Control and Performance in International Joint Ventures

University of RichmondUR Scholarship Repository

Management Faculty Publications Management

5-1997

Control and Performance in International JointVenturesHans Mjoen

Stephen TallmanUniversity of Richmond, [email protected]

Follow this and additional works at: http://scholarship.richmond.edu/management-faculty-publications

Part of the Business Administration, Management, and Operations Commons, and theInternational Business Commons

This Article is brought to you for free and open access by the Management at UR Scholarship Repository. It has been accepted for inclusion inManagement Faculty Publications by an authorized administrator of UR Scholarship Repository. For more information, please [email protected].

Recommended CitationMjoen, Hans, and Stephen Tallman. "Control and Performance in International Joint Ventures." Organization Science: A Journal of theInstitute of Management Sciences 8, no. 3 (May/June, 1997): 257-74. doi:10.1287/orsc.8.3.257.

http://robins.richmond.edu/?utm_source=scholarship.richmond.edu%2Fmanagement-faculty-publications%2F11&utm_medium=PDF&utm_campaign=PDFCoverPages

http://robins.richmond.edu/?utm_source=scholarship.richmond.edu%2Fmanagement-faculty-publications%2F11&utm_medium=PDF&utm_campaign=PDFCoverPages

http://scholarship.richmond.edu?utm_source=scholarship.richmond.edu%2Fmanagement-faculty-publications%2F11&utm_medium=PDF&utm_campaign=PDFCoverPages

http://scholarship.richmond.edu/management-faculty-publications?utm_source=scholarship.richmond.edu%2Fmanagement-faculty-publications%2F11&utm_medium=PDF&utm_campaign=PDFCoverPages

http://scholarship.richmond.edu/management?utm_source=scholarship.richmond.edu%2Fmanagement-faculty-publications%2F11&utm_medium=PDF&utm_campaign=PDFCoverPages



http://network.bepress.com/hgg/discipline/623?utm_source=scholarship.richmond.edu%2Fmanagement-faculty-publications%2F11&utm_medium=PDF&utm_campaign=PDFCoverPages

http://network.bepress.com/hgg/discipline/634?utm_source=scholarship.richmond.edu%2Fmanagement-faculty-publications%2F11&utm_medium=PDF&utm_campaign=PDFCoverPages

mailto:[email protected]

Control and Performance in InternationalJoint Ventures

Hans Mjoen • Stephen TallmanNexia DA, Oslo, Norway

Dauid Eccles School of Business, University of Utah,Salt Lake City, Utah 84112

The incidence of international joint ventures and alliances seems to be increasing as a means toachieving strategic flexibility. Although IJVs have been the subject of much "theorizing" they

remain empirically under-researched. This paper seeks to increase knowledge of IJVs by focusing onthe relationship between partner specific resources, equity share, control (over specific resources oractivities) and performance. The paper develops a distinction between control and equity demonstrat-ing that control consists of several specific dimensions. Of particular interest is the application of amodel of bargaining power and its role in determining equity and the distinction between transactioncost economics and firm specific resources and their relation to control and equity.

Arie Y. Lewin

AbstractThe authors examine the meaning of control in internationaljoint ventures (IJVs) and the relationships of potential meansof control in such organizations to the performance satisfac-tion of the foreign partner. They propose a conceptual modelthat provides both a traditional ownership-focused internal-ization perspective on those issues and an integrated ap-proach combining a broader transaction cost interpretationof control with a resource input-based bargaining powermodel. A set of simultaneous structural equations with en-dogenous explanatory variables provides multiple possiblepaths from various resource and power inputs through differ-ent means of control to perceived performance satisfaction.In such a model, intermediate variables act both as depen-dent and independent variables; thus the complex theoreticalinteractions of the variables are modeled more comprehen-sively and realistically than in single-equation models.

To test the model and compare the theoretical relation-ships, the authors used data from a survey of managers inNorwegian multinational firms having at least one IJV. Forstructural equation modeling with latent variables, they usedthe LISREL VII program that simultaneously fits the mea-sured variables to the latent variables and provides a maxi-mum likelihood solution for the structural equation system.The results clearly reject the traditional internalization ap-proach to IJV governance that relies strictly on ownership

share to delineate degree of control. However, relative re-source input has a strong relationship to relative bargainingof the parent companies, which then drives equity share,control over specific activities, and perceptions of overallcontrol of the IJV. That result supports a bargaining-power-based model of IJV control. The relatedness of the strategicresources of the parent and the joint venture also drivesspecific control, implying that although transaction risk isimportant to governance, governance is provided by specificcontrol rather than ownership level. Perceptions of perfor-mance are strongly and positively related to overall control.

Those results suggest that specialized control providesboth protection and exploitation of key resource inputs and isgained through increased bargaining power. Higher levels ofspecific control result in a perception of overall control andthereby satisfaction with perceived levels of IJV performanceamong foreign parent company managers. Interestingly, tra-ditional exogenous determinants of IJV control and perfor-mance such as government mandates, cultural similarity, andinternational experience levels fail to provide significant ef-fects. Rather, the focus is on endogenous aspects of theparent-IJV relationship, suggesting that the key to parentfirm satisfaction with an IJV is control over operations thatuse key strategic inputs from the parent.(Joint Ventures; Alliances; Intemational; Control)

1047-7039/97/0803/0257/$05.00Copyright © 1997. Institute for Operations Researchand the Management Sciences ORGANIZATION SCIENCE/VOI. 8, No, 3, May-June 1997 257

HANS MJOEN AND STEPHEN TALLMAN Control and Performance in Intemational Joint Ventures

IntroductionThe international economic environment is changingrapidly from a condition of relatively static competitionbased on comparative advantages of nations to one ofintense technology-based, time-sensitive competition.Major changes are occurring in political and economicsystems around the world. Countries are entering theindustrial economy from the ranks of both the formerplanned economies and the less developed economies.Political and economic alliances of nations are develop-ing in new regions or evolving into closer relationships.In response, to protect their access to markets andresources, international companies are trying to gainentry to the new national and bloc economies whilestriving to enhance their positions in the traditionalindustrial triad.

At the same time, global competition, world-scaletechnology, and converging demands from customers inmany lands are forcing multinational firms to investwidely. The new technologies are so expensive, yet sofleeting, that companies in such industries as electron-ics and Pharmaceuticals must have global distributionfrom the time of introduction of a new product. Other-wise, another global competitor will move first—a criti-cal step when product life cycles are measured in a fewshort years. Recovering huge investments in technologyin ever-shorter time frames necessitates a broad pres-ence in global markets.

How are multinational companies to cope with in-creasing uncertainty coupled with a burgeoning de-mand for global products? From a strategic perspec-tive, greater uncertainty in the environment suggests aneed for flexible commitments and parsimonious appli-cation of resources. Organizationally, such a strategy isfacilitated by the use of alliances in place of solooperations. Strategic alliances reduce the resourcecommitment in any one location, enabling a single firmto cover more of the globe with the same assets.Because of the risks inherent to sharing ownership andcontrol of operations, alliances have long been treatedas less than optimal organizational solutions. In timesof turbulence, however, the risks from misjudging thedirection of the environment or missing an opportunityfor market entry appear to outweigh the risks of di-vided loyalties and leadership. When international firmsmust be able to operate effectively everywhere andanywhere while remaining ready to make immediateadjustments in their worldwide systems, even the largestmultinational companies need the added resources, thenew access, the political connections, or the latesttechnologies that can be gained from cooperative ar-

rangements. Such alliances make possible the bold butcarefully hedged strategies necessary in uncertain times.

International alliances take a variety of forms. Con-tractual partnerships requiring long-term commitmentsand intensive interaction are increasing in number(Contractor and Lorange 1988). Shared-equity interna-tional joint ventures (IJVs), more capable of transmit-ting the complex competencies needed for competitiveadvantage, are perceived as increasingly importantstrategic weapons for competing within a firm's coremarkets and technologies (Harrigan 1986). Equity hold-ings give partners freedom to access information, mon-itor performance, and control and observe the controlof operations, which can be precluded in a contractualmode. A joint owner can protect its share of theresidual income and have some input to decision mak-ing without constant renegotiation, but the same free-doms and powers are accorded to its ally. Equity ven-tures, because they set up new organizations, can in-crease the loyalty of managers and workers who areremoved from the parents and can encourage transferof organizationally embedded implicit knowledge andgroup skills. However, the threat of differential learn-ing rates (Hamel 1991) to continued competitive ad-vantage increases the need for control of activities inIJVs to protect the performance possibilities of bothjoint venture and parent.

Of critical concern in all models pertaining to jointventures—domestic or international—and other al-liances is the issue of ownership and control of theventure and the implication for the overall perfor-mance of the venture. Parental control of ventureactivities implies that the parent firm can ensure themost effective use of whatever strategic resources itshares with the IJV, a great concern in turbulentenvironments. Control also implies that the strategicresources of one parent can be sheltered from the kindof casual exposure to the other parent by which com-petitive advantage may be lost to a potential competi-tor.

Control in IJVs traditionally has been modeled byrelative degree of ownership, but new work on allianceforms, networking, relational contracting, and otherorganizational models suggests that ownership may notbe the optimal means of control in every situation andmay even be a minor issue in governance. We thereforeused structural equation modeling to test empirically amodel that integrates consideration of strategic re-sources, equity, control, and performance, along withthe effects of other variables historically proposed tohave an influence on the performance of international

258 ORGANIZATION SCIENCE/VOI. 8, No. 3, May-June 1997


joint ventures. Two primary research questions wereaddressed in our study:

1. How is the locus of control in international jointventures defined and determined?

2. What is the relationship between control andperformance levels of the IJV, as perceived by theforeign parent company?

Theoretical Perspectives on Controland PerformanceIn a joint venture, each of the joint venture parentsinvests resources with the intention of getting as muchas possible in return for the investment. Resource-based theory suggests that certain "strategic resources"(Chi 1994) generate competitive advantage and supe-rior performance (Barney 1991), and that alliancesenable firms to create new resource bundles that cangenerate additional rents not otherwise available toeither parent. Resource models also suggest that orga-nizational learning aimed at acquiring the partner'sstrategic resources is a major factor in IJV formation(Hamel 1991, Kogut 1988). Because the most valuableresources involved in a joint venture are often intangi-ble, a crucial factor determining the return a parentcan expect from a joint venture is the amount ofcontrol that parent can impose on the venture. Con-trolling resource application may determine the actualrent extraction, and controlling leakages of proprietaryknowledge prevents uncompensated transfer of capa-bilities and breakdown of the IJV. Several theories canbe used as a framework for investigating equity owner-ship and other control issues in IJVs. Transaction costtheory and bargaining power theory are the primarytheoretical lenses we use to analyze the relationshipsamong strategic resources, equity, control, and perfor-mance of the joint venture from the perspective of theinternational parent firm.

Transaction Cost EconomicsTransaction cost economics (TCE) describes theboundary conditions of the firm. The theory is com-monly used in explaining when joint ventures might bea more efficient governance mode than wholly ownedsubsidiaries or market transactions (Beamish and Banks1987; Buckley and Casson 1988; Hennart 1988, 1991;Kogut 1988). TCE models suggest that a firm canprotect its strategic resources by internalizing transac-tions where important proprietary know-how might beexposed to other firms. IJVs arise when complete in-ternalization is not efficient, but markets fail to trans-

fer intermediate goods adequately. Beamish and Banks(1987) state that, to justify a joint venture as a gover-nance form, the benefit of joining forces with anotherfirm must exceed the increased governance costs andpotential leakage of tacit knowledge associated withhaving a partner. The transaction costs model of IJVformation relies heavily on the role of intangibleknowledge or "know-how." Hennart (1988, 1991) notesthat joint ventures arise when two or more firms desireto combine their inputs, but that the transfer of thoseresources has high market transaction costs, typicallybecause they are know-how resources, so an equitytransaction is preferred. However, when neither firmcan afford (or desires) to acquire all of the other, orboth sets of resources are so embedded in their organi-zations that the market fails in both cases, or thestrategic opportunity is time-sensitive, a completetakeover may not be desirable and an equity jointventure is the preferred subsidiary form. Hennart(1991) notes, though, that protecting know-how fromthe partner is difficult. Kogut (1988) and Buckley andCasson (1988) suggest that the joint residual ownershipstatus of shared equity holdings makes each firmhostage to the other in considering misappropriatingresources. Hence, the more critical the strategic re-sources transferred to the venture are to the parent,the more likely the parent is to desire whole ownershipor, if that is not possible, the highest possible level ofownership.

Those findings have implications for the model wepropose. Most empirical literature also suggests thatthe commitment of parent strategic resources to aventure will be protected through equity ownershipand overall control (Bleeke and Ernst 1991, Hennart1991, Stopford and Wells 1972). Because equity posi-tion often determines the composition of the board ofdirectors, and the board usually appoints high-levelexecutives, the partner with a dominant equity positionhas the ability to exercise more control. Hence, thetotal path from strategic resource similarity betweenthe parent and the IJV to performance should runfrom strategic resources through equity share and over-all control to performance. In the traditional owner-ship-focused TCE model, all of the following hypothe-ses should be supported.

HI. Commitment of the parent's strategic resources tothe IJV is related positively to equity share in the IJV.

H2. Equity share is related positively to overall con-trol.

ORGANIZATION SCIENCE/VOI. 8, No. 3, May-June 1997 259


H3. Overall control is related positively to perfor-mance.

Bargaining Power TheoryThe preceding (naive) TCE model has been successfulin determining the relative use of wholly owned sub-sidiaries versus IJVs (Gatignon and Anderson 1988,Hennart 1991), but it has had less success in differenti-ating levels of joint ownership and control (Gatignonand Anderson 1988). The reason may relate to variableselection (does spending level really indicate compe-tency?), methodology (generally either single-equationlogit or multiple regression models), or conceptuallimitations. In studies on ownership and control, thepresumption seems to be that the joint venture part-ners will get the levels of equity and control theyprefer (Anderson and Gatignon 1986, Gatignon andAnderson 1988, Hennart 1988). Those studies fail totake into account the bargaining process that is aprecursor to an agreement in any contract involvingtwo or more parties. The bargaining power model ofthe IJV (Blodgett 1991a, b; Gray and Yan 1992; Lecraw1984) centers around how a parent can use its re-sources and capabilities to gain control of the IJV,whether through a higher equity share, a higher levelof overall management control, and/or a higher levelof control of specific activities. The partner with thestrongest bargaining position typically can negotiate fora higher "overall"level of control if desired. Again, theexact meaning of "control" varies across researchers.Alternative measures are compared here.

Although the international bargaining power litera-ture has mainly emphasized the bargaining situationbetween a multinational firm and a host government(Blodgett 1991b, Fagre and Wells 1982, Gomes-Casseres 1990, Lecraw 1984), it can be valuable forinvestigating the bargaining process between the IJVparents. Basically, a bargaining process occurs eachtime two or more individuals, groups, or organizationshave a conflict of interest, and when they want toresolve their differing goals because doing so would bemutually beneficial (Bacharach 1981). The outcome ofa bargaining situation between two joint venture part-ners will be satisfactory to the partners only if thecontributions and expected benefits of the two firmsare kept in balance (Robinson 1969). In our study, thebargaining process of interest is the one that occurswhen two potential joint venture partners negotiate forownership and control. Gomes-Casseres (1990) sug-gests that each party in the bargaining process has apreconceived notion of its potential power position

(what it wants). The relative bargaining positions of theparties will determine the final outcome of the negotia-tions (what they can get).

The study of bargaining power in organizations hasseveral theoretical foundations. Some authors definebargaining power as the ability to affect outcomes or toget things done (Mintzberg 1983, Salancik and Pfeffer1977)—what might be considered strategic or overallcontrol. Others hold that power stems from control ofcritical resources (Blodgett 1991b, Pfeffer 1981). Thefocus on control of strategic resources and structuraldependencies emphasizes that relative bargaining posi-tions are a result of the resources each of the IJVparents supplies to the venture. We suggest that therelative value of parent resources committed to the IJVdrives the final outcome. Previous research has identi-fied the resources that have the most important impacton the bargaining positions of the IJV parents.

In a study of IJVs in China, Yan and Gray (1994)investigated bargaining power and control. They referto resource dependency theory (Pfeffer 1981), propos-ing that control of critical resources constitutes powerbases that can tilt the negotiations in one party's favor.Realizing that an IJV consists of three entities, theysuggested that the important dependencies are be-tween the parents and the joint venture, not betweenthe parents. Yan and Gray used five components ofbargaining power in their analysis: technology, marketaccess and marketing skills, local knowledge, manage-rial expertise, and equity share. Their findings showthat the proposed components had a positive impacton the bargaining strength of the partner possessingthem, although the statistical significance was generallylow. Of particular interest in their model is the presen-tation of equity as an input to the bargaining process,not an outcome.

Blodgett (1991a, b) identified five factors influencingthe bargaining power of JV parents: government sua-sion, technology, local knowledge/marketing skills,control of intrasystem transfer (both the supply ofinputs to the venture and distribution of output fromthe venture), and financial capital provided for theventure. Financing, the factor most closely associatedwith ownership share, is not considered a crucial sourceof bargaining power for the firm because "financing isthe most fungible, least tacit, of the assets" (Blodgett1991a, p. 47). Yet equity shares are likely to be tiedclosely to capital inputs.

While acknowledging that IJV partners may at timesprefer lower levels of particular control types, we be-lieve the literature suggests that greater bargainingpower generally is related to higher levels of equity.


HANS MJOEN AND STEPHEN TALLMAN Control and Performance in IntemationaUoint Ventures

overall control, and specific control. Hence, we makethe following hypotheses.

H4. A parent's relative resource contribution is re-lated positively to its bargaining power.

H5. The bargaining power of a parent is related posi-tively to the parent's equity share.

H6. The bargaining power of a parent is related posi-tively to the parent's specific control.

H7. The bargaining power of a parent is related posi-tively to the parent's overall control.

An Integrated Approach to IJV TheoryThe preceding models and theories typically have beentreated separately in the literature, although severalconstructs are central to both models. Treating theo-ries separately is convenient in theory development,but we need a broader understanding to developknowledge of a phenomenon. By developing an inte-grated model of IJV performance that includes con-structs and links derived from each of the theoriesdiscussed, we should be able to test the relationshipamong the constructs more rigorously and thus con-tribute to a better understanding of both the theoriesand the influence of the constructs on joint ventureperformance.

Key to managing an IJV is the integration, exploita-tion, and protection of strategic resources, whetherfrom a resource-based (Conner 1991), organizationallearning (Hamel 1991, Kogut 1988), or transaction costviewpoint. The underlying mechanism for managingresources or capabilities is control. Joint venture con-trol is the central issue in the model we develop. Weuse three separate control constructs. One is the equityshare held by each of the parents, because that con-struct is commonly used in the literature and alsobecause equity share is a way of influencing strategiccontrol over the joint venture. Another is the overall orstrategic (Yan and Gray 1995) control of the venture assuggested by Killing (1983). The third is control overspecific operational activities, as suggested by severalresearchers (Geringer 1986, Geringer and Hebert 1989,Hill and Hellriegel 1994, Schaan 1983).

Many studies on equity and control are based on theassumption that equity ownership ratio can be used asa proxy for control. Blodgett (1991b) and Stopford andWells (1972) conceptually distinguished between equityand management control, but operationalized controlas ownership. Darrough and Stoughton (1989) made nodistinction between the two. When Gatignon and

Anderson (1988) used transaction cost analysis as thetheoretical foundation for a multinational's choice ofentry mode, they used equity share as the dependentvariable. That operationalization is convenient becausedata are much more readily available for equity levelsthan for a construct such as control. Note that transac-tion cost theory actually centers around governance,which is compatible with the possibility of specificcontrol mechanisms, but virtually all TCE-based empir-ical models focus on equity, relating partial to wholeownership and thus to degree of overall control. Bring-ing specific control into the TCE model requires amodified approach, concerned less with equity owner-ship rights per se and more with the risks of specifictechnology leakages in IJVs.

One must recognize, however, that equity positionand control are two conceptually different constructs.Gaining equity share is but one way of trying to obtaincontrol of the operation of a joint venture. Control isthe main issue in the theories behind multinationalsand joint ventures (Beamish and Banks 1987; Dunning1979; Hennart 1982, 1988, 1991; McManus 1972). Hav-ing a larger equity position generally means havingmore power on the board of directors (which usuallyappoints high-level executives) and thus greater abilityto control the IJVs strategy. However, when bargain-ing power is considered directly in an integrated model,control may be negotiated separately from equity share.Also, we must acknowledge Yan and Gray's (1994)position that equity share is but one input to theprocess of defining control of the IJV. Therefore, over-all control must be treated as a separate, althoughrelated, issue from ownership share.

Why does a parent want to control a joint venture?One reason is that control implies the ability to deter-mine how best to use the capabilities of the venture.Geringer and Hebert (1989, p. 236) suggest another.

In turn, exercising control over some or all of the activities ofan international joint venture helps protect the firm frompremature exposure of its strategy, technological core or otherproprietary components to outside groups.

Schaan (1983) found that some joint venture parentschoose to target specific areas of control rather thanoverall control of the venture. In a clinical study ofjoint ventures in Mexico, he found that success wasrelated to fit among objectives, selective focus of con-trol, and mechanisms for applying control. Geringer(1986) found support for the idea that parents maychoose to exercise control over a specific scope of thejoint venture activities rather than overall control. Bybargaining to control specific areas of the IJVs activi-

ORGANIZATION SCIENCE/VOL 8, No. 3, May-June 1997 261


ties to which it contributes key strategic resources, aparent firm can manage the effective application of itsassets and skills and control the exposure of its owncrucial capabilities, even with a relatively small share ofthe overall equity and control. Overall control of theIJV suggests that the parent, rather than focusing onits own contribution, is interested in controlling thestrategic direction of the combined bundle of comple-mentary strategic resources represented by the IJVitself. Hamel (1991) posits that some parents may bemore concerned about the success of the venture andothers with the value addition to the parent. Theoverall/specific control division reflects those twostrategic resource positions.

Equity share is hypothesized (H2) to be a specificmeans of controlhng the overall direction of the IJV.Equity share may also affect specialized control, as theIJV board could select managers for key operationalareas. However, the literature suggests (Geringer 1986)that specialized control is a separate issue from owner-ship share. Control of strategic activities may be nego-tiated when the IJV is formed, as a condition of theagreement, thus removing key appointments from thepurview of the IJV board or top managers. Hence, thebargaining power input is the critical issue for deter-mining specialized control (H6). If equity is an input tobargaining power, per Yan and Gray (1994), it mayhave an indirect affect on the negotiation of special-ized control. Although we believe the literature sup-ports a separation of equity and specialization, wepresent the following hypothesis in the alternative formfor consistency.

H8. Equity share is related positively to control overspecific activities of the IJV.

If specific control can indeed protect strategic re-sources, though, opportunism costs can be limitedwithout formal ownership. Control over know-how byone IJV parent can be assured through either overallcontrol of the joint venture or control of specific activi-ties of importance to that parent. The degree of spe-cialized control should influence the parents' overallfeelings of control of the IJV, and eventually theirperceptions of performance of the venture.

H9. Exposure of parent company strategic resourcesin an IJV is related positively to control over specificactivities.

HIO. Control over specific activities is related posi-tively to overall control.

The integrated model also supports H3 by suggestingthat overall control is related positively to perfor-mance. However, it does not eliminate the possibilityof a more direct performance effect, one more in linewith the results of Schaan (1983), Geringer (1986), andHill and Hellriegel (1994).

H l l . Control over specific activities is related posi-tively to perceived performance of the IJV.

Other Issues in IJV FormationOur hypotheses relate to the specifics of organizationalmodels of control and performance in IJVs. Previousstudies have shown that a variety of exogenous factorsmay affect those concerns, but we treat such factors asmoderating variables and do not test formal hypothe-ses. However, we anticipate certain effects from asurvey of the literature. A key issue in much of theearly IJV literature was the effect of government regu-lation on ownership position. Typically, governmentrestrictions limit the equity share held by the foreignparent. Some studies indicate that a firm's level ofinternational experience will affect its ownership pat-terns (Davidson 1980, Erramilli 1991, Hennart 1991,Kogut and Singh 1988b). International experienceshould be related to the equity share held in the jointventure, although a clear consensus on sign is notapparent.

Another concept related to the experience factor issociocultural distance, a "particularly potent form ofinternal uncertainty" (Gatignon and Anderson 1988, p.311). The more unfamiliar the multinational firm iswith the values and operating methods of the foreignparent, the more it would tend to avoid high-equityentry modes. Therefore, cultural distance on the partof the multinational parent is thought to be relatednegatively to the equity share held in the joint venture.The literature has focused on the effects of the exoge-nous variables on ownership, and we follow that pat-tern. An interesting issue, but one that would addunwanted complexity to our model, is the impact ofexogenous factors on specialized control and directlyon overall control.

Killing (1983) found that many U.S.-Japanese jointventures formed in the 1960s failed because of culturaldifferences. Good communication can be difficult todevelop if the managers are from different organiza-tions, and "particularly difficult if the parent firms areof different nationality and of markedly different cor-porate culture" (Killing 1983, p. 57). Geringer andHebert (1991) confirmed those findings. We expect



cultural distance to be related negatively to perfor-mance of the joint venture.

Model Specification, Data Collection,and Research MethodsModelGoldberger (1973) identified three situations in whichstructural equations are important and regular regres-sion parameters alone fail to give the relevant informa-tion: (1) when the observed variables contain measure-ment errors and the relationship of interest is amongtrue variables, (2) when there is simultaneous causationamong the observed variables, and (3) when importantexplanatory variables have not been observed. As suchconditions were present in our study, we used struc-tural equation modeling with latent variables to ana-lyze the relationships among our variables. Fornell,Lorange and Roos (1990) used structural equationmodeling to examine cooperative venture formation ona similar data set, seeking to reduce the impact ofstatistical noise on a similarly complex strategic model.

Structural equation modeling using LISREL VII(Joreskog and Sorbom 1981, 1989) handles two basicproblems that arise in scientific inference in the socialand behavioral sciences, what the observed measure-ments really measure (by the measurement model) andhow one can infer complex causal relationships amongvariables that are not directly observable, but are re-flected in fallible indicators (by the structural equationmodel) (Joreskog and Sorbom 1989, p. 2). The mea-surement model addresses how the unobserved (orlatent) variables are measured through the observedvariables, describing the validity and reliability of theobserved variables. The structural equation model ad-dresses the causal relationships among the unobservedvariables. An independent or exogenous unobservedvariable always acts as a presumed cause and never asa presumed effect, whereas a dependent (or endoge-nous) unobserved variable is always presumed to bedirectly caused or influenced by another variable(Hayduk 1987). The LISREL program estimates theunspecified parameters in the model to get the bestpossible fit between the covariance structure of theobserved data and the covariance structure of theconceptual model.

The ten constructs in our structural model are socio-cultural distance (CULTURE), previous internationalexperience (EXPERINC), relatedness of parent andIJVs rent-yielding strategic resources (STRATRES),the relative contribution of resources to the venture byeach partner (RELCONT), bargaining power in the

negotiation process (BARGPWER), regulation limitingequity share (LAWS), actual equity share (EQUITY),control over specific JV activities (SCONTROL),'overall control (CONTROL), and perceived perfor-mance of the JV (PERFORM). The parent firm's rela-tive contribution is defined as the total of its contribu-tions of resources minus the resource contributions ofthe other parent in each of several operational areas.

A system of five simultaneous equations was used toestimate the coefficients of the structural model andtest the hypotheses.^ The first equation states thatbargaining power is influenced by the relative resourcecontribution of the parent firm (H4).

BARGPWER = y,4 * RELCONT + (1)

The second equation states that equity share is influ-enced by bargaining power, cultural distance, interna-tional experience, relatedness of strategic resources,and legal requirements limiting equity share (H5, HI).

EOUITY = 2̂1 * BARGPWER -I- yji * CULTURE

+ 722 * EXPERINC -I- 723 * STRATRES

+ 725*LAWS-t-^2- (2)

The third equation states that control over specificactivities is influenced by bargaining power, equityshare, and relatedness of strategic resources (H6, H8,H9).

SCONTROL = iSj, * BARGPWER -I- ̂332 * EQUITY

+ 733 * STRATRES -I- ^3. (3)

The fourth equation states that overall control isinfluenced by bargaining power, equity share, and con-trol over specific activities (H7, H2, HIO).

CONTROL = 1641 * BARGPWER + 1842 * EOUITY

+ /343 * SCONTROL + U • (4)

The fifth equation states that perceived performanceof the IJV is influenced by control over specific activi-ties, overall control, cultural distance, and relatednessof strategic resources (Hll, H3).

PERFORM = )353 * SCONTROL + 3̂54 * CONTROL

-I- 751 * CULTURE + 753 * SRATRES

+ ^5- (5)



A direct role for strategic resources in determiningperceived performance is suggested by resource-basedtheory (Barney 1991). That term in Equation (5) pro-vides a control for the possible direct influence.

Figure 1 shows the constructs used in the model andthe hypothesized links between them, as well as thesignificant empirically derived coefficients. The direc-tions and signs of the links between the constructsrepresent specific hypotheses about the relationshipbetween the constructs.

SurveyThe population of interest was Norwegian firms in-volved in international joint ventures. Because mostempirical testing has been done with U.S. firms, astudy involving Norwegian multinationals provided achance to test the generalizability of previous findings.The respondent of interest was the person in the

Norwegian parent firm who was involved in setting upa specific international joint venture and also wasinvolved in the daily activities of the venture. Aftercontacting more than 100 of the largest NorwegianMNEs, we selected a total of 147 IJVs involving 52large Norwegian multinational firms in industries suchas marine engineering, construction, oil, aluminum,shipping, and Pharmaceuticals.^ The information de-partments of those firms were asked to contact arespondent for each of the IJVs. Sample statistics forthe various input measures are reported in Appendix 1.

Respondents involved with 107 IJVs agreed to par-ticipate in the study; 95 were contacted through per-sonal interviews and 12 were contacted by mailing theinstruments. The 12 respondents receiving the mailsurvey were in remote areas where personal interviewswould have been costly. Five of the respondents didnot answer all questions because of the sensitivity of

Figure 1 The Theoretical iViodei with Hypotheses and Significant Coefficients

LAWS

-.18,1 tailed test


HANS MJOEN AND STEPHEN TALLMAN Control and Perfomiance in Intemational Joint Ventures

the information that would be revealed. In total, surveyquestionnaires from 37 firms about 102 IJVs wereeompleted and used in the study. Only one side of theIJV triangle was examined. The Norwegian parentswere asked for their impressions of their contributions,control, and perception of the performance of the IJV.They also assessed the partner's relative positions orcontributions on some variables, as in the Fornell et al.(1990) study. Although obtaining direct input from allplayers might be ideal, Geringer and Hebert (1991)showed that parents have generally accurate picturesof their partner's position. The use of respondentsfrom only one partner is a limitation on the study, butsurveying the local partners and foreign-based IJVswas not practicable given the size of the sample andthe anonymity of the IJVs.

InstrumentA questionnaire was designed exclusively for the study.One approach to the problem of difficult to specifymeasures is to use multiple indicator measures to esti-mate the theoretical constructs as latent variables. Ac-cording to Hughes, Price and Marrs (1986), that ap-proach has two advantages. First, such models providea method for both estimating structural relationshipsamong unobservable constructs and assessing the ade-quacy with which those constructs have been mea-sured. Second, such models require a way of thinkingabout theory construction and data analysis that pro-vides for more precise testing as well as a betterunderstanding of the data. Seven of our ten constructswere measured with multiple indicators. To ensure thatat least two indicators would have high enough com-posite reliabilify, the pretest survey included a mini-mum of five indicators for each of the seven constructs.Cultural distance was measured with seven indicators,international experience with five, relatedness of par-ent and IJV strategic resources with thirteen, bargain-ing power of the parent with nine, control over specificactivities with eight, overall control with twelve, andperceived performance with nine.'* Those 63 indicatorswere measured on 6-point Likert scales, with codingranging from one (strongly disagree) to six (stronglyagree).

The other constructs were measured with a singleindicator. The influence of government restrictions wasmeasured on a 6-point Likert scale ranging from zero(no government restrictions) to five (strong governmentrestrictions). The level of equify held by the parent wasmeasured as the percentage of total equify.' Finally,the contribution of each partner was measured on aninterval scale ranging from zero to 100 percent.

Data Analysis MethodIn two-step approach to structural equation modelingsuggested by Anderson and Gerbing (1988), we usedconfirmatory factor analysis to establish the measure-ment model, then tested the full structural model.

Measurement Model. The first part of the question-naire consisted of 49 items developed to measure sevenunderlying constructs. Because the items were de-signed exclusively for our study, we had no assurancethat they would indeed load on the seven underlyingconstructs. Therefore, we used exploratory factor anal-ysis to eliminate items until a smaller set of items wasfound with loadings greater than 0.5 on a set of sevenfactors. Finally, for parsimony, a smaller number ofmeasured variables that loaded highly on the latentvariables was retained (see Appendices 2 and 3).

Table 1 reports the standardized pattern coeffi-cients, z-statistics, and reliabilify values for the mea-surement model. The standardized coefficients are allrelatively high and highly significant. The reliabilify ofthe individual indicators ranges from 0.408 to 0.986,with only one value below 0.5. Those results show ahigh level of reliabilify for the individual indicators.The free coefficients are all significant at the 0.001level, suggesting that convergent validify has beenachieved. The composite reliabilify values for the con-structs range from 0.770 to 0.951, well above the 0.6cutoff value suggested by Bagozzi and Yi (1988).

An evaluation of the measurement model using thepure ^ ^-statistic did not indicate that the data fit themodel well (;v-̂ = 182.21, d.f. 128, P = 0.001). How-ever, that statistic often suggests rejection of a model.Another indicator of how well the data fit the model,the ratio of ;̂ '̂ to degrees of freedom, is 1.424. A valueof less than three for the ratio indicates a good fit(Carmines and Mclver 1981). The goodness-of-fit index(GFI) is 0.867 for the model. Usually, a value above 0.9is considered good, and a value between 0.8 and 0.9 isconsidered moderate (Bentler and Bonett 1980). Thegoodness-of-fit measure adjusted for degrees of free-dom (AGFI) is lower at 0.782. That those goodness-of-fit indicators are not higher may reflect the size andcomplexify of the model as much as measurementdifficulties, as the simpler Cronbach's alpha indicatorsof reliabilify are 0.95 for cultural distance, 0.86 forinternational experience, 0.89 for relatedness of corecompetencies, 0.88 for relative bargaining power, 0.92for control over specific activities, 0.78 for overall con-trol, and 0.94 for performance. Overall, the resultsindicate that the measurement model is appropriatefor an investigation of the full structural model.



Table 1 Coefficients, Z-Statistics, and Reliability Vaiuesfor the Measurement Model

Construct / IndicatorStandardized

CoefficientIndicator ConstructReliability Reliability

CULTUREucultdifunation

ucultdis

EXPERiNCeingtradeoperyrsestryrs

STRATREScrskillscourtech

(item 17)(Item 44)

(item 30)

(item 16)(item 2)(item 47)

(item 7)(item 11)

0.9510,9490,890

0,9290,9020,639

0,9930,802

—^

19.39*15.74*

—^

10.91*7,28*

—'̂

7,89*

00.904

0,9010,792

0,8630,8140.408

0,9860,643

RELCONTbpower

0,951

0,870

0,896

1.000(composite) 1,000 —'^ 1,000

^Coefficients of leading indicator for each construct were set to 1,0to establish scale for the construct,*p < 0.001,

Structural Model. Figure 1 is the a priori structuralmodet we tested, with significant parameter vatues andrelevant hypotheses indicated. The futl resutts of thestructural model are reported in Table 2. Results forthe specific hypotheses are given in Table 3. Themaximum likelihood procedure was used for parameterestimation.

The value for x^ with 147 degrees of freedom is280.05 {P = 0.000), indicating that the estimated modelis significantly different from the proposed model.

Table 2 Magnitude and Significance of HypothesizedStructural Reiationships (a Priori Model)and Total Path Effects

Following/ Hypothe-

Leading sized Para-Standardized TotaiConstructs Sign meter Direct Effect t Effect f

BARGPWERRELCQNT •+

EQUITYCULTUREEXPERiNC + / -STRATRES +RELCQNTLAWSBARGPWER -I-

SCONTRQLu\ws

iitigat

BARGPWERbwepower (item 23)bwestrng (item 4)

EQUiTYeqityshr (% equity)

SCQNTROLswemgers (item 49)sgenstaf (item 29)

CQNTROLoothctrl (item 10)octrlprt (item 32)

PERFORMpsatisf (item 48)pmetobj (item 1)pprofit (Item 43)

1,000

0.9550.830

1,000

0,9840.867

0.8260.774

0.9520.9080,894

—^

8,02*

—^

11.35*

—^6,91*

—=

15,77*15,02*

1,000

0,9120,689

1,000

0,9680,752

0,6820,599

0,9060.8240.799

1.000

0.883

1.000

0.919

0,770

0,940

STRATRESCULTURE

EXPERINC

RELCONT

LAWSBARGPWEREQUITY

CQNTRQL

CULTUREEXPERiNCSTRATRESRELCONTiJ\WSBARGPWEREOUiTYSCONTRQL

PERFQRMCULTUREEXPERiNCSTRATRESRELCQNTU\WS

BARGPWEREQUiTYSCQNTROLCQNTRQL

723

733

/331

0,48*

0,100,10

-0,13

-0,110,28*

0,44*

4,47 0.48*** ,4,49

0,870.98

-1.33

-0.992.58

0.12

•)'53

0,25*0,060,62*

-0,18^

0,01

0.090,41*''

0.15

0.532.36

0.100.10

-0.130.13*

-0,110.28*

0,870,97

-1,282,31

-1,002,59

4.10 0.43***0.010.010,10

-0,011.79 0.21*1.22 0,12

0,010,010,26**0,19*

4,010.710,771,99

-0,772,161,22

0.700,763,052.73

-0.01 -0.772,39 0,40*** 3,240,63 0,13 1,204,58 0.62*** 4,58

0.010.160.09*

-0.010.18*0.060.34*0,41*

-1,800,771,572,30

-0,782.581.242,902.36

*P < 0.05 (two-tailed test).

**P < 0.01 (two-tailed test).

***P < 0,001 (two-tailed test),

^p < 0,05 (one-tailed test).

V < 0.01 (one-taiied test).

266 ORGANIZATION SCIENCE/VO1. 8, No. 3, May-June 1997


However, GFI of 0.80 indicates a moderate fit. TheAGFI is tower at 0.72. The ratio of ;̂ '̂ to degrees offreedom is 1.892, giving support for the overatt modet.The normed fit index is 0.83, indicating a moderate fit(Bentter and Bonett 1980). Further, both the compara-tive fit index and the incrementat fit index are 0.91,indicating a high degree of fit. The root mean squaredresiduat (RMSR) is 0.092. A vatue betow 0.050 for thatmeasure is generatty considered good.

Additional support for the overatl modet is found byexamining the squared muttipte correlations for theequations for the five endogenous constructs in themodet. The R^ values for those equations are: bargain-ing power (0.23), equity share (0.11), control over spe-cific activities (0.27), overalt controt (0.56), and perfor-mance (0.25). Those vatues are generatty high. Theoveratt coefficient of determination for the five equa-tions tatcen together is 0.589, providing further supportfor the theoretical modet.

Taken together, the preceding statistics suggest amoderate fit between the data and the theoreticallyderived model. One reason the fit measures are nothigher is the size of the model. The model is large andcomplex, with 10 constructs and 20 indicators. Randomerror associated with every indicator and each struc-tural equation included in a model wilt affect theoveratt fit of the model. In view of the complexity of

the integrated model, the overall fit seems adequatefor testing the structural linkages among the constructsin an exploratory study.

The results for the linkages among the constructs arereported in Table 2. Student's ?-tests were used todetermine the significance of the links. Two-tailed testsof significance are indicated in the tabte. One-taitedtests atso were used where there was a directionathypothesis. Of the 15 direct tinks in the a priori model(see Figure 1), seven turned out to be significant atP < 0.05 in a two-tailed test and one additional signifi-cant relationship was indicated with a one-tailed test.

FindingsWe found no support for HI and H2, which indicatesthat the total path from relatedness of core competen-cies to overall control and performance is not relatedto equity share. The ability to influence overall control(CONTROL) has a significant positive effect on theperformance (PERFORM) of the joint venture, sup-porting H3 (P < 0.01). Those findings are contrary tothe traditional transaction cost perspective on the roleof ownership (hierarchy) as providing parent companycontrol over strategic resources in IJVs.

The relative resource contribution of the partners(RELCONT) has a significant positive effect on the rela-

Table 3 Results of Hypothesis Testing

HypothesisProposed

SignActualSign

SignificanceLevel

(2-tailed test)

Formal hypotheses1: STRATRES is related positively to EQUITY2: EQUITY is related positively to CQNTRQL3: QQNTRQL is related positively to PERFQRM4: RELCQNT is related positively to BARGPWER5: BARGPWER is related positively to EQUITY6: BARGPWER is related positively to SCQNTRQL7: BARGPWER is related positively to CQNTRQL8: EQUITY is related positively to SCQNTRQL9: STRATRES is related positively to SCQNTRQL

10: SCQNTRQL is related positively to CQNTRQL11: SCQNTRQL is related positively to PERFQRMAdditional relationships

LAWS is related negatively to EOUITYEXPERiNC is related to EQUITYCULTURE is related negatively to EQUITYCULTURE is related negatively to PERFQRM

+/-

-0.13+0.06+0.41+0.48+0.28+0.18+0.25+0.12+0.44+0.62+0.09

-0.11+0.10+0.10-0.18

n.s.n.s.P < 0.05 (0.01,1-tail)P < 0.001P < 0.05n.s. (0.05,1-tail)P < 0.05n.s.P < 0.001P < 0.001n.s.

n.s.n.s.n.s.n.s. (0.05,1-tail)



tive bargaining power of the parents (BARGPWER){P < 0.001), giving strong support to H4. The relativebargaining power of the parents (BARGPWER) isreiated significantiy and positiveiy to equity share(EQUITY) {P < 0.01), control over specific activities(SCONTROL) {P < 0.05), and overaii controi (CON-TROL) (P < 0.01). Those resuits support H5, H6, andH7. The equity share owned by the parents (EOUITY)is not reiated significantly to control of specific activi-ties of the IJV (SCONTROL), so that hypothesis is notsupported.

The preceding four significant iinks and the supportof H3 affirm the importance of the bargaining powermodei. The model states that the bargaining power ofthe parent is infiuenced by its relative contribution tothe venture, and that this bargaining power in turngives the parent the abiiity to achieve a higher equityshare as weli as controi over specific activities andoveraii controi, and thereby higher perceived perfor-mance.

The relatedness of the strategic resources of theparent and the joint venture (STRATRES) has a signif-icant positive effect on controi over specific activities(SCONTROL), strongly supporting H9 (P < 0.001).Controi over specific activities (SCONTROL) in turnhas a highly significant positive effect on overaii con-trol (CONTROL), strongiy supporting HIO (P < 0.001).SCONTROL is not significantiy directiy related toPERFORM, so that Hl l is not supported. The directpath from STRATRES to PEREORM, suggested byresource-based theory, is not significant.

The preceding two significant links and the successof H3, indicate the importanee of modeis invoivingprotection and direction of the parent firm's resourceswithin the IJV. Resource-based theory suggests thatthe rent-yieiding potentiai of strategic resources is tiedto the parent's organizationai forms. Learning modeis(Hamel 1991) suggest that if those resources are notcontrolled closeiy, they are at risk. A governance, ratherthan ownership, approach to transaction costs suggeststhat the parent will act to reduee the risk of oppor-tunistic misappropriation of its key resourees by itspartner. Einally, the parent's perception of perfor-mance wili rise with greater controi (aithough notnecessariiy greater ownership) of the IJV.

In addition to the direet effects among the con-structs, we found seven signifieant totai effects. Totaieffects ineiude both direet effects between a pair ofconstructs and the indirect effects between two con-structs. According to Olsen, Granzin and Biswas (1992,pp. 13, 14) indirect effects are effects that oeeur when:

. . . alternative paths of influence lead through other con-structs intervening between the pair in question. Thus, thetotal effects give a more comprehensive indication of theimpact of one construct on another.

The reiative contribution of the parent (RELCONT)has a signifieant positive totai effeet on four othereonstruets in the modei: equity share (EOUITY) of theparent, the parent's control of speeifie activities(SCONTROL), overaii eontrol (CONTROL), and per-formance of the joint venture (PEREORM). Thesefindings support the importanee of developing resoureedependencies in the IJV to aequire eontrol of bothseleet operations and overall direction.

The reiatedness of strategic resourees of the parentand the joint venture (STRATRES) has a highiy signif-ieant positive total effeet (P < 0.001) on overaii eontroiof the venture (CONTROL). Combined with the pre-ceding findings, this resuit suggests that the importaneeof resource contributions both to the IJV and to thecontributing parent infiuenees the desire for eontroi ofthe venture.

Both the reiative bargaining power of the parent(BARGPWER) and the abiiity to eontroi speeifie aetiv-ities (SCONTROL) have a signifieant positive totaleffeet {P < 0.01) on the performance of the joint ven-ture (PEREORM). Interestingiy, aithough the direeteffeet from SCONTROL to PEREORM is oniy 0.09,the total effect of 0.34 is highly significant. Hence, theadditionai path through overaii CONTROL piays animportant role in linking the two eonstruets. This resuitis at odds with Hiil and Heliriegei's (1994) finding thatoveraii eontroi had no influenee on perceived perfor-mance but eontrol over speeifie funetions had a posi-tive and direet relationship.

Traditionaiiy, the roies of variables considered to beimportant in international business studies have beenmodeied through their effects on equity share. Wemaintained that tradition in investigating our moderat-ing variabies. The equity share of the parent (EOUITY)was not infiueneed by the parent's internationai experi-enee (EXPERINC), euitural distance (CULTURE), orthe iegai requirements of the host country. The nega-tive effeet of CULTURE on PEREORM is signifieantoniy in a one-taiied test. Most respondents eonsideredthe Norwegian parents to be experieneed in interna-tional operations. The iow varianee of that variablemay aeeount for its iaek of significance, and may aisoaeeount for the nonsignifieanee of euiturai differences.Littie government interferenee was found in the major-ity of eases, whieh limited the varianee of that variable

268 ORGANIZATION SOIENOE/VOI. 8, No. 3, May-June 1997

HANS MJOEN AND STEPHEN TALLMAN . Control and Performance in IntemationaUoint Ventures

dramatieaiiy.*" Tiie roie of CULTURE in determiningperformanee is periiaps tiie resuit of oider, betterestabiisiied, more produetive IJVs being in euituraiiyeiose Western European eountries.'

DiscussionOur a priori modei is a eompiex one integrating severaitiieoretieai eoneepts used in otiier studies on interna-tionai and domestie joint ventures. Tiie tiieoretieaifoundations for tiie overaii patiis as weil as tiie individ-uai iinks of tiie strueturai modei are based primariiy ontransaetion eost tiieory and the bargaining powermodei, but aiso on resouree-based theories of the firm.

Transaction Cost TheoryTraditionai studies on transaetion eost theory havesuggested that strategie resourees affeet performaneethrough overaii control provided by ownership rights.The studies typieaiiy do not ineiude any eontroi eon-struet beyond equity share (Darrough and Stoughton1989, Gatignon and Anderson 1988). Our resuits indi-eate that relative degree of ownership aione does notprovide a signifieant link between vaiue of resoureeinputs and eontroi of the IJV or its speeifie aetivities.Rather, eontroi seems to be a direet manageriai fune-tion reiated to both speeifie aetivities and strategiedireetion. Therefore, a minority partner eould wellhave effective eontroi of an IJV, or at least of keyaetivities. The role of ownership as the key to IJVeontroi appears to be highiy questionabie, a distinetdeparture from traditional market-hierarehy models.

Our resuits are eonsistent with those of more sophis-ticated approaches to transaction cost modeiing(Hennart 1988) inciuded in our integrated modei, sug-gesting that reiatedness of parent strategic resourees tothe joint venture encourages eontroi over specific aetiv-ities. Speeiaiized controi of key activities resuits infeeiings of overaii eontroi, and thereby higher per-ceived joint venture performanee. Governanee effi-eieney and the need to avoid potentially opportunisticpartner behavior—the essence of transaetion costmodeis—are supported, though not in the form ofinternaiization or ownership. A role for TCE is re-tained, but the eosts work intimately with concernsabout power and influence in deeiding issues of controiand uitimateiy of performance satisfaction.

Bargaining Power Theory and ResourcesBargaining power theory suggests that the reiative bar-gaining power of the joint venture parents is deter-mined by the relative importanee of the resourees they

bring into the joint venture. Further, inereased bar-gaining power gives a parent the opportunity to ievymore controi on the joint venture. Increased bargain-ing power of one joint venture parent has a significantpositive impact on ail of our measures of eontrol: theequity share of the parent, eontroi over specific aetivi-ties, and overall joint venture eontroi. Those resultsstrongly support the bargaining power model as ap-piied to joint ventures (Biodgett 1991a, b; Gray andYan 1992; Yan and Gray 1994). The strongest linkfrom bargaining power is to equity share, with the linkto overaii control second strongest and the link tospeeifie controi third strongest. Why might bargainingpower be used to inerease equity share if equity sharedoes not infiuence eontroi significantiy? Much of theprofit of a joint venture is divided according to therelative equity shares of the parents. The concern forequity in itseif indieates a eoncern for profit distribu-tion based on resource contribution. Such an emphasison the earning potentiai of asset depioyment ratherthan just the risks is aiso suggested by resouree-basedstrategy (Conner 1991). The totai effeets of reiativecontribution and bargaining power on joint ventureeontroi and performanee are positive and signifieant.The resuits show that this total effect was channeledthrough specific and overall joint venture eontrol.

The Integrated ModelOur results support our proposed integrated modei, inthat the paths to perceived joint venture performaneefrom relative resource eontribution and bargainingpower and from relatedness of strategie resources gothrough eontrol over speeifie activities and overaii con-troi. Those linkages support the concept that the par-ent is most eoneerned with the earning potential of itsown capabiiities, both in the IJV and for itseif. Re-souree-based modeis (Barney 1991) suggest that rent-yieiding resources are organizationaliy embedded, andmay yieid profits only in a particuiar organizationalcontext. Henee, larger eontributions of strategic re-sourees drawn direetiy from the parent's capabiiitieswouid require organizationai eontrol. If the parent'seore skiils are perceived to be at risk in the IJV, assuggested by Hamel (1991), the more simiiar the IJVsstrategie resources are to those of the parent, the morethe parent wouid want to controi those particuiar activ-ities.

The SCONTROL construct appears to provide aeonneetion between a bargaining power approaeh een-tering around the firm's need to retain eontrol of itskey assets whiie gaining controi of the venture's opera-



tions and a modified transaetion eost approaeh focus-ing on the need for governanee of strategic resources.That is, greater risk of iosing eontroi of key assetsproduees a need for greater control of the joint ventureorganization, but not neeessariiy for more internaiiza-tion. Our proposed modei integrates those dynamieapproaches to governanee and controi in the speeiai-ized eontroi/overaii eontroi/pereeived performanceconstruct. Use of high bargaining power to gain bothoverall and specific control and the eonelusive roie ofoverall eontrol in determining pereeived performanceindicate that using, as weii as protecting, resourees iscriticai to pereeptions of performanee. Finaiiy, ai-though the direct iink from strategie resourees to per-ceived performance is not signifieant, the overaii im-portance of resouree eoneerns is supported. Both thebargaining power and modified transaetion cost pathsbegin with strategic resourees and their rent-earningpotentiai, supporting a resouree-based perspeetive(Conner 1991) and emphasizing the need for integrat-ing theoreticai constructs.

Other IssuesThe hypothesized negative iink from government re-strictions to equity is not supported. That finding isinteresting as other researchers have suggested thatsuch restrietions shouid have a strong impact on equitylevel. Testing the same model on a sample of firms in aeountry with a more defined roie in internationai poli-tics might have resulted in a significant negative linkfrom government restrietions to equity share. Norwaymay be too smail and nonaggressive to attraet foreigngovernment reguiation. Also, countries are generaiiymoving toward fewer investment restrictions, as seen inMexico, Russia, and Eastern Europe.

The effect of internationai experienee on the equityievel preferred by the parent has mixed support in theliterature. Gatignon and Anderson (1988) found thatinexperieneed firms generaiiy would choose low-eon-troi entry modes, but other researehers (Davidson 1980,Kogut and Singh 1988) did not find support for thatobservation. Our results faii to support a link betweeninternationai experienee and equity share heid by theNorwegian parent. All respondents in our study feltthat their firm had many years of international experi-ence. The highly skewed distribution of the data is themost iikeiy reason for the nonsignifieant resuits.

A significant iink between euitural distance and eq-uity ievei is not supported by the data. That finding isnot surprising, beeause of eonfiieting arguments in

other studies suggest that the reiationship betweeneuiturai distanee and equity share is not ciear-cut, thegeneraiiy iow variation in cultural differences in thesampie, and the experienee of the Norwegian firms.More experieneed firms ean be expeeted to be pre-pared to cope with foreign euitures. Cuiturai differ-ences are reiated negativeiy to perceived performance,perhaps beeause of differenees in eognitive maps andoperating methods.

ConclusionsThe modei we deveioped provides an integrated ap-proach to the study of the reiationships between strate-gic resources, equity, eontroi, and performance. Con-siderations from resouree-based strategy, transactioncost theory, and bargaining power modeis of IJV oper-ations were examined simuitaneousiy. The importanceof strategic resouree inputs is indieated in severai paths.Transaction cost theory is supported by the importanttotai path from eore competencies of the firm throughcontroi over speeifie activities and overaii eontrol topereeived performanee. The bargaining power model issupported by the finding that the reiative importanceof the resources contributed by the joint venture par-ents has an impaet on the bargaining power of theparents, and that the bargaining power is used toinerease ievel of equity, eontrol over specific aetivities,and overaii control. The eomplexity of our modei showsthat iarge modeis integrating severai theories ean beinvestigated by structurai equation modeiing. Integrat-ing theories into a singie modei provides a broaderperspective on IJVs and does not require one theory tobe ehosen over the other, but aiiows the key eoneeptsfrom each theory to be mutuaily supportive.

The limitations of the study are reiated iargeiy to itssampie. Aithough modeis deveioped from studies offirms in iarge eountries appear reievant to firms in asmail industrialized nation, our results should be eon-firmed through studies of firms in a iarger home eoun-try. Another key issue is that we surveyed only oneparent in the IJV trianguiar reiationship. That ap-proaeh is consistent with most research, and findingsby Geringer and Hebert (1991) suggest that partnersdo indeed understand eaeh other's positions. Neverthe-iess, future work ineiuding the compiete IJV triadwouid be desirable.

In a highiy competitive environment, the need toproteet and properiy expioit eompetitive advantageseems to be of paramount importance. Our study shows



that the parent firms of IJVs are vitually concernedwith control of the joint venture, and in particular withcontrolling activities most closely related to their ownstrategic resources. The more important the relativecontribution of key resources, the more ways the par-ent will use to gain control of the IJV, and the morecontrol the parent has over its resources, the higher itsperception of the joint venture's performance. How-ever, control seems to be evolving away from simpleownership rights to a complex relationship of manage-rial control mechanisms. The multinational firm's needto both protect and apply its rent-yielding, embeddedknow-how resources in its joint ventures must be con-sidered a major challenge for international managers.

AcknowledgmentThe authors thank Kent Granzin and Dawn Carlson for their helpwith the empirical analysis.

Appendix 1 Sample Statistics for Measured Variables

Variable Name & Item Mean S.D, Min. Max.

equityshr {% equity of MN) 44.14 14.14 10.0crskills (item 7) 3.99 1.32 1.0courtech (item 11) 4,01 1.45 1.0bwestrng (item 4) 3.84 1,14 1,0bwepowerdtem 23) 3.62 1.29 1.0swemgers(item 49) 3.81 1.68 1.0sgenstaf (item 29) 3.62 1.73 1.0psatisf (item 48) 4.36 1.55 1.0pmetobj (itemi) 4.68 1.39 1.0pprofit (item 43) 4.10 1.69 1.0

RELCONT Inputs:MN firm %

technologymarketingiocal knowledgeinput supplydistributionfinanciai capital

Local firm %technologymarketingiocal knowledgeinput supplydistributionfinancial capitai

56.0349,7433,2737.5041.6344.38

35.8641.3754.1537,7243.1753,03

31,1329,6731.1329.6329.0719.60

29.6528,9432.3730.1530.4220.55

000000

000000

75.06.06.06.06,06,06,06,06.06.0

99.099.099.099.099.099,0

99.090.099.099.099,099,0

Appendix 2

IndicatorName

Factor Loadings for Expioratory

Factors

1 2 3 4 5

Factor Anaiysis

6 7

PPROFITPMETOBJPSATISFPECORETNPMETEXPPNOTiNVPNOTSUCCPNOPRTNRCRSKILLSCRKNOWHOCOURTECHCCORESKLCOURKNOWCWEGOODUCULTDIFUNATIONUCULTDISUEATDIFFOPRTNINFOOTHCTRLOMOSTDECOCTRLPRTOOVERALLOEXPLANSSGENSTAFSWEMGERSSMGDEPTSSCTRPRTSELNGTRADEOPERYRSESTRYRSEiNVESSEBWESTRNGBWEPOWERBWINNERS

0.92280,89700,8927

-0.85030.8353

-0,7980-0,7801-0.7065

0.85560.84700.84450.83040,76710,6999

0.93640.91590.90690.7928

0.7891-0.69250.6609

-0.63420.55690.5338

0.4645

0.4657

0,83790.78110.66960,5823 0.4192

0.88560.86460.84090.8336

0,82520.81520,7493

Appendix 3 Survey Items in the Finai iVIeasurement iVIodei

1. Cultural distance (CULTURE) was measured with three indi-cators. The indicators retained for the measurement model were:

a. There are significant cultural differences between us and theother JV partner (item 17).

b. Their national culture is quite different from ours (item 44).c. There are many cultural dissimilarities between us and the

other JV partner (item 30).

2. International experience (EXPERINC) was also measured withthree indicators:

a. We have a long tradition of international operations (item 16).b. We have been operating in foreign countries for many years

(item 2).

ORGANIZATION SCIENCE/VOI. 8, No. 3, May-Jutie 1997 271


c. Investing abroad has been part of our strategy for many years(item 47).

3. The relatedness of strategic resources of the parent and the JV(STRATRES) was measured with two indicators;

a. The JV to a large degree utilizes skills that we contributed(item 7).

b. Much of our technology is used in the JV (item 11).

4. The bargaining power of the JV parent (BARGPWER) wasmeasured with two indicators:

a. We were the most powerful firm during the initial negotiations(item 23).

b. In the initial negotiation process we felt we were the strongestpartner (item 4).

5. The control over specific activities (SCONTROL) of the JV wasmeasured with two indicators:

a. We provided most of the managers in the area of our expertise(item 49).

b. The departments where our contributions are used generallystaffed by us (item 29).

6. The overall control of the JV (CONTROL) was also measuredwith two indicators:

a. The other JV partner influences a great deal of control overdaily activities (item 10, reversed).

b. The overall managerial control generally resides with our JVpartner (item 32, reversed).

7. Finally, the performance of the JV (PERFORM) was measuredwith three indicators:

a. We are satisfied with the performance of the JV (item 48).b. The JV has met the objectives for which it was established

(item 1).c. The JV has been a profitable investment (item 43).

8. Three constructs were measured with single indicators:a. The relative contribution of each partner (RELCONT) was

measured with a composite index. The parent was asked whatpercentage (range zero to 100) of the following inputs were providedby each of the partners (1) technological know-how, (2) marketingknow-how, (3) knowledge of local environment, (4) supply of inputs(raw materials), (5) distribution of outputs, and (6) financial capital.The contribution of the partner was substracted from the contribu-tion of "our" firm to give a relative contribution. Each of the sixinputs was given equal weight. The composite index was constructedby adding the six relative contributions together.

b. The effect of government restrictions on equity share (LAWS)was measured on a 6-point Likert scale ranging from no restrictionsto strong restrictions.

c. The final measure, equity share of "our" firm (EQUITY), wasmeasured as a percentage ranging from zero to 100.

EndnotesAlthough SCONTROL refers to control of the types of activitiesmeasured for RELCONT, it is measured by general statementsabout control of areas of special interest, so does not directlymeasure degree of control of specific activities (see Appendix 3).

The system of equations proposed is recursive, with four exogenouscausal latent variables and a number of zero restrictions on coeffi-cients, and thus is overidentified. The estimation confirms that, as noproblems of underidentification or nonconvergence arise.^The subject IJVs were identified by the Norwegian parents' infor-mation departments and not identified by name. Although all parentfirms were not independent and/or publicly owned, the size distribu-tion of those that were {n = 20), based on assets, was as follows:

Currency Mean S.D.

NKr(Million)$US(Million)

11778.1511706.98

20554.663029.69

Note that the performance latent variable reflects qualitative assess-ments of performance from the perspective of the Norwegian parent.Fornell et al. (1990) used multiple qualitative measures in anotherstructural equation model. Geringer and Hebert (1991) support thecorrelation of qualitative and quantitative performance measures,and the ability of participants in the IJV triad to accurately reflecteach other's assessments of performance.We used a continuous measure of percentage of equity holdings.

Other studies use categorical distributions. On such a measure, wehad 47 minority ownership, 35 equal ownership, and 20 majorityownership cases in the final sample.The experience, culture, and legal variables were measured as

described in Appendix 3. Experience and culture are latent variablescomposed of three highly correlated measured variables each. Asindicators of the overall levels of those measures, mean scores on thethree measured variables were calculated. Mean composite experi-ence was 5.101, S.D. 1.12; mean composite cultural distance was3.06, S.D. 1.54 (l-to-6 scales). LAWS was a single variable with mean1.33 and S.D. 1.75 (O-to-5 scale). Only 19 firms reported significantregulatory effects, and they included minority, equal, and majorityequity positions.

ReferencesAnderson, E. and H. Gatignon (1986), "Modes of Foreign Entry: A

Transaction Cost Analysis and Proposition," Joumal of Intema-tional Business Studies, 17, 3, 1-26.

Anderson, J. C. and D. W. Gerbing (1988), "Structural EquationModeling in Praetice: A Review and Recommended Two-StepApproach," Psychological Bulletin, 103, 3, 411-423.

Bacharach, S. B. (1981), Bargaining, San Francisco, Jossey-Bass.Bagozzi, R. P. and Y. Yi (1988), "On the Evaluation of Structural

Equation Models," Joumal of the Academy of Marketing Science,16, 1, 74-94.

Barney, J. B. (1991), "Firm Resources and Sustained CompetitiveAdvantage," Joumal of Management, 17, 1, 99-120.

Beamish, P. W. and J. C. Banks (1987), "Equity Joint Ventures andthe Theory of the Multinational Enterprise," Joumal of Interna-tional Business Studies, 18, 2, 1-15.

Bentler, P. M. and D. G. Bonett (1980), "Significance Tests andGoodness of Fit in the Analysis of Covariance Structures,"Psychological Bulletin, 88, 588-606.



Bleeke, J. and D. Ernst (1991), "The Way to Win in Cross-BorderAlliances," Harvard Business Reuiew, 69 (November/December),127-135.

Blodgett, L. L. (1991a), "Toward a Resource-Based Theory of Bar-gain Power in International Joint Ventures," Joumal of Global

Marketing, 5, 1/2, 35-54.

(1991b), "Partner Contributions as Predictors of Equity Sharein International Joint Ventures," Joumal of Intemational Busi-ness Studies, 22, 1, 63-78.

Buckley, P. J. and M. Casson (1988), "A Theory of Cooperation inInternational Business," in F. J. Contractor and P. Lorange(Eds.), Cooperative Strategies in Intemational Business, Lexington,MA: Lexington Books, 31-53.

Carmines, E. and J. Mclver (1981), "Analyzing Models with Unob-served Variables: (Analysis of Covariance Structures," in G.Bohrnstedt and E. Borgatta (Eds.), Social Measurement: CurrentIssues, Beverly Hills, CA: Sage, 65-115.

Chi, T. (1994), "Trading in Strategic Resources: Necessary Condi-tions, Transaction Cost Problems, and Choice of ExchangeStructure," Strategic Management Joumal, 15, 4, 271-290.

Conner, K. R. (1991), "A Historical Comparison of Resource-BasedTheory and Five Schools of Thought Within Industrial Organi-zation Economics: Do We Have a New Theory of the Firm?"Joumal of Management, 17, 121-154.

Contractor, F. J. and P. Lorange (1988), "Why Should Firms Coop-erate? The Strategy and Economic Basis for Cooperative Ven-tures," in F. J. Contractor and P. Lorange (Eds.), CooperativeStrategies in Intemational Business, Lexington, MA: LexingtonBooks, 3-30.

Darrough, M. N. and N. M. Stoughton (1989), "A Bargaining Ap-proach to Profit Sharing in Joint Ventures," Joumal of Business,

62, 2, 237-270.

Davidson, W. H. (1980), "The Location of Foreign Direct Invest-ment Activity: Country Characteristics and Experience Effects,"Joumal of Intemational Business Studies, 12, 9-22.

Dunning, J. H. (1979), "Explaining Changing Patterns of Interna-tional Production: In Defense of the Electric Paradigm," OxfordBulletin of Economics and Statistics, November, 269-295.

Erramilli, M. K. (1991), "The Experience Factor in Foreign MarketEntry Behavior of Service Firms," Joumal of Intemational Busi-ness Studies, 22, 3, 479-501.

Fagre, N. and L. T. Wells (1982), "Bargaining Power of Multination-als and Host Governments," Joumal of Intemational BusinessStudies, 3, 9-23.

Fornell, C , P. Lorange and J. Roos (1990), "The CooperativeVenture Formation Process: A Latent Variable Structural Mod-eling Approach," Mfl/!age/?je«f 5c(e/!ce, 36, 10, 1246-1255.

Gatignon, H. and E. Anderson (1988), "The Multinational Corpora-tion's Degree of Control Over Foreign Subsidiaries: An Empiri-cal Test of a Transaction Cost Explanation," Joumal of Law,Economics, & Organization, 4, 1, 305-336.

Geringer, J. M. (1986), "Criteria for Selecting Partners for JointVentures in Industrialized Market Economies," Ph.D. disserta-tion. University of Washington.

and L. Hebert (1989), "Control and Performance of Interna-tional Joint Ventures," Joumal of Intemational Business Studies,20, 2, 235-254.

and (1991), "Measuring Performance of InternationalJoint Ventures," Joumal of Intemational Business Studies, 22, 2,249-264.

Goldberger, A. S. (1973), "Structural Equation Models: AnOverview," in Structural Equation Models in the Social Sciences,

A. S. Goldberger and O. D. Dunean (Eds.), New York: SeminarPress.

Gomes-Casseres, B. (1990), "Firm Ownership Preferences and HostGovernment Restrictions: An Integrated Approach," Joumal of

Intemational Business Studies, 21, 1, 1-22.

Gray, B. and A. Yan (1992), "A Negotiations Model of Joint VentureFormation, Structure and Performance: Implications for GlobalManagement," Advances in Intemational Comparative Manage-ment, 7, 41-75.

Hamel, G. (1991), "Competition for Competence and InterpartnerLearning within International Strategic Alliances," Strategic

Management Joumal, 12, 83-103.

Harrigan, K. R. (1986), Managing for Joint Venture Success, Lexing-ton, MA: Lexington Books.

Havduk, L. A. (1987), Structural Equation Modelling with LISREL,

Baltimore, MD: The Johns Hopkins University Press.

Hennart, J. (1982), A Theory of Multinational Enterprise, Ann Arbor:MI: University of Michigan Press.

(1988), "A Transaction Costs Theory of Equity Joint Ventures,"Strategic Management Joumal, 9, 361-374.

(1991), "The Transaction Cost Theory of Joint Ventures: AnEmpirical Study of Japanese Subsidiaries in the United States,"Management Science, Yt, 483-497.

Hill, R. C. and D. Hellriegel (1994), "Critical Contingencies in JointVenture Management: Some Lessons from Managers," Organiza-

tion Science, 5, 594-607.

Hughes, M. A., R. L. Price and D. W. Marrs (1986), "Linking TheoryConstruction and Theory Testing: Models with Multiple Indica-tors of Latent yar'\2!o\e%" Academy of Management Review, 11, 1,128-144.

Joreskog, K. G. and D. Sorbom (1981), LISREL—User's Guide,Uppsala, Sweden: University of Uppsala.

and (1989), LISREL 7—A Guide to the Program and

Applications, Uppsala, Sweden: University of Uppsala.

Killing, J. P. (1983), Strategies for Joint Venture Success, New York:

Praeger.

Kogut, B. (1988), "Joint Ventures: Theoretical and Empirical Per-spectives," Strategic Management Joumal, 9, 319-332.

and H. Singh (1988), "The Effect of National Culture on theChoice of Entry Mode," Joumal of Intemational Business Studies,19, 3, 411-432.

Lecraw, D. J. (1984), "Bargaining Power, Ownership, and Profitabil-ity of Transnational Corporations in Developing Countries,"Joumat of Intemational Business Studies, 15, 1, 27—43.

Long, J. S. (1983), Confirmatory Factor Analysis: A Preface to LISREL,Beverly Hills, CA: Sage Publications.



McManus, J. (1972), "The Theory of the Multinational Firm," in

G. Paquet, (Ed.), The Multinational Firm and the Nation State,

Don Mills, Ontario, Canada: Macmillan.

Mintzberg, H. (1983), Power in and Around Organizations, Englewood

Cliffs, NJ: Prentice Hall.

Olsen, J. E., K. L. Granzin and A. Biswas (1993), "Influencing

Consumers' Selection of Domestic vs. Imported Products: Impli-

cations for Marketing Based on a Model of Helping Behavior,"

Joumal of the Academy of Marketing Science, 21, 307-321.

Pfeffer, J. (1981), Power in Organizations, Marshfield, MA: Pitman.

Robinson, R. (1969), "Ownership Across National Frontiers," Indus-

trial Management Reuiew, 1, Fall, 41-65.

Salancik, G. R. and J. Pfeffer (1977), "Who Gets Power—and How

They Hold on to It: A Strategic Contingency Model of Power,"

Organizational Dynamics, 5, 3-21.

Schaan, J. L. (1983), "Parent Control and Joint Venture Success:

The Case of Mexico," Ph.D. dissertation. University of Western

Ontario.

Stopford, J. M. and L. T. Wells (1972), Managing the Multinational

Enterprise, New York: Basic Books.

Yan, A. and B. Gray (1994), "Bargaining Power, Management Con-

trol, and Performance in United States-Chinese Joint Ventures:

A Comparative Case Study," Academy of Management Joumal,

37, 6, 1478-1517.

and (1995), "The Exercise of Management Control in

International Joint Ventures," Paper revised and resubmitted to

Academy of Management Joumal, available on request from the

authors.

Accepted by Arie Y. Lewin; received April 1994. This paper has been with the authors for two reuisions.


Documents

Control and Performance in International Joint Ventures