predic-tiverelevance(Q2).AccordingtoHairetal.(2014),themain criteria for the structural model evaluationare thecoefficient ofdetermination(R2)andthelevelandsignificanceofthepath coefficients(β).Tocalculatethem,thepathweightingscheme andabootstrappingtechniquewereusedwith78observations and 500 randomsamples to estimatethe t-values inorderto assessthe significance.Forsocial scienceresearches,R2 val-uesof0.26,0.13and0.02areconsideredstrong,moderateand weak,respectively(Cohen,1998).
Continuing Fichman and Kemerer (1997), in addition to the full model, we have evaluated two nested models (con-trolmodelandtheoreticalmodel).Intotal,thesethreemodels were accessed toevaluate the true impact andthe additional explanatorypowerofthetheoreticalvariablesafterthevariance explainedbythecontrol.Thefullmodelincludesallthisstudy variables,thecontrolmodelincludesonlythecontrolvariables, and thetheoretical model includesthe hypothesized relation-ships.Comparisonsamongstthethreemodelsaresummarized inTable3.
TheR2valueresultsforthefullmodel(includingcontrol vari-ables)indicatethatthevarianceof36%inRKTwasexplained bythemodel.Thisresultisconsideredsubstantialandprovides evidence thatthe model iscapableof explainingthe depend-entvariable(Cohen,1998).Whencomparingtheresultsofthe adjusted R2 (33%) with the sensitivity analysis on statistical power,itisfoundaR2valuewellabovetheminimumdetectable bythemodel,whichis15%.
A comparison between the full model and control model (location andage) shows that the control model explains an incrementalvarianceonR2of19%onthedependentvariable (RKT).Thedeltabetweenthecontrolmodelandthefullmodel was(R2=0.17).Thisresultsuggeststhat,despitehaving pre-sentedamoderateresult,controlvariablesalonedonotprovide asolidbasisthroughwhichonecanunderstandandpredictRKT patterns.
Table2
Evaluationofthemeasurementmodel.
Variables Mean S.D. AVE C.R. C.A. 1 2 3 4 5 6
1.Age 10.8 11.1 1.00 2.Country – – 0.13 1.00 3.Complexity 3.7 1.1 0.61 0.89 0.86 −0.18 0.19 0.78 4.Autonomy 3.2 1.0 0.55 0.88 0.90 0.07 0.17 0.03 0.74 5.Embeddedness 3.0 1.0 0.53 0.90 0.87 0.08 0.28 0.02 0.07 0.73 6.ReverseTransfer 3.0 1.0 0.54 0.87 0.83 −0.15 0.35 0.27 0.30 0.41 0.73 Note1:Inboldonthediagonal,therearevaluesofthesquarerootoftheaveragevarianceextracted.
Note2:AVE,averagevarianceextracted;C.R.,compositereliability;C.A.,Cronbachsalpha. Note3:AVEbenchmarks:0.5;compositereliability:0.7;Cronbach’salpha:0.6.
Table3
Significancetestresultsofthestructuralmodelpathcoefficients.
H Pathfrom To Fullmodel Controlmodel Theoreticalmodel Effectsize(f2)
β p-Values β p-Values β p-Values
Age RKT −0.182 0.01** −0.212 0.00*** Country RKT 0.195 0.04* 0.408 0.00*** H1 Autonomy RKT 0.188 0.04* 0.252 0.01** 0.05 H2 Complexity RKT 0.246 0.01** 0.262 0.00*** 0.08 H3 Embeddednes RKT 0.351 0.00*** 0.395 0.00*** 0.15 ReversetransferR2 0.36 0.19 0.30 R2 0.17 0.06
Valuesoftwerecalculatedthroughbootstrappingwith500resamplesand78casespersample.
* p≤0.05. ** p≤0.01. ***p≤0.001
Comparing the full model and the theoretical model, the incrementalvariancederived bythemodelisaround 30%for RKT.Resultsindicatethatthetheoreticalmodelinthisstudyis substantiveenoughtoexplainthevarianceintheresearchmodel. However,controlvariableswereresponsibleforaconsiderable proportionofthevarianceintheR2valueofRKT.Asthe pre-dictedpathsforthestructuralmodel,allthehypothesizedwere statisticallysignificant.Theconfidencelevel inthe prediction modelwasmeasuredbytheindicatorQ2whichmustbehigher thanzero.TheQ2valuetoconstruct‘RKT’is0.171ensuring themodelpredictiverelevance(Hairetal.,2014;Henseleretal., 2009).
Theeffectsize(f2)measuresthemagnitudeofan indepen-dent variableon a dependent variable(Tabachnick & Fidell, 2007).The exogenous constructs omission of the model can beusedtoassessinwhichcasetheseomittedconstructshave substantialimpactontheendogenousconstructs.Cohen(1998) providedvaluesof0.02,0.15and0.35consideredweak, mod-erateandstrong,respectively.The f2 isalsocalculatedbyR2 included=f2−R2excluded/1−R2included(Hairetal.,2014). Following,Table3showsthesignificanceresultsofeachpath amongstthelatentvariablesandtheeffectsize.
