You are on page 1of 28

Knowledge Transfer in Multinational Corporations

Author(s): Dana B. Minbaeva


Source: MIR: Management International Review, Vol. 47, No. 4 (2007), pp. 567-593
Published by: Springer
Stable URL: http://www.jstor.org/stable/40658222 .
Accessed: 25/06/2014 00:19

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Springer is collaborating with JSTOR to digitize, preserve and extend access to MIR: Management
International Review.

http://www.jstor.org

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
mir
mirvol.47,2007/4,
pp.567-593 - ^^ •

MonoQontQiit
ImfHnufKHHH K6VI6W
© Gabler Verlag 2007

Dana B. Minbaeva

Knowledge Transferin Multinational


Corporations

Abstractand Key Results

■ Thispaperdevelopsandtestsa modelthatanalyzesthejointeffect offourdeter-


minants ofknowledge transfer - characteristics
ofknowledge, characteristicsof
bothknowledge sendersandreceivers, andtherelationships betweenthem- on
thedegreeofknowledge transfer fromheadquartersto subsidiaries.
■ Theresults analysischallengetheviewthatthesuccessofknowl-
ofthestatistical
edgetransferis exclusivelya functionofthecharacteristics of thatknowledge.
To fullyunderstandtheprocessofknowledge transfer,itis importantto include
oftheindividuals
characteristics involvedinthetransferprocess as wellas char-
ofthecontextin whichknowledge
acteristics transfer
takesplace.

Keywords
Multinational
KnowledgeTransfer, Corporations

Author
Dana B. Minbaeva,AssistantProfessorin HumanResourceManagement, CenterforStrategic
Management andGlobalization,
CopenhagenBusinessSchool,Copenhagen,
Denmark.
receivedJuly2005,revisedMarch2006,finalrevisionreceivedOctober2006.
Manuscript

mirvol. 47, 2007/4 567

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

Introduction

A reviewoftherecent literature
onknowledge transferinmultinational corporations
(MNCs) showsa tendency tofocusonknowledge transfer as an aggregateconcept.
Indeed,"notwithstanding the criticalityof internalknowledgetransfers within
MNCs ... verylittlesystematic empiricalinvestigation into the determinants of
intra-MNC knowledgetransfers has so farbeenattempted" (Gupta/Go vindarajan
2000,p. 474). Thispaperaimstofillthatgap inthecase ofknowledge transferred
fromtheheadquarters of MNCs to theirsubsidiaries by developingand testinga
modelthatanalyzesthejointeffect offourdeterminants onthedegreeofknowledge
transfer.
Severaltheoreticalorientationshavemadevaluablecontributions totheanalysis
of knowledgetransfer. Amongthemare theresource-based viewof thefirm,the
dynamiccapabilitiesapproach,organizational learningtheoryand theemerging
knowledge-based view. These contributions have been reviewedto identify ap-
to the
proaches understanding knowledge transferprocessandclassifyitsdetermi-
nants.Thereviewidentified over90 determinants ofknowledge transfer.
Usingthe
signalingmetaphor thathasbeenbySzulanski(2000),thedeterminants areclassi-
fiedintofourgroups:characteristics of knowledge, characteristicsof knowledge
senders(disseminative capacity),characteristicsof knowledge receivers(absorp-
tivecapacity),and characteristics of therelationships betweensendersand re-
ceivers.Fourhypotheses addresstherelationship betweeneachofthedeterminants
andthedegreeof knowledgetransfer. Hypotheses are testedusingdatacollected
from92 subsidiaries ofDanishMNCs in 11 countries.
Theresultsofthisstudysupport thethesisthatthesuccessofknowledge trans-
feris notexclusivelya function of thecharacteristics of knowledge;ratherthe
characteristicsof the sendersand receiversthemselves play an important role.
Organizations whose employees have superior skillsand willingness toboth absorb
and shareknowledgeachievesuperior knowledgetransfer results.The degreeof
knowledge transferis even higher when close relationships betweensendersand
receiversareestablished.

Determinantsof Knowledge Transfer:LiteratureReview

Atthemostgenerallevel,twometaphors haveguidedknowledge transfer


research.
Thefirst
seesknowledge as a processofcommunication1,
transfer whilethesecond
viewstransfer intermsofcostandbenefit:
primarily thehigherthecostoftransfer,
willoccur(Attewell1992).Few studiesin theknowledge-
theslowerthetransfer

568 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
in Multinational
KnowledgeTransfer Corporations

Figure1. KnowledgeTransfer:
A SchematicDiagram

y/ ,'' ' ? ! i Knowledge elhit'ieteristies '' J ! ^^V

// Sender -j ¿ Knowledge > !


; |i ► ^> ^> |_| ^ ''
/ / S i¡
! ! y/ y/ > ¡ Receiver ^ '

'^. Oliili iii/alion;il context ,-'/


^'"-^ ReUnionsliips between senders ,-''s^

In Hold Clements o'' know ledge transfer

///
Italies harriersdeterminantsassociated with the lour elements of know ledge transfer

basedviewusethefirst metaphortostructure thesubsequent analysisoftheknowl-


edgetransfer process.Inherreviewofknowledge transferstudies,
however, Argote
(1999) uses itto examinehowcharacteristics oftherelationship betweenorgani-
zations,theorganizations' thefeatures
characteristics, oftheknowledge beingtrans-
ferred, and the dimensions of the transfer process affectthe actual knowledge
transfer.Eisenhardt andSantos(2002) explorehowknowledgetransfer withinan
organization dependsuponthecharacteristics ofknowledge, thesender, therecipient,
and theirmutualrelationships. Szulanski(1996, 2000, 2003) definesknowledge
transferas a processofdyadicexchangesofknowledge betweenthesenderandthe
receiver,where the of
effectiveness transfer to
depends someextenton thedispo-
sitionandabilityofthesourceandrecipient, onthestrength ofthetiebetweenthem,
andonthecharacteristics oftheobjectthatis beingcreated(Szulanski2003). This
paperadoptssimilarapproachandspecifiesthebasicelements ofa transfer:source,
message,recipient and There
context. arebarriersassociatedwitheach ofthenamed
elements (Figure1), which arethe characteristicsofknowledge, characteristics
of
knowledge receivers (absorptivecapacity), characteristics
of knowledgesenders
(disseminative capacity)andcharacteristicsoftherelationshipbetweensendersand
receivers.
Severalconceptualresearchers havecontributed toourunderstanding ofsome
ofthenamedtransfer elements. To takesomeexamples,tacitness as a sourceofam-
biguityhasbeendevelopedbyReedandDeFillippi(1990),thenotionofabsorptive
capacityhasbeenanalyzedbyZahraandGeorge(2002),andtheconceptofknowl-
edge sharinghostility has beenexploredby Hustedand Michailova(2002). Em-
piricalresearchers have examinedtherelationships betweentwoormoreelements
ofparticular transfersandtheiroutcomes.Forexample,somestudieshavefocused

mirvol. 47, 2007/4 569

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

on dependency on thecharacteristics of thetransferred knowledge(e.g.,Zander/


Kogut 1995, Simonin 1999a, 1999b, Pak/Park 2004), knowledge sources(e.g.,Foss/
Pedersen2002), absorptive capacity(e.g.,Lyles/Salk 1996, Lane/Lubatkin 1998,
Lane et al. 2001, Minbaevaet al. 2003, Pak/Park 2004), and the organizational
contextin whichthetransfer takesplace (e.g., Simonin1999a, 1999b,Bresman
et al. 1999,Pak/Park 2004). Few studiesincludeall fourelementsof thetransfer
process (Szulanski 1996, Gupta/Go vindaraj an 2000).
The reviewedstudiesare limitedby at leastfivedimensions. First,theyhave
focusedon selectedfunctional expertise, such as technological, marketing, and
R&D knowledge, ratherthanon generalorganizational knowledge.As Simonin
indicated, "studiesof knowledgetransfer turnalmostexclusivelyto technology
transfer whenempiricalinvestigation is in order"(Simonin1999b,p. 596). Other
of
componentsorganizational knowledge, suchas knowledge ofmanagement systems
andpractices, or internal processeshavebeenlargelyneglectedin themajority of
thereviewedstudies.In thispaper,I followGuptaand Govindarajan (2000), fo-
cusingonthetransfer ofseventypesofknowledge: marketing know-how, distribution
know-how, packaging design/technology, product design,processdesign,purchasing
know-how, andmanagement systems andpractices.
Second,studiesof MNC knowledgetransfer have exploredhow knowledge
transfer within an organization dependsuponvariousdeterminants. However,only
twoofthereviewedempiricalstudiesintroduced andempirically investigated the
nameddeterminants a
using single model. Szulanski's study(1996) offersa com-
prehensive taxonomy ofbarriers tointra-firm knowledge transfer.
Thedeterminants
explored inthatstudy arecharacteristics oftransferred knowledge (causalambiguity
andunproveness), characteristics ofknowledge recipients(lackofmotivation, lack
ofabsorptive capacity and lack ofretentive capacity), characteristicsof knowledge
sources(lackofmotivation andperception as unreliable), andcharacteristics ofthe
context (barren organizational contextand arduous relationships).Gupta and Govin-
darajan(2000) conceptualized knowledgeflows(in and outof a subsidiary) as a
function ofthevalueofthesourceunit'sknowledge stock(modeofentry, subsidiary
size,andrelativeeconomiclevel),motivational disposition ofthesourceunit(in-
centivefocus),existenceandrichness oftransmission channels(formalintegrative
mechanisms andlateralsocialization mechanisms), andabsorptive capacityofthe
target unit.The authorsofbothstudiescall forfurther researchon thejointeffect
of knowledgetransfer determinants. This studyconsidersthejointeffectof four
determinants ina singleeclecticmodel,whichalso makesitpossibletoidentify the
relative importance of each determinant (as was recommended by Szulanski 1996).
A thirdlimitation ofthereviewedstudiesis relatedto reliability andtheper-
ceptualnatureof knowledgetransfer measures.Mostdataon knowledge-related
outcomeshas beencollectedusingperceptual scales and thedata's validitywas
oftenlimitedbytheuse ofonlyone respondent perMNC - usuallya CEO at the
MNCs' headquarters. To minimize theriskofcommonmethodbias,future studies

