Birkinshaw SMJ 1997 - Theme 4

7/30/2019 Birkinshaw SMJ 1997 - Theme 4

1/24

Entrepreneurship in Multinational Corporations: The Characteristics of Subsidiary InitiativesAuthor(s): Julian BirkinshawSource: Strategic Management Journal, Vol. 18, No. 3 (Mar., 1997), pp. 207-229Published by: John Wiley & SonsStable URL: http://www.jstor.org/stable/3088128

Accessed: 26/09/2008 06:53

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=jwiley.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access to Strategic

Management Journal.
http://www.jstor.org/stable/3088128?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/action/showPublisher?publisherCode=jwileyhttp://www.jstor.org/action/showPublisher?publisherCode=jwileyhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3088128?origin=JSTOR-pdf


2/24

Strategic Management Journal, Vol. 18:3, 207-229 (1997)

ENTREPRENEURSHIPN MULTINATIONALCORPORATIONS: HECHARACTERISTICS FSUBSIDIARYINITIATIVESJULIAN BIRKINSHAWInstitute of International Business, Stockholm School of Economics, Stockholm,Sweden

Thispaper defines initiative as a key manifestationof corporate entrepreneurship, nd examinesthe types of initiative exhibited in a sample of six subsidiaries of multinationalcorporations.From a detailed analysis of 39 separate initiatives, our distinct types are identified,which werefer to as 'global,' 'local,' 'internal' and 'global-internal hybrid,' to correspondto the locusof the marketopportunitywhence each arose. Two importantconclusions are indicated.First,entrepreneurshipat the subsidiary level has the potential to enhance local responsiveness,worldwide learning and global integration, a much broader role than previously envisioned.Second, the use of contextual mechanisms to create differentiatedsubsidiary roles has itslimitationsbecause each initiativetype is facilitated in differentways. ? 1997 by John Wiley &Sons, Ltd.

INTRODUCTIONThe ability of the large multinational corporation(MNC) to leverage the innovative and entrepre-neurial potential of its dispersed assets is a funda-mental strategic imperative (Bartlett and Goshal,1989). While some studies have explicitly con-fronted the challenge of entrepreneurship inMNCs (e.g., Ghoshal, 1986), research has tendedto focus on either the organizaationof the existingactivities of the MNC or on corporateentrepreneurship as a generic managerial issue(Ghoshal, 1986: 6). As stated by Hedlund andRidderstrale (1992: 5) the dominant theme inprior MNC research-particularly from a theo-retical perspective-has been 'on the exploitationof givens (i.e., existing product-marketcombinations) rather than on the creation of nov-elty.' The need for research that explicitly linksMNC management with studies of corporateentrepreneurshipis therefore substantial.The current paper examines initiatives in MNC

Key words: multinationalcorporation; subsidiary;initiative;corporate ntrepreneurshipCCC 0143-2095/97/030207-23? 1997by JohnWiley & Sons, Ltd.

subsidiaries. An initiative is defined here as adiscrete, proactive undertaking that advances anew way for the corporation to use or expand itsresources (Kanter, 1982; Miller, 1983). An initia-tive is essentially an entrepreneurial process,beginning with the identification of an opportunityand culminating in the commitment of resourcesto that opportunity. In this sense, the term isnarrowerthan related constructs such as 'internalcorporate venture' (Burgelman, 1983a) whichinvolve both the initiative and the ongoing man-agement of the resultantbusiness activity. Severalprior studies have used the initiative construct(e.g., Burgelman, 1991; Cohen and Machalek,1988; Sathe, 1985).Subsidiary is defined here to be any operationalunit controlled by the MNC and situated outsidethe home country. This definition ensures thatthe artificial notion of a single parent-subsidiaryrelationship is avoided. The reality in most MNCstoday is that subsidiaries have a multitude oflinkages with other corporateentities in the homecounty and worldwide (Ghoshal and Bartlett,1991), but academic research has-for the mostpart-continued to work on the basis of a singleparent-subsidiary relationship. By working at the

Received 5 May 1995Final revision received 23 April 1996


3/24

208 J. Birkinshawinitiative level this study breaks new ground inthat it documents activities at a sub-subsidiaryunit of analysis.Why study initiative in MNC subsidiariesrather than in the parent company? The simpleanswer is that despite the compelling logic fortapping into local markets through the subsidiarynetwork (Bartlett and Ghoshal, 1986), many cor-porations appear to neglect the creative potentialof their subsidiaries. Thus, subsidiary initiativemay be a relatively rare and underresearchedphenonemon but its potential value to the MNC ishigh. While the most common form of subsidiaryinitiative is probably the identification and pursuitof a new product opportunity in the local market,this paper takes the concept further by showingthat three other forms of subsidiary initiative canalso be identified. The paper puts forward aconceptual framework, based on the networktheory of the MNC (Ghoshal and Bartlett, 1991)through which subsidiary initiative can be betterunderstood, and then describes the results of adetailed inductive study of 39 initiatives. Themajor contribution of the paper is the finding thatsubsidiary initiative has the potential to enhancelocal responsiveness, worldwide learning andglobal integration, a much broader role than pre-viously envisioned.This paper is organized as follows. First, theliterature on corporate entrepreneurship isreviewed, in broad terms and then specifically interms of MNC subsidiary management. Second,the network theory of the MNC is used to builda conceptual framework in which three types ofsubsidiary initiative are identified. Each type ofinitiative is also identified in the extant literature.Third, the research methodology is described.Fourth, the findings from the study arediscussed-this section includes the identificationand description of a fourth type of initiative, anda systematic description of the salient character-istics of all four types. Finally, the implicationsof the study for MNC management theory andfor corporate entrepreneurshipare discussed.

CORPORATE ENTREPRENEURSHIPAND INITIATIVESCorporateentrepreneurship s receiving increasinglevels of research attention, and was the focus ofa recent special issue of the Strategic Manage-

ment Journal (Guth and Ginsberg, 1990). Inbroad terms, three forms of corporateentrepreneurshipcan be identified (Stopford andBaden-Fuller, 1994): (1) the creation of newbusiness activities within the existing organiza-tion; (2) the transformationor renewal of existingorganizations; and (3) the enterprise changing therules of competition in its industry. The focus ofthis study is on the first of these forms: thecreation of new business activities within theexisting enterprise.There is a broad recognition, however, that thegeneration of new business activities or 'newcombinations' (Schumpeter, 1934) alone does notconstitute entrepreneurship.A research and devel-opment group, for example, has a clear mandateto innovate, but the behavior expected of itsemployees falls within established norms andguidelines. Entrepreneurship uggests more: a pre-disposition towards proactive and risk-takingbehavior (Covin and Slevin, 1991; Miller, 1983);use of resources beyond the individual's directcontrol (Kirzner, 1973; Stevenson and Jarillo,1990); or a 'clear departure from existing prac-tices' (Damanpour, 1991: 561). Kanter (1982)proposed the following distinction between 'ba-sic' and entrepreneurialactivities:

Basic accomplishments .. are part of theassignedob andrequire outine ndreadilyavail-able meansto carrythem out. In contrast nno-vative accomplishments re strikinglyentrepre-neurial.They are sometimeshighly problematicandgenerally nvolveacquiring ndusingpowerand influence. 1982: 97)On the basis that within-firm corporateentrepreneurship nvolves a departurefrom exist-ing practices or 'a new way for the corporationto use or expand its resources' (Kanter, 1982),the literaturesuggests two distinct models, whichwill be termed focused and dispersed corporateentrepreneurshiprespectively. Initiative, the focalconstruct in this research, is a manifestation ofthe dispersed approach.Focused corporate entrepreneurship (alsocalled corporate venturing) works on the premisethat entrepreneurshipand management are funda-mentally different processes that require differentmodes of organization to occur effectively (Burnsand Stalker, 1961; Galbraith, 1982; Kanter, 1985).This is typified by the New Venture Division,whose mandate is to identify and nurture new


4/24

Entrepreneurship in Multinational Corporationsbusiness opportunities for the corporation(Burgelman, 1983a; Kuratko, Montagno andHorsby, 1990; Sykes, 1986). The new venturedivision is typically a semi-autonomous entitywith little formal structure,integration across tra-ditional functional areas, availability of 'patientmoney,' and management support for risk takingand creativity (Galbraith, 1982; Kanter, 1985;Kuratkoet al., 1990; Quinn, 1985; Sathe, 1985).There are many examples of corporations thathave pursued this approach to corporateentrepreneurship, including 3M, Kodak, andExxon (Ginsberg and Hay, 1995; Sykes, 1986).Note that the mandate of a new venture divisionis fundamentally broader and more ambiguousthan that of a research and development group,where the set of tasks and responsibilities can befairly narrowly defined. In Schollhammer's terms(1982), the new venture division is a case of'incubative' entrepreneurship while the R&Dgroup is 'administrative' entrepreneurship.Dispersed corporate entrepreneurship (alsocalled intrapreneurship)rests on the premise thatevery individual in the company has the capacityfor both managerial and entrepreneurialbehaviormore or less simultaneously. Rather than hivingoff separate groups or divisions to be entrepre-neurial, while the rest are left to pursue theongoing managerial tasks (Galbraith, 1982), thedispersed approach sees the development of anentrepreneurialculture or posture as the key ante-cedent to initiative (Covin and Slevin, 1991;Goshal and Bartlett, 1994; Kanter, 1985; Stopfordand Baden-Fuller, 1994). The design of an'organic' (Bums and Stalker, 1961) or 'integra-tive' (Kanter, 1985) organization creates the facil-itating conditions, but entrepreneurship s actuallydriven by the actions of employees who-forwhatever reason-choose to pursue risky oruncertain ventures 'for the good of the organiza-tion' (Barnard, 1938: 200). The challenge forcorporate management is to instill the personalinvolvement and commitment in its employeesthat drives entrepreneurship(Ghoshal and Bart-lett, 1994).Dispersed corporate entrepreneurshipthereforeassumes a latent dual role for every employee,consisting of (a) the management of ongoingactivities and (b) the identification and pursuit ofnew opportunities (Kirzner, 1973; Penrose, 1959;Stevenson and Jarillo, 1990). The advantage ofthis approach over the focused approach is that

