learning and protection of proprietary assets in strategic alliances: building relational capital

Strategic Management JournalStrat. Mgmt. J.,21: 217–237 (2000)

LEARNING AND PROTECTION OF PROPRIETARYASSETS IN STRATEGIC ALLIANCES: BUILDINGRELATIONAL CAPITAL

PRASHANT KALE1, HARBIR SINGH2* and HOWARD PERLMUTTER2

1University of Michigan Business School, Ann Arbor, Michigan, U.S.A.2The Wharton School of Business, University of Pennsylvania, Philadelphia,Pennsylvania, U.S.A.

One of the main reasons that firms participate in alliances is to learn know-how and capabilitiesfrom their alliance partners. At the same time firms want to protect themselves from theopportunistic behavior of their partner to retain their own core proprietary assets. Mostresearch has generally viewed the achievement of these objectives as mutually exclusive. Incontrast, we provide empirical evidence using large-sample survey data to show that whenfirms build relational capital in conjunction with an integrative approach to managing conflict,they are able to achieve both objectives simultaneously. Relational capital based on mutualtrust and interaction at the individual level between alliance partners creates a basis forlearning and know-how transfer across the exchange interface. At the same time, it curbsopportunistic behavior of alliance partners, thus preventing the leakage of critical know-howbetween them.Copyright 2000 John Wiley & Sons, Ltd.

INTRODUCTION

Studies on alliances confirm a significant increasein their use as a strategic device (Hergert andMorris, 1987; Anderson, 1990; Ahuja, 1996).Firms use alliances for a variety of reasons: togain competitive advantage in the marketplace,to access or internalize new technologies andknow-how beyond firm boundaries, to exploiteconomies of scale and scope, or to share riskor uncertainty with their partners, etc. (Powell,1987; Bleeke and Ernst, 1991). Learningalliances, in which the partners strive to learn orinternalize critical information or capabilities fromeach other, constitute an important class of suchalliances (Prahalad and Hamel, 1990; Hamel,1991; Khanna, Gulati, and Nohria, 1998). Yet,these alliances also raise an interesting dilemma,

Key words: strategic alliances; relational capital;learning; proprietary assets*Correspondence to: Professor H. Singh, Wharton School ofBusiness, University of Pennsylvania, Suite 2000, SteinbergHall–Dietrich Hall, Philadelphia, PA 19104, U.S.A.

CCC 0143–2095/2000/030217–21 $17.50Copyright 2000 John Wiley & Sons, Ltd.

as a firm that uses them also risks losing itsown core proprietary capabilities to its partners,especially when these partners behave opportu-nistically.

The transaction costs literature has emphasizedthe relevance of partner opportunism in inter-organizational relationships. Building upon it,subsequent literature on learning alliances dubbedthem as a ‘learning race’ (Khannaet al., 1998)in which partners often engaged in opportunisticattempts to outlearn each other. This ‘race’ cre-ates a significant tension for firms. On the onehand, alliances may help a firm absorb or learnsome critical information or capability from itspartner. On the other, they also increase thelikelihood of unilaterally or disproportionatelylosing one’s own core capability or skill to thepartner. Thus, firms are faced with the challengingtask of managing the balance between ‘trying tolearn and trying to protect.’ In contrast to thetransaction cost literature, recent alliance researchhas highlighted the existence, and importance, ofinter-personal relationships and trust in allianceor exchange situations (Ring and Van de Ven,

218 P. Kale, H. Singh and H. Perlmutter

1992; Gulati, 1995; Zaheer, McEvily, and Per-rone, 1998). We use this work to develop thenotion of relational capital, which refers to thelevel of mutual trust, respect, and friendship thatarises out of close interaction at the individuallevel between alliance partners. We suggest thatrelational capital can help companies successfullybalance the acquisition of new capabilities withthe protection of existing proprietary assets inalliance situations. On the one hand, relationalcapital facilitates learning through close one-to-one interaction between alliance partners. On theother hand, it minimizes the likelihood that analliance partner will engage in opportunisticbehavior to unilaterally absorb or steal infor-mation or know-how that is core or proprietaryto its partners.

Conflict is inherent in alliances because ofpartner opportunism, goal divergence (Doz, 1996)and cross-cultural differences, and using explicitmechanisms to manage conflict will help firms todeal with these difficulties. There has been ageneral tendency in the alliance literature to linkformal governance mechanisms with the man-agement of conflicts (Williamson, 1985). Butmore recently, there is recognition that a combi-nation of contractual and organizational mecha-nisms (formal and informal) is more effective inmanaging conflict (Doz, 1996; Dyer and Singh,1998). In the context of alliances—areas rife withpotential opportunism—organizational mecha-nisms to manage conflict can be particularlyimportant in addressing the dilemma discussedearlier. First, effective conflict management couldenable partners to build relational capital that notonly facilitates learning but also limits partneropportunism. Second, these processes can help inprotecting proprietary assets owing to their abilityto monitor and control partner behavior.

Relational capital, which is seen to be soimportant at the dyadic level in alliances, canbe equally important in the context of alliancenetworks. Scholars (Gulati and Gargiulo, 1999)have argued that strong interpersonal ties betweentwo organizations provide channels through whichpartners learn about other firms’ competenciesand reliability. From this perspective, relationalcapital that rests upon close interpersonal ties atthe dyadic level can also play an important rolein creating and building larger alliance networks.First, it increases the probability that partners willform more alliances with each other in the future.

Copyright 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J.,21: 217–237 (2000)

Second, it allows each firm to form new allianceswith other firms based on referrals that its part-ners are ready to provide for it. Eventually, alarger and richer alliance network can evolve onthe basis of strong relational capital at the dyadiclevel between two partners.

LITERATURE REVIEW ANDRESEARCH QUESTION

Strategic alliances can be defined as purposivestrategic relationships between independent firmsthat share compatible goals, strive for mutualbenefits, and acknowledge a high level of mutualdependence (Mohr and Spekman, 1994). Gulati(1995) defines an alliance as any independentlyinitiated interfirm link that involves exchange,sharing, or co-development.

Three streams of research typify most of theacademic work on alliances. The first stream thatattempts to explain the motivations for allianceformation has put forth three rationales: strategic,transaction costs related, and learning related.Strategic considerations involve using alliances toenhance a firm’s competitive position throughmarket power or efficiency (Kogut, 1988). Trans-action cost explanations view alliance formationas a means to reduce the production and trans-action costs for the firms concerned (Williamson,1985; Hennart, 1988). Learning explanations viewalliances as a means to learn or absorb criticalskills or capabilities from alliance partners. Thesecond stream of research focuses on the choiceof governance structure in alliances. Informedlargely by transaction cost economics, it arguesthat governance in alliances mirrors the underly-ing transaction costs associated with an exchange,and that equity-based structures are more likelyunder conditions of high transaction costs(Pisano, Russo, and Teece, 1988; Pisano, 1989).The third stream of research examines the effec-tiveness and performance of alliances. It seeksto identify factors that enhance or impede theperformance of either the alliance itself, or of thealliance’s parent firms that are engaged in one(Beamish, 1987; Harrigan, 1985; Koh and Venka-traman, 1991; Merchant, 1997).