570 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

maybenefitfromcollectingdataon knowledgetransfer fromotherrespondents,


suchas subsidiaryrepresentatives.Alternatively,
knowledge measurescan
transfer
be testedforinter-rater as
reliability, suggested and
byGupta Govindarajan (2000).
Guptaand Govindarajan obtaineddata on thesame variablesdescribing knowl-
edge-relatedoutcomesfromsupervisors atHQ andtheirimmediate subordinatesat
thesubsidiarylevel.Theythenlookedatthecorrelation betweenthecorporate-level
measuresofknowledge-related outcomes(i.e.,knowledge inflowsandoutflows to
andfrom and
subsidiaries) measures comingfrom thesubsidiaries.
The correlations
wereexpectedto be positiveand significant. Guptaand Govindarajan concluded
thattheinter-rater
reliabilitytestis usefulwhenarguingforthereliability of the
perceptual measures of knowledge-related outcomes,especially when they are
centralto thestudy.In thisstudy,therecommended reliability is
inter-rater test
conducted usingresponses on some from
variables bothHQ and subsidiaries.
Fourth,thereviewedstudiesmaybe criticizedfornotclearlyfixingthelevel
of analysis.Knowledgetransfer withinMNCs occursalongmultipledimensions
and in multipledirections.Since MNCs are complex,multi-dimensional entities,
thelevelofanalysisandtheapproachchosento analyzethedeterminants ofMNC
knowledge transfermustbe specified.GuptaandGovindarajan (2000) arguethat
knowledge transferwithinMNCs mightbe studiedon at leastthreelevels:
• Nodal,whentheanalysisfocuseson thebehaviorofindividualunits,
• Dyadic,whentheanalysisfocuseson thejointbehaviorofunitpairs,and
• Systemic,
whentheanalysisfocuseson thebehavioroftheentirenetwork.
GuptaandGovindarajan decidedto limittheirinvestigationtothenodalleveland
concentrateon thebehaviorof theorganizational units.Giventherelativedearth
ofpreviousempiricalworkon thesubject,theyarguedthatit wouldbe usefulto
first
concentrate atthenodallevelofanalysis- "thesimplest levelfeasible"(Gupta/
Govindarajan 2000,p.49 1), andlaterconsider thedyadicandsystemic
investigating
levels.Thisstudyadaptsa similarapproach.
theresultsoftheempiricalstudieshavebeencontroversial.
Fifth, The lackof
consistency in theresults may be attributed
to the factthatthe conceptualization
andoperationalization ofthenameddeterminants differ
substantially fromonestudy
to another.As an example,one maycomparetheoperationalization ofone of the
determinants - thecharacteristicsofknowledge receivers.Szulanski(1996) devel-
opeda construct measureforabsorptive capacityconsisting ofninemeasuresthat
the
capture ability to value, assimilateand apply new knowledge.Amongthem
were:theexistenceof a commonlanguagebetweenrecipients, a vision,shared
information, a divisionof rolesand responsibilities, ability solveproblems,
the to
and thenecessaryskillsand competencies (CronbachAlpha0.83). Accordingto
GuptaandGovindarajan (2000),absorptivecapacityis anorganizational-levelphe-
nomenon anda function oforganization'sfamiliaritywiththe incoming knowledge
("thefirm'slevelofpriorrelatedknowledge" inCohenandLevinthal1990,p. 128)

mirvol. 47, 2007/4 571

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

and inter-unit homophily ("thedegreeto whichtwoor moreindividualswhoin-


teractaresimilarin certainattributes, suchas beliefs,education, social statusand
thelike"inRogers1995,p. 18). Absorptive was
capacity operationalized interms
oftwovariables:modeofentry, andtheproportion oflocalnationalstoexpatriates
withinthe subsidiary'stop management team.This exampleof a significant
differencein operationalization illustrates thattheconceptualization of variables
describing knowledge transfer is far from unified.
Severalstepsweretakento selectthecentraldeterminants forthisstudy.First,
representative empiricalstudies (including those that investigatedthenameddeter-
minants in a singlemodel)werereviewedandtheirconclusions on thepreviously
identified
determinants ofthetransfer weresummarized. Thefollowing studieswere
reviewed:Zanderand Kogut(1995), Szulanski(1996), Lyles and Salk (1996),
Moweryet al. (1996),Lane andLubatkin(1998),Bresmanet al. (1999), Simonin
(1999a,1999b),Laneetal. (2001),Subramaniam andVenkatraman (2001),Fossand
Pedersen(2002),Minbaevaetal. (2003),Schulz(2003),Pak andPark(2004),Cho
andLee (2004),andBjorkman et al. (2004). Onlythosedeterminants thatcouldbe
categorized as barriers
on the signaling metaphor - the characteristics
of knowledge,
sendersandtheirrelationships
receivers, - wereselected.Second,fromthechosen
determinants, thosethatconsistently provedsignificant forknowledge relatedout-
comeswereidentified, as werethoseforwhichcontroversial resultswerereported.
determinants
Finally, thatweredefined conceptually butwerenotstudied empirically
ordidnothaveanysupport werealsonoted.Forexample,ofall theknowledge char-
acteristics,
only tacitnessshowed a consistently significant negativeeffecton knowl-
edgetransfer.2 The resultson complexity andspecificity variedamongthestudies.
Availabilitywasdefined theoretically (Winter 1987)butwasnotverypopularamong
theempiricalresearchers.3 In thisstudy,all fourknowledgecharacteristics were
chosenas central determinants ofknowledge transfer.Similarreasoning wasbehind
theinclusion oftheabilityandmotivation ofknowledge receiversandsenders. Among
thedeterminants reflecting characteristics ofknowledge sendersandreceivers, only
theabilityofknowledge receivers was previously proventobe significant. Further-
more,theresults on themotivation ofknowledge receivers werecontroversial.
Inthefollowing sections, each offour determinants ofknowledge transfer (char-
acteristics
of knowledge, characteristics of knowledgesenders,characteristics of
knowledgereceivers,and characteristics of the relationsbetweenknowledge
sendersandreceivers) areconceptualized andhypotheses aredevelopedaccording
totheireffect on thedegreeofknowledge transfer.

Natureand Characteristics of Knowledge

Alongwithcomplexity theconceptof tacitknowledgehas been


and specificity,
central
totheempirical
literature
onknowledge stickiness.
Itderivesfromthework

572 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

ofPolanyi,whoarguedthat"we can knowmorethanwe can tellandwe can know


nothing without relying uponthosethings whichwemaynotbe abletotell"(Polanyi
1966,p. 4). On this view, tacitknowledgeneednotbe articulated in orderto be
transferred. Theindividual mustsimplypossesstheappropriate toolstounderstand
information fromtheoutsideworld.Thisargument has beentakenfurther to sup-
port the claim that all knowledge can be transferred - it is just a matter of costs
(Nelson/Winter 1982).Many researchers see tacitness as a main source of ambiguity
(Reed/DeFillippi 1990,Szulanski1996,Simonin1999a,1999b)andtherefore count
itamongthebarriers toknowledge transfer alongwithcomplexity andspecificity.
In identifying thisbarrier moreprecisely, twogeneralapproaches havebeentaken.
Some,likeWinter (1987) andBresmanet al. (1999), haveunderstood tacitnessin
termsof therelativearticulability of knowledge,whileothers,like Zanderand
Kogut(1995),havetakenittodenotetherelativecodifiability ofknowledge. Tac-
itnessis thusdefinedintermsofhowdifficult itis to articulate andcodifya given
domainofknowledge.
In general,tacitnesshas beenfoundto havea negativeimpacton knowledge
transfer.Forinstance, ZanderandKogut( 1995) established thatthedegreeofknowl-
edgearticulation and the of
difficulty teachingcapabilitiesinfluence thespeedof
capability transfer. Theyfoundthata higher degreeoftacitness decreasesthespeed
oftransfer sincetacitknowledge is hardtoarticulate withformal languageorexpress
directly.Simonin (1999a, 1999b) addressed the role of tacitness bysuggesting that
thedegreeoftacitness influences knowledge transfer outcomes its
through impact
onknowledge ambiguity. He foundtheeffect oftacitness onambiguity, andthereby
on knowledge transfer, tobe consistently significant.
The secondcharacteristic of knowledgeis complexity. Accordingto Winter
(1987), "the complexity /simplicity dimension has to do with theamountofinfor-
mationrequired tocharacterize theitemofknowledge in question"(Winter1987,
p. 172). Complexity resultsfromhavinga largenumberof interdependent skills
andassets(Reed/DeFillippi 1990).Therearea limited number ofempirical findings
oncomplexity andknowledge transfer. Amongthemis a studybySimonin(1999a,
1999b), who found that is
complexity negatively relatedto knowledgetransfer.
to
According him,complexity refers to the number of "interdependent technolo-
gies,routines, individuals, andresources linkedtoa particular knowledge"(Simonin
1999b,p. 600). Overall,complexknowledge is expectedtobe difficult to transfer
sinceit drawsuponmultiplekindsof interrelated competencies - the higherthe
of
degree interdependency, thelower the of
degree knowledge transfer.
Originally, thethirdcharacteristic - specificity - referred to transaction costs
assetspecificity. ReedandDeFillippi(1990) definedspecificity as transaction-spe-
cificskillsandassetsthatareutilizedinproduction processesandintheprovision
ofservicesforparticular customers. Simonin(1999b)considered specificity innar-
rowerterms,operationalizing it as durableinvestments in specializedequipment
and facilities, and in skilledhumanresources.He foundspecificity insignificant

mir vol. 47, 2007/4 573

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

Figure2. ConceptualModel

ofknowledge
Characteristics
I H,(-)