Theresultssupporttwoofthethreehypothesesstatements. Hypothesis H1 shows that autonomy has a positive and significant effectonreversetransfer(β=0.19, p≤0.05).The effectsize (f2) of 0.05 indicates that theconstruct subsidiary autonomyhasaweakeffectontheendogenouslatentvariable RKT(Cohen,1998).Hypothesis H2 statesthat the lowerthe
complexityofsubsidiary’sR&Dknowledge,thelargertherate of reverse technology transfer to headquarters. Surprisingly, this study’s results showed that knowledge complexity has a significant, but positive effect on reverse transfer (β=0.25, p≤0.01).Thisrelationship ischaracterizedby aweakeffect (0.08)ontheendogenouslatentvariable‘RKT’(Cohen,1998). Finally,theresultsshowedthatsubsidiaryembeddednesshasa significantandpositiveeffect(0.15)onRKT,whichconfirms H3 hypothesis (β=0.35, p≤0.001). This relationship is characterizedbyamoderatetostrongeffectontheendogenous latentvariable‘RKT’(Cohen,1998).
Withregardtothecontrol variables,thelocalizationeffect waspositiveandsignificant(β=0.19,p=0.05)forRKT, indi-catingthatsubsidiarieslocatedindevelopingcountriesaremore likelytotransferknowledge inreversedirection.Alsofor the subsidiaryagevariablethecoefficientissignificant(β=−0.18, p=0.01)butthenegativesignindicatesthatRKTismorelikely to occur from young subsidiaries, confirmingthe findings of Bezerraetal.(2015).
Discussion
Despiteambiguous evidenceabout RKTin Brazil(Fleury & Fleury, 2011), this study found that Brazilian subsidiaries withahighautonomydegreearemorecapableoftransferring knowledgebacktoheadquarters,confirmingourhypothesisH1. An argument on the positive effect of autonomy for RKT is basedontheideathat thesubsidiariesindependence provides
greater accesstolocalknowledgedatabases,knowledge from local partners andpossibilities to innovate(Andersson etal., 2002;Ciabuschi&Martín,2012;Gupta&Govindarajan,1991; Cantwell&Mudambi,2005).Hence, subsidiaryautonomy is recognizedasanimportantpredictorofreverseknowledge trans-ferinthecontextofEMNEs.Autonomyempowerssubsidiaries toexploretheirownbusinessandmarketopportunitiessothat they can make use of external sources to their competitive advantage. Taking into account that Brazilian multinationals are still at an early stage of internationalization, it is a new phenomenon the fact that their subsidiaries have been seek-ing for autonomyandindependence fromtheir headquarters’ decisions.
Thispaperidentifiedthatknowledgecharacteristicsand sub-sidiarycharacteristicsdeterminetherateofreverseknowledge transferfromsubsidiariestoemergingmarketMNEs.First,from theknowledgecharacteristicsviewpoint,itwaspossibletoshow thattheknowledgecomplexitylevelhasapositiveimpactonthe extentofRKT.Thisfindingiscontrarytothisstudy’shypothesis (H2),whichsuggestedthatthelowerthesubsidiaryknowledge complexity,thegreatertheRKT.Itissuspectedthatoneofthe reasonsforthisintriguing,butinterestingresult,mayberelated totheknowledge complexityparadox,because,while knowl-edgetransferencountershighercostsproblems,itisthe most compensatory type of knowledge to the headquarters. Thus, itissuspectedthat theBrazilianmultinationalstrytotransfer the mostcomplex knowledge developedintheir subsidiaries, regardlessofthecomplexitylevelsassociated,whichincludes theinvolvementtoagreaterextent,oftheheadquarterssothat thistype of transfer actually materializes (Nairet al., 2015). Sucharesultisalsoinlinewiththeframeworkoflearningand effective leverage(Mathews, 2006)of theEMNE’s resources andnetworksabroad(LLLchart).Otherpossibleexplanationis theeffectofsubsidiary’srole.Forexample,moreinnovative sub-sidiariesmighttransfermorecomplexR&Dknowledge,which suggeststhatimplementerandcontributorsubsidiariesmaynot transfer(ortransfertoalesserextent)complextypeknowledge. Insummary,althoughtheinitialH2wasnotsupported,thisresult providesanopportunitytosuggestthatinnovativesubsidiaries mayengageincomplexknowledgetransferandthusbecomea competitiveplayer.
OurresultsalsosupportthehypothesisH3whichproposes thatlocalembeddednessimpactstherateofRKT.Itwasfound thatembeddednesswithsuppliersandcustomers,inotherwords, localbusinessnetworksincreasethepossibilityofgainingaccess tonew knowledge,whichcan subsequently be transferredto EMNEs. This paper confirms that subsidiaries from emerg-ing market multinationals become internationalized in order toexploreknowledgeandexistingcapabilitiesinforeign mar-kets as well as to develop new knowledge and capabilities throughknowledgeavailableinthesubsidiaries’host environ-ment(Narula,2012).Forsubsidiariesisessentialtobeembedded inlocalbusinessnetworkstoobtaindistinctiveknowledge devel-opment.Newconnectionswithlocalnetworksallowsubsidiaries toperforminnovative tasksfor headquarters,instead of tasks limited to adaptation of products and processes to the local market(Borini&Fleury,2011).
Withregardtothefirstcontrolvariable(location),theresults indicatedthat subsidiarieslocatedindevelopedmarkets,such asNorthAmericaandEurope,areprobablytheonesthatmost transferknowledgetotheirheadquarters.Thisresultisinline withseveralcontributionsintheliteraturewhichstatethatthe innovationcapacityofsubsidiarieslargelydependsonthehost countriesadvantages(Gupta&Govindarajan,2000;Cantwell &Mudambi,2005;Yangetal.,2008).
BasedonpreviousfindingsaboutEMNEs,twoperspectives canbepresented.Thefirstperspective,ledbyCuervo-Cazurra andGenc(2008),Ramamurti(2008),KhannaandPalepu(2011), Cuervo-Cazurra (2012) and Ramamurti (2012), argues that EMNEs have a newtype of capability,unlike the traditional MNEscapabilities,whichisrelatedtotheabilityofcopingwith the institutional deficiencies towhichtheyare exposed. This current advocates that emerging MNEs, for having operated inenvironmentspresentingdifficultconditions,suchas under-developed premises, corrupt bureaucracies, poor educational institutions and unstable governments, have the “advantages of adversity.”The second perspective,led byauthorssuch as Mathews (2006) andChild andRodriguez (2005),argue that MNEsplacetheirsubsidiariesindevelopedcountriesasaway toleveragetheirproductive,technologicalandmarketing effi-ciency, following an asset-seeking strategy, looking for their competitiveadvantagesincrease.Therefore,thepreferencesof emerging MNEs for developing markets exemplify their ten-dency to explore the “institutional voids”. However when it comes to subsidiaries that transferknowledge in the reverse direction, theyare morelikelytobeincountrieswherethere arebetterinfrastructureconditions,businesssupportinstitutions andfavorablelegalenvironment.
Regarding the second control variable (age), the results surprisingly indicatedthat therewas asignificant correlation, thoughnegative,betweenageandRKT.Thus,theyoungerthe subsidiary, the more likelythe existence of RKT. Apossible explanationforthisunexpectedresultisthefactthatexperience leads to efficiency gains, but on the other hand, in environ-mentswherechangesoccurveryrapidly,theadjustmentbetween organizational capabilities and market demands declines, as the subsidiariesgrow older,having inviewthatmore mature companiestakelongertoincorporatethemostcurrent techno-logical developments (Sørensen &Stuart, 2000).It is inthis perspectivethatageandaccumulatedskillscanbecome disad-vantageswhencomparedtoyoungersubsidiaries.Particularly, this occurs withregard tothe company’s ability to adapt or developmajortechnologicalchanges(Sørensen&Stuart,2000; Tushman & Anderson, 1986). With respect to the group of emerging MNEs,younger subsidiariesmay be more influen-tial in the headquarters’ knowledge exactly because they are abletobemoreagileanddynamicinrelationtotechnological developments.
Conclusions,limitationsandfurtherresearch
Thispaperexplainedreverseknowledgeflowsinsubsidiaries of emerging market multinationals and tested the impact of threedeterminantsinBrazilianmultinationals(Govindarajan&
Ramamurti,2011).Hence,severalcontributionstothe knowl-edge flow of RKT in Brazilian MNEs are suggested. First, incomparison with traditional MNEs, BrazilianMNEs have a higher interest in reverse technology transfer, due to the higherimportanceofsubsidiariesforheadquarters.Second, sub-sidiariesofBrazilianMNEswilltransferproducts’knowledge justwithabasicandintermediateleveloftechnological com-plexity(Ariffin &Figueiredo, 2003;Iammarinoetal., 2008). Third,ontheprocessofRKTinforeignsubsidiariesofBrazilian MNEs,thisworkexploredtheimpactofknowledgecomplexity characteristicsaswellsubsidiarycharacteristics,i.e.autonomy andembeddedness.TheresultsshowedthatRKTispositively affectedby knowledge complexity,subsidiary autonomy and embeddednessofforeignsubsidiarieswithcustomersand sup-pliers. Fourth, it was assessed the effect of the subsidiary’s location and age on the RKT.The results indicate that sub-sidiaries located in developed countries are more likely to transfer knowledge in reverse direction as well as younger subsidiaries. Thispaper’s empirical implications suggest that subsidiarieswithhigheraccesstolocalknowledgewillbebetter positionedtoacquirenewknowledgeandconsequentlytransfer itback toheadquarters.Theexternal embeddednesshasbeen indicatedasanimportantdeterminantofRKT.Fromthe view-pointofpracticalimplications,itisnecessarythatsubsidiaries investinmechanismsofrelationshipandknowledgeexchange
to establish strong collaborations with local partners. These findings mayalso beusefulfor policy makersin as muchas understandingtheinnovationtransferpatternisakeycomponent ofacountry’sinnovationsystem.
Animportantlimitationofthisstudyisthatthisresearchis limited tothe narrowcontextofBraziliansubsidiaries,which thereforeimposeslimitstotheresultsgeneralization.Second, thesamplesizeandsamplecompositionturnitdifficulttomake far-reaching generalizations of its results. Third, the survey methodprovidesasnapshotthatreducestheinformationsource credibility,theaccesstotherightpeople,theresponsescontrol, andtheutilizationofonlyonerespondentbycompany.Fourth, itschoiceofcontrolvariables,whichcouldhavecoveredother aspects, possibly stakeholdersinthe achievedresult. Finally, it isassumed somerestrictions relatedtothe unit of analysis andtheinformationfromheadquarters.Furtherresearchescould explore the autonomy and integration degree of subsidiaries fromemergingmarketsmultinationals.
Conflictsofinterest
Theauthorsdeclarenoconflictsofinterest.
AppendixA. Operationaldefinitionofmodelvariables
Indicators Authors
Dependentvariables
Reverseknowledge transfer(RKT)
1.1Developmentofnewproductionprocess;1.2Developmentofnewequipmentand/ortools;1.3 Developmentofnewproducts;1.4Know-howandexpertiseintheformofplans,models,instructions, guides,formulas,specifications,designs,plans,technicaldrawings,and/orprototypestodesignnew products;1.5Resultsofresearchintonewmaterialsandspecifications;1.6Resultsofresearchand development(R&D)intonewproductgenerations.
AriffinandFigueiredo (2003);BellandPavitt (1995);Iammarinoetal. (2008);Lall(1992);Yang etal.(2008)
Independentvariables
Complexity 2.1Itsunderstandingrequirespriorlearningfromotherrelatedtechnologicalknowledge;2.2Its understandingrequiresalargeamountofinformation;2.3Itistheproductofmanyinterdependent routines,individualsandresources;2.4Itincludesmanydifferentskillsorcompetencies;2.5Itis technologicallysophisticatedanddifficulttodeploy;2.6Itiscomplex(vs.simple)
Simonin(2004);Zanderand Kogut(1995)
Autonomy 3.1Implementationofchangesinproductsandservices;3.2Developmentofnewproductsand services;3.3Implementationofchangesinproductionprocesses;3.4Entryintonewmarketsinthe country;3.5Procurementandsupplychainmanagement;3.6.ManagementofPurchasingandSupply Chain;3.7Hiringandfiringofthesubsidiaryworkforce.
GhoshalandNohria(1989); Birkinshawetal.(1998); Rabiosi(2011) External embeddedness (withcustomers, suppliers)
4.1Customers/suppliershasfullyparticipatedinthedevelopmentoftechnologicalknowledgeinthe subsidiary;4.2Customers/suppliersshowedimportantinitiativesforthedevelopmentoftechnological knowledgeinthesubsidiary;4.3Customers/supplierssatisfiedtherequirementsindeveloping technologicalknowledgeinthesubsidiary;4.4Thetechnologicalsubsidiaryknowledgewaspartially developedwithinthisCustomers/suppliers’premises;4.5Thecooperationwithcustomers/suppliers hasbeencharacterizedbyfrequentinteractions.
LaneandLubatkin(1998),
Anderssonetal.(2005)and
Najafi-Tavanietal.(2013)
Moderatingvariables
Subsidiary’slocation 5.1Low-incomecountries(0);5.2High-incomecountries(1) CantwellandMudambi (2005)
Subsidiary’sage 6.1subsidiariesunder10yearsold(0);6.2Subsidiarieswithover10yearsold(1) AmbosandSchlegelmilch (2007);Rabiosi(2011)
References
Ambos,T.,Ambos,B.,&Schlegelmilch,B.(2006).Learningfromforeign sub-sidiaries:Anempiricalinvestigationofheadquartersbenefitsfromreverse knowledgetransfers.InternationalBusinessReview,15(3),294–312.
Ambos,T.C.(2015).Reverseknowledgetransfer.ThePalgraveencyclopedia ofstrategicmanagement.PalgraveMacmillan.
Ambos,B., & Schlegelmilch,B. B.(2007). Innovation and controlin the Multinationalfirm:Acomparisonofpoliticalandcontingencyapproaches. StrategicManagementJournal,28(5),473–486.
Andersson,U.,Forsgren,M.,&Holm,U.(2002).Thestrategicimpactof exter-nalnetworks:Subsidiaryperformanceandcompetencedevelopmentinthe multinationalcorporation.StrategicManagementJournal,23(11),979–996.
Andersson,U.,Björkman,I.,& Forsgren,M.(2005).Managingsubsidiary knowledgecreation:Theeffectofcontrolmechanismsonsubsidiarylocal embeddedness.InternationalBusinessReview,14,521–538.
Andersson,U.,Forsgren,M.,&Holm,U.(2007).Balancingsubsidiaryinfluence inthefederativeMNC:Abusinessnetworkview.JournalofInternational BusinessStudies,38(4),802–818.
Ariffin,N.,&Figueiredo,P.(2003).Internationalizationofinnovative capabili-ties:Counter-evidencefromtheelectronicsindustryinMalaysiaandBrazil. InPaperforDRUIDsummerconference(pp.12–14).
Aulakh,P.S.(2007).Emergingmultinationalsfromdevelopingeconomies: motivations,pathsandperformance.JournalofInternationalManagement, 13(3),235–240.
Bartlett,C.A.,&Ghoshal,S.(1989).Managingacrossborders.The transna-tionalsolution.Boston,MA:HarvardBusinessSchoolPress.
Bartlett,C.A.,&Ghoshal,S.(2000).Transnationalmanagement(3rded.).New York:McGrawHill.
Bell,M.,&Pavitt,K.(1995).Thedevelopmentoftechnologicalcapabilities.In Trade,technologyandinternationalcompetitiveness(1sted.).Washington, DC:WorldBank.
Bezerra,M.A.,&Borini,F.M.(2015).Theimpactofsocialandrelational contextsoninnovationtransferinforeignsubsidiaries.InternationalJournal ofLearningandIntellectualCapital,12.
Bezerra, M. A., Borini, F. M., & Maclennan, M. L. (2015). Reverse TransfersofInnovationandNationalDevelopment:Evidencefrom Brazil-ianSubsidiaries.Journalof TechnologyManagement Innovation,10(4), 1–8.
Birkinshaw,J.,&Hood,N.(1998).Multinationalsubsidiaryevolution: Capa-bilityandcharterchangeinforeign-owedsubsidiariescompanies.Academy ofManagementReview,23(4),773–795.
Birkinshaw,J.,Hood,N.,&Jonsson,S.(1998).Buildingfirm-specific advan-tages in multinational corporations: The role of subsidiary initiative. StrategicManagementJournal,19,221–241.
Birkinshaw,J.(1997).Entrepreneurshipinmultinationalcorporations:The char-acteristicsofsubsidiaryinitiatives.StrategicManagementJournal,18(3), 207–229.
Borini,F.M.,&Fleury,M.T.L.(2011).Developmentofnon-local compe-tencesinforeignsubsidiariesofBrazilianmultinationals.EuropeanBusiness Review,23,106–119.
Borini,F.,Oliveira,M.M.,Jr.,Silveira,F.S.,&Concer,R.O.(2012).Thereverse transferofinnovationofforeignsubsidiariesofBrazilianmultinationals. EuropeanManagementJournal,30(3),219–231.
Borini,F.M.,Costa,S.,Bezerra,M.A.B.,&Oliveira,M.M.,Jr.(2014).Reverse innovationasaninducerofcentresofexcellenceinforeignsubsidiariesof emergingmarkets.InternationalJournalofBusinessandEmergingMarkets, 6(2),163–182.
Borini, F. M.,Costa, S.,& Oliveira, M.M., Jr. (2016). Reverse innova-tionantecedents.InternationalJournalofEmergingMarkets(Print),11, 175–189.
Cantwell,J.,&Mudambi,R.(2005).MNEcompetence-creatingsubsidiary mandates.StrategicManagementJournal,26,1109–1128.
Cantwell,J.,&Piscitello,L.(1999).Theemergenceofcorporateinternational networksfortheaccumulation ofdispersed technological competences. ManagementInternationalReview,39,123–147.
Ciabuschi,F.,&Martín,M.O.(2012).Knowledgeambiguity,innovationand subsidiaryperformance.BalticJournalofManagement,7(2),143–166.
Child,J.,&Rodriguez,S.B.(2005).TheinternationalizationofChinesefirms: Acasefortheoreticalextension?ManagementandOrganizationReview, 1(3),381–410.
Chin,W.W.,&Newsted,P.R.(1999).Structuralequationmodelinganalysis withsmallsamplesusingpartialleastsquares.InR.HHoyle(Ed.),Statistical strategiesforsmallsampleresearch(pp.307–341).ThousandOaks,CA: SagePublications.
Chu, R.A.,&Wood, T., Jr.(2008).Culturaorganizacional brasileira pós-globalizac¸ão:globaloulocal?RevistadeAdministra¸cãoPública,42(5), 969–991.
Cohen,J.(1998).Statisticalpoweranalysisforthebehavioralsciences. Hills-dale,NJ:LawrenceErlbaumAssociates.
Cohen,W.M.,&Levinthal,D.A.(1990).Absorptivecapacity:Anew perspec-tiveonlearningandinnovation.ASQ,35,128–152.
Cooper,D.R.,&Schindler,P.S.(2003).Businessresearchmethods(8thed.). Boston:McGraw-HillIrwin.
Criscuolo,P.(2005).Inter-firmreversetechnologytransfer:Thehomecountry effectofR&Dinternationalization.IndustrialandCorporateChange,18(5), 869–899.
Criscuolo, P., &Narula,R.(2007).Usingmulti-hubstructures for interna-tionalR&D:Organisationalinertiaandthechallengesofimplementation. ManagementInternationalReview,47(5),639–660.
Cuervo-Cazurra, A.,& Genc, M.(2008). Transforming disadvantagesinto advantages:Developing-countryMNEsintheleastdevelopedcountries. JournalofInternationalBusinessStudies,39(7),957–979.
Cuervo-Cazurra,A.(2012).BusinessrelationshipsbetweenLatinAmericaand Asia:ThenextresearchFrontier.Globalization,Competitiveness,and Gov-ernability,6(1),16–22.
Dunning,J.H.(1993).Multinationalenterprisesandtheglobaleconomy. Wok-ingham,England/Reading,MA:AddisonWesley.
Faul,F.,Erdfelder,E.,Buchner,A.,&Lang,A.G.(2009).Statisticalpower analysesusingG*Power3.1:Testsforcorrelationandregressionanalyses. BehaviorResearchMethods,41(4),1149–1160.
Fichman,R.G.,&Kemerer,C.F.(1997).Theassimilationofsoftwareprocess innovations:Anorganizationallearningperspective.ManagementScience, 43(10),1345–1363.
Figueiredo,P.N.(2005).Acumulac¸ãotecnológicaeInovac¸ãoindustrial: con-ceitos,mensurac¸ãoeevidênciasnoBrasil.SãoPauloemPerspectiva,19(1), 54–69.
Figueiredo,P.N.(2011).Theroleofdualembeddednessintheinnovative perfor-manceofMNEsubsidiaries:EvidencefromBrazil.JournalofManagement Studies,48(2),417–440.
Figueiredo,P.N.,&Brito,K.(2011).TheinnovationperformanceofMNE subsidiariesandlocalembeddedness:Evidencefromanemergingeconomy. JournalofEvolutionaryEconomics,21(1),141–165.
Fleury,A.(Ed.).(2010).GestãoEmpresarialparaainternacionaliza¸cãodas empresasbrasileiras.Ed.Atlas.
Fleury, A.,&Fleury,M.(2011).Brazilianmultinationals.Competencesfor internationalizationCambridge.
Foss,N.J.,&Pedersen,T.(2002).TransferringknowledgeinEMNs:Therole ofsourcesofsubsidiaryknowledgeandorganizationcontext.Journalof InternationalManagement,8,49–67.
Fornell, C., & Larcker, D. (1981). Evaluating structural equation models withunobservablevariablesandmeasurementerror.JournalofMarketing Research,18(1),39–50.
Frost,T.S.,&Zhou,C.(2005).R&Dco-practiceandreverseknowledge inte-grationinmultinationalfirms.JournalofInternationalBusinessStudies,36, 676–687.
Frost, T., Birkinshaw, J., & Ensign, P. (2002). Centers of excel-lence in multinationalcorporations.StrategicManagement Journal,23, 997–1018.
Galina,S.V.R.,&Moura,P.G.D.(2013).InternationalizationofR&Dby Brazilianmultinationalcompanies.InternationalBusinessResearch,6(8), 55–67.
Gammelgaard,J.,Holm,U.,&Pedersen,T.(2004).ThedilemmasofMNC sub-sidiaryknowledgetransfer.InV.Mahnke,&T.Pedersen(Eds.),Knowledge flows,governanceandthemultinationalenterprise.NewYork:Palgrave Macmillan.
Ghoshal,S.,&Bartlett,C.A.(1988).Creation,adoption,anddiffusionof innova-tionsbysubsidiariesofmultinationalcorporations.JournalofInternational BusinessStudies,19,365–388.
Ghoshal,S.,&Nohria,N.(1989).Internaldifferentiationwithinmultinational corporations.StrategicManagementJournal,10,323–337.
Govindarajan,V.,&Trimble,C.(2012).Reverseinnovation:Aglobalgrowth strategythatcouldpre-emptdisruptionathome.Strategy&Leadership, 40(5),5–11.
Govindarajan,V.,&Ramamurti,R.(2011).Reverseinnovation,emerging mar-kets,andglobalstrategy.GlobalStrategyJournal,1(3–4),191–205.
Guillén,M.F.,&García-Canal,E.(2009).TheAmericanmodelofthe multina-tionalfirmandthenewmultinationalsfromemergingeconomies.Academy ofManagementPerspectives,23,23–35.
Grant,R.(1996).Towardaknowledge-basedtheoryofthefirm.Strategic Man-agementJournal,17,109–122.
Grabher,G.(1993).Theweaknessofstrongties:Thelock-inofregional net-works?InG.Grabher(Ed.),Theembeddedfirm–Onthesocioeconomics ofindustrialnetworks(pp.255–277).London:Routledge.
Granovetter,M.(1985).Economicactionandsocialstructure:Theproblemof embeddedness.AmericanJournalofSociology,91(3),481–510.
Gupta,A.K.,&Govindarajan,V.(2000).Knowledgeflowswithinthe multina-tionalcorporation.StrategicManagementJournal,21,473–496.
Gupta,A.,&Govindarajan,V.(1991).Knowledgeflowsandthestructureof controlwithinmultinationalcorporations.AcademyofManagementReview, 16(4),768–792.
Hair,J.F.,Hult,G.T.M.,Ringle,C.M.,&Sarstedt,M.A.(2014).Primer onpartialleastsquaresstructuralequationmodeling(PLS-SEM).Sage: ThousandOaks.
Håkanson, L., & Nobel, R. (2001). Organizational characteristics and reverse technology transfer. Management International Review, 41(4), 395–420.
Henseler,J.,Ringle,C.M.,&Sinkovics,R.R.(2009).Theuseofpartialleast squarespathmodelingininternationalmarketing.InR.R.Sinkovics,&P. N.(Org.)Ghauri(Eds.),Newchallengestointernationalmarketing(pp. 277–319).
Holm,U.,&Pedersen,T.(2000).Thedilemmaofcentresofexcellence: Contex-tualcreationofknowledgeversusglobaltransferofknowledge.Copenhagen BusinessSchoolDepartmentofInternationalEconomicsandManagement, 3,8–20.
Hui,B.S.,&Wold,H.O.(1982).Consistencyandconsistencyatlargeofpartial leastsquaresestimates.InK.G.Jöreskog,&H.Wold(Eds.),Systemsunder indirectobservations:Causality,structure,prediction,PartII(pp.119–130). Netherlands:North-HollandPublishingCompany.
Iammarino,S.,Padilla-Pérez,R.,&VonTunzelmann,N.(2008).Technological capabilitiesandglobal–localinteractions.Theelectronicsindustryintwo Mexicanregions.WorldDevelopment,36(10),1980–2003.
Immelt,J.R.,Govindarajan,V.,&Trimble,C.(2009).HowGEisdisrupting itself.HarvardBusinessReview,56–65.
Jansen,J.,VanDenBisch,F.,&Volberda,H.(2005).Managingpotentialand realizedabsorptivecapacity:Howdoorganizationalantecedentsmatter? AcademyofManagementJournal,48(6),999–1015.
Joreskog,K.G.,&Wold,H.(1982).TheMLandPLStechniquesformodeling withlatentvariables:Historicalandcompetitiveaspects.InK.G.Joreskog, &H.Wold(Eds.),Systemsunderindirectobservation,Part1(pp.263–270). North-Holland:Amsterdam.
Khanna,T.,&Palepu,K.G.(2011).Winninginemergingmarkets:Aroadmap forstrategyandexecution.Boston,MA:HarvardBusinessPress.
Kogut,B.,&Zander,U.(1992).Knowledgeofthefirm,combinativecapabilities, andthereplicationoftechnology.OrganizationScience,3,3.
Lane,P.J.,&Lubatkin,M.(1998).Relativeabsorptivecapacityand interorga-nizationallearning.StrategicManagementJournal,19,461–478.
Lall,S.(1992).Technologicalcapabilitiesandindustrialisation.World Devel-opment,20(2),165–186.
Luo,Y.,&Tung,R.(2007).Internationalexpansionofemergingmarket enter-prises.JournalofInternationalBusinessStudies,38(4),481–498.
Lohmöller, J.-B.(1989). Latent variable path modeling with partial least squares.pp.1989.Heidelberger:Physica-Verlag.
Lyles,M.A.,&Salk,J.E.(1996).Knowledgeacquisitionfromforeignparents ininternationaljointventures:AnempiricalexaminationintheHungarian context.JournalofInternationalBusinessStudies,27,877–903.
Mathews,J.(2006).Dragonmultinationals:Newplayersinthe21stcentury. Asia-PacificJournalofManagement,23,2–27.
Meyer,K.,Mudambi,R.,&Narula,R.(2011).Multinationalenterprisesand localcontexts:Theopportunitiesandchallengesofmultipleembeddedness. JournalofManagementStudies,48(2),235–252.
Michailova,S.,&Mustaffa,Z.(2012).Subsidiaryknowledgeflowsin multi-nationalcorporations:Researchaccomplishments,gaps,andopportunities. JournalofWorldBusiness,47(3),383–396.
Minbaeva, D.B. (2007).Knowledgetransferin multinationalcorporations. ManagementInternationalReview,47,567–593.
Mudambi,R.,Mudambi,S.M.,&Navarra,P.(2007).Globalinnovationin MNEs:Theeffectsofsubsidiaryself-determinationandteamwork.Journal ofProductInnovationManagement,24,442–455.
Mudambi,R.,&Navarra,P.(2004).Isknowledgepower?Knowledgeflows, subsidiarypowerandrent-seekingwithinEMNs.JournalofInternational BusinessStudies,35(5),385–406.
Mudambi,R.,Piscitello,L.,&Rabbiosi,L.(2014).Reverseknowledgetransfer inMNEs:Subsidiaryinnovativenessandentrymodes.LongRangePlanning, 47(1–2),49–63.
Najafi-Tavani,Z.,Giroud,A.,&Andersson,U.(2013).Theinterplayof net-workingactivitiesandinternalknowledgeactionsforsubsidiaryinfluence withinMNCs.JournalofWorldBusiness,4(1),122–131.
Nair,S.R.,Demirbag,M.,&Mellahi,K.(2015).Reverseknowledge trans-ferfromoverseasacquisitions:AsurveyofIndianMNEs.Management InternationalReview,55(2),277–301.
Narula,R.(2012).DoweneeddifferentframeworkstoexplaininfantMNEs fromdevelopingcountries?GlobalStrategyJournal,2(3),188–204.
Nohria, N., & Ghoshal, S. (1994). Differentiated fit and shared values: Alternativesformanagingheadquarters-subsidiaryrelations.Strategic Man-agementJournal,15,491–502.
Nohria, N.,& Ghoshal, S.(1997). Thedifferentiated network:Organizing multinationalcorporationsforvaluecreation.SanFrancisco:Jossey-Bass Publishers.
Nunnally,J.C.,&Bernstein,I.H.(1994).Psychometrictheory(3rded.).New York:McGraw-Hill.
Powell,W.W.,Koput,K.W.,&Smith-Doerr,L.(1996).Interorganizational collaborationandthelocusofinnovation:Networksoflearningin biotech-nology.AdministrativeScienceQuarterly,41,116–145.
Rabiosi,L.(2011).Subsidiaryrolesandreverseknowledgetransfer:An inves-tigationoftheeffectsofcoordinationmechanisms.JournalofInternational Management,17,97–113.
Rabiosi,L.(2008).Theimpactofsubsidiaryautonomyonreverseknowledge transfer:Movingfromsingletointerdependentexplanations.Workingpaper. CopenhagenBusinessSchool.CenterforStrategicManagementand Glob-alization.
Rabiosi,L.,&Santangelo,G.D.(2011).Parentcompanybenefitsfromreverse knowledgetransfer:TheroleoftheliabilityofnewnessinMNEs.In Pro-ceeedingsofDRUID2011.
Ramamurti,R.(2008).Whathavewelearnedaboutemerging-marketMNEs? Insightsfromamulti-countryresearchproject.Emergingmultinationals: Outward FDI from emerging and developing economies. Copenhagen, Denmark:CopenhagenBusinessSchool.
Ramamurti,R.,&Singh,J.V.(Eds.).(2009).Emergingmultinationalsin emerg-ingmarket.CambridgeUniversityPress.
Ramamurti,R.(2012).Whatisreallydifferentaboutemergingmarket multina-tionals?GlobalStrategyJournal,2(1),41–47.
Reddy,P.(2011).Globalinnovationinemergingeconomies.NewYork: Rout-ledge.
Reed,R.,&DeFillippi,R.(1990).Causalambiguity,barrierstoimitation,and sustainablecompetitiveadvantage.AcademyofManagementReview,25, 88–102.
Ringle,C.M.,Sarstedt,M.,&Schlittgen,R.(2014).Geneticalgorithm segmen-tationinpartialleastsquaresstructuralequationmodeling.ORSpectrum, 36(1),251–276.
Roth,K.,&Morrison,A.J.(1992).Implementingglobalstrategy.Characteristics ofglobalsubsidiarymandates.JournalofInternationalBusinessStudies, 23(4),715–735.
Simonin,B.L.(2004).Anempiricalinvestigationoftheprocessofknowledge transferininternationalstrategicalliances.JournalofInternationalBusiness Studies,35,407–427.
Simonin,B.L.(1999).Transferofmarketingknow-howininternationalstrategic alliances:Anempiricalinvestigationoftheroleandantecedentsof knowl-edgeambiguity.JournalofInternationalBusinessStudies,30,463–490.
Szulanski,G.(1996).Exploringinternalstickiness:Impedimentstothetransfer ofbestpracticewithinthefirm.StrategicManagementJournal,17,27–43.
Sørensen,J.B.,&Stuart,T.E.(2000).Aging,obsolescenceandorganizational innovation.AdministrativeScienceQuarterly,45,81–112.
Tabachnick,B.G.,&Fidell,L.S.(2007).Usingmultivariatestatistics(5thed.). Boston:AllynandBacon.
Tenenhaus,M.,Vinzi,E.V.,Chatelin,Y.,&Lauro,C.(2005).PLSpath model-ing.Computationalstatisticsanddataanalysis,48,159–205.
Tsai,W.,&Ghoshal,S.(1998).Socialcapitalandvaluecreation:Theroleof intrafirmnetworks.AcademyofManagementJournal,41,464–476.
Tsai,W.(2001).Knowledgetransferininter-organizationalnetworks:Effects ofnetworkpositionandabsorptivecapacityonbusinessunitinnovationand performance.AcademyofManagementJournal,44(5),996–1004.
Tsai,W.(2002).Socialstructureof“coopetition”withinamultiunit organiza-tion:Coordination,competition,andintraorganizationalknowledgesharing. OrganizationScience,13,179–190.
Tushman, M. L., & Anderson, P. (1986). Technological discontinuities and organizational environments.AdministrationScienceQuarterly,31, 439–465.
Uzzi, B. (1996). Thesources and consequences of embeddedness forthe economic performance of organizations: Thenetwork effect. American SociologicalReview,61,674–698.
Uzzi,B.,&Gillespie,J.J.(2002).Knowledgespilloverincorporate finan-cingnetworks:Embeddednessandthefirm’sdebtperformance.Strategic ManagementJournal,23(7),595–618.
Van Wijk, R., Jansen, J. J. P., & Lyles, M. (2008). A inter- and intra-organizationalknowledgetransfer:Ameta-analyticreviewandassessment ofitsantecedentsandconsequences.JournalofManagementStudies,45(4), 830–853.
Vernon-Wortzel,H.,&Wortzel,L.H.(1998).Globalizingstrategiesfor multi-nationalsfromdevelopingcountries.ColumbiaJournalofWorldBusiness, Spring,23(1),27–35.
Yamin,M.,&Otto,J.(2004).PatternsofknowledgeflowsandMNEinnovative performance.JournalofInternationalManagement,10,239–258.
Yang,Q.,Mudambi,R.,&Meyer,K.E.(2008).Conventionandreverse knowl-edgeflowsinmultinationalcorporations.JournalofManagement,34(5), 882–903.
Zander, U.,&Kogut, B.(1995). Knowledgeandthespeedof thetransfer andimitationoforganizationalcapabilities:Anempiricaltest.Organization Science,6(1),76–92.