Despite their different emphases, existingalliance research has begun to focus increasinglyon the phenomenon of learning in alliance situ-ations. Learning in terms of accessing and acquir-

Learning and Protection of Proprietary Assets in Alliances 219

ing critical information, know-how, or capabilitiesfrom the partner is oft stated to be one of theforemost motivations for alliance formation(Hamel, 1991; Khannaet al., 1998). Alliancesare seen not only a means of trading accessto each others’ complementary capabilities—whatmight be termed quasi-internalization—but alsoas a mechanism to fully acquire or internalizepartner skills. Yoshino and Rangan (1995) statethat such learning is always an implicit strategicobjective for every firm that uses alliances. Giventhe importance that firms place on formingalliances to exploit learning opportunities,researchers have begun to examine various factorsthat might impact the learning process (Khannaet al., 1998) and learning success (Hamel, 1991).For example, it has been argued that equity-based governance structures are better suited forlearning critical know-how or capabilities fromthe partner (Mowery, Oxley, and Silverman,1996). Such alliances are especially seen as effec-tive vehicles for learning tacit know-how andcapabilities as compared to nonequity-based con-tractual arrangements because the know-howbeing transferred or learnt is more organi-zationally embedded (Kogut, 1988). Using case-based research, Hamel (1991) also shows thatfirms that possess a strong learning intent andcreate an appropriate learning environment winthe so-called ‘Learning Race.’ Khannaet al.,(1998) extend this stream of research to providean excellent analytical framework that describesthe dynamics of the learning process in such a‘Learning Race.’ They show that firms’ incentivesto learn are driven by their expected pay-offsthat have complex, interdependent and dynamicstructures. Learning success is determined by theamount of resources that firms allocate to learnfrom their alliance partner. The resource allo-cation is itself dependent upon the expected pay-offs associated with such learning. The magnitudeof these pay-offs is also linked to the degreeof overlap between alliance scope and parentfirm scope.

We believe that there is sufficient opportunityto extend current research on learning alliances.Current alliance research has failed to sufficientlyaddress, theoretically and empirically, animportant dilemma that often exists in learningalliances. Participants in learning alliances wouldnot only like to access some useful informationor know-how from the partner, but also inter-


nalize some complementary capabilities and skillspossessed by the partner. At the same time, theywould also like to protect some of their own coreproprietary capabilities from being unilaterallyabsorbed or appropriated by the partner. Thusthere is an underlying tension between ‘trying tolearn and trying to protect.’ The dilemma arisesbecause conditions that might facilitate the learn-ing process are likely to expose firms to thedanger of losing some of their crown jewels tothe partner. The NUMMI alliance between Gen-eral Motors and Toyota is a classic example ofsuch an alliance (Badaracco, 1988). GeneralMotors was keen to learn some of Toyota’smanufacturing management practices through thealliance, whereas Toyota wanted to learn how tomanage U.S. labor and how to run a manufactur-ing plant in the United States from GM. However,both partners were also keen to prevent leakageof some of their core proprietary skills to theother. Toyota was keen to protect its skills ofsmall car design and effective supplier man-agement and GM its capabilities of managingdealerships in the United States.

Current alliance research fails to sufficientlyexamine how firms can balance the apparentduality or tension between learning and protect-ing. In this context, we seek to address the fol-lowing question: What factors enable a firm tonot only learn critical skills or capabilities fromits alliance partner(s), but also protect itself fromlosing its own core proprietary assets or capabili-ties to the partner? In the following sections, wedevelop hypotheses that address these questionsand test the hypotheses using large-sample surveydata from alliances of U.S.-based firms.

Before we move on to the next section, wewould like to stress a few important points aboutthe learning phenomenon in alliances. Learningin alliance situations can be of several kinds andwe focus on just one of them in our paper.First, we have learning that essentially involvesaccessing and/or internalizing some critical infor-mation, capability, or skill from the partner. Thisis the kind of learning that has been most referredto in the alliance literature and our paper exam-ines the tension associated with balancing someof the dynamics involved in such learning. Suchlearning is often a private benefit that potentiallyaccrues to firms that participate in alliances(Khannaet al., 1998). Second, researchers havealso referred to learning wherein the alliance


partners in the context of their existing alliance‘learn’ how to manage the collaboration processand work better with each other as their relation-ship evolves (Doz, 1996; Arino and de la Torre,1998). It involves learning about the partners’intended and emergent goals, how to redefinejoint tasks over time, how to manage the allianceinterface, etc. Such learning is equally criticalto sustaining successful cooperation in alliances.Third, we have learning that looks at how anindividual firm might learn how to manage itsalliances better, and build what has been referredto as alliance capability (Anand and Khanna,2000; Kale and Singh, 1999). Alliance capabilityas referred to above may be built over time byaccumulating more alliance experience, i.e. byforming more and more alliances (Anand andKhanna, 2000). However, it could also bedeveloped by pursuing a set of explicit processesto accumulate and leverage the alliance man-agement know-how associated with the firm’sprior and ongoing alliance experience (Kale andSingh, 1999). Our paper focuses only on thefirst type of learning, namely the accessing andinternalizing of critical information or capabilitiesfrom alliance partner(s). Here, we do not examinethe other two, equally important types of learningin alliances. Thus henceforth, whenever we talkabout learning in alliances, we are essentiallyreferring to learning that involves the acquisitionor internalization of some critical information,know-how, or capability possessed by the partner.

THEORY AND HYPOTHESES

Relational capital

The alliance literature has focused extensively onpartner opportunism and most researchers haveadopted the theoretical stance informed by trans-action cost economics to examine this aspect(Hennart, 1988; Kogut, 1988; Pisano, 1989; Wil-liamson, 1991). Firms’ concerns about opportun-istic behavior by their partners are likely to leadto high transaction costs and it has been suggestedthat firms can adopt appropriate contractual agree-ments or governance structures to address theseconcerns. Using transaction cost economics,scholars have identified two sets of governanceproperties through which equity alliances caneffectively alleviate the transaction costs involved.One is the property of a ‘mutual hostage’ situ-


ation, in which shared equity aligns the interestsof the partners involved. Since partners arerequired to makeex ante commitments to anequity alliance, their concern for their investmentsreduces the possibility of opportunistic behaviorover the course of the alliance (Pisano, 1989).Second, in equity alliances, the investing partnerscreate a hierarchical supervision not only tooversee the day-to-day functioning of the alliance,but also to address contingencies as they arise(Kogut, 1988).