a greater diversity of opportunitieswill be sensed,because the entrepreneurialcapability is dispersedthroughoutthe organization, rather than restrictedto a new venture division. The major disadvan-tage of this approach is that managerial responsi-bilities typically 'drive out' entrepreneurialresponsibilities (Hedlund and Ridderstrale, 1992;Kanter, 1986) because they are more clearlydefined and have more immediate rewards. Unlessit is well managed the dispersed approach canactually inhibit entrepreneurship Drucker, 1985).Initiative, as used in this paper, is the primarymanifestation of dispersed corporateentrepreneur-ship. The initiative process is bounded by theidentification of an opportunity at the front endand the commitment of resources to the undertak-ing at the back end. Note that the long-termsuccess of the resultant business activity is asecondary issue. The entrepreneurialchallenge isto move from an idea to a commitment ofresources; the managerial challenge is to makethe resultant business activity profitable. It isimportant,moreover, to recognize that the focusedand dispersed approaches are complementaryrather than alternative. For example, an oppor-tunity identified in a subsidiary may be nurturedand developed in the new venture division; equa-lly, an innovation by the new venture divisionmay inspire further innovation by an operatingdivision. Clinical evidence suggests that success-ful entrepreneurial companies such as 3M andHP do indeed exhibit both approaches (Kanter,1985; Peters and Waterman, 1982; Pinchott,1985).

INITIATIVE IN MNC SUBSIDIARIESParent-subsidiary relationships in MNCs havebeen studied for many years (Martinez and Jar-illo, 1989). Most early research focused narrowlyon facets of the parent-subsidiary relationshipsuch as centralization, formalization, coordinationand control (Brandt and Hulbert, 1977; Cray,1984; Hedlund, 1981; Negandhi and Baliga, 1981;Picard, 1980). More recently, new conceptualiza-tions of the MNC such as the heterarchy(Hedlund, 1986) or the transnational corporation(Bartlett and Ghoshal, 1989) enabled a moreholistic understandingof the subsidiary as a semi-autonomous entity within a differentiated system.Within this broad school of thought, two distinct

209


5/24

210 J. Birkinshawviews of the subsidiary can be discerned, withdirect parallels to the two types of corporateentrepreneurship described in the previous sec-tion.The first perspective viewed the subsidiary ashaving a 'role' in the MNC. Bartlett and Ghoshal(1986) made the observation that national subsidi-aries can take one of four generic roles, basedon the strategic importance of the local environ-ment and the competence of the subsidiary. Theyfurthersuggested that the MNC's structure shouldreflect this heterogeneity, so that certain subsidi-aries receive, for example, much greater strategicautonomy than others. This study and others inthe same genre (e.g., Ghoshal and Nohria, 1989;Gupta and Govindarajan, 1991; Jarillo and Marti-nez, 1990) shared a number of underlying charac-teristics: first, an implicit parent company per-spective, in that subsidiaries were modeled interms of 'relative capabilities' (vs. sistersubsidiaries); second, the belief that the subsidi-ary's role was determined by the parent companyand essentially assigned to the subsidiary in ques-tion; and third, the notion that the subsidiary'srole was enacted through the definition of anappropriateset of coordination and control mech-anisms (its structural context in Bower's, 1970,terms).This model is entirely consistent with thedescription of focused corporate entrepreneurshipabove. Certain subsidiaries are given the responsi-bility for innovating or pursuing initiatives, whileothers are given implementational roles. Theseroles are enacted through the structural contextof the MNC. As shown by Ghoshal and Bartlett(1988), autonomy, local resources, normativeintegration and interunit communication are asso-ciated with creation (of innovations) in subsidi-aries, but negatively associated with adoptionand diffusion.The second perspective focused directly on thesubsidiary level of analysis. This perspectiveenvisioned a much greater element of strategicchoice on the part of subsidiary managementthan the subsidiary role perspective. Thus, thesubsidiary's strategy was constrained (rather thandefined) by the structuralcontext, and subsidiarymanagers had considerable latitude within theimposed constraints to shape a strategy as theysaw fit. This body of research was predominantlyCanadian, stretching back to Safarian's (1966)work on the foreign ownership of Canadian

industryand with key contributionsfrom Crookell(1986), D'Cruz (1986), Poynter and Rugman(1982) and White and Poynter (1984, 1990). Inpart because of the high levels of foreign owner-ship in Canadian industry, academic thinking haspushed towards subsidiary managers utilizingtheir strategic discretion rather than simplyresponding to parentaldecree. White and Poynter(1984: 69), for example, noted that subsidiarymangers 'Will have to adjust their strategies tosuccessfully deal with changed circumstances ...Through the careful development of local capa-bilities the subsidiary manager can contribute tothe evolution of the Canadian subsidiary's strat-egy.' This is consistent with the dispersedapproach to corporate entrepreneurship.As sug-gested by White and Poynter (1984) and others,creativity and innovation should be endemic tothe national subsidiary as the driver of its strat-egy. The subsidiary has ongoing managerialresponsibilities but at the same time it has theresponsibility to respond to entrepreneurialopportunities as they arise.In summary, entrepreneurship n MNC subsidi-aries is a subject that has received limitedresearch attention but that can be informed bythe broader literature on corporate entrepreneur-ship. As with the literature on corporateentrepreneurship,the implication here is not thatone model of subsidiary management is superiorto the other in terms of entrepreneurialcapabilitybut that the two are complementary. In particular,proponentsof the subsidiary role perspective havemade the observation that complete control of thenational subsidiary through contextual mech-anisms is neither possible nor desirable (e.g.,Prahalad and Doz, 1981). There is clearly aninteresting trade-off between control and auton-omy in the parent-subsidiary relationship, andthe fact that subsidiary 'role' research favorscontrol and subsidiary 'strategy' research favorsautonomy is essentially a function of the opposingperspectives of parent and subsidiary managers.

CONCEPTUAL DEVELOPMENTAs defined above, an initiative is viewed as adiscrete, proactive undertaking that advances anew way for the corporation to use or expand itsresources. For the purposes of this research, thisdefinition was subject to two additional con-


6/24

Entrepreneurship in Multinational Corporationsstraints. First, the entrepreneurial thrust had tocome from subsidiary managers, rather than thoseat head office. Second, the initiative had to leadto internationalresponsibilities for the subsidiary,such as exporting intermediate products to affili-ates or managing a product line on a globalbasis. This condition was set to exclude trivialinitiatives. MNCs are becoming increasinglyglobal in the configuration and coordination oftheir value-adding activities (Porter, 1986), andsubsidiaries are likewise recognizing the interde-pendence of their activities with those of theglobal network. Particularly in a country such asCanada where the national market is small andthe cross-border flows with the United Stateslarge, the likelihood of any new activity attractingcorporate support is substantially enhanced whenit has international scope.Conceptual frameworkNotwithstanding the question of personal moti-vation, the origin of an initiative lies in theidentification of an opportunity to use or expandthe corporation's resources. In Kirzner's (1973)words, it is an 'alertness to hitherto unnoticedmarket opportunities' that stimulates the entrepre-neur to act. In similar fashion, Stevenson andJarillo (1990: 23) saw entrepreneurshipas '... aprocess by which individuals-either on theirown or inside organizations-pursue oppor-tunities without regard to the resources they cur-rently control.'From the perspective of the MNC subsidiary,the notion of a market opportunity is usuallyunderstood in terms of its national market. Thetraditional role of the subsidiary was first to adaptthe MNC's technology to local tastes, and thento act as a 'global scanner,' sending signals aboutchanging demands back to head office (Vernon,1966, 1979). More recently, it has been recog-nized that subsidiaries often have unique capa-bilities of their own, as well as critical links withlocal customers and suppliers. In such situations,the subsidiary's ability to pursue local oppor-tunities itself, and subsequently to exploit themon a global scale, is an important capability(Bartlett and Ghoshal, 1986; Harrigan, 1983;Hedlund, 1986).To view market opportunity solely in terms ofthe subsidiary's local relationships is, however,somewhat restricting. There is a growing body of

research that models the MNC as an interorgani-zational network (Forsgren and Johanson, 1992;Ghoshal and Bartlett, 1991), in which the subsidi-ary has multiple linkages to other entities bothinside and outside the formal boundaries of theMNC. Viewed in this way, the national subsidiarysits at the interface of three 'markets:' (1) thelocal market, consisting of competitors, suppliers,customers, and regulatorybodies in the host coun-try; (2) the internalmarket,which is comprised ofhead office operations and all corporate-controlledaffiliates worldwide; and (3) the global market,consisting of competitors, customers and suppliersthat fall outside the local and internal markets.This conceptualization is depicted in Figure 1.Global market relationships, of course, do notexist in all cases, but increasingly subsidiariesare taking on specialized roles and responsibilitieswithin the MNC that give them access to inter-national customers and suppliers (Roth and Mor-rison, 1992). Again using Kirzner's definition,three types of initiative are immediately sug-gested, namely local market initiative, internalmarket initiative, and global market initiative.These are defined by the locus of the marketopportunity in each case.How much evidence is there for these typesof initiative? No systematic research appears tohave been done on either internal market orglobal market intiatives, while Ghoshal (1986) isthe only detailed prior study on local marketinitiatives. The remainder of this section, then,will pull together the limited body of researchthat exists to provide a grounding for the currentstudy. In addition to the locus of opportunity (bywhich the types are defined), three additionalsets of characteristics will be considered: thefacilitating conditions (i.e., those elements of thesubsidiary's structural context that foster anenvironment in which initiative can occur), theintiative process; and the intended outcome.Figure 2 illustrates these elements. This frame-work is very similar to those used in severalother process research studies, including Mintz-berg, Raisinghani, and Theoret (1976) and Nutt(1993). Note that the relationships between theseelements is not one of linear causality. Facilitatingconditions and process, for example, probablyinteract with and reinforce one another, ratherthan one defining the other. Thus, the elementscould best be described as configurations (Meyer,Tsui, and Hinings, 1993), which are 'tight con-

211


7/24

212 J. Birkinshaw

GLOBAL MARKET Local^y* ~ ^V~^~ ~ MarketCustomers\ Competitors InitiativesGlobalcustomers \andsuppliers Suppliers Government\ i Government

Global- (#- Market'I- _~ - -' -

'^ Initiatives

Internal/INTERNAL; MARKET - \ Markett ~~~/~~ ~ ~~/~ , / . \x Initiatives

ParentCompany

Figure1. Conceptualmodel of the national ubsidiary nd threetypes of initiative

stellations of mutually supportive elements'(Miller, 1986: 236).Local market initiativesThere are several well-documented cases of localmarket initiatives in the literature: examplesinclude Philips UK's development of Teletexttechnology (Bartlett and Ghoshal, 1986) and AlfaLaval U.S.'s invention of the milking machine in1917 (Hedlund and Ridderstrale, 1992). Thesetwo cases were both inspired by local productand/or market needs and subsequently exploitedon a global scale.'In terms of facilitating conditions, Ghoshal's(1986) detailed study of innovation in MNCsidentified local resources, local autonomy, norma-tive integration, subsidiary-HQ communication,and intrasubsidiarycommunication as factors thatwere positively associated with the 'creation'process. No other work has systematicallyaddressed this issue, though several Canadianstudies have hinted at the conditions in whichlocal market initiatives occur (e.g., Etemad andDulude, 1986). Some insight into the initiative'Note that in theory local market initiatives could remainwithin the local market, e.g., if a product is designed specifi-cally for a segment of the local market. In practice, thesample used in this study exhibited no incidences of localinitiatives that stayed local: all went on to become inter-national products.

LOCUS FOPPORTUNITY(Local,Global,nternal)

FACILITATINGONDITIONS

INITIATIVEROCESS

INITIATIVEUTCOME

Figure2. Organizingramework or processstudy

process can also be inferred from these studies.It appears to be quite protracted, involving con-siderable effort from subsidiary management todevelop the concept in the first place, and evenmore effort to get it accepted in other countries.However, this is not significantly different fromthe generic innovation process (e.g., Howell andHiggins, 1990), or from the process that can beinferred from prior studies of global or internalmarket initiatives (see below). There is thus con-siderable scope for new insight in this area.Finally, the intended outcome, according to Bart-lett and Ghoshal (1986), is to enhance world-wide learning.


8/24

Entrepreneurship in Multinational CorporationsGlobal market initiativesThese are driven by unmet product or marketneeds among nonlocal suppliers and customers.In theory, the subsidiary could potentially interactwith any customer or supplier in the world, butevidence in the literature suggests that suchinitiatives probably occur as extensions to existingrelationships.2 Consider the case of Litton Sys-tems Ltd. (Science Council of Canada, 1980).Litton developed an international business in the1960s (through a local market initiative) aroundan inertial navigation system. On the basis ofits worldwide customer base it then identifiedadditional opportunities in related areas, and wenton to develop products such as air traffic controlsystems and radar systems. These latter productintroductions were global initiatives, because thelocus of opportunity was outside the subsidiary'slocal market.With the exception of the Science Councilof Canada study, no evidence of global marketinitiatives was found in the literature. This isperhaps surprising, in that there are severalexamples of subsidiary companies taking on inter-national product responsibilities (e.g., Hedlund,1986; Roth and Morrison, 1992), which wouldnaturally provide the circumstances in whichglobal initiatives could transpire. In terms of thecharacteristics of global market initiatives, theScience Council of Canada suggested (a) thatlocal autonomy, local resources and existing inter-national responsibilities were facilitating con-ditions, and (b) that the intended outcome wasto leverage the subsidiary's existing capabilitiesinto related areas. However, these should beviewed as very tentative propositions becausethey arose from only three case studies. Therewas, unfortunately,no substantive insight into theinitiative process, except insofar as it appearedto mirror the generic process described above.Internal market initiativesThe concept of an internal market initiative issomewhat unusual, in that it arises through 'mar-ket opportunities' identified in the corporate sys-2 This point is substantiated by the body of literature oninternationalization which sees international relationshipsgrowing incrementally through experience and existing con-tacts (e.g., Johanson and Vahlne, 1977).

tem. The best way of explaining the concept isthrough an example: the quote below is withregard to Dow Chemical's internal market (Whiteand Poynter, 1990: 56):The internal product sourcing relationshipbetweena manufacturinglantanda commercialdepartmentan be 'challenged'at any time ...If the U.S. marketfor polyethylenecould besourcedat lowerout-of-pocket ost from a DowCanadaplant than elsewhere, that adjustmentwouldbe made.The initiative in this case would be the Cana-dian plant challenging the incumbent in-housemanufacturerfor the rights to polyethylene pro-duction, on the basis that the incumbent waseither not operating efficiently or was operatingin a country where factor costs put it at a com-parative disadvantage vis-a-vis the challengingplant. This type of initiative is thus subtly differ-ent from the other two types, because it isdirected towards cost reduction rather than rev-enue enhancement.The concept of an internal market, in whichdivisions or subsidiaries of a single companypursue competitive exchange relationships withone another, is well established in academic writ-

ing (e.g., Galunic and Eisenhardt, 1994; Halal,1993) and is widely used in many large corpora-tions. As before, however, the internal marketinitiative is implicit in prior research. Galunicand Eisenhardt (1994), for example, discussedcompetition from other subsidiaries as a stimulusfor charterloss; and White and Poynter's descrip-tion of the horizontal organization (above)included a clear example of subsidiary initiativethat was not labeled as such. The characteristicsof internal market initiatives can thus be inferredfrom the existing literature as follows: (a) localresources, some decentralization of decision mak-ing, good relationships with the parent companyand shared decision premises as facilitators ofinitiative; (b) an entrepreneurialselling process,again very similar to that seen in other types ofinitiative; and (c) efficiency in global operationsand desire for local value-added as the intendedoutcomes. Table 1 represents a summary ofprior research.In conclusion, this section set forth a concep-tualization of subsidiary initiative that is consist-ent with the emerging body of literatureon inter-organizational networks, but which at the same

213


9/24

214 J. BirkinshawTable1. Previousresearchon threeforms of subsidiarynitiativeLocus of market Local market nitiative Internalmarket nitiative Global marketnitiativeopportunitydefinition)Facilitating actors * Local resources, ocal * Local resources, ome * Localresources,ocalautonomy,normative decentralizationGalunic autonomy, xistingintegration,ub.-HQ andEisenhardt, 994) internationalcommunication,ntra-sub.* Local resources,good responsibilitiesSciencecommunicationGhoshal, relationswith parent Council of Canada,1980)1986) (ScienceCouncil ofCanada,1980)* Horizontalnetworkandshareddecisionpremises(WhiteandPoynter,1990)Process * No discernibledifferencebetweenprocessmodelsin all threetypes. Genericmodelindicatesa protracted elling processby subsidiarymanagemento parentmanagementEtemadand Dulude, 1986;Science Council of Canada,1980)Intendedoutcome * Enhanceworldwide * Efficiencyof global * Leverageexistinglearning Bartlettand operations nd local subsidiary apabilitiesntoGhoshal,1986) value-addedScience relatedareas(Science* Maximizeglobal Councilof Canada,1980; Council of Canada,1980)innovation Harrigan, White and Poynter,1990)1983)

time suggests a much broader opportunity set forthe subsidiary than previously thought. The threeinitiative types have been identified to varyingdegrees in the literature, so our a priori expec-tations are not sufficiently clear that researchpropositions can be explicated. The objective ofthe remainder of this paper is thus twofold: (1)to assess the validity of this conceptualization;and (2) to describe the facilitating conditions, theprocess, and the outcomes of the types of subsidi-ary initiative that are identified.

RESEARCH METHODOLOGYSampling methodologyDefining the research sample posed two methodo-logical problems. The first problem was that manyMNC subsidiaries have never undertaken initia-tives, so it was necessary to identify a sample ofsubsidiaries that had undertaken intiatives.3 The

3Note that this does not impart a bias to the results: theresearch objective was to understand the characteristics ofsubsidiary initiatives, not to understandwhy some subsidiariesundertake them while others do not.

second problem was identifying a sample ofinitiatives from the sample subsidiaries. Thedecision was made to research the entire set ofinitiatives in a small number of companies, onthe grounds that the quality of data was a criticalelement of this study. The alternative of a onecompany-one initiative study would also havebeen possible, but would have entailed very fewrespondents per initiative, and therefore less rich-ness in the data. If a single initiative had beenstudied for each subsidiary there would have beena danger of selection bias towards the respon-dents' more memorable-but not necessarilyrepresentative-initiatives. A third problem wasanticipated but did not materialize, namely theidentification of initiatives that met the definitionsposed at the outset. Kanter (1982: 99) noted thatthe delineation between an entrepreneurialactivity(i.e., an initiative) and a managerial activity ismuch easier in practice than it is in words, andthis study bore out her observation. Likewise, thestated need for all initiatives to lead to inter-national responsibilities never created anymethodological problems, in that every initiativeidentified, even the local market type, quickly(i.e., within 24 months) led to internationalsales.


10/24

Entrepreneurship in Multinational CorporationsThe following sampling methodology wasadopted. Using the Financial Post 500 index, alist of subsidiaries was drawn up with the follow-ing restrictions: (1) Canadian subsidiaries of U.S.-owned MNCs; (2) with 1992 sales revenues of

greater than $200 million; and (3) participatingin a global industry (Morrison, 1990). Thesecriteria were selected on the basis that a caseresearch design should intentionally limit the vari-ance in the sample to make cross-case analysismeaningful (Parkhe, 1993; Yin, 1984). Therewere also strong a priori reasons for the size andglobal industry restrictions, in that the strategicimperatives for smaller subsidiaries (which maybe sales-only) and subsidiaries in nonglobalindustries (with limited cross-border trade flows)are not germane to the management of the matureMNC. One additional criterion was alsodeveloped, namely clear evidence of initiative. Bypicking out those subsidiaries with an establishedrecord of success in taking the initiative or gain-ing international responsibilities we were able toavoid approaching companies that did not exhibitthe necessary levels of entrepreneurship.Approxi-mately 40 subsidiaries met all these criteria.Subsidiaries from this sample were thenapproached on a convenience basis. Two of fourinitial selections agreed to participate fully. Asresearch in these two subsidiaries progressed,additional subsidiaries were approached, and overthe course of 9 months four more were added(with a further two declining to participate). Thedecision to stop at six sites was based on theprinciple of redundancy (Yin, 1984), in that nosubstantive additional insights appeared to beforthcoming towards the end of the study. Thedata collection period lasted from November 1993to August 1994. Some basic facts about the sub-sidiary sample are displayed in Table 2.4In each sample company, initiatives were iden-tified through discussions with senior managers.The major initiatives, particularly those that hadresulted in new internationalresponsibilities, wereidentified immediately, but some careful investi-gation was required to pick out smaller or less-4The six subsidiaries in the final sample appeared to betypical of the larger sample of 40 subsidiaries. Their averagesales revenues in 1993 were $550 million compared to $440million in the larger sample. Furthermore, nformal interviewswere subsequently conducted in a further 12 of the 40 compa-nies, with qualitatively very similar findings to thosereported here.

successful initiatives. It was only towards the endof the research that some of the more interestingfailures were uncovered, presumably becauserespondents' comfort level with the researcherincreased with time. Between 3 and 10 initiativeswere identified in each company to give a totalsample size of 39 initiatives (Table 3).Data collection methodsThe two primary sources of data were (1) semi-structured interviews with subsidiary and parentcompany managers, and (2) a questionnaire filledout by the key individual for each initiative.These data were complemented by business plansand other company documents, and secondarydata compiled through a CD ROM library search.Interviews followed a carefully prepared protocol,with a mixture of specific questions ('What wasthe proposed dollar investment in this project?')and open-ended questions ('How did this initia-tive arise in the first place?). For each initiativethe intention was to speak to every manager whowas actively involved. Thus, between 2 and 10managers were interviewed for each initiative. In65 percent of cases one or more parent managersinvolved in the initiative were interviewed. Inter-views were taped and transcribed,and notes werealso taken. A total of approximately 1500 pagesof data were assembled through this process.The questionnaire was put together towards theend of the research, as a means of validatingthe qualitative interview findings. Questions weredrawn up to measure the emerging constructs,such as the nature of the market opportunity orthe level of selling by subsidiary management.Following a careful preparationprocess, in whichfour academicians and three subsidiary managers(not in the sample) provided feedback on wordingand design, the questionnaire was mailed to thekey respondent for each of the 39 initiatives.Thirty-five questionnaires were returned, and inthe four remaining cases the questionnaire wasfilled in by an independent research assistantfollowing a careful reading of the interview tran-scripts.55 More specifically, the research assistant completed a ques-tionnaire for every initiative (i.e., all 39) based on his readingof the transcripts.Interrater eliability for the 35 questionnaireswhere both the key respondent and the research assistant hadcompleted them was 0.64 (using Cohen's kappa; Perreaultand Leigh, 1989). Consistent with Nunnally (1978) this is

215


11/24

216 J. BirkinshawTable2. Subsidiary ample

ApproximateizeCompany Principalndustry (1993 revenues) Numberof interviewsA Industrial nd consumerproducts $600 milliona 22B Chemicals $450 million 19C Computerhardware/software $500 million 15D Computerhardware/software $800 million 19E Industrial roducts $420 million 14F Engineering ystems $420 million 11aFigures are in Canadian dollars.

Data analysisFive constructs were measured using question-naire data.6 The nonparametric Kruskal-Wallistest was conducted on these measures to deter-mine whether there were any significant differ-ences between the means of the different initiativetypes. Qualitative data were analyzed using theprocedures recommended by Miles and Huberman(1984), which emphasize the use of tables anddiagrams for reducing and visualizing data. Theanalysis was undertaken by both the researchers,and discrepancies reconciled through discussion.The qualitative findings were summarized in theform of a case history, and sent to the leadrespondent to verify that the case was a fairportrayalof what actually happened. Specific con-structs were then abstractedfrom the case studiesand compared to the quantitative findings. Theresults section (below) provides both sets of dataso that the convergence between the qualitativeand quantitative techniques can be judged.acceptable for the early stages of a research program. It wastherefore decided that the research assistant's responses werea reasonable proxy in the four cases where no responsewas forthcoming.6 Questions, all on 5-point Likert scales except the last one,were as follows. AUTONOMY: (a) What level of strategicautonomy did the Canadian subsidiary have?; PROVENRESOURCES: (a) To what extent did the subsidiary haveprevious internationalresponsibilities? (b) To what extent didthe subsidiary have a track record of success getting projectsapproved? PARENT-SUBSIDIARY COMMUNICATION:(a) What was the extent of the relationship between theparent and subsidiary management? INTERNAL SELLING:(a) What was the extent of the selling process followed bysubsidiary management? (b) How active was the Canadianpresident's involvement in pursuing the initiative?APPROVAL: (a) What was the immediate outcome (explicitapproval, implicit approval, rejection)?

RESEARCH FINDINGSTypes of subsidiary initiativeThe conceptual framework suggested that threetypes of initiative should be identifiable, on thebasis of the locus of the market opportunity(i.e.,local, internal and global). This proposition wasconfirmed to the extent that all three putativetypes were in evidence, but in addition thereemerged in the course of the research a fourthtype that was essentially a hybrid between theinternal market and global market forms. Thishybrid type took the form of subsidiary manage-ment identifying and bidding for an embryoniccorporate investment. For example, in one casesubsidiary management read in the corporatelong-range plan that a new plant was scheduled4 years hence. They recognized a fit with theCanadian subsidiary's capabilities so they built acase for making the investment in Canada andsolicited support for their proposal at head office.In this case the initial market opportunity wasidentified by head office managers but subsidiarymanagement proactively pursued it. Thus, therewas a heavy element of internal selling, to pers-uade head office management that the subsidiarywas the appropriate site for investment, but theinitial market opportunity was clearly global.The hybrid case can best be understood interms of the locus of opportunity vs. the locusof pursuit,7 in that its locus of opportunity wasglobal but its locus of pursuit was internal. In allthe other three forms the locus of opportunitycoincided with the locus of pursuit. This state of7We are indebted to an anonymous reviewer for suggestingthis distinction.


12/24

Entrepreneurship in Multinational CorporationsTable 3. Initiative sampleSub. Initiative Outcome Interviews

Initial proposal to manufacturein OntarioProposal for plant extension in ManitobaProposal to bring conversion to Ontario (a)Proposal to bring conversion to Ontario (b)Proposal to bring finishing to CanadaProposal to consolidate production in CanadaBid for major new tape plantRestructureof sales and marketing organizationBid for plant extensionNew library security productBid for new chemical facility in CanadaIncremental investment in Canadian plantBusiness management shifted to CanadaToll manufacturebrought in-houseBid for new formulation of major productNew dispensing product for major productProposed technological innovation in CanadaNew digital screen technologySystem controller productCutting equipment innovationElectronics for defense missileAirport terminal product innovationNew generation terminal productSecond generation missile electronicsCommunication network productLocal hardware company acquisitionIndustrial terminal product innovationNew high-technology terminalRegional product development centerSoftware development centerNew information protocol centerExport of valve to EuropeExport of switch manufacture to U.S.Rationalization of North American productionDesignation of two SBUs in CanadaSoftware system for building controlsSoftware/hardwaresystem for oil flowNew PC network management productLocal high-tech company acquisition

SuccessSuccessFailureSuccessSuccessSuccessSuccessSuccessSuccessMixedFailureSuccessSuccessSuccessSuccessSuccessFailureSuccessSuccessSuccessSuccessSuccessSuccessFailureFailureMixedMixedSuccessSuccessMixedSuccess

SuccessSuccessSuccessSuccessMixedSuccessSuccessSuccess

22232338521457522763342225661033451061656

affairs serves to underline the complexity of thesubsidiary's role within the 'interorganizationalnetwork' of the MNC. Rather than just focusingon one type of market opportunity at a time,there is also the need to reconcile emerging globalopportunities with internal capabilities. We mightargue that this is the job of the corporate center,

but the evidence shows that the proactive subsidi-ary can also take on parts of that role itself.Table 4 lists the number of cases found of eachinitiative type, with a description to facilitateunderstanding. In terms of methodology, thefourth type (the hybrid) emerged during the datacollection process. Then, during the data analysis

AAAAAAAAAABBBBBBBCCCCCCCCDDDDDDEEEEEFFF

217


13/24

218 J. BirkinshawTable 4. Four types of initiative

Number of Number ofInitiative type cases failures DescriptionLocal market initiative 13 2 Seek to develop a new product, market, or process through

opportunities that are first identified in the subsidiary's homemarketInternal market initiative 12 1 Promote the redistributionof existing corporate assets orresources such that they are more efficiently deployed. In theCanadian context the objective is typically to reconfigure abranch plant in the light of the North American Free TradeAgreementGlobalmarket nitiative 9 1 Seek to buildon an existingmandateor provencapability omeet a perceived nternationalroductor marketopportunityHybrid nitiative 5 1 Seeks to attracta global investmentwhich has already inprinciple)receivedcorporate upport

stage each initiative was assigned by the leadresearcher to one of the four types. The researchassistant replicated the process, on the basis ofthe interview transcripts, and made the sameassignment as the lead researcher in 37 of 39cases. The remaining two were discussed andagreement was eventually reached regarding theappropriate type. Next, a discriminant analysiswas undertaken on the questionnaire data, tocheck whether the four types could be dis-tinguished on the basis of the 'drivers' of initia-tive.8 This analysis yielded Wilks lambdas of0.06, 0.23 and 0.54 respectively for the threecanonical discriminant functions, representing avery high level of separation between the fourgroups. In sum, 33 of the 39 cases were correctlyclassified by this procedure, or 85 percent of thetotal. This is an excellent result, and providesfurther confirmation of the validity of thetypology.9

8The seven initiative drivers were: (1) desire to consolidateoperations with those of the parent; (2) creation of a NorthAmerican free-tradeenvironment; (3) routine product or busi-ness upgrade on existing product; (4) business opportunitydefined by a parent request for proposal; (5) a product-market opportunity arising through interaction with local cus-tomers; (6) a product-market opportunity arising throughinteraction with international customers; and (7) desire bysubsidiary management to enhance local value-added.9As discriminant analysis assumes multivariate normality andequivalent variance-covariance matrices, neither of which waswholly present given the small sample size, this result shouldbe interpretedwith caution.

Facilitators of initiativeLocal market initiativesThe 13 cases of local market initiatives suggestedan interesting duality to the roles of subsidiaryautonomy and parent-subsidiarycommunicationasfacilitatorsof initiative that had not been identifiedbefore. At the formative stage autonomy had to behigh and communication correspondingly low sothat subsidiary resources could be applied to theopportunitywithout head office interference;at themore advanced stage of viability a higher level ofcommunicationand a lower level of autonomyweremore appropriate, n that subsidiary managementhad typically to achieve sponsorship for the busi-ness in question from a U.S.-based operatingdivision. In one case a subsidiarymanagerbuilt a$20 million business from nothing in 4 years, buttook a further 4 years to find a 'home' withone of the major corporate business groups. Hisobservation was that autonomy had helped him tomove quickly early on, but was a liability as heattemptedto enlarge the business and integrate itinto the corporatenetwork.In terms of resources, it became apparent inthe course of the study that proven resources(i.e., those which were recognized by parentmanagement) were a more important facilitatorof intiative than resources per se.'0 However, in

0 This finding is consistent with process models of strategysuch as Bower (1970) in which resource allocation decisionsare typically made on the basis of the individual's trackrecordrather than on purely economic or technological arguments.


14/24

Entrepreneurship in Multinational Corporationsthe specific case of local market initiatives, thesample companies only exhibited moderate levelsof proven resources. Respondents commented thatthe subsidiary had to have sufficient experienceand/or expertise to pursue opportunities as theyarose, but without the high levels of provenresources that were necessary to succeed withsome of the other types.Respondents observed that internal marketinitiatives were facilitated most effectively by thecredibility of the subsidiary in the eyes of theparent company, which was a function, in turn,of the subsidiary's high level of proven resources.In cases where the subsidiary had already builtup a number of world product mandates throughprior initiatives, subsequent initiatives progressedmuch more smoothly. Credibility was also a func-tion of a high level of parent-subsidiary com-munication and, by the same token, a fairly lowlevel of autonomy in that subsidiary managershad to be working very closely with their U.S.counterparts. As described by one respondent:

It is awfully important hat we have a closeassociation with the U.S. management].We aretalking frequentlyabout what are the issues intheirbusiness,what are theirproblems,what arethe opportunitieshat we can offer to help themsolve theirproblems.Thatis importanto do.One facilitator of internal market initiatives thathad not been identified in prior research was theglobal orientation (Perlmutter, 1969) of the seniormanagement in the United States. Some wereessentially ethnocentric in their approach, whichcreated enormous obstacles for subsidiary man-agement; others took a geocentric attitude whichstreamlined the entire process, as the followingquote suggests.[The generalmanager]had a kind of 'let's dothe right thingthen tell everyone'attitude.Let'snot be politicalaboutthis, let's collaborateanddo the right thing.What'srightto do hereis notwhat'srightfor our own camp,it's what'srightfor the corporation. o we got backingon thisand got him interested.He then adopted t andbecame the overall mentorfor it, and kept upthe momentum.Global market initiatives appeared to be facili-tated by high autonomy, a high level of provenresources, and a correspondingly low level ofparent-subsidiary communication. The impor-tance of autonomy was underlined by one case

in which the subsidiary had achieved great suc-cess in building a viable international business,but where the parent company had then curtailedits autonomy because of corporate financialdifficulties. Pursuing intiatives suddenly becamea time-consuming and frustratingprocess, as thisquote suggests:

The basic dilemmafacing [the generalmanagerof the subsidiary] s lack of investment.If hewants $100,000 to develop a productthe cus-tomer is payingfor he has to make a coupleofvisits to head office, which might take threemonths. By the time approval s granted,theopportunity as passed.Proven resources here referred to a history ofsuccessful initiatives and an accumulation over

time of specialized and valued capabilities. Allthe subsidiaries undertaking global market initia-tives in this study were essentially building onexisting international responsibilities or worldmandates (Etemad and Dulude, 1986). Withregard to parent-subsidiary communication, mostrespondents felt that initiative was facilitated bylow levels of communication. This is not tosuggest that communication is damaging, but thatit is a low priority and is liable to be limited ina high-autonomy subsidiary. This quote is indica-tive of the level of parent-subsidiary communi-cation that was observed:

... [The head office boss] was looking at thenumbers, nd 'other ncome' was quitelarge.Hesaid 'What's that?' and [his colleague] said'that's the electronicsgroupup in Canada.'Somy head office boss called me and said, 'Wedon'tknow whatyou'redoingup there,butkeepit up.' Isn'tthatrepresentativef therelationship!Hybrid initiatives were facilitated by very simi-lar factors to internal market initiatives. That is,the credibility of the subsidiary with head office

decision-makers was felt to be critical, and thiswas typically a function of moderate to highlevels of proven resources, strong parent-subsidiary communication and relatively lowautonomy. One subsidiary president commentedon the nature of the bid process in his company:You end up with a couple of sites that comepretty close and one that will have a minoradvantage economically, but sitting in anoperatingcommitteein the States, what reallyswings you is the credibilityof the organizationthat'saskingfor the order.

219


15/24

220 J. BirkinshawThe fact that the market opportunity in thehybrid initiative was global had little bearingon the facilitating conditions, because the entireprocess was internal to the MNC. The evidence,in fact, suggested that hybrid initiatives required

the highest level of 'selling' of all four types,which in turn necessitated a high level of ongoingparent-subsidiary communication.

Triangulation of qualitative and quantitativedataThe questionnaire data that tapped into the threemain constructs under discussion is displayed inTable 5. The nonparametric Kruskal-Wallis testwas conducted to determine whether there wereany significant differences between the means ofthe four initiative types. Despite the small samplesize (between 5 and 13 observations for eachtype), two significant results were recorded, bothconsistent with the qualitative data. First, the highlevel of autonomy in global market initiativeswas confirmed, in relation to all other types.Second, parent-subsidiary communication wasdifferent (at a marginal level of significance)across the types, with local and global initiativesexhibiting low levels of communication andinternal and hybrid initiatives exhibiting high lev-els. The two measures selected as proxies forproven resources (existing international responsi-

bilities and a record of success with initiatives)were not able to distinguish between initiativetypes, though there was a suggestion that provenresources were slightly lower in local marketinitiatives. While consistent with the qualitativedata, this finding is at odds with Ghoshal's (1986)observation that local innovations are associatedwith high levels of local resources.Taken together, the quantitative and qualitativedata provided several importantnew insights intothe facilitators of subsidiary initiative. Internalmarket and hybrid initiatives exhibited a higherlevel of integration (in terms of parent-subsidiarycommunication and low autonomy) than previousstudies suggested, and appeared to rely on geo-centrically minded parent company managers tobe successful; local market initiatives appearedtobe facilitated by a careful balance between auton-omy and integration;and global market initiativeswere exhibited only where the subsidiary wasvery autonomous. The fact that autonomy wasapparently so critical to the global market typeis interesting, because a priori one might notexpect local and global market initiatives to besignificantly different. This evidence suggests thatsubsidiaries can not easily build world mandatebusinesses while at the same time remaining inte-grated with the rest of the corporation, but it isat odds with several studies that have suggestedthis is a desirable combination (e.g., Bartlett andGhoshal, 1989; Roth and Morrison, 1992).

Table5. Questionnaire ataon initiative acilitatorsGlobal-internalLocal market Internal market Global market hybrid Kruskal-Wallis Pairs significantlyinitiative initiative initiative initiative ANOVA (F/sig.) different

Subsidiary 3.0 2.6 4.4 2.8 11.2/0.01 Global with allautonomy othersSpecializedresources:(a) Existing 3.2 3.7 4.1 3.6 3.1/0.38 Noneinternationalresponsibilities(b) Recordof 2.9 3.8 3.4 4.0 4.3/0.23 Nonesuccess withinitiativesParent-sub. 3.1 3.9 3.0 4.2 7.4/0.06 Localandcommunication hybridAll measures on 5-point likert scales where 5 = high; 1= low.


16/24

Entrepreneurship in Multinational CorporationsInitiative processesThe qualitative findings shed considerable lighton the initiative process, though in broad termsthey were consistent with expectations. Two dis-tinct processes were identified: one internallyfocused and the other externally focused. Theinternally focused process was exhibited byinternal market initiatives and hybrid initiatives,in that formal corporate approval was necessaryfor resources to be made available. There wastherefore a high level of selling, first of all bymiddle-level managers to their superiors in thesubsidiary, and subsequently by the top subsidiarymanagers to their superiors in head office. Thisprocess is encapsulated by the following quoteby a middle-level manager regarding an internalmarket initiative:

I said really we should make a play for [thisbusiness], and startedto build the argument.soundedout the [U.S. businessmanager]:Whatare the possibilitieshere? What about runningthe businessfrom Toronto?What do you thinkabout it?' He basically thought it had merit, andhe coachedme. But my sales effort was not totry to convince U.S. people beyond my sphereof influence, t was reallyto get the peoplehereconvinced,to providethem a positionthat theycould then embellish. So I worked with themand ultimately the CanadianPresident]was theguy to say 'We would like to do this' at a verysenior level in head office.While much of the internal selling took theform of an upward progression through the hier-archy (Bower's 1970 'impetus'), there was alsosome evidence of horizontal selling as well. Manymiddle managers in the subsidiary were on globalbusiness teams, and were thus in a position toinfluence their head office and internationalpeers,and subsidiary top management were typicallyconnected into an extensive lateral network

through which support for initiatives could bebuilt. The following quote illustrates the extensiveselling process undertaken by one subsidiaryCEO:First he had to get approval for the initiative)from the operations committee, who reportdirectlyto the chairman.Then he went to thesectormeetings,whereyou hadthe divisionVPs.There were three of them. ... He then went to acouple of othercorporatebodies, typical placeswhereyou would showcasethis kind of thing-the marketing ouncil, the technicalcouncil as

well, which is a huge group of the laboratorymanagers.So havingcascaded t down he triedto pick large bodies where he would get to thelevel below division VP.The concept of a horizontal organization super-

imposed on the traditionalM-form has been docu-mented by several academics (Bartlett and Gho-shal, 1993; Hedlund, 1986; White and Poynter,1990). In terms of subsidiary initiatives, it was avaluable source of support, though typically onlyas a complement to the vertical chain of com-mand through which resource allocation occurred.Were there any differences between theobserved process in internal market initiatives andhybrid initiatives? The distinguishing character-istic was the level of involvement of parent com-pany management, in that hybrid initiativesalways had parent management support in prin-ciple from the start, whereas internal marketinitiatives had to build their own support. Thiscreated a rather subtle difference in process:internal market initiatives were iterative, involv-ing several rounds of credibility building withparent management and refining of proposals;hybrid initiatives were 'take-it-or-leave-it' pro-posals in which parent management typically hadto choose between several directly competingcourses of action.The local market and global market initiativeswere externally focused. Head office approval wastypically implicit, so the majority of the efforton the part of subsidiary management was dedi-cated to building a viable product for local andglobal customers respectively. In the case of localmarket initiatives the subsidiary usually tookresponsibility for developing the concept usinglocal development funds or bootlegged resources,without the support-and sometimes even withoutthe knowledge-of the parentcompany. As statedin the previous section, approval from the parentwas only sought on average several years laterwhen the business had become viable. In the caseof global market initiatives the subsidiary almostalways had authority to invest in new projectswithin the subsidiary's existing charter (Galunicand Eisenhardt, 1994). The appropriate parentcompany division was kept informed of all suchinvestments, but they did not intervene in theprocess. Championing was thus relegated to aninternal subsidiary activity, as middle managerssought to convince the subsidiary general man-ager that their initiative should be pursued.

221


17/24

222 J. BirkinshawIn general terms, then, the initiative processtook one of two forms, neither of which corre-sponded precisely to the classic formulation ofBower (1970). Where resource allocation was acentralized phenomenon (i.e., with internal market

and hybrid initiatives), horizontal systems com-plemented the vertical chain of authority as ameans of building legitimacy and momentum forthe initiative. Where resource allocation wasdecentralized, selling occurred primarily withinthe subsidiary and head office approval wasimplicit. It is not clear the extent to which thisfinding is specific to the sample, which was delib-erately skewed to favor initiative-taking subsidi-aries. Bartlett and Ghoshal (1993) observed thattheir 'new' organizational model, which includeda very similar extension to the Bower model ofresource allocation, was induced from a sampleof leading-edge companies, some of which werethe same as used in the current study. Thus thereis some suggestion that these processes may notbe representative of the population of MNCs.Future research will be necessary to examinethis question.Triangulation of qualitative and quantitativedataThe questionnaire data that tapped into the twoconstructs under discussion are displayed inTable 6. The nonparametric Kruskal-Wallis testwas conducted on the two internal selling meas-ures; the approval process data were nominal so

they are presented in raw form. The results pro-vide strong support for the qualitative findingsdiscussed above. Both local and global initiativetypes exhibited a lower level of internal sellingthan their internal and hybrid counterparts, andthis difference was significant for the subsidiarypresident's selling activity. Equally, the approvalprocess was predominantly implicit for the localand global types (13 out of 19 cases), and explicitfor the internal and hybrid types (14 out of 15cases). In sum, there appears to be a very clearsplit between the two broad processes.Initiative outcomesSetting aside the five that were not successful,the 34 remaining initiatives all led to increasedsales, investment from head office, and new jobs.As Table 7 shows, however, the differences inthese measures between types were mostly small.The only substantive difference, in fact, was thatthe internal market and hybrid initiative typesboth involved higher levels of capital investmentthan the other two. This is presumably a functionof the resource allocation system in the MNC.Where subsidiary managers had the authority toapprove the initiative (as with local and globalmarket types) they proceeded in an incrementalfashion, investing relatively small sums each time.Where approval was centralized at the headoffice, investments were typically larger. As withthe data on facilitating conditions and the initia-tive process, this process emphasized the simi-

Table6. Questionnaire ata on initiativeprocessGlobal-internalLocal market Internal market Global market hybrid Kruskal-Wallis Pairs significantlyinitiative initiative initiative initiative ANOVA (F/sig.) different

Internal selling: 3.6 4.1 3.3 3.8 3.9/0.27 None(a) by subsidiarymanagement(b) by subsidiary 2.8 3.8 2.4 3.8 7.3/0.06 Local andpresident internal;Global andinternalNatureof approval 4 explicit 10 explicit 2 explicit 4 explicit - Local andprocess 8 implicit 1 implicit 5 implicit 0 implicit globaldifferent1 rejected 1 rejected 2 rejected 1 rejected to internalandhybridAll measures on 5-point likert scales (except approval process) where 5 = high; 1 = low.


18/24

Entrepreneurship in Multinational CorporationsTable7. Initiativeoutcomes

Local market Internalmarket Globalmarketinitiatives initiatives initiatives Hybrid nitiativesAveragenew $2.2 million $4.8 million $1.7 million $8.6 millioninvestmentnsubsidiary s a resultof approvalAveragenew sales for $5.2 million $10.5 million $7.8 million $9.2 millionsubsidiarywithin twoyearsLong termoutcomes Local value-added Local value-added Specialized Local value-addedstatedby respondents Customer Competitive capability Optimally ocated(subjective) responsiveness subsidiary Local value-added new facilityNew businessfor operations DevelopmentofMNC Integrated roduction centerofexcellence

larity between the hybrid and internal marketinitiative types.Qualitatively, the outcomes of the four types ofinitiative were markedly different. Local marketinitiatives led, in the first instance, to an enhancedservice to local customers, but as they developedthey led to new business opportunities for theMNC as a whole. Two of the 13 cases in thisstudy became 'blockbuster' products (with rev-enues in 1994 of $50 million and $110 millionrespectively), while the rest led to niche busi-nesses, or businesses that lasted a few years andwere then phased out. Local market initiativescan therefore be viewed as enhancing worldwidelearning as well, in that opportunities identifiedin the Canadian market were addressed and thenapplied in other countries. More broadly, localmarket initiatives are also instrumental to theimperative of corporate adaptation and renewal,in that they provide the variety that the MNC'ssystems can then select against (Burgelman,1991). Without the diversity of opportunities andideas that local market initiatives represent, theMNC's ability to adapt to changing environmentaldemands would be severely constrained.Internal market initiatives, as predicted, led toa rationalization of activities between Canada andthe United States, and hence a more efficientcorporate system. Typically a branch plant wentfrom producing a broad range of products for the

Canadian market to one or two products on aNorth American or global basis. Overall volumesstayed the same (initially at least), but both U.S.and Canadian plants increased their export sales.There were also cases of rationalizationin admin-istrative functions: in one case, for example, prod-uct management was relocated to Canada to bemore closely integratedwith the associated manu-facturing. Both outcomes are symptomatic of theshift in MNCs towards a geographical concen-tration of value activities (Porter, 1986).The outcome of the global market initiativesin the study was the maintenance and develop-ment of a specialized corporate capability. Thatis, each initiative sought to build a new productor market around an existing business line usingthe distinctive capabilities resident in that subsidi-ary. The term 'center of excellence' was used byrespondents in this regard, with the implicationthat the parent company and other subsidiariesalso stood to benefit from those capabilities. Interms of the corporate objectives identified at theoutset, then, this is another facet of worldwidelearning. Tangentially, it does suggest that theconcept of worldwide learning is multifaceted,with at least two separate characteristics: (1) thetransfer of information about customer needswithin the corporatenetwork, as achieved throughlocal market initiatives; and (2) the transfer ofproprietary technology and other capabilities

223


19/24

224 J. Birkinshawwithin the corporate network, as achieved throughglobal market initiatives. Both appear to beimportant strategic imperatives for the MNC.Finally, hybrid initiatives had a similar out-come to the internal market type, in that they ledto a geographically concentrated value activityserving the North American or global market.The four successful cases identified in this studymade use of both the comparative advantages ofCanada (relative to the United States) such ascheap power and a low-cost, low-turnover work-force, and also the specialized capabilities of thesubsidiary. This is a facet of the global integrationimperative, but it is actually superiorto the ration-alization process that internal market initiativespromote because it results (in theory at least) inthe positioning of the value-adding activity at theoptimum global location, rather than an existinglocation. Table 8 provides a summary of the find-ings: it is similar in format to Table 1, so thefindings can be easily compared to the a prioriexpectations.

DISCUSSIONMNC subsidiary managementA major contribution of this study is its docu-mentation of an internal subsidiary phenomenon,in contrast to most previous research that hasconcerned itself more with aspects of the parent-subsidiary relationship. Many recent studies(Bartlett and Ghoshal, 1993; Birkinshaw, 1995;Humes, 1993; Quelch, 1992) have suggested thatthe parent-subsidiary relationship is multifaceted,in that it varies across business units and operatesat multiple levels of management. By focusinghere on the initiative as the unit of analysis theproblem of defining a parent-subsidiary relation-ship was circumvented. Certain initiatives werefound to be specific to a single plant so involveda single relationship with a U.S. manufacturingdirector; others involved the whole spectrum ofbusiness units and all their relationships with theirU.S. counterparts. In all cases, however, it waspossible to examine a generic process basedaround the nature of the initiative, rather than a

Table8. Summaryof findings romcurrent tudyInternalmarketLocal market nitiative initiative Globalmarket nitiative Hybrid nitiative

Facilitating ? Low parent-sub. * High parent-sub. * High autonomy * High parent-sub.conditions communication communcation ? Low parent-sub. communication(quantitative) * Low autonomy communication * Low autonomy* High autonomyat * Strongproven (quantitative) * Strongprovenfirst resources,hence * Strongproven resources,hence* Moderateproven credibility resources qualitative) credibilityresources qualitative) Geocentricperspective * Geocentricperspectivein parentcompany in parentcompany(all qualitative) (all qualitative)Process * Low to moderate * High internal elling * Low internal elling * High internal ellinginternal elling * Explicitapproval * Implicitapproval * Explicitapproval* Implicitapproval processb qualitative processa qualitative processb qualitativeprocessa qualitative andquantitative) andquantitative) and quantitative)and quantitative)Intended * New business for the * Rationalizationf * Enhancement nd * Optimum lobaloutcome MNC;local existingactivities; internationaleverage siting of new value-opportunityeveraged increasedefficiency of an existing product addingactivityworldwide (qualitative) line or business (qualitative)(qualitative) (qualitative)aThe global market initiative process involved somewhat less internal selling and greater levels of implicit approval than thelocal market initiative process.bThe internal market and hybrid initiative processes differed only in subtle ways. Hybrid initiatives involved significantlyearlier parent involvement than local market initiatives, but the magnitude of the investment in question coupled with thecompetitive nature of the bid meant that a stronger selling effort was typically observed. The internal market initiative process,by contrast, tended to be slightly more incremental in nature, often taking several years to come to fruition.


20/24

Entrepreneurship in Multinational Corporationssomewhat artificial conceptualization of theparent-subsidiary relationship.This research also embraced a broader concep-tualization of subsidiary initiative than had pre-viously been identified. Taking Ghoshal (1986)as the definitive piece of work in this area,subsidiary initiatives can be focused either onlocal market opportunities ('local for localinnovations') or on global market opportunities('local for global' or 'global for global'innovations). This research has shown, inaddition, that initiatives can be internally focused,towards a rationalization of existing activities orthe promotion of new ones. Viewed in this way,the subsidiary suddenly has the potential toenhance the local responsiveness, global inte-gration and worldwide learning capabilities of theMNC. This is a significantly broader role thanprevious research has suggested. The implicationis that the MNC that is able to harness thefull entrepreneurial capabilities of its subsidiariesstands to gain competitive advantage.How can the subsidiary's entrepreneurial capa-bilities be most effectively harnessed? The firstchallenge is to create an appropriate structuralcontext, that is, one that facilitates entrepreneur-ship. Ghoshal's (1986) research showed that cet-eris paribus high autonomy, specialized resources,high normative integration and high interunitcommunication were associated with subsidiaryinitiative. This study suggested a more complexset of relationships. Autonomy, for example, wasshown to be positively associated with local andglobal market initiatives and negatively associatedwith internalmarketand hybrid initiatives.11Like-wise, the other facets of structuralcontext actuallyvaried between initiative types as well. The impli-cation is that a single structural context cannotfacilitate all four types of initiative. If a subsidiaryis highly integrated with its parent, for example,it can easily pursue internal market and hybridinitiatives, but less easily undertake local orglobal market initiatives.The implicit trade-offs that the parentcompanyfaces in shaping the subsidiary's structuralcontextare reduced when one recognizes that the subsidi-ary is itself differentiated. One division of thesubsidiary can be closely integrated with its par-'' Qualitatively, the evidence suggested a causal relationshipbetween automony and initiative in the directions indicated,though this cannot be verified with the quantitative data.

ent; another may be largely autonomous. GECanada, for example, has 11 divisions each oneof which has a unique relationship to its respec-tive parent division in the United States. Further-more, the subsidiary's structural context and itsassigned role are not cast in stone. Over time asuccessful initiative-taking subsidiary wouldexpect to impact its own strategic context(Burgelman, 1983b) and hence its perceived rolewithin the MNC. One of the subsidiaries in thisstudy, for example, built a new business fromscratch in Canada. As the division in questiongrew it developed an internationalcustomer baseand a unique set of capabilities so that eventuallyit operated as a stand-alone global business. Overthis period its emphasis shifted from local marketinitiatives to global market initiatives, and corre-spondingly its structural context also changed toaccommodate its new role.In sum, the idea that subsidiary roles can bedifferentiated through contextual mechanisms(Bartlett and Ghoshal, 1986) is a powerful one,but not without its limitations. This study hasshown that context needs to be differentiated atthe sub-subsidiary level (typically the division,business unit or plant) if the full scope of initia-tive types is to be facilitated. It has also suggestedthat a more dynamic approachto role and contextmanagement is appropriate,given that the subsidi-ary's opportunity set and internal capabilities arecontinually evolving.The second challenge facing the MNC that isattempting to enhance the entrepreneurial capa-bilities of its subsidiaries is to develop an entre-preneurial culture, i.e., one that motivates itsemployees to take the initiative. The review ofthe corporate entrepreneurship iteratureidentifiedthis as a key imperative, but the current studywas not able to shed light on it. By focusing onsix companies that had all successfully pursuedinitiatives, it would be reasonable to deduce thatall had relatively entrepreneurialcultures. Futureresearch, in which a sample of initiative-takingsubsidiaries is compared with a sample of nonin-itiative-taking subsidiaries, will be necessary tounderstandexactly what an entrepreneurialculturemeans for a MNC.Corporate entrepreneurshipEntrepreneurship in this paper was defined asalertness to market opportunity. While this con-

225


21/24

226 J. Birkinshawceptualization is well accepted in the literature(Kirzner, 1973; Stevenson and Jarillo, 1990), tra-ditional usage was extended by modeling 'mar-ket' to include entities both internal and externalto the MNC. This insight, in turn, led to arecognition that there were two distinct entrepre-neurial processes at work in the sample compa-nies: (1) 'internal entrepreneurship'(internal andhybrid initiatives) in which initiatives were sub-ject to corporate selection mechanisms such aslegitimacy and approval; and (2) 'externalentrepreneurship'(local and global initiatives) inwhich initiatives were subject to environmentalselection mechanisms such as customer accept-ance and survival. The concept that the corpo-ration can create its own variation-selection-retention mechanism is, of course, not new (e.g.,Burgelman, 1991), but the data in this case pro-vided strong empirical support for what is arelatively underresearchedphenomenon.To what extent did a sample of MNC subsidi-aries make this research on corporateentrepreneurship a special case? The Canadiansetting is interesting (and rich in data) becausethe recent transition to North American FreeTrade, and the perceived threat to Canadian oper-ations, has induced many Canadian subsidiariesto actively look for ways to add value. Further-more, the problems of building relationships andgaining credibility with decision-makers are allexacerbated by the geographical and (relativelyminor) cultural distances between subsidiary andhead office. In addition, the concepts of 'local'and 'global' initiative are clearly designed toapply to foreign subsidiaries and not domesticentities. Notwithstanding these facts, our positionis that the subsidiary context implies a differenceof degree rather than of kind. It is a smallstretch (conceptually and physically) to move asubsidiary plant in London, Ontario to Buffalo,New York, so we suggest that the issues illumi-nated here are no less portable. While the primaryfocus of the research was obviously on MNCsubsidiary management, the implications for cor-porate entrepreneurshipare substantial.

CONCLUSIONThe objective of this paper was to document anempirical investigation of subsidiary initiatives,and to understand them in terms of the existing

theory of the MNC. The key finding was thatfour types of subsidiary initiative can be iden-tified. Previous research by Ghoshal (1986) hadindicated that subsidiary initiatives can be focusedeither on local market or global market oppor-tunities. This research demonstrated, in addition,that initiatives can be internally focused, towardsa rationalization of existing activities or towardsthe promotion of new ones. This finding has twoimplications: first, it suggests that the subsidiaryhas the potential to drive the local responsiveness,global integration and worldwide learning capa-bilities of the MNC, a much broader role thanpreviously recognized; second, it indicates thatthe differentiationof subsidiaryroles throughcon-textual mechanisms (Bartlett and Ghoshal, 1986)has its limitations. Both of these implications areexplored in the discussion.This study had a number of limitations. First,the sample was drawn from a single country,which was appropriate given the state of knowl-edge about subsidiary initiatives (Parkhe, 1993;Yin, 1984) but also limiting with regard to exter-nal validity. Subsequent research in other coun-tries and with parents of other nationalities is theimportant next step in building knowledge aboutsubsidiary initiatives. Second, the sample wasselected to include only those subsidiaries withsome record of success with initiatives. This,again, was appropriate given the need to under-stand the characteristics of initiatives, but it begsthe question 'What factors are responsible forpromoting initiative in some subsidiaries and sti-fling it in others?' Now that the types of initia-tives are better understood a follow-up studyaddressing this question can be conducted.In terms of the methodology, three limitationswere evident. The first was the challenge ofcollecting data on failed initiatives. For a varietyof reasons managers were reluctant to dwell ontheir less auspicious moments, so a small numberof failed initiatives may have been missed in thesample companies. This problem could potentiallybe mitigated by spending longer in the company,but there are clear trade-offs in terms of dimin-ishing returns and potentially alienating time-con-strained managers. A real-time study would alsosolve this problem, but would be very time inten-sive. Equally problematical were those initiativesthat died out before they took off. Rather thanbeing failures per se, such initiatives simply didnot register as important in the minds of the


22/24

Entrepreneurshipin Multinational Corporationsrespondents. Again, a real time study would bethe only way to circumvent this concern. Finally,the questionnaire was deliberately short to ensurea high response rate, but this meant that one-and two-item construct measures were used. Forfuture research greater attention needs to be givento the development of a valid and reliablemeasurement instrument.Future research, then, is recommended in twodirections. The first thrust should be towardsa comprehensive understanding of the initiativephenomenon in other MNC settings, and in alarger sample of subsidiaries. Of interest hereis not only the generalizability of the initiativecharacteristics identified in the current study, butalso the characteristics of subsidiaries that exhibitinitiatives. The second research thrust should bedirected towards the personal motivation ofemployees to pursue initiatives. While the currentstudy identified the intended outcomes behindinitiatives, it was unable to inform the discussionof why certain individuals choose to pursueentrepreneurialopportunities while others do not.If we are to build a complete model of subsidiaryinitiative, the critical element may be the personalmotivation and the spark of creativity that setsthe whole process in motion.

ACKNOWLEDGEMENTSThe thoughtful comments of Nick Fry, PaulBeamish, SumantraGhoshal, Allen Morrison andRod White are gratefully acknowledged. Financialassistance was provided by the Plan for Excel-lence, Richard Ivey School of Business. Researchassistance from Laura MacLellan and NigelOwens is also acknowledged.

REFERENCESBarnard, C.I. (1938). The Functions of the Executive.HarvardUniversity Press, Cambridge, MA.Bartlett, C. A. and S. Ghoshal (1986). 'Tap yoursubsidiaries for global reach', Harvard BusinessReview, 64(6), pp. 87-94.Bartlett,C. A. and S. Ghoshal (1989). Managing acrossBorders: The Transnational Solution. HarvardBusi-ness School Press, Cambridge, MA.Bartlett, C. A. and S. Ghoshal (1993). 'Beyond the M-Form: Toward a managerial theory of the firm',Strategic Management Journal, Winter SpecialIssue, 14, pp. 23-46.

Birkinshaw, J. M. (May-June 1995). 'Is the countrymanager an endangered species?', InternationalExecutive, pp. 279-302.Bower, J. L. (1970). Managing the Resource AllocationProcess. Irwin, Homewood, IL.Brandt,W. K. and J. M. Hulbert(Winter 1977). 'Head-quarters guidance in marketing strategy in the multi-national subsidiary', Columbia Journal of WorldBusiness, 12, pp. 7-14.Burgelman, R. A. (1983a). 'A process model of internalcorporate venturing in the diversified major firm',Administrative Science Quarterly, 28, pp. 223-244.Burgelman, R. A. (1983b). 'A model of the interactionof strategic behavior, corporatecontext and the con-cept of strategy', Academy of Management Review,8(1), pp. 61-70.Burgelman, R. A. (1991). 'Intraorganizationalecologyof strategy making and organizational adaptation:Theory and field research', Organization Science,2(3), pp. 239-262.

Bums, T. J. and G. Stalker (1961). The Mangement ofInnovation. Tavistock Publications, London.Cohen, L. E. and R. Machalek (1988). 'A generaltheory of expropriativecrime: An evolutionary eco-logical model', American Journal of Sociology, 94,pp. 465-501.Covin, J. G. and D. P. Slevin (Fall 1991). 'A concep-tual model of entrepreneurshipas firm behavior',EntrepreneurshipTheory and Practice, pp. 7-25.Cray, D. (1984). 'Control and coordination in multi-national corporations',Journal of InternationalBusi-ness Studies, 15(3), pp. 85-98.Crookell, H. H. (1986). 'Specialization and inter-national competitiveness'. In H. Etemad and L. S.Dulude (eds.), Managing the Multinational Subsidi-ary. Croom Helm, London, pp. 102-111.Damanpour, F. (1991). 'Organizational innovation: Ameta-analysis of effects of determinants and moder-ators', Academy of Management Journal, 34,pp. 555-590.D'Cruz, J. R. (1986). 'Strategicmanagement of subsidi-aries'. In H. Etemad and L. S. Dulude (eds.), Man-aging the Multinational Subsidiary. Croom Helm,London, pp. 75-89.Drucker, P. (1985). Innovation and Entrepreneurship.Harper& Row, New York.Etemad, H. and L. S. Dulude (1986). Managing theMultinational Subsidiary. Croom Helm, London.Forsgren, M. and J. Johanson (1992). Managing Net-works in International Business. Gordon & Breach,Philadelphia, PA.Galbraith,J. (Winter 1982). 'Designing the innovatingorganization', Organizational Dynamics, pp. 5-25.Galunic, D. C. and K. M. Eisenhardt (1994). 'Theevolution of intracorporatedomains: Division charterlosses in high-technology, multidivisional corpora-tions', working paper, INSEAD.Ghoshal, S. (1986). 'The Innovative Multinational: ADifferentiated Network of Organizational Roles andManagement Processes', unpublished doctoral dis-sertation, Harvard Business School.Ghoshal, S. and C. A. Bartlett (1988). 'Creation,adop-tion and diffusion of innovations by subsidiaries of

227


23/24

228 J. Birkinshawmultinationalcorporations', Journal of InternationalBusiness Studies, 19(3), pp. 365-388.Ghoshal, S. and C. A. Bartlett (1991). 'The multi-national corporation as an interorganizationalnet-work', Academy of Manag

Documents

Birkinshaw SMJ 1997 - Theme 4