Characteristics
ofknowledge H2(+) .
receivers
(absorptive
capacity) ^Q ee ^
knowledge
Tj /,x transfer
Characteristics
ofknowledge "3 'T)
senders(disseminative
capacity) I

oftherelationships
Characteristics H ,+x '
betweenknowledgesendersand
receivers

and suggestedthat"theconstruct's lackofeffectneedsto be further investigated


forothertypesofcompetencies" (Simonin1999b,p. 612). Followingthisadvice,
specificity in thisstudyreflects thedegreeto whichknowledgeis aboutspecific
functional expertise.Organizations perform differentfunctional activities,and
specific functional is and
knowledge developed integrated aroundthese activities.
Specificity shouldbe positivelyrelatedtointernal knowledge transfer,sinceMNC
unitsareoftenintegrated vertically aroundthefunctions theyperform. Zanderand
Kogut (1995) used thedimension to
"systemdependence" capture a similar char-
acteristicofknowledge. Thisincluded itemsrelatedtothedegreeofmanufacturing's
dependenceon otherfunctions, and theyhypothesized "systemdependence"as
being negativelyrelatedto the probability of transfer. In Winter(1987), the
dimension "an elementof a systemvs. independent" is relatedto thespecificity
dimension. Overall,functionalknowledge (suchas production,marketing, andtech-
nologicalknow-how) should be able to "stand alone" withoutbeing partof the
a
interrelated knowledge systemand should be easierto transfer.
In thisstudy, onedimension was addedtothelistofknowledge characteristics
influencing thedegreeofknowledge transfer- thecharacteristic
ofknowledge that
refersto the"notobservablein use vs. observablein use" dimension in Winter's
taxonomy (Winter1987)."In general,thequestionatissueinvolvestheopportuni-
tiesthatuse makesavailableto someonewhowishesto discovertheunderlying
knowledge"(Winter1987,p. 172; emphasisadded).Availability does notneces-
sarilymeantacitness. Knowledgemaynotbe articulated butnewpersonnel maybe
briefedaboutknowledgesourcesso thattheycan getsuchknowledge"uponre-
quest".On theotherhand,knowledge maybe keptinexplicitform(blueprints) but

574 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

employeesmayhesitateto sharesuchdocuments withnewcomers forvariouspo-


liticalreasons(highknowledgehostility). is expectedto be
Overall,availability
positivelyassociatedwiththedegreeofknowledge transfer.
Tacitness,complexity, and availabilitycomprisea firstgroupof
specificity
independent conceptualmodel(see Figure2). It is
variablesin themulti-faceted
expected that:
L The higherthe degreeof knowledgetacitness,complexity,
Hypothesis non-
and the
specificity non-availability, lowerthe of
degree knowledge
tothesubsidiary.
transfer

Characteristics of Knowledge Receivers

Animplicit consensusexistsabouttheimportance ofknowledge receiverbehavior


withrespect totheabsorption oftransferred The
knowledge. inability ofknowledge
receivers toabsorbnewknowledge (low absorptivecapacity) is one ofthemostof-
tencitedimpediments tointernal
knowledge transfer (e.g.,Cohen/Levinthal 1990,
Lyles/Salk 1996,Szulanski1996,Lane/Lubatkin 1998,Gupta/Go vindarajan 2000,
Lane et al. 2001). CohenandLevinthaldefineabsorptive capacityas the"ability
to recognizethevalueof newexternalinformation, assimilateit,and applyit to
commercial ends"(Cohen/Levinthal 1990,p. 128). Theyassumethatabsorptive
capacitytendsto developcumulatively, is path-dependent and buildson existing
knowledge: "absorptivecapacityis morelikelytobe developedandmaintained as
a byproduct of routineactivitywhentheknowledgedomainthatthefirmwishes
toexploitis closelyrelatedtoitscurrent knowledge base" (Cohen/Levinthal 1990,
p. 150).
Debates on the conceptof absorptivecapacityand its development have
emerged.The subjecthas beendiscussedin suchfieldsas strategic management
(e.g.,Lane/Lubatkin 1998),R&D andinnovation (e.g.,Mowery/Oxley 1995),the
resource-based view(e.g.,Laneetal. 2001),andorganizational learning(e.g.,Kim
2001). The resultsimplythattheabsorptive capacityof knowledgereceiversis a
majordeterminant oftheknowledge transferprocess:thegreater theabsorptive ca-
the the of
pacity, greater degree knowledge transfer
(e.g., Szulanski 1996,Lyles/Salk
1996,Gupta/Govindarajan 2000).
FollowingCohenandLevinthal(1990) andKim(2001),thisstudydefinesab-
sorptive capacityas havingtwoelements: priorknowledge andintensity ofeffort.4
Priorknowledgeincludesbasic skills,a sharedlanguage,relevantpriorexperi-
ence and up-to-date informationon knowledgedomains(Cohen/Levinthal 1990,
Szulanski1996,2003).Thetermrefers totheexisting individual unitsofknowledge
availablewithin theorganization(Kim2001). Employees needtohavecombinations
ofskillsthatenablethemtofind,acquire,manage,share,andapplyknowledge that
theorganization needs.

mir vol. 47, 2007/4 575

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

The secondelementofabsorptive capacity,as proposedbyKim(2001), is the


intensity of effort.Employees'intensity of effortis wellunderstood bycognitive
processtheories, such as theexpectancy theory of work motivation(Vroom1964).
Overall,motivated employeeswantto contribute to organizationalperformance.
Eventhoughtheorganization mayconsistofindividuals withsignificant learning
abilities,theorganization's abilitytoutilizetheabsorbedknowledge willbe low if
employeemotivation is low orabsent(Baldwinet al. 1991).
Bothaspectsof absorptive capacity(abilityand motivation) mustbe present
to optimally facilitate
theabsorption of knowledgefromotherpartsof theMNC
(Minbaevaetal. 2003). As Vroomstated,"moreis tobe gainedfromincreasing the
motivation ofthosewhoarehighin abilitythanfromincreasing themotivation of
thosewhoarelow in ability...Moreis gainedfromincreasing theabilityofthose
whoare highlymotivated thanfromincreasing theabilityof thosewhoare rela-
tivelyunmotivated" (Vroom1964,p. 203). Accordingly, bothaspectsofabsorptive
capacity- abilityand motivation ofemployees - should be highin ordertofacili-
tateknowledge transfer.
Theabilityandmotivation ofknowledge receivers thesecondgroup
represents
of independent variablesin theconceptualmodelpresented Figure2. It is ex-
in
pectedthat:
2. Thehighertheabilityandmotivation
Hypothesis ofthesubsidiary's
employees
toabsorbknowledge (absorptivecapacity), higher degreeof
the the
knowledgetransferto thesubsidiary.

Characteristics of Knowledge Senders

To date,researchon knowledgesenders'behaviorin relationto knowledge trans-


ferhas been largelytheoretical and case-based(e.g., Husted/Michailova 2002,
Michailova/Husted 2003,Cabrera 2003, Lindsay etal. 2003,Davis etal. 2005). Al-
the
though importance ofthisdeterminant hasbeen illustrated,substantial empirical
supportis largelyabsent.The notableexceptions arestudiesbySzulanski(1996),
Simonin(1999),andGuptaandGovindarajan (2000). However,thesepresent dif-
conclusions
ferent onthebehaviorofknowledge senders, whichis hardly surprising
sincetheyhadreachedno consensuson theappropriate definitionandmeasureof
theconcept. InGuptaandGovindarajan (2000), the source'smotivational disposition
is measuredas theextentto whicha subsidiary president's bonusis network-fo-
cusedrather Simonin(1999a, 1999b)measured
thansubsidiary-focused. theextent
to whichknowledgeprovidersprotecttheircompetencies usingtwo items:(1)
whether thepartner hasintentionalprocedures, routines, and policies torestrictthe
sharingof relevantinformation its
concerning marketing skillsand know-how; and
(2) whether thepartneris veryprotectiveof itsmarketing skillsand know-how.

576 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

Szulanski(1996) developeda complexmeasureof 13 itemsto capturea source's


lackofmotivation to shareknowledge.
In thissection,thisconstruct is re-operationalized. In particular, it is argued
thatthedecisiontotransfer knowledge is largelyindividual andis drivenbyatleast
twobehavioral factors:theabilityandthewillingness ofknowledge senderstoshare
knowledge.
Cabrera(2003) reviewedthemainsociologicaland psychological theoriesto
identifyfactors related toknowledge-sharing behavior (see also Cabrera et al. 2006).
She identified ninefactors thatcouldinfluence suchbehavior.Trustwas foundto
positivelypredict knowledge sharing withinand betweenworkunits.A feelingof
obligation to share was
knowledge positively related totheknowledge-sharing be-
haviorofindividuals. Normsthatencourageopenexchangesofknowledge among
organizational members willleadtoa greater degreeofknowledge sharing. A strong
sense of groupidentity also influencesindividualknowledge-sharing behavior.
Furthermore, individuals willbe likelyto sharetheirknowledge iftheyperceivea
clearbenefit (reward)fordoingso. Thus,theperceivedcostofsharing knowledge
is positivelyassociatedwiththeknowledge-sharing behaviorof individuals. The
perception thatothersarewillingto sharetheirknowledge(reciprocity) is an im-
portant factor indetermining whether anindividual choosestosharehis/her knowl-
edge with others - one will only be motivated to share knowledge ifone believes
thattheparticular pieceofknowledge is worthsharing. Therefore, beliefsregarding
variousindividual competencies andskillsarepositively associatedwithindividual
knowledge-sharing behavior.Personality traits- particularly extroversion, agree-
ableness,conscientiousness, andopenness- arepositively associatedwithknowl-
edge-sharing behavior. According toCabrera(2003), normsareexpectedto influ-
enceknowledge-sharing intentions directly,whiletheotherfactors areexpectedto
influence knowledge-sharing intentionsindirectly through theirimpacton knowl-
edge-sharing attitudes.
Otherrecentcontributions havebeenmadebyHustedandMichailova(2002)
andMichailovaandHusted(2003). Theyarguethatknowledgesenders'behavior
dependsontheirwillingness toshareknowledge withotherorganizational members
on request.The decisionnotto shareis individual, andis oftenrationalandwell-
justifiedfrom theperspective oftheknowledge sender(although thatis notpreferred
anddesiredfrom anorganizational perspective). Husted and Michailova (2002) out-
linesix reasonsforknowledgesenders'hostility towardssharingknowledge:

• Potential
lossofvalue,bargaining power,andprotection
ofindividual
competitive
advantageduetoa strong ofpersonalownership
feeling oftheaccumulated,
"hard
won"knowledge.
• Reluctancetospendtimeon knowledge Knowledgesendersmaynotbe
sharing.
in knowledgesharingsincethetimeand resourcesspenton it could
interested
be investedin activities
thataremoreproductivefortheindividual.

mirvol. 47, 2007/4 577

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

• Fearofhosting"knowledge parasites".Knowledgesendersmaybe reluctant to


sharetheirknowledge withsomeonewhohasinvested littleornoeffort
inhis/her
ownknowledge development.
• Avoidanceof exposure.By notsharingknowledge, individualsprotectthem-
selvesfromexternal assessments ofthequalityoftheirknowledge.
• Strategyagainstuncertainty. totheuncertainty
Due oftheknowledge receiver's
and
perception interpretation of shared knowledge,
knowledge sendersmaybe
highlycautious aboutrevealingthe relevant
knowledge.
• Highrespectforhierarchyandformal power.Knowledgesenders maybereluctant
to sharecrucialknowledge forfearoflosinga positionofprivilegeandsuperi-
ority.
MinbaevaandMichailova(2004) termthebehaviorofknowledge sendersas "dis-
seminative capacity".Theyargue thatability and of
willingness organizational
actorsto sharetheirknowledgeare crucialto thesuccessof knowledgetransfer.
Valuableknowledge is oftentacitin nature.Transferring tacitknowledge requires
teaching(Winter1987). Moreover, knowledgesharingis markedby different in-
terpretationsof thesame idea, false and
starts, disruptions(Zellmer-Bruhn 2003).
Therefore,knowledgesendersshouldhave well-developed abilitiesto articulate
andcommunicate knowledge. These abilitiescouldbe acquiredthrough education,
observation,
training, andinvolvement. On theotherhand,knowledge sendersmay
be capablebutunwilling toshareknowledge forthereasonsoutlined byHustedand
Michailova(2002) andMichailovaandHusted(2003). Thegreater an individual's
influenceontheworkcarriedout,howitis doneandbywhom,thegreater thesense
ofresponsibility theindividual tendstofeelforthese decisions and thegreater com-
mitment knowledge sendersexhibit.
The abilityandmotivation ofknowledge sendersformthethirdgroupofinde-
pendent variables in Figure 2. It is expected that:
of theknowledgesendersto
3. The highertheabilityand motivation
Hypothesis
shareknowledge capacity),thehigherthedegreeof
(disseminative
knowledgetransferto thesubsidiary.

Characteristics of the Relations between Knowledge Senders and Receivers

Ingeneral, knowledge
transferring acrossorganizational forvarious
unitsis difficult
reasons:a unit'sownexperience is likelyto be more relevant than theexperience
mayexistincapabilities,
ofotherunits;differences structure
culture, ortechnology;
andthereareinherent intheexperience
differences withknowledge transfer(Argote
1999).Thus, intra-MNC knowledge transferis possibleonly when close relation-
shipsareestablishedbetweensendersandreceivers. Theremustbe communication
for
bridges,possibilities dialogue acrossorganizational hierarchy, conditions for

578 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

teamlearning, and systemsto captureand sharelearningwithintheorganization


(Levitt/March Senge 1990,Argyris/Schon
1988, 1996).
Thereare severalempiricalstudieson internal MNC knowledgetransfer that
support theseideas.Szulanskiclaimedthatsince"intrafirm exchangesof knowl-
edgeareembedded inanorganizationalcontext" (Szulanski1996,p. 3 1),"a transfer
of knowledge, especiallywhentheknowledgetransferred has a tacitcomponent,
mayrequire numerous individual exchanges"(Szulanski1996,p. 32). Bresmanet
al. (1999) showedthatinterpersonal communication, suchas visitsandmeetings,
weresignificant ofinternational
facilitators knowledge transfer.GuptaandGovin-
darajan(2000) considered notonlytheexistenceof communication channelsbut
also therichnessof communication links,capturedas informality, openness,and
density of communication. The resultsprovided almost completesupportforthe
prediction thattheexistence andrichnessoflateralinter-unitintegrationmechanisms
(channelslinking a focalsubsidiarytotherestofanMNC) arepositively associated
withtheknowledge transferintothatsubsidiary.The findings arein linewithpre-
viousworkofGalbraith (1973) andEgelhoff (1993).
Hansen( 1999) concludedthattheabsenceofdirectrelations andextensive com-
munication amongpeoplefromdifferent departments inhibitsknowledge transfer,
whilestrong inter-unit
relationsfacilitate
it.In hisrecentworkon knowledgenet-
works(Hansen2002),headvancedhisprevious findingsbyfocusing onthequestion
of whysomebusinessunitsare able to benefitfromknowledgeresidingin other
partsofthecompany whileothersarenot.He hypothesized thattheextentofdirect
relationsintheknowledge network mayimpactthedifficulties oftransferring non-
codifiedknowledge."The directrelationsin the knowledgenetworkprovide
immediate accessandhenceareespeciallyusefulfora focalteaminquiring about
opportunities" (Hansen2002,p. 234). He found thatunits thatwere more involved
intheirrespective knowledge network wereable to acquiremoreknowledge from
otherdivisionsinthenetwork. Accordingly, itis expectedthat:
Hypothesis4. Thehigherthedegreeofinvolvement ofthefocalsubsidiary
innet-
workrelations
withotherMNC units,thehigherthedegreeofknowl-
edgetransfer
to thesubsidiary.
The characteristics
oftherelationships
betweenknowledgesendersandreceivers
form thefourth
groupofindependent variablesinthemulti-faced
conceptualmodel
presentedinFigure2.

Data

TheHermesCD DirectfromKOB (Kobmandstandes OplysningsBureau)wasused


toconstruct
thedataset.5Thedatabasequerywas initiated thosefirms
byselecting

mir vol. 47, 2007/4 579

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

thatwereheadquartered inDenmark, andthenthesamplewasreducedtothosethat


hadtwoormoresubsidiaries abroad.Theprocedure resulted ina listthatwascross-
checkedwiththeBörsen5006to ensurethatthepopulationwas as completeand
relevant as possible.In a fewcases thecomparison provedusefulgiventhatsome
firmsomitthefactthattheyareDanish-owned in theirKOB profile.However,as
Denmarkhasrelatively fewlargeMNCs,thislistis comprehensive intermsofthis
study. the
Limiting sample on thebasisof turnover or employees provedineffective,
as manyMNCs havetheirparententitiesconstructed as a separateunitwithfew
employees andlittleturnover. Conglomerates, suchas NovoNordisk, wouldthere-
forebe excludedifthesearchwas limitedtofirmswith100+ employees.
The MNCs includedin thesamplewerefurther limitedto thosewhosesub-
sidiariesemploymorethan30 employees, as small-scalecompaniesingeneral, and
smallsubsidiaries in particular, do notutilizea widerangeof formalHRM prac-
tices(Miner/Crane 1995).Atthisstage,aneffort wasmadetoscanthechosenMNCs
andeliminate thosethatdidnotfulfill thecriteria of"atleasttwosubsidiaries with
morethan30 employees each".Thecompaniesweretherefore contacted viaphone,
theirwebsitesandannualreports werescanned,andtheresultsofthepreviouspro-
jectswereexamined.
Intotal,358DanishMNCswereincluded intheoriginal sample.Eachcompany's
headquarters was contacted with a request for the contact information forat least
twosubsidiaries withmorethan30 employees thatcouldbe approached. Sometime
was allowedbeyondthereturn datebeforefollow-up procedures were initiatedby
telephone.
49 companies werethenexcludedfrom thesurvey forvariousreasons,including
closureof thesubsidiary, bankruptcy, mergers, or theirstatusas a smallfamily-
ownedcompany oronewithlessthan30 employees inthesubsidiaries. Thesample
sizetherefore decreased to 309 MNCs. 58 completed were
questionnaires received,
while109 MNCheadquarters refused toparticipate inthesurvey. Amongthereasons
fornon-participation were"notinterested", "do nothavetime","matter ofinternal
policy", and "under serious organizational change". 142 MNCs headquarters did
notrespond.
Additional subsidiary contacts wereobtainedfromtheforeign commercial sec-
tionsof theDanishEmbassies.Detailson 255 subsidiary contactsfromthe 142
MNCswhohadnotresponded wereobtainedviatheDanishEmbassiesinGermany,
Sweden,USA, China,andRussia.The majority ofthesubsidiaries includedinthe
sample were located in Germany, Sweden and UK. According to the ForeignSub-
sidiariesSurvey(FAS) survey(DanskIndustry 2003), due to the predominance of
smallandmediumsizedcompanies, DanishMNCs concentrated their investments
inNordiccountries, theUK andGermany, wherethecultural, linguistic andregu-
latorybarriers arerelatively modest. Further, according tothe latter,Danish MNCs
primarily choose to set up businesses in the countries, which takethe majority of
Danishexports. ThatmakestherestoftheWestern Europe and North America the

580 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

secondmostpopulardestination ofDanishsubsidiaries. The samplealso includes


subsidiaries locatedin France,Spain,Portugaland USA. Furthermore, according
to the2003 surveyoftheConfederation ofDanishIndustries, EasternEuropeand
Asia arecurrently important areas.Despiteofthesignificant barrierssuchas local
legislation,lack of cooperation withdomesticauthorities, languageand cultural
differences, thecommitment of DanishMNCs in theseregionsis quiteconsider-
ableandgrowing rapidly.Russia andChinawerechosentorepresent theseregions
(Dansk Industri 2003).
The finaldatasetconsistedof 305 Danishsubsidiaries. Questionnaires were
addressedto theHRM Manager/General Manager of the focal If
subsidiary. the
approached manager was unable to the
complete survey, he/she could forward the
questionnaire to a senior or middle managerwith sufficient knowledgeregarding
thethemesofthisstudy.
A web-basedsurveywas chosenfordatacollectiondue to thetimeand cost
considerations. The respondents wereapproached bya coverlettersentvia email,
whichexplained thepurposeofthesurvey, detailedtheresearch processandanalysis
procedures, offered follow-upreports and related workingpapers,and provided
straightforward directionscompleting thequestionnaire. Inaddition,a websitewas
established to backup thesurvey.Respondents wereinvitedto visittheweb site
andreadmoreonthesurveysubjectsandrelatedthemes. A linktothequestionnaire
was providedwithinthetextofthecoverletterandthesurveywas onlyavailable
through thatlink,whichdecreasedtheriskofpotential error.
The strategy outlinedaboveresulted inachievinga responserateof30 percent
(92 outof305 subsidiaries). 20 responding subsidiaries werelocatedin Germany,
17 in theUSA, 15 in Russia,14 in China,10 in Sweden,6 in theUK, 6 in France
andone eachin SriLanka,Philippines, SpainandPortugal.The responseratesin
variouscountriesreflectsthe generalgeographicaldistribution of Danish sub-
sidiariesabroad.
As explainedearlier,subsidiaries wereincludedinthesampleonlyiftheyem-
ployed more than 30 employees. 30 outof92 subsidiaries employed morethan100
employees. More than half of thesubsidiariesin Russiawerelarge.25 percentof
thetotalsubsidiary samplerepresented themanufacturing sector,andthemajority
ofproduction subsidiaries werelocatedinUSA andAsia.Therestofthesubsidiaries
weremainlysales andmarketing-oriented, whichis notsurprising, sincesales ac-
tivities
arestilldominant inall Danishsubsidiaries abroad.However,according to
the2003 surveyofDanishSubsidiariesin ForeignCountries, thereis a switchto-
wardsforeign production activities,supportedbytheincreased investments inEast-
ernEuropeandAsia.Thesurveyalso indicated that,inrecentyears,DanishMNCs
haveincreased thenumber ofR&D activities locatedabroad.In 1996,thedistribu-
tionofworkers insubsidiaries was:management, administration andIT- 14percent;
of
production goods and services - 41 percent;sales and marketing - 44 percent;
andR&D - 1 percent. In 2002,thepicturewas different: management, administra-

mirvol. 47, 2007/4 581

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

tionandIT - 13 percent; ofgoodsandservices- 37 percent;


production sales and
- 46
marketing percent; and R&D - 4 percent.The conclusion of theFAS survey
was thatmostofthemultinational R&D efforts are stilllocatedin Denmarkor in
neighboring Amongthesubsidiaries
countries. includedinthisdataset,therewere
onlyfivesubsidiarieswhereR&D comprised morethan15 percentof activities:
twolocatedinChina,one in Russia,one in Swedenandone in theUK.
The subsidiarieswereestablished through variousmodesof entry. Onlyone
thirdof thesamplewereGreenfield. The Europeansubsidiaries wereownedby
sharedcapital,whiletheresthad a majority of foreigncapitalin theownership
packages.All subsidiarieshadsomeexperience in working internationally,
except
thoselocatedin Sweden.

Measures

Degree of Knowledge Transfer(DoKT)

Thedegreeofknowledge transferwas definedinthequestionnaire as theextentto


whichsubsidiary employees received knowledge transferredto their subsidiary
fromtheMNC headquarters.
wasadoptedfrom
Theoperationalization GuptaandGovindarajan (2000). Data
was collectedon thefollowingitems:marketing know-how, distributionknow-
how,packagingdesign/technology, productdesigns,processdesigns,purchasing
know-how, andmanagement systemsandpractices.Subsidiary respondents were
askedto evaluatethe of
degree knowledge from
transferred headquarters their
to
foreachaspectusinga five-point
subsidiaries scale,where1 indicated
Likert-type
verylow use of knowledgeby thesubsidiary and 5 indicatedsubstantialuse of
knowledge.Responses were to a
averaged yield composite index reflecting the
of
degree knowledge transferred
to thefocal The
subsidiary. Cronbach Alpha was
0.77.

Characteristics of Knowledge (KnChar)

forthecharacteristics
Themainsourcesofinspiration ofknowledge measureswere
thetaxonomicdimensionsofWinter(1987) and studiesby Zander (1991), Kogut
andZander(1992),andZanderandKogut(1995).
Respondentswereaskedto clarifytheextentto whichMNC organizational
knowledgecould as (1) easyto codify,i.e., information
be characterized is often
in and
manuals, procedures;
provided blueprints, (2) complex,i.e., knowledgeis

582 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

abouthighlyinterdependent routines,individuals, and technologies;(3) specific,


i.e.,knowledge in itscontextis aboutspecificfunctional expertise; (4) avail-
and
able,i.e.,always availablefor and easy accessibleby new personnel.Respondents
wereaskedtoindicatetheirperception foreachoftheseitemson a 5-pointLikert-
typescaleranging from"tono orlittleextent"to "a verygreatextent".Theunder-
standing of theterms was checkedduringthepilotstage.Additional explanations
andexamplesweregivenon thesurvey'sdesignedwebsitetoensureequal under-
standing amongall respondents.
To yielda compositeindexforknowledge fourmeasureswere
characteristics,
averaged:non-codifiability (reverse to "easy to codify"),complexity, non-speci-
ficity(reverseto "specific"),and non-availability (reverseto "available").The
CronbachAlphawas 0.637.

Characteristics of Knowledge Receivers (AbsCap)

Sevenitemswereusedtocapturethisconstruct, fiveofwhichwereadoptedfrom
a previousstudy(Minbaevaet al. 2003). Respondents wereaskedtoevaluatesub-
sidiaryemployees'job-related abilitiesandoverallcompetencies,andjob-related
motivation, involvement andjob satisfaction.
Twoitemsmeasuring theabilityand
of
willingness subsidiary employees to absorbnew knowledge were added,which
werenotmeasuresofindividual motivation butratheroftheoverallwillingnessof
subsidiary employees toabsorbknowledge. Allquestionsuseda 5-pointLikertscale
ranging from"verylow" to "outstanding". A comparatively highCronbachAlpha
(0.79) allowed theuse ofthecomposite indexforabsorptivecapacityofknowledge
receivers.

Characteristics of Knowledge Senders (DisCap)

Thisconstructcapturesability andwillingnessofknowledge senderstoshareknowl-


edge.Respondents wereaskedtoappraisetheabilityandmotivation ofknowledge
sendersat headquartersto transfer knowledgeto their In
subsidiary. thiscase, the
useoftheperceptual measurewas necessary (Minbaeva2004). Thetheoretical rea-
soningbehind thisconstructis attitudinal
and If
perceptual. knowledge senderswere
askeddirectly abouttheirbehaviorregarding knowledgesharing, theirresponses
wouldbe unreliable. No one wouldopenlyacknowledge thattheydo notwantor
arenotable to shareknowledge. Sincethisstudyfocuseson theiractualbehavior,
knowledge senders'subordinates wereaskedabouttheirperception ofknowledge
senders'abilityandmotivation to shareknowledge.
The questionsused a 5-pointLikertscale rangingfrom"verylow" to "out-
standing".The CronbachAlphawas 0.66.

mir vol. 47, 2007/4 583

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

Characteristics of Relations between Knowledge Senders and Receivers


(NetMNC)

One itemwas usedtocharacterize theclosenessofrelationships betweenthefocal


subsidiary andtherestoftheMNC. Subsidiary respondents wereaskedtoindicate
theirsubsidiaries'degreeof involvement in MNCs network, usinga Likert-type
scale of 1 (verylow) to 5 (verystrong).
To dealwithreliability ofmeasures,theinter-ratereliabilitytest8wasconducted
priortothe analysis(Gupta/Go an
vindaraj 2000).Responses on some variableswere
obtainedfromboththeMNC headquarters andtheiroverseassubsidiaries. One of
thequestionnaires was senttotheheadquarters ofDanishMNCs andaddressedto
a SeniorHRM Manager/Personnel Director,whowas askedtonamea subordinate
(an HRM Manager/Personnel Directorofthe subsidiary), withwhomhewasclosely
collaborating. In turn,thepersonindicatedwas approachedwitha similarques-
tionnaire. In 21 cases (about20 percent ofthetotalpopulation), twogroupsofre-
spondents answered the same questionsabout receivers'ability motivation
and to
absorbknowledge. the
Although subsidiary hada
respondents tendency to ratethe
abilityand motivation of knowledgereceivershigherthantherespondents from
headquarters, theoverallcorrelation betweentheratingsfromHQ andtheirrela-
tivesubsidiaries arepositiveandsignificant9.The significant correlationsbetween
measuresofreceivers'abilityandmotivation (p < 0.01 and p < 0.05 respectively)
givean estimate ofthereliability(orconsistency) betweentheraters.

Results

Descriptive data(meanvalues,standard minimum


deviation, andmaximum values)
on all variablesareprovidedinTable 1. The correlation forall variablesis
matrix
presented in Table2. In thecorrelation
matrix,someassociationsbetweenthein-
dependent variables
are (10
apparent correlations degreesoutof10pos-
ofdifferent

Table 1. DescriptiveStatistics
Variable Minimum Maximum Mean SD
KnChar 1.50 5.00 2.9701 0.68022
AbsCap 2.14 4.86 3.6530 0.51133
DisCap 1.50 5.00 3.1957 0.75940
NetMNC 1.00 5.00 3.2603 1.14294
DoKT 1.00 5.00 2.8850 0.80667

584 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
in Multinational
KnowledgeTransfer Corporations

Matrix
Table 2. Correlation
12 3 4 5
1. KnChar 1.000
2. AbsCap -0.204f 1.000
3. DisCap -0.212* 0.344*** 1.000
4. NetMNC -0.271* 0.309** 0.208f 1.000
5. DoKT -0.198* 0.333** 0.291** 0.360** 1.000

***p<0.001,**p<0.01,*p<0.05,fp<0.1,n = 92

sible).However,noneofthecorrelations - forall vari-


indicatesmulticollinearity
ablesr< 0.5 (tolerance - -
0.824 0.884; VIF 1.170 1.214).Variablesmeasuring the
disseminative of
capacity knowledge senders andthe absorptivecapacityof knowl-
edge receiverswerestrongly positivelycorrelated(p < 0.001). As expected,the
variablemeasuring knowledgecharacteristics was negatively correlatedwithall
otherindependent variablesandthedegreeofknowledge The threeother
transfer.
independent variableswerecorrelated andsignificantly
positively withthedepen-
dentvariable(p < 0.01).
A regression analysiswas usedtosimultaneously assessrelationshipsbetween
eachindependent variableandthedependent measure.Table3 presents theresults
oftheregression analysisontheimpactofthecharacteristicsofknowledge (KnChar),
thecharacteristicsofknowledge receivers(AbsCap), thecharacteristics
ofknowl-
edge senders (DisCap) and the characteristics
of the relationshipsbetween the
sendersandreceivers (NetMNC)onthedegreeofknowledge transfer(DoKT). Un-
standardized coefficients are reported.
As recommended by Hansenand Lovas

Table 3. RegressionAnalysisfortheDegreeof KnowledgeTransfer


Modell Model 2 Model 3 Model 4
ß s.e. ß s.e. ß s.e. ß s.e.
Constant 3.594*** 0.387 1.647* 0.757 1.245 0.786 0.239 0.832
KnChar -0.237t 0.126 -0.162 0.124 -0.127 0.124 -0.041 0.130
AbsCap 0.473** 0.160 0.386* 0.167 0.422* 0.176
DisCap 0.192+ 0.114 0.222f 0.119
NetMNC 0.161* 0.081
R-square 0.039 0.128 0.157 0.276
Adjusted 0.028 0.108 0.127 0.233
R-square
F 3.524+ 6.266** 5.225** 6.464***
***p<0.001,**p<0.01,*p<0.05,fp<0.1,n = 92

mir vol. 47, 2007/4 585

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

Table 4. Regression Analysis forthe Degree of Knowledge TransferwithStandardized Variables

ß (standardized) s.e.
Constant 2.895*** 0.085

ZscoreCKnChar) -0.028 0.088

Zscore(AbsCap) 0.216* 0.090

Zscore(DisCap) 0.169f 0.090

Zscore(NetMNC) 0.184* 0.093


R2 0.276
F 6.464***

***p<0.001, *p<0.05, +p<0.10, n = 92

(2004),thevariablestesting themaineffects ofthefourdeterminants wereentered


sequentiallyinModels1-3.Theresults confirm priorresearch:thehigher thedegree
of knowledgenon-codifiability, complexity, and non-availability,
non-specificity
thelowerthedegreeofknowledge transfer.However,thiseffect disappearswhen
otherdeterminants areintroduced tothemodel.
Sincethisstudyinvestigates thejointeffect ofknowledge transfer determinants
on thedegreeofknowledge transfer,theactualhypotheses testing was carriedout
inModel4. Themodelwas statistically significant(p < 0.001) withan R-squareof
0.276 (adjustedR-square:0.233). Thismeansthatabout23 percentofthedegree
ofknowledge transfer variancecan be explainedbythejointvarianceoftheinde-
pendent variables.
Theresults oftheregression analysisdidnotsupport hypothesis 1. As predicted,
theimpactofknowledge characteristicson thedegreeofknowledge transferhada
negativesign, but the resultswere not statistically
significant.
Hypothesis 2 was supported bytheanalysis.The characteristics ofknowledge
receivershada strong positiveeffect onthedegreeofknowledge transfer(p < 0.05).
The modelalso providedan examination of therelationship betweenthecharac-
ofknowledge
teristics sendersandthedegreeofknowledge transfer (hypothesis 3).
As predicted,theeffect oftheformer onthedegreeofknowledge transfer waspos-
itivebutonlyslightly significant(p < 0.10). The finalhypothesis - hypothesis 4-
statedthattheclose relationships betweenknowledgesendersandreceiverswere
expectedto havea positiveimpacton thedegreeof knowledgetransfer. The hy-
pothesisreceived full - the
support impact of the of
characteristics the relationships
betweenknowledgesendersandreceiverson thedependent variablewas positive
andsignificant (p < 0.05).
FollowingSzulanski(1996),therelativeimportance ofeachindependent vari-
able was also checked.First,all independent variableswerestandardized (z-trans-
formed). Then,a regression was runwiththestandardized variables,andthestan-
dardizedbetacoefficients in theregression analysis were compared.As Szulanski

586 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

indicates,"thelargertheabsolutevalue of a coefficient,
themoreimportant is
thecontributionofthecorrespondingvariable"(Szulanski1996,p. 36). Theresults
oftheregression arepresentedinTable4, andsuggestthatthevariablemeasuring
of knowledgereceivershas thestrongest
thecharacteristics effecton thedegree
of knowledgetransfer,closelyfollowedby therelationshipsbetweenknowledge
sendersand receiversand thecharacteristics
of knowledgesenders.Knowledge
aretheleastimportant
characteristics determinant, whenotherfactorsarepresent.

Discussion and Concluding Remarks

This paperstartedwitha reviewof conceptualand empiricalstudiesthathave


contributed to theunderstanding of intra-MNC knowledgetransfer determinants.
Using thesignaling metaphor (suggested by Szulanski 2000), the determinants were
classifiedintofourgroups:characteristics ofknowledge, characteristics ofknowl-
edgesenders(disseminative capacity),characteristics ofknowledge receivers(ab-
sorptivecapacity),andcharacteristics oftherelationships betweensendersandre-
ceivers.Fourhypotheses werepresented to addresstherelationship betweeneach
ofthedeterminants andthedegreeofknowledge transfer.
Thehypotheses werethen
testedusingdatafrom92 Danishsubsidiaries locatedin 11 countries. The results
ofthestatistical analysischallengetheviewthatthesuccessofknowledge transfer
is exclusivelya functionofthecharacteristics ofknowledge. Theanalysisindicated
thatthecharacteristics of knowledgesenders,knowledgereceivers(thestrongest
predictor),andtherelationships between themareimportant determinants ofknowl-
edge transfer.The are
findings presented in detailbelow.
Thereviewoftherelevant literatureshowedthatknowledge characteristics
have
beenincludedas independent variablesinmanyempiricalstudiesofMNC knowl-
edge transfer. Hypothetically, knowledgecharacteristics (non-codifiability, com-
plexity, and
non-specificity non-availability) should have a negativeimpact the
on
of
degree knowledge transfer. The variable measuring knowledgecharacteristics
wasindeedrelatednegatively tothedegreeofknowledge transfer (p < 0.10). How-
ever,thevariablelostitsimportance whenotherindependent variableswereadded
tothemodel.Thefindings areinteresting giventhatmanyresearchers haveoffered
excessivepraisefortheimportance ofknowledge characteristicsandpaidlittleat-
tentiontootherdeterminants oftheknowledge transfer
process.On theotherhand,
theabsenceoftheresultson knowledgecharacteristics is in linewithrecentsug-
gestionsthatwhileknowledge characteristicsareimportant determinants ofknowl-
edgetransfer,theyarenotnecessarily theonlyones.SzulanskiandCappettawrote:
"information stickinessinvolvesnotonlyattributes oftheinformation butat-
itself,
tributesof and choicesmadeby information seekersand information providers"
(Szulanski/Cappetta 2003,p. 528).

mir vol. 47, 2007/4 587

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

Whenemphasizing thebehavioral characteristics ofindividuals, suchas ability


andmotivation, respondents shed light on the dual role ofemployees intheprocess
ofknowledge transfer.They are the primary actors in the process. Their behavior
in their
and, particular, ability and motivation to absorb and share knowledgeare
keyto successful knowledge transfer. Previously, thebehaviorofbothknowledge
receivers andknowledge senderswereseldomincludedinthesamemodel.In few
exceptional cases,thoseweretreated as cognitive barriersdistinct from motivational
factors. both
Presumably, ability and motivation must be present to achieve a higher
degreeofknowledge since
transfer, ability without motivation is likely to result
in
poorperformance, andviceversa.
EvenMNCs withhighlyskilledandmotivated employeeswillnotbe effective
in knowledgetransfer if theyare unsuccessful in buildingthenecessaryinfra-
structureforlearning andsupport. Knowledgetransfer willbe higher within certain
organizational contextswherebridgesbetweenindividualsand possibilitiesfor
dialogueacrossthe organizational hierarchy existto facilitatecommunication.
Therefore, thedegreeofknowledgetransfer to thefocalsubsidiary was expected
tobe greater ifthesubsidiary was involvedinnetwork relationswithotherunitsof
thesameMNC. In linewithpreviousstudies,thevariablecapturing theextentof
therelationships withinMNC positively affected thedegreeofknowledge transfer.
As recommended byHansenandLovas (2004),fourdeterminants wereentered
inthesamemodel,whichallowedtesting oftherelativeimportance ofeachdeter-
minant in predicting knowledge transfer (see also Szulanski 1996). The resultsof
theanalysisrevealedthatamongthefourdeterminants, absorptive capacityhadthe
strongest effecton thedependent variable,closelyfollowedby characteristics of
therelations among senders and receivers, and disseminative capacity.
Thestudyrepresents aneffort todevelopandtesta conceptual modelofknowl-
edge transfer. As some concepts were relatively new, new scales weredeveloped
andone of thedeterminants was capturedusinga singlemeasure,theresultscan
be perceivedas an illustration oftheproposedconceptual model.Thereliability of
perceptual measures was checked through an inter-raterreliability test as recom-
mendedbyGuptaandGovindarajan (2000). Unfortunately, I wasabletocollectus-
able data onlyon fouritemsfromthesubsidiary managersand theirimmediate
supervisors. This is obviouslynot enough.Ideally,the testshouldhave been
conducted on all itemsthatareusedto measurebothdependent andindependent
variables.Regrettably, theexistingdatadoes notallowus to do it.Still,eventhe
limited resultsoftheinter-rater reliability testcanbe usefulforensuring somekind
ofreliability ofat leastsomemeasures.Futurestudiesshouldtryto obtaindatato
conductinter-rater testsformorethanone determinant.
reliability
Another limitationofthestatistical analysishereis theuseofcontrolvariables.
Previousempirical studiesonknowledge transfer identified someotherfactors that
couldinfluence knowledge transferto a subsidiary. Among them are size(Lyles/Salk
1996,Bresmanet al. 1999,Lane et al. 2001,Foss/Pedersen 2002,Minbaevaet al.

588 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Knowledge Transferin MultinationalCorporations

2003), industry characteristics (Lane/Lubatkin 1998,Gupta/Go vindarajan 2000,


Lane et al. 2001,Subramaniam/Venkatraman 2001, Minbaevaet al. 2003), mode
of entry(Foss/Pedersen 2002, Martin/Salomon 2003), and previousexperience
(Simonin1999a,1999b).Degreesoffreedom werenotavailabletoincludeallcontrol
variablesatearlierstagesofstatistical analysis.However, all ofthehypotheses were
testedaftercontrolling fortheeffect ofsize,industry, modeofentry, andprevious
experience. Controlvariableswerefirstintroduced alone and thentogether with
independent variables but none of them showed significant effects. There were no
significant variations ineitherdirection oftherelationships ortheregression coef-
ficientsofindependent variablesaftercontrolvariableswereintroduced. Thislack
ofsignificance, however, is owedmainlytothelimitedsamplesize andwe should
by no means conclude that thelinkssuggestedareunimportant.
The hypotheses werealso testedwithdummy variablesrepresenting countries
wheresubsidiaries werelocatedandno significant changeswereobserved.How-
ever,theremaybe a substantively significant relationship whichthecurrent analy-
sis did notshowdue to thesamplesize percountry andgeographical distribution
ofthesubsidiaries. Indeed,severalempiricalstudiesreported culturaldistanceas
one oftheobstaclesforMNC internal knowledge transfer. One ofthemostrecent
ones by Cho and Lee (2004) foundthattheculturalsimilarity betweena South
Koreansubsidiary andothermembers oftheMNC network is positively relatedto
intra-network knowledgesharing in MNCs. Yet from the measures used by the
authors itisdifficult totellwhether are
they measuring similarity of nationalcultures
or corporate cultures:twomeasureswereused.If it is aboutthedegreeto which
corporate culture unitesall employeeswithin a corporation, thenthisvariableis re-
lated(butunfortunately notcapturedwiththemeasuresused in thispaper)to the
fourth determinant oftheproposedmodel,i.e., characteristics oftherelationsbe-
tweenknowledge sendersandreceivers (barrierassociatedwiththeorganizational
contextelementin Figure1). The topicis veryinteresting, butto approachit the
datashouldbe collectedfrommultiplerespondents withinthesamecorporations
(tocontrol forcorporate cultureeffect) andfrommultiple organizations withinthe
samecountry (to control for national culture effect).
Inability to capturethesubsidiary's strategicroleis anotherlimitation of this
paper. It may be suggested that thecharacteristics of the relations between head-
quarters and a R&D subsidiary might differfrom those between headquarters and
a subsidiary acting as a sales agent.Unfortunately, I do not have data to test
for the
strategic role of the subsidiary: there are only five cases when R&D activities
com-
prisedmorethan15 percent ofoperation.
Overall,generalizing thefindings of anycross-sectional studyis constrained
by thefact that the statisticalanalysis is basedon samples collected atspecifictimes,
inspecificplacesandfromspecificindividuals. Clearly, there is a need fora similar
study with a much and
largersample geographical representation. limitations
Data
aside,however, theresults ofthisstudysupport thethesisthatthesuccessofknowl-

mirvol. 47, 2007/4 589

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

edgetransfer is notexclusivelya functionofthecharacteristics ofknowledge. The


study confirms intuitionbehindHansen and Lovas (2004), who wrote: "to advance
researchoh thistopic,studiesneed to shedthepasttendency of analyzingone
determinant totheexclusionofothers"(Hansen/Lovas 2004,p. 820). Indeed,ifwe
aretofullyunderstand theprocessofknowledge itis important
transfer, toinclude
determinants thatarerelatednotjustto knowledge characteristics,butalso to the
individualsinvolvedinthetransfer processandthesituations in whichknowledge
transfertakesplace.
Themodelthathasbeendevelopedandtestedherecouldalso be usedas a basis
forfuture empirical testson whatmanagers can do toenhanceorfacilitate knowl-
edge transferbetween headquartersand thefocal subsidiary. example,studies
For
on humanresourcemanagement (HRM) practicesfacilitating knowledgetransfer
oftenblack-boxthelinkbetweentheformer andthelatter(e.g.,Lyles/Salk1996,
Minbaevaetal. 2003,Bjorkman etal. 2004). To understand thelink,Guest(1997)
recommends considering mediating variablesthatshouldbe determined theoreti-
callydepending on theoutcomes the HRM practicesaim to achieve. This study
offersatleastthreedeterminants ofknowledge transferthatcouldbe considered as
mediating variables. By employing HRM practices thateffect theability and mo-
tivationofemployees toabsorbandshareknowledge, andthosethatfacilitate close
relationsbetweenMNC units,MNCs ensureinternal knowledgetransfer. More-
over,theresultsofthecurrent analysisshowthatdealingwithone determinant is
notlikelyto enhanceknowledgetransfer unlessotherdeterminants are also con-
sidered.

Endnotes

1 Thismetaphor is basedoncommunication theoryandis oftenreferredtoas a signalingmetaphor.


The termwas introduced byShannonandWeaver(1949) in "The Mathematical TheoryofCom-
munication".
2 In somestudies(forexample,see Szulanski1996),researchers useda notionorcausalambiguity,
whichis a compositemeasureofknowledgecharacteristics.
3 Theexception is anempirical studybyZanderandKogut( 1995) withtheirteachability dimension.
4 Thepriorknowledge base (oremployees'ability)andintensity (or employees'motiva-
ofefforts
tion)arerelatedto theconceptsofpotential andrealizedabsorptivecapacityproposedbyZahra
andGeorge(2002). Potentialabsorptive capacityis expectedtohavea highcontent ofemployee
abilitywhilerealizedabsorptive capacityis expectedto have a highcontentof employeemoti-
vation.
5 The KOB datasetis a comprehensive, continuouslyupdateddatasetofdomesticandinternational
Danishfirms(www.kob.dk).
6 Börsenis theDanishbusinesssector'sglobal,nationaland regionalnewspaper. Everyyearthe
newspaper publishesan annualstatusreporton Danishbusinesses(www.borsen.dk).
7 Nunnally(1967) recommended a Cronbachalphaequal to or greaterthan0.60 as theminimum
valuesufficientforresearchpurposes.Sincethisstudyis amongthefirstattempts todevelopand

590 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
KnowledgeTransfer
in Multinational
Corporations

empirically testtheframework foranalyzingknowledgetransfer inMNCs,maybe itcouldbe ac-


ceptabletoconsidertheCronbachAlphaof0.63 forcharacteristics ofknowledgeand0.66 forthe
characteristicsofknowledgesenders,bothareleastdevelopedin theliterature. Thatwas donein
otherstudies,whichtriedto tacklethistaskbefore.Forexample,in studyaimedat developinga
framework ofcross-border knowledge transferbyPakandPark(2004),publishedinMIR 4thquar-
ter,theauthorsusedtwoscales (outofeight)withCronbachAlpha0.6212 and0.6532. Szulans-
ki's (1996) studyon knowledgestickiness, publishedin StrategicManagement JournalWinter
SpecialIssue,also reported twoscales withCronbachAlphabelow0.70: "unproven knowledge"
- 0.67 and"sourceis notperceivedas reliable"- 0.64. However,itis clearlyon oftheweaknesses
ofthepaper,whichis discussedin thefinalsectionofthepaper.
8 The inter-rater (or inter-observer
reliability reliability) testis one of fourgeneralclasses of reli-
abilityestimates andthebestone toestimatereliability whenthemeasureis an observation. It is
used to assess thedegreeto whichdifferent respondents give consistent estimatesof thesame
phenomenon. Whenthemeasureis a continuous (as in ourcase), thebestwayis to calculatethe
correlation betweentheratingsoftworespondents.
9 In a similarexercise,Arvidsson(1999) andDenrellet al. (2004) findthatinter-rater reliabilityof
marketing capabilityevaluationis relativelylow: theresultsshowperception gaps in around70
percent ofthecapability evaluations. Atthesametime,Arvidsson (1999) foundthatthereis much
greater agreement betweenevaluationsoffinancialperformance thanevaluationsofcapabilities:
corporate andsubsidiary managers agreeon howwellthesubsidiaries areperforming infinancial
termsin around60 percentof thecases. Thus,perception gaps exist,butnotalways.However,
methods originallyusedbyArvidsson (1999) andlaterreported inDenrelletal. (2004) areslightly
different fromthemethodsuggestedby Guptaand Govindarajan(2000) and replicatedin this
study.Arvidssoncollectedevaluationsof a particular subsidiary's marketorientation fromthree
differentsources:thesubsidiary's ownmanagers;corporate managersresponsibleformarketing
activitiesof particularunits;and managersat othersubsidiariesthantheone beingevaluated.
Thosemanagerswerechosenbasedon thefactthattheyall (to certainextent)have a formalre-
sponsibility formarketing activities:globallyforcorporatemanagersand locallyforsubsidiary
managers. Notice,therespondents (corporateandsubsidiary managers) werenotnecessarily work-
ingtogether andmighthaveaccess to different amountsof information.

References

Argyris,C./Schon, D., OrganizationalLearningII, ReadingMassachusetts, Addison-Wesley 1996.


Argote,L., Organizational Learning:Creating,Retaining, and Transferring Knowledge,Boston:
KluwerAcademic1999.
Arvidsson,N., TheIgnorant MNE. TheRoleofPerception Gaps inKnowledgeManagement, Stock-
holm:SwedishSchoolofEconomics1999.
Attewell,P.,Technology DiffusionandOrganizational Learning:The Case ofBusinessComputing,
Organization Science,3, 1, 1992,pp. 1-19.
Baldwin,T./Magjuka, R./Loher,B., The Perilsof Participation:Effectsof Choice of Trainingon
TraineeMotivation andLearning,PersonnelPsychology, 44, 1, 1991,pp. 51-65.
Bjorkman, J./Barner-Rasmussen, W./Li,L., ManagingKnowledgeTransfer in MNCs: The Impact
of Headquarters ControlMechanisms, Journalof International BusinessStudies,35, 5, 2004,
pp. 443-455.
Bresman,H./Birkinshaw, J./Nobel,R., KnowledgeTransfer in International Journal
Acquisitions,
ofInternational BusinessStudies,30, 3, 1999,pp. 439-462.
Cabrera,E., Socio-psychologicalAspectsof KnowledgeSharingin Organizations, Proceedingsof
the 7thConference on International HumanResourceManagement,University of Limerick,
Ireland2003.

mir vol. 47, 2007/4 591

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
Dana B. Minbaeva

Cabrera,A./Collins,W./Salgada,J.,Determinantsof Individual Engagementin Knowledge Sharing,


InternationalJournal of Human Resource Management, 17, 2, 2006, pp. 245-264.
Cho, K./Lee, J.,Firm Characteristicsand MNC's Intra-networkKnowledge Sharing,Management
InternationalReview, 44, 4, 2004, pp. 435-455.
Cohen, W./Levinthal,D., Absorptive Capacity: A New Perspective on Learning and Innovation,
AdministrativeScience Quarterly,35, 1, 1990, pp. 128-152.
Davis, J./Subrahmanian, E./Westerberg,A., The "Global" and the "Local in Knowledge Manage-
ment,Journal of Knowledge Management, 9, 1, 2005, pp. 101-1 12.
Dansk Industri,Danske VirksomhedersEtableringeri Udlandet,Copenhagen: Dansk Industri2003.
Denrell, J./Arvidsson, N./Zander,U., Managing Knowledge in the Dark: An Empirical Study of the
Reliabilityof Capability Evaluations, Management Science, 50, 11, 2004, pp. 1491-1503.
Egelhoff, W., InformationProcessing Theory and the Multinational Corporation, in Ghoshal,
S./Westney,D. (eds.), Organization Theory and the Multinational Corporation, New York:
St. Martin's Press, 1993, dd. 182-210.
Eisenhardt,K./Santos,F., Knowledge-baseView: A New TheoryofStrategy?in: Pettigrew, A./Thomas,
H./Whittington, R. (eds.), Handbook of Strategyand Management,SAGE Publications,2002.
Foss, N./Pedersen,T., Transferring Knowledge in MNCs: The Role of Sources of SubsidiaryKnowl-
edge and Organizational Context,Journal of InternationalManagement, 8, 1, 2002, pp. 49-67.
Galbraith,J.,Designing Complex Organizations, Reading Massachusetts, Addison-Wesley 1973.
Guest, D., Human Resource Management and Performance: A Review and Research Agenda,
InternationalJournal of Human Resource Management, 8, 3, 1997, pp. 263-276.
Gupta, A./Govindarajan,V., Knowledge Flows withinMNCs, Strategic Management Journal, 21,
4, 2000. dd. 473-496.
Hansen, M., The Search-TransferProblem: the Role of Weak Ties in Sharing Knowledge Across
OrganizationSubunits,AdministrativeScience Quarterly,44, 1, 1999, pp. 82-1 11.
Hansen, M., Knowledge Networks:Explaining EffectiveKnowledge Sharing in MultiunitCompa-
nies. Organization Science. 13. 3. 2002. dd. 232-248.
Hansen, M./Lovas, B., How Do MultinationalCompanies Leverage Technological Competencies?
Moving fromSingle to InterdependentExplanations, Strategic Management Journal, 25, 8/9,
2004, pp. 801-822.
Husted, K./Michailova, S., Diagnosing and FightingKnowledge Sharing Hostility,Organizational
Dynamics, 31,1, 2002, pp. 60-73.
Kim,L., AbsorptiveCapacity,Co-operation,and KnowledgeCreation:Samsung'sLeapfroggingin Semi-
conductors,in Nonaka, I., Nishiguchi,T. (eds.), KnowledgeEmergence:Social, Technical,and Evo-
lutionaryDimensionsof KnowledgeCreation,Oxford:OxfordUniversityPress,2001, pp. 270-286.
Kogut, B. /Zander,U., Knowledge of the Firm, Combinative Capabilities, and the Replication of
Technology,Organizational Science, 3, 3, 1992, pp. 383-397.
Lane, P./Lubatkin,M., Relative Absorptive Capacity and InterorganizationalLearning, Strategic
Management Journal, 19, 5, 1998, pp. 461-477.
Lane, P./Salk,J./Lyles,M., AbsorptiveCapacity, Learning, and Performancein InternationalJoint
Ventures,StrategicManagement Journal,22, 12, 2001, pp. 1139-1161.
Levitt,B./March,J.,Organizational Learning,Annual Review of Sociology, 14, 1988, pp. 319-340.
Lindsay,V./Chadee,D./Mattsson,J./Johnston, R./Millett,B., Relationships,the Role of Individuals
and Knowledge Flows in the Internationalizationof Service Firms, International Journal of
Service IndustryManagement, 14, 1, 2003, pp. 7-35.
Lyles, M./Salk,J.,Knowledge Acquisition fromForeign Parentsin InternationalJointVentures:An
Empirical Examination in the HungarianContext,Journalof InternationalBusiness Studies, 27,
5, 1996, pp. 877-903.
Martin,X./Salomon, R., Knowledge TransferCapacity and its Implications for the Theory of the
MultinationalCorporation,Journalof InternationalBusiness Studies, 34, 4, 2003, pp. 345-356.
Michailova, S. /Husted,K. Knowledge Sharing Hostilityin Russian Firms, California Management
Review,45, 3, 2003, pp. 59-77.
Minbaeva, D./Pedersen,T./Bjorkman,I./Fey,C./Park,H., MNC Knowledge Transfer,Subsidiary
AbsorptiveCapacity and Knowledge Transfer,Journal of InternationalBusiness Studies, 34, 6,
2003, pp. 586-599.

592 mir vol. 47, 2007/4

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions
KnowledgeTransfer
in Multinational
Corporations

Minbaeva,D., HRMPracticesand KnowledgeTransfer inMNCs,Ph.D. Dissertation, Copenhagen:


CopenhagenBusinessSchoolPress2004.
Minbaeva,D./Michailova, S., KnowledgeTransfer and Expatriation Practicesin MNCs: The Role
ofDisseminative Capacity,EmployeeRelations,26, 6, 2004,pp. 663-679.
Miner,J./Crane, D., HumanResourceManagement:TheStrategicPerspective, New York:Harper-
CollinsCollegePublishers1995.
Mowery,D./Oxley,J.,InwardTechnologyTransferand Competitiveness: The Role of National
Innovation Systems,CambridgeJournalofEconomics,19, 1, 1995,pp. 67-94.
Mowery, D./Oxley, J./Silverman, B., StrategicAlliancesandInterfirm KnowledgeTransfer, Strate-
gic Management Journal,17,Special Issue Summer,1996,pp. 77-91.
Nelson,R./Winter, S., An Evolutionary Theoryof EconomicChange,Cambridge,Massachusetts:
HarvardUniversity Press1982.
Nunnally,J.C, Psychometric theory, New York:McGraw-Hill1967.
Pak,Y./Park,Y, A Framework ofKnowledgeTransfer inCross-Border JointVentures:AnEmpirical
TestoftheKoreancontext, Management International Review,44, 4, 2004,pp. 435-455.
Polanyi,M., TheTacitDimension,New York:Doubleday1966.
Reed,R./DeFillippi, R., Causal Ambiguity, Barriersto Imitation,andSustainableCompetitive Ad-
vantage,AcademyofManagement Review.15, 1, 1990,pp. 88-102.
Rogers,E., Diffusion ofInnovations, New York:FreePress1995.
Schulz,M., PathwayofRelevance:ExploringInflowsofKnowledgeintoSubunitsofMultinational
Corporations, Organization Science,14,4, 2003,pp. 440-459.
Senge,P., The FifthDiscipline:TheArtand Practiceof theLearningOrganization, New York:
Doubleday1990.
Shannon,C./Weaver, W., The MathematicalTheoryof Communication, Chicago: University of
IllinoisPress1949.
Simonin,B., Transfer of Marketing Know-Howin International StrategicAlliances:An Empirical
Investigation of theRole and Antecedents of KnowledgeAmbiguity, Journalof International
BusinessStudies,30, 3, 1999a,pp. 463-490.
Simonin,B., Ambiguity and theProcessof KnowledgeTransfer in StrategicAlliances,Strategic
Management Journal,20, 7, 1999b,pp. 595-623.
Subramaniam, M./Venkatraman, N., Determinants of Transnational New ProductDevelopment
Capability:Testingthe Influenceof Transferring and DeployingTacit OverseasKnowledge,
StrategicManagement Journal,22, 4, 2001,pp. 359-378.
Szulanski,G., ExploringInternalStickiness:Impediments to theTransfer of Best PracticeWithin
theFirm,StrategicManagement Journal,17,Special Issue Winter, 1996,pp. 27-43.
Szulanski,G., Appropriability and theChallengeof Scope: Bank One RoutinizesReplication,in
Dosi, G./Nelson, R./Winter, S. (eds.), TheNatureand DynamicsofOrganizational Capabilities,
New York:OxfordUniversity Press2000.
Szulanski,G.,Sticky Knowledge: BarrierstoKnowing intheFirm,London:SAGE Publications 2003.
Szulanski,G./Cappetta, R., Stickiness:Conceptualizing, Measuringand Predicting Difficultiesin
theTransfer of KnowledgewithinOrganizations, in Easterby-Smith, M./Lyles,M. (eds.), The
BlackwellHandbookofOrganizational Learningand KnowledgeManagement, Oxford:Black-
wellPublishing 2003.
Vroom,V., Workand Motivation, New York,LondonandSydney:JohnWileyandSons 1964.
Winter,S., Knowledgeand Competenceas StrategicAssets,in Teece, D. (ed.), The Competitive
Challenge,Massachusetts, Cambridge:BallingerPublishing Company1987.
Zahra,S./George, G.,Absorptive Capacity:A Review,Reconceptualization, andExtension,Academy
ofManagement Review,27, 2, 2002,pp. 185-203.
Zander,U., Exploitinga TechnologicalEdge: Voluntary and Involuntary Dissemination of Tech-
nology,Ph.D. Dissertation, StockholmSchoolof Economics1991.
Zander,U./Kogut, B., Knowledgeand theSpeed of theTransfer and Imitation of Organizational
Capabilities,Organization Science,6, 1, 1995,pp. 76-92.
Zellmer-Bruhn, M. Interruptive EventsandTeamKnowledgeAcquisition, Management Science,49,
4, 2003,pp. 514-528.

mir vol. 47, 2007/4 593

This content downloaded from 185.44.78.76 on Wed, 25 Jun 2014 00:19:07 AM


All use subject to JSTOR Terms and Conditions

You might also like