Numerous researchers have criticized the trans-action cost economics perspective on alliances forits singular focus on partner opportunism and itsadvocating the use of contractual agreements orequity to resolve it. This approach fails to capturean important element in alliance partnerships,namely the role of interfirm trust and the evolu-tion of interpartner relationships (Gulati, 1995).‘Trust’ has been referred to in several ways inthe literature. First, it is considered ‘a type ofexpectation that alleviates the fear that one’sexchange partner will act opportunistically’(Bradach and Eccles, 1989). Offering a slightlydifferent emphasis, Madhok (1995), suggests thattrust between exchange partners has two compo-nents: a structural component which is fosteredby a mutual hostage situation, and a behavioralcomponent, which refers to the degree of confi-dence that individual partners have in thereliability and integrity of each other. Similarly,Gulati (1995) differentiates knowledge-based trustfrom deterrence-based trust. Knowledge-basedtrust emerges between two firms as they interactwith each other and learn about each other, todevelop trust around norms of equity. Deterrence-based trust is based on utilitarian considerationswhich lead a firm to believe that a partner willnot engage in opportunistic behavior owing tothe costly sanctions that are likely to arise. Over-all, there is an emerging consensus amongalliance scholars that mutual trust creates thebasis for an enduring and effective relationshipbetween contracting firms. For example, Gulati(1995) shows how trust enables firms to reducedependence on equity structures to govern therelationships. Zaheeret al., (1998) demonstratehow trust reduces negotiating costs in alliancesand also enhances alliance performance.

Trust between organizations has often beenconceived as the agglomeration of trust betweenindividuals in the two organizations. Numerous


examples highlight the existence of stable obliga-tory relationships based on trust between individ-ual members of the partnering firms. Accounts ofthe industrial districts in Italy (Piore and Sabel,1984), of subcontracting relationships in theJapanese textile industry (Dore, 1983), and thosein the Japanese automobile industry (Dyer, 1996)highlight this aspect. The premise is that as firmswork with each other trust is built among individ-ual members of the contracting firms because ofthe close personal ties that develop between them(Macaulay, 1963). Such trust is based upon closeinteraction and relationships that develop at thepersonal level. It is akin to the knowledge-basedtrust referred to by Gulati (1995) or thebehavioral-based trust referred to by Madhok(1995). A history of close relationships helpsindividual members develop such trust in theircounterparts in the partnering firm. Relationalexchange theory (Dore, 1983) in economic soci-ology also discusses how personal relationshipsbased on trust arise and exist between firms.Palay (1985) and Ring and Van de Ven (1992)have also pointed out the important role of per-sonal connections and relationships between con-tracting firms. We refer to such mutual trust,respect, and friendship that reside at the individ-ual level between alliance partners as relationalcapital. Relational capital, as defined, residesupon close interaction at the personal levelbetween alliance partners. We believe thatrelational capital has important performanceimplications for the alliance partners. More speci-fically, we argue that it may significantly impactthe ability of a firm to successfully manage thedual objectives of learning from its alliance part-ners and also protecting its own core proprietaryassets from them.

The role of relational capital in learningalliances

Learning from the alliance partner primarilyinvolves the acquisition of two types of knowl-edge: (a) information and (b) know-how (Kogutand Zander, 1992). Information is defined aseasily codifiable knowledge that can be trans-mitted without loss of integrity, once the syntacti-cal rules required for deciphering it are known.It includes facts, axiomatic propositions, and sym-bols. On the other hand, know-how involvesknowledge that is tacit, sticky, complex, and dif-


ficult to codify (Nelson and Winter, 1982; Szulan-ski, 1996). Von Hippel (1988) defines know-howas the accumulated practical skill or expertise thatallows one to do something smoothly andefficiently.

Firms that wish to learn critical information orknow-how from their alliance partner must firstunderstand where the relevant information or ex-pertise resides in its partner and who possessesit (Dyer and Singh, 1998). Close personal inter-action between the partnering entities enablesindividual members to develop this understanding.Learning or transfer of such know-how is thencontingent upon the exchange environment andmechanisms that exist between the alliance part-ners. Know-how, as discussed earlier, is generallymore sticky, tacit, and difficult to codify thaninformation and thus more resistant to easy trans-fer, both within and across firms (Szulanski,1996). But von Hippel (1988) and Marsden(1990) have argued that close and intense inter-action between individual members of the con-cerned organizations acts as an effective mech-anism to transfer or learn sticky and tacit know-how across the organizational interface. Learningsuccess also rests upon an iterative process ofexchange between the member firms and theextent to which personnel from the two firmshave direct and intimate contact to further anexchange (Arrow, 1974; Badaracco, 1991). Asocial exchange approach provides the basis forsuch interaction and exchange. Strong relationalcapital usually engenders close interactionbetween alliance partners. It can thus facilitateexchange and transfer of information and know-how across the alliance interface.

A firm is also able to learn from alliancepartners more easily when the level of trans-parency or openness between them is high(Hamel, 1991; Doz and Hamel, 1998). The pri-mary hindrance to such openness or transparencyis the mutual suspicion of opportunistic behaviorbetween alliance partners which generally causesthem to be less willing to share informationand know-how with each other. Existing researchsuggests that mutual trust between partnersreduces the fear of such opportunistic behavior(Gulati, 1995; Zaheeret al., 1998), allowing forgreater transparency between the exchange. Build-ing upon it, we could argue that trust-basedrelational capital can contribute to a freer andgreater exchange of information and know-how


between committed exchange partners. This isbecause decision-makers do not feel that theyhave to protect themselves from the others’opportunistic behavior (Blau, 1977; Jarillo, 1988,Inkpen, 1994). Without its existence, the infor-mation and know-how exchanged would be lowin accuracy, comprehensiveness, and timeliness.

Overall, we believe that strong relational capitalbetween alliance partners facilitates greater learn-ing across the alliance interface. Thus,

Hypothesis 1a: The greater the relationalcapital that exists between the alliance part-ners, the greater will be the degree of learn-ing achieved.

Nevertheless, certain scholars have suggested thatpronouncements such as ‘build relationships tocreate harmony and learning’ are fraught withcomplications owing to the inherent contradictionamong the different strategic objectives that firmsseek in alliances (Yoshino and Rangan, 1995).A potential danger in alliance situations is therisk of unilaterally losing core proprietary know-how or capabilities to the partner (Hladik, 1988).A firm derives its competitive strength from itsproprietary assets and will be protective aboutlosing them to the alliance partner. Partnershipsare fraught with hidden agendas driven by theopportunistic desire to access and internalize thepartner’s core proprietary skills much faster thanthe partner. These ‘learning races’ often leave afirm in a Catch-22 situation: if it contributes toolittle to building the relationship, the alliance maybe doomed to fail (Khannaet al., 1998); on theother hand, if it contributes too much and tooopenly, its partner will gain the upper hand(Doz, 1988).

Although the transaction cost perspective rec-ommends a variety of contractual mechanisms toguard against partner opportunism, scholars fromother perspectives have suggested alternate meansfor minimizing it. Dyer and Singh (1998) proposealternatives of self-enforcing agreements, whichare sometimes referred to as ‘private ordering’ inthe economics literature or ‘trust’ in the sociologi-cal literature. Sociologists, anthropologists, andlegal scholars have long argued that informalsocial controls supplement and often supplantformal controls (Macaulay, 1963; Granovetter,1985). These self-enforcing agreements rely onrelational capital or reputation as governance


mechanisms and are often a more effective andless costly means of protecting specialized invest-ments and proprietary assets (Sako, 1991; Hill,1995). Relational capital creates a mutual confi-dence that no party to an exchange will exploitothers’ vulnerabilities even if there is an oppor-tunity to do so (Sabel, 1993). This confidencearises out of the social controls that such capitalcreates. Partners in an alliance often specify whatis core or proprietary to each party and developinformal or formal codes of conduct to restrictbehavior or action that leads to the appropriationof such assets. Relational capital reduces the ten-dency of alliance partners to break such informalexisting agreements that might be in place. Partiesto the exchange make a good-faith effort not totake excessive and unilateral advantage of theother, even when the opportunity is available.Thus overall, we can argue that trust-basedrelational capital can counteract the potential ofopportunistic or self-serving behavior by thealliance partner(s) and thus mitigate the possi-bility of losing one’s core proprietary assets tothe partner.

Hypothesis 1b: The greater the relationalcapital between alliance partners, the greaterwill be the ability to protect core proprietaryassets from each other.

Conflict management

A critical aspect of any partnership is the poten-tial for conflict between the alliance partners andhow they deal with it. Conflict often exists in anyalliance relationship on account of the inherentdependencies involved in such interactions. Giventhat a certain amount of conflict is expected, howsuch conflict is managed is important (Borys andJemison, 1989), as the impact of conflict reso-lution on the relationship can be productive ordestructive (Deutsch, 1969).

A number of factors are associated with man-aging conflicts integratively. Integrative conflictmanagement entails joint management of conflictwith mutual concern for ‘win-win’ for all con-cerned (Bazerman and Neal, 1984). It engendersa communication- and contact-intensive processof conflict management. Strong two-way com-munication is a key element of successful conflictresolution (Cummings, 1984). MacNeil (1981)and others acknowledge the importance of honest


and open lines of communication to the continuedgrowth of close ties and resolution of potentialconflict situations. Our fieldwork also shows theimportance that experienced managers give toeasy and open communication for addressingalliance-related conflicts. Besides communication,readiness to engage in joint problem solving anda mutual concern for ‘win–win’ outcomes willoften produce mutually satisfactory solutions.Joint problem solving fosters closer collaborationbetween the alliance partners, thereby creating amore conducive environment for futurecooperation. On the other hand, use of destructiveconflict resolution techniques like domination,coercion (Deutsch, 1969), and an attitude portray-ing a ‘win–lose’ perspective are seen ascounterproductive and are likely to strain thefabric of the alliance.

Sometimes the method of conflict managementis institutionalized, with partners setting up formaljoint mechanisms to ‘monitor’ potential conflictsituations. Monitoring not only provides eachpartner with a better understanding of mutualconcerns but also enables prompt recognition ofpotential conflict situations. An equally importantelement of most conflicts is the organizational orcultural distance between the alliance partners(Harrigan, 1988b; Parkhe, 1993). Attempts toaddress cultural obstacles in an explicit and inte-grative manner should lower the potential forconflict and enhance the likelihood of alliancesuccess.

We believe that an integrative process of con-flict management significantly impacts both thenature of the relationship that exists betweenalliance partners and the specific outcomes ofinterest, namely learning and the protection ofproprietary assets. An integrative method of con-flict resolution engenders feelings of proceduraljustice between the alliance partners, wherebypartners view the decision process to be fair andjust. Such feelings affect individuals’ higher-order attitudes of trust and commitment (Kimand Mauborgne, 1998) as well as lead to thedevelopment of positive psychological feelingstowards individuals from the other side. Ourfieldwork with companies like Hewlett Packardor Corning also demonstrates the importance ofintegrative conflict management towards build-ing a stronger relationship between alliance part-ners. Thus, effective and integrative conflictmanagement can be an important catalyst in


building relational capital between the alliancepartners.

The communication- and contact-intensiveprocess of conflict management also aids thelearning process. Learning from the alliance part-ner is strongly conditioned by the closeness ofinteraction between the partners, especially thedegree to which personnel from the partner firmshave direct and intimate contact with each other.Two-way communication and joint problem solv-ing, both of which are key aspects of managingconflicts integratively, involve close interactionbetween individuals across the alliance interface.Thus it creates a potentially useful channel tolearn or transfer critical information or know-howbetween them. Second, perceptions of proceduraljustice that result from integrative conflict man-agement induce easier exchange of knowledgeand ideas between the partners (Kim and Mau-borgne, 1998). Thus,

Hypothesis 2a: The greater the extent towhich conflicts are managed in an integrativefashion, the greater will be the relational capi-tal between alliance partners.

Hypothesis 2b: The greater the extent towhich conflicts are managed in an integrativefashion, the greater will be the degree oflearning achieved.

Integrative conflict management can also impacteach firm’s ability to protect its proprietaryassets. Conflicts in alliances are often centeredupon issues of asymmetrical contributions byrespective alliance partners and the returns tothem (Khanna et al., 1998). The communi-cation-intensive process of conflict managementhelps alliance partners to clearly define whateach partner contributes or gets from therelationship and what is ‘off-limits.’ Contact-intensive mechanisms help alliance partners tomonitor not only potential conflict situations butalso instances of opportunistic (or secretive)attempts by either party to unilaterally absorbcore proprietary assets of the other. As discussedearlier, integrative conflict management alsoengenders feelings of procedural justice andtrust between the partners to minimize selfishor opportunistic behavior on the part of anypartner. Collectively, these attributes of in-tegrative conflict management enable alliance


partners to better protect their core proprietaryassets from each other. Thus,

Hypothesis 2c: The greater the extent towhich conflicts are managed in an integrativefashion, the greater will be partners’ abilityto protect core proprietary assets from eachother.

On the other hand, an integrative approach toconflict management requires partner firms toengage in close and intense interaction at multiplelevels across the alliance interface. Communi-cation is also undertaken more closely, frequently,and openly to recognize and eliminate potentialconflict situations. All of these activities may notbode well for the firm’s ability to control theflow of important and critical information andknow-how across the alliance interface. Althoughinstitutionalized means of monitoring conflict mayalleviate this threat partially, they may still beineffective at preventing unwanted leakage andthe loss of some important proprietary know-howto the partner firm.

Controls

Organizational fit: Compatibility andcomplementarity

In studying alliances, academics and practitionershave usually emphasized some of theex antestructural characteristics of the alliance (Harrigan,1988b). Specific importance has been given tothe organizational fit between alliance partners,with the following dimensions of fit beingregarded the most critical: complementarity andcompatibility between the partners (Harrigan,1988b; Tucchi, 1996).

Complementarity between the alliance partnersrefers to the lack of similarity or overlap betweentheir core businesses or capabilities—the lowerthe similarity, the greater the complementarity(Mowery et al., 1996b). Harrigan (1988a) showsthat ventures and partnerships are more likely tosucceed when partners possess complementarymissions and resource capabilities. Comp-lementarity ensures that both partners bring differ-ent but valuable capabilities to the relationship.It also creates the potential for each firm to learnfrom its partner. Moweryet al., (1996) find thatcomplementarity (i.e., less overlap) between the


alliance partners correlates positively with inter-partner learning across the alliance interface.

Researchers have also argued that compatibilityof partners is an important aspect of fit thataffects alliance outcomes. In a study of 90 jointventures, Geringer (1988) demonstrates how part-ner compatibility correlates with alliance success.He also discusses how firms employ nine firm-specific related criteria to assessex antepartnercompatibility along several dimensions. Compati-bility of partners has been assessed in severalways: operating strategy, corporate cultures, man-agement styles, nationality (Parkhe, 1993), andat times even firm size. Compatibility betweenpartners fosters the ‘chemistry’ between them. Italso facilitates the reconciliation of differencesbetween partners (De la Sierra, 1995) to enableopen and easy exchange between them. Compati-bility between the partners allows the firms toactually capitalize on the learning potentialoffered by the complementarity of capabilitiesbetween them. Overall, fit in terms of compati-bility and complementarity is expected to posi-tively impact both relational capital and learningbetween partners.

Alliance governance (equity vs. nonequity).Asmentioned earlier, a large body of the allianceliterature based on the transaction cost perspectiveexplains the presence and impact of equity inalliances. The presence of equity not only alignsthe interests of the partner firms but also providesa basis for monitoring partner behavior (Kogut,1988; Hennart, 1988; Pisano, 1989) so as toreduce the possibility of opportunistic behaviorby any of the partner(s). Alignment of interestsdue to equity is expected to result in much closerinteraction between the partners. This interactionshould facilitate learning and exchange of infor-mation and know-how, especially of tacit knowl-edge, across loosely connected firms (Badaracco,1991). Various studies have shown that equityarrangements promote greater interfirm knowl-edge transfers than do mere contractual ones(Kogut, 1988; Moweryet al., 1996). In addition,since equity alleviates the hazard of partneropportunism, equity alliances are expected tominimize the likelihood of a firm losing its coreproprietary know-how to the partner.

Prior alliances. Current research has highlightedthe important role of trust in alliances (Gulati,


1995; Dyer and Singh, 1998; Zaheeret al., 1998).Since trust itself is difficult to observe and meas-ure, researchers have used a factor that likelyproduces trust as its proxy, namely prior alliancesbetween the firms (Gulati, 1995). The underlyingintuition is that two firms with prior alliances arelikely to trust each other more than other firmswith whom they have had no alliances (Ring andVan de Ven, 1989). By generating a high degreeof trust and interaction, repeat alliances shouldfacilitate a high degree of learning and infor-mation or know-how exchange between partners.At the same time, the presence of a prior cooper-ative history between the two firms also limitsthe possibility of opportunistic behavior betweenthem, thus reducing the threat that one of thefirms will lose its core proprietary assets to itspartner.

Nationality. If alliance partners are of differentnationalities, problems related to cultural differ-ences, opinions, beliefs, and attitudes are accentu-ated. Language can also be a problem, especiallyif the interface managers cannot speak the part-ner’s language (Killing, 1982). Harrigan (1988b)finds differences in national origins to have asignificantly negative relationship with expectedoutcomes. Parkhe (1993) also finds that allianceoutcomes and performance are strongly linked topartner nationalities. Specific to learning, Moweryet al. (1996) argue that the forbidding barriersof culture, language, educational background, anddistance with cross national partners should resultin lower levels of learning and knowledge trans-fer. These barriers have also been noted to accen-tuate partner tendencies to engage in opportunisticbehaviors (Reich and Mankin, 1986).

Age. We have also included a control to capturethe impact of alliance duration on the variablesof interest. This is because it could be arguedthat the greater the duration of the alliance, thegreater would be the learning from the alliancepartner. At the same time, longer duration wouldalso increase the likelihood of losing one’s pro-prietary assets to the partner firm.

RESEARCH METHODOLOGY

To understand the dynamics in learning alliances,we not only studied extant literature in the areas


of strategic alliances and organizational learningbut also supplemented this knowledge withfieldwork in a few companies. We used thesetwo sources to develop the theoretical model thataddresses the research question. This exercise alsoprovided richness of contextual detail permittinggrounded specification of the framework and con-structs. We then collected data that would allowus to test our framework and hypotheses.

Data collection and sample

The level of analysis is an alliance between twopartners. Alliance-related data on aspects such asrelational capital or conflict management arealmost impossible to get through archival sources.One could collect these data through interviewswith or surveys of managers who are responsibleor knowledgeable about their firm’s alliance(s).Although in-depth interviews provide a rich tap-estry of information, it was beyond the scope ofthis project to collect data through interviewsfrom a large sample. Instead, we decided tocollect the data through survey questionnairesadministered to relevant managers across a largesample of alliances formed by U.S.-based com-panies.

Given our research question, it was necessaryto study firms that have engaged in alliances andthat operate in industries where alliances are acritical means of competing. Past research showsthat industries such as pharmaceuticals, chemi-cals, computers (hardware and software), elec-tronics, telecommunications, and services fallwithin this category (Culpan and Eugene, 1993).To select the sample, we first identified com-panies with more than $50 million annual salesfor the year 1994, in each of these industries. Wethen identified appropriate respondents in each ofthese firms. Although most survey-based studieson alliances have generally relied on sending thesurveys to the CEO (Mohr and Spekman, 1994;Simonin, 1997), our fieldwork suggested thatthere may be other people in companies to whomwe could send the questionnaire. For example,companies often have executives with focalcorporate responsibility for strategic alliances,corporate development, or mergers and acqui-sitions. These executives are more directlyresponsible or knowledgeable about their firms’alliances. These individuals, whom we refer toas the primary recipients, were identified in two


different ways. First, we used secondary datasources, such as the Standard & Poors’ digest oncompany executives, to create a preliminary listof executive names and contact details. In caseswhere we did not have enough information, wecalled up the company to collect or reconfirmthis information. In some cases we were directedto send the survey to an executive or managerwho was different from the one we had in ouroriginal list. We dropped cases where we failed toget sufficiently clear information. We eventuallymailed our survey to 592 companies.

The primary recipient in each company wasrequested to select any one alliance that the com-pany had been involved with and forward thequestionnaire to a manager who was directlyassociated with that alliance. This latter individualwas the primary respondent to the survey ques-tionnaire. In certain cases, the primary recipientselected an alliance for which he/she was alsothe primary respondent. We received responsesfrom 278 companies, of which 212 were completein all respects. With respect to the companies’sales and employees, no significant differenceswere observed between the respondent and nonre-spondent groups.

Measurement

Multi-item scales were used to collect data onmost of the key constructs. Since little empiricalprecedent existed to develop these measures, werelied on extant literature and our fieldwork toselect individual items for our scales. Simplicityin scoring was sought by using a balanced 7-point Likert-type scale that is easy to master.Basically, each respondent was asked to indicatethe extent to which he/she disagreed or agreedwith the given statement, such that 1= StronglyDisagree and 7= Strongly Agree. We pretestedthe survey instrument with a small group ofmanagers from different companies before send-ing out the final version. Pretesting helped usmodify the language suitably and reject itemsthat were difficult to understand, or involvedunnecessary repetition. The Appendix providesdetails of individual items used to measure eachtheoretical construct.

The dependent variables, ‘learning’ and ‘pro-tection of proprietary assets,’ and the keyexplanatory variables, ‘conflict management’ and‘relational capital,’ are all measured using multi-


item scales. Among the controls, ‘partner fit’ interms of complementarity and compatibility isalso measured with a multi-item scale. However,for the rest of the control variables, we relied oncategorical measures to obtain the responses. Foralliance structure, respondents had to indicatewhether the alliance was an equity alliance ornot and the responses were coded as ‘Yes= 1’and ‘No= 0.’ Similarly, respondents provided a‘Yes/No’ answer to indicate the existence of prioralliances between the partners, where existenceof prior alliances was coded as ‘1’ and ‘0’ other-wise. For partner nationality, respondents had togive a ‘Yes/No’ response to whether the alliancepartners belonged to same nationality and thecoding was such that ‘Yes= 1’ and ‘No = 0.’Alliance duration is a simple count of the numberof years since the alliance was formed.

RESULTS AND ANALYSES

The analyses have been conducted in multiplestages such that results from these can collectivelyhelp assess the proposed framework and hypoth-eses. When multiple-item scales are used to meas-ure latent constructs and a composite score basedon these items is used in further analyses, it isimportant to assess the validity and reliability ofthe scales used (Gerbing and Anderson, 1988).Selection of scale items on the basis of priorliterature, fieldwork, and pretesting of the surveyinstrument helped ensure content or face validity.To assess scale reliability, we computed Cronbachalphas for each multiple scale item and foundthese to be well above the cut-off value of 0.7in each case (Nunnally, 1978). Table 1 providesthe results of this analysis. Table 2 provides thedescriptive statistics and correlation matrix of thekey variables.

We first used ordinary least-squares regression

Table 1. Reliability of scales used to measure latentconstructs

Construct Cronbacha Items Valid N

Partner fit 0.8165 4 239Relational capital 0.9063 5 252Conflict 0.9160 6 231management


Table 2. Descriptive statistics and correlation matrix

Variable Mean S.D. PP CM RC DUR LER PC

PP 4.33 1.76 1.00 0.51 0.49 0.13 0.41 0.39CM 4.16 1.68 1.00 0.67 0.10 0.64 0.56RC 4.00 1.63 1.00 0.18 0.68 0.45DUR 3.70 3.88 1.00 0.07 −0.02LER 4.13 1.89 1.00 0.39PC 3.98 1.58 1.00

*Figures in italics are significant at the 0.05 levelPP, partner fit; CM, conflict management; RC, relational capital; DUR, alliance duration; LER, learning; PC, protection ofproprietary assets or crown jewels

Table 3. OLS regression

Model 1a/1b—dependent variable: Learning from the alliance partnerModel 2a/2b—dependent variable: Protection of proprietary assets

Explanatory variables Model 1a Model 1b Model 2a Model 2b

Relational capital 0.432* 0.498* 0.401** 0.328**Conflict management 0.374* 0.335** 0.186** 0.184***Partner fit 0.129 0.110Previous alliances 0.077 −0.067Alliance duration 0.035 −0.037Partner nationality 0.112 0.101Alliance governance 0.124 −0.120R2 0.594 0.647 0.316 0.354Number of observations 212 178 200 178

*p , 0.01; **p , 0.05; ***p , 0.10

to test the hypotheses. Separate models, theresults of which are shown in Table 3, wereestimated for each of the two dependent variables:degree of learning achieved and protection ofproprietary assets.

The results of Models 1a and 1b in Table 3provide strong support for Hypotheses 1a and 2b.Relational capital shares a significant and positiverelationship with the degree of learning achieved.These results underscore the importance of havingstrong relational capital with the alliance partnerin order to enhance learning in alliance situations.Conflict management also has a significant andpositive relationship with the dependent variable.A communication- and contact-intensive processof managing conflicts relates positively to learn-ing success. Despite relatively high correlationbetween the two explanatory variables, concernsabout unstable regression coefficients are mini-mized since each of them has a strong and sig-nificant relationship with the respective dependent


variables. Tests for multicollinearity also showthat each of these variables has significantexplanatory power by itself and that the extentof collinearity is well within generally acceptablelimits. The tolerance values for each explanatoryvariable are well above the cut-off value of 0.1,and the variance inflation factor values are wellbelow the cut-off value of 10 (Hairet al., 1998).Of the control variables, we observe that onlypartner fit is marginally significant in explainingvariation in learning success.

Results of Models 2a and 2b, which have ‘pro-tection of proprietary assets’ as the dependentvariable, provide support for Hypotheses 1b and2c. The significant and positive relationshipbetween relational capital and protection of pro-prietary assets highlights the importance of infor-mal self-enforcing governance mechanisms inalliances. Relational capital, based on mutualtrust, friendship, and respect between the alliancepartners, effectively curbs partner opportunism to


protect against leakage of core proprietary assets.None of the other variables, including the con-trols, shows any significant relationship with pro-tection of proprietary assets. This result is quitesurprising, given the emphasis placed by priorresearch on aspects like equity governance orprior ties.

Instead of conducting the analyses separatelyas above, we can use methods that combinethese techniques as well as provide additionaladvantages. Structural modeling is one such tech-nique that can be used. It consists of two stages:(a) a measurement model that assesses reliabilityand validity of the scales used to measure eachlatent construct, and (b) a structural model thatlays out and estimates multiple dependentrelationships between the constructs of interest.The true value of structural equation modelingcomes from the benefit of analyzing the structuraland measurement models simultaneously. Anadditional advantage of this technique lies inits ability to estimate a series of dependencerelationships, wherein one dependent variablebecomes the explanatory variable in subsequentrelationships. It also allows researchers to assessthe impact of explanatory variables on two ormore dependent variables, at the same time (Hairet al., 1998).

In our theory section, we had suggested thatconflict management and partner fit could haveboth a direct impact on the two dependent vari-ables (learning and the protection of proprietaryassets), as well as an impact on the relationalcapital between partners. Thus, relational capitalwould be a dependent variable with respect toconflict management and partner fit and anexplanatory variable with respect to learning andprotection of proprietary assets. Structural model-ing is well equipped to handle such multipledependent relationships. We also believe that inalliance situations firms face the tension of tryingto achieve the two focal objectives, learning andprotecting proprietary assets, simultaneously.Thus instead of estimating separate models forthe relationships between the explanatory vari-ables and each of the dependent variables, asdone earlier, we can use structural modeling toestimate the two sets of relationships si-multaneously. Finally, since we are measuringeach of the theoretical constructs using a numberof manifest items, the measurement model canalso help us examine the validity and reliability


of these constructs, even as we examine thedependence relationships between them.

Figure 1 provides the path diagram for themodel that includes the multiple dependentrelationships that we propose and Tables 4 and 5provide the equations for the measurement andstructural models based on the path diagram.

In the model that we estimate we have omittedall control variables, except partner fit, for severalreasons. First, structural modeling is better suitedto examine relationships between constructs thatare measured using interval or ratio scales. Mostcurrent techniques are not well suited toadequately handle categorical explanatory vari-ables such as all of our controls, with the excep-tion of partner fit. Second, our initial analysesshow that none of those controls is in any waysignificantly related to the dependent variables.Thus, dropping them from our model should notconstitute severe problems. We estimate themodel using the maximum likelihood estimationprocedure of LISREL 7, which is robust, efficient,and unbiased, when the assumption of multi-variate normality is met (Joreskog and Sorbom,1988). Results of the analysis are discussed inthe following section.

Overall model fit

The first step in structural modeling is to assessoverall model fit with one or more goodness-of-fit measures. Goodness-of-fit is a measure of thecorrespondence of the actual or observed input

Figure 1. Path diagram (structural modelling)


Table 4. Measurement model

(a) Measurement model (exogenous constructs)

Exogenous Exogenous constructs Errorindicator

j1 j2

X1 = l11j1 +d1X2 = l21j1 +d2X3 = l31j1 +d3X4 = l41j1 +d4X5 = l12j2 +d5X6 = l22j2 +d6X7 = l32j2 +d7X8 = l42j2 +d8X9 = l52j2 +d9X10 = l62j2 +d10

X1–X4: indicators for ‘partner fit’ (j1) corresponding to PP1–PP4 in Figure 1.X5–X10: indicators for ‘conflict management’ (j2) corre-sponding to CM1–CM6 in Figure 1.l11–l62: parameters estimating the relationship betweenmanifest indicators and latent constructs.d1–d10: error terms for indicators X1–X10.

(covariance or correlation) matrix with that pre-dicted from the proposed model. If the proposedmodel has acceptable fit, by whatever criteriaapplied, the researcher has not ‘proved’ the pro-posed model, but has only confirmed that it isone of the several possible acceptable models(Hair et al., 1998).

The first measure we report is the likelihoodratio chi-square statistic. For the proposed model,we get a chi-square of 316.19 (d.f.= 177). If themodel is to provide a satisfactory representationof the data, it is important for the chi-squarevalue to be nonsignificant (p . 0.05). The sig-nificance level of 0.02 for the chi-square of ourmodel is close to the usually acceptable thresholdof 0.05, indicative of partially acceptable fit. Thesecond measure we report is the normed chi-square (Joreskog, 1969), where the chi-square isadjusted by the degrees of freedom to assessmodel fit. Models with adequate fit should havea normed chi-square less than 2.0 or 3.0(Carmines and McIver, 1981). With a normedchi-square of 1.78, the proposed model providesa fairly satisfactory representation of the data.The third measure reported is the GFI index,which is the most popular goodness-of-fit measureprovided by LISREL analysis (Joreskog and Sor-


(b) Measurement model (endogenous constructs)

Endogenous Endogenous constructs Errorindicators

h1 h2 h3

Y1 = l11h1 +e1Y2 = l21h1 +e2Y3 = l31h1 +e3Y4 = l41h1 +e4Y5 = l51h1 +e5Y6 = l12h2 +e6Y7 = l22h2 +e7Y8 = l32h2 +e8Y9 = l13h3 +e9Y10 = l23h3 +e10

Y1–Y5: indicators for ‘relational capital’ (h1) correspondingto RC1–RC5 in Figure 1.Y6–Y8: indicators for ‘Learning’ (h2) corresponding toL1–L3 in Figure 1.Y9–Y10: indicators for ‘protection of proprietary assets’ (h3)corresponding to PC1–PC2 in Figure 1.l11–l23: parameters estimating the relationship betweenmanifest indicators and latent constructs.e1–e10: error terms forY1–Y10.

Table 5. Structural model

Endogenous Exogenous Endogenous Errorconstructs constructs constructs

j1 j2 h1 h2 h3

h1 = l11j1 + l12j2 + z1h2 = l21j1 + l22j2 + b21h1 + z2h3 = l31j1 + l32j2 + b31h1 + z3

h1 = construct representing ‘relational capital’h2 = construct representing ‘learning’h3 = construct representing ‘protection of proprietary assets’j1 = construct representing ‘partner fit’j2 = construct representing ‘conflict management’g11–g32 = parameters estimating the relationship betweenexogenous and endogenous constructsb21–b31 = parameters estimating the relationship betweenvarious endogenous constructsz1–z3 = error terms

bom, 1988). It is a nonstatistical measure rangingin value from 0 (poor fit) to 1.0 (perfect fit).We get a GFI of 0.89 for our model, which issufficiently close to the generally acceptable levelof 0.90 (Hair et al., 1998. We also assessed theincremental fit of the model compared to the nullmodel by examining the Normed Fit Index. The


Normed Fit Index of 0.91 is above the desiredthreshold level of 0.90. Overall, the differentgoodness-of-fit measures indicate partial support

Table 6. (a). Measurement model: Parameter esti-mates

Construct Parameter t-statisticindicators estimate

Partner fit (PP)PP1 0.912 14.22*PP2 0.870 13.92*PP3 0.453 6.23*PP4 0.619 9.11*

Conflictmanagement(CM)CM1 0.857 14.68*CM2 0.889 15.63*CM3 0.848 14.24*CM4 0.891 15.85*CM5 0.735 11.69*CM6 0.848 14.21*

Relational capital(RC)RC1 1.00RC2 0.810 9.41*RC3 0.872 11.36*RC4 0.883 11.73*RC5 0.851 10.47*

Learning (L)L1 1.00L2 0.921 14.24*L3 0.835 9.08*

Protection ofproprietary assets(PC)PC1 1.00PC2 0.683 6.412*

*p-value, 0.001

(b). Measurement model: Construct reliability

Construct Reliability estimates

Partner fit 0.826Conflict management 0.912Relational capital 0.902Learning 0.905Protection of prop. assets 0.848

Note: Threshold levels for acceptability—constructreliability . 0.70


for the proposed model. Although not perfect, thelevel of fit seems sufficient enough to proceedwith the assessment of the measurement andstructural models.

Measurement model fit

In the measurement model, the first step is toexamine the loading of the manifest indicatorson the underlying theoretical constructs and tofocus on nonsignificant loadings, if any. As wesee in Table 6a, all the indicators are significantlyrelated with their respective underlying constructs(t-values. 2.0 andp , 0.05).

Since none of the indicators have a loadingthat is so low or nonsignificant that they shouldbe deleted, we can proceed to assess the validityand reliability of the construct scales. The sig-nificance of the factor loadings provides supportfor the convergent validity of the respective scales(Anderson and Gerbing, 1988). Discriminant va-lidity was assessed by comparing a model withthe correlation between two explanatory con-structs constrained to equal one with an uncon-strained model. A significantly lower chi-squarefor the model with unconstrained correlation pro-vides support for discriminant validity (Joreskog,1971). Table 6b provides results for scalereliability. We see that the reliability estimatesexceed the suggested level of 0.70, in all cases.Together, the results suggest that the manifestindicators are significant and reliable measures ofthe latent constructs being used. Our analysisalso revealed significant correlation (p , 0.05)between the measurement errors for some of theindicators within constructs (e.g.,d1 and d2; d5and d6; d7 and d8; e2 and e3; e5 and e6).Correlated measurement errors suggest the exis-tence of consistent response bias across certainindicators within constructs that needs to be con-trolled for while estimating the model.

Structural model fit

Having assessed the overall model fit and themeasurement model, we can now examine thetheoretical relationships between the underlyingconstructs. The most obvious examination in thestructural model involves the significance of theestimated coefficients. Structural modelingmethods provide not only estimated coefficientsbut also standard errors andt-values for each


coefficient. Table 7 contains the results for thevarious structural equations.

1. Both relational capital and conflict man-agement show a statistically significant (t–value. 2.0 andp-value, 0.05) and positiverelationship with ‘learning.’ This result pro-vides support for the results of the multipleregression conducted earlier.

2. For ‘protection of proprietary assets,’ conflictmanagement is the only significant explanatoryvariable (t = 2.318 andp = 0.020). Relationalcapital is, however, just outside the signifi-cance range.

3. Besides having a positive and significantrelationship with the two core dependent vari-ables, conflict management also has a positiveand significant association with the relationalcapital that exists between the alliance partners(t = 3.50 and p = 0.001). This result mayexplain why the relationship between relationalcapital and protection of proprietary assetsbecomes less significant when we use multi-stage structural modeling as compared to usingordinary OLS regression.

Obtaining an acceptable level of fit suggests thatthe proposed model explains or fits the data quitesatisfactorily. However, other models, based onalternate theories, may provide equal or better fit.Thus, a stronger test of the proposed model is totest competing models that estimate other theo-retically plausible relationships between the con-structs. In our case, we estimated two other com-peting models. In the first of these models, weconsidered conflict management to be anendogenous construct rather than an exogenousone the way we have hypothesized. This is

Table 7. Structural model: Parameter estimates

Construct relationship Parameter t-statistic p-valueestimate

Partner fit→ learning 0.037 0.826 0.409Conflict management→ Learning 0.290 2.629 0.009Relational capital→ Learning 0.607 4.003 0.000Partner fit→ Protecting prop. assets 0.090 1.568 0.119Conflict management→ Protecting prop. assets 0.332 2.318 0.020Relational capital→ Protecting prop. assets 0.130 1.493 0.131Partner fit→ Relational capital 0.026 0.907 0.364Conflict management→ Relational capital 0.519 3.500 0.001


because theoretically it could be plausible toargue that better relational capital betweenalliance partners would allow them to manageconflicts more integratively. To test this alterna-tive argument we estimated a model whereinwe introduced a unidirectional relationship fromrelational capital to conflict management, whileretaining most of the other relationships in ourproposed model. However, this model, with a GFIof 0.83 and a significant chi-square (x2 = 385,p , 0.00), was an inferior fit as compared to theoriginal model. We also estimated a modelwherein we dropped both the intermediaterelationships from partner fit and conflict man-agement to relational capital while retaining allthe direct relationships between the explanatoryand dependent variables. This model too, with aGFI of 0.63 and a significant chi-square(x2 = 487, p , 0.00), indicated poor fit. Theinferior fit of the other models increased theoverall acceptance of the proposed model.

Statistically, it is possible to estimate severalmore models to examine which of them explainsthe data best. However, in this paper our primarygoal in using structural modeling is to assess thebasic adequacy of a model that simultaneouslyaccounts for the multiple dependent relationshipsthat we theoretically propose, rather than toexpost identify the best-fitting model that had notbeen theoretically proposedex ante. It is likelythat other interesting and important relationshipsmay exist among some of our constructs. Forexample, it can be argued that success with learn-ing and/or protection of core assets influencesrelational capital or the ability to manage con-flicts. However, these relationships address verydifferent questions from the one posed here andfuture research would need to develop the theo-


retical arguments associated with these relation-ships in greater detail before estimating the corre-sponding models.

DISCUSSION

Overall our results provide some importantinsights into the dynamics and implications ofalliance management. Although most extant litera-ture emphasizes structural factors such as partnerfit and equity to explain alliance success, theresults of this study highlight the need to paygreater attention to how a firm manages thealliance, post formation, especially with regard tobuilding relational capital and managing conflicts.These aspects of alliance management play agreater role in explaining and determining keyalliance objectives such as learning and protectingcritical capabilities and skills—objectives thatquite often have been regarded as mutually exclu-sive. Learning, especially the acquisition ofdifficult-to-codify competencies, is best achievedthrough wide-ranging, continuous and intensecontact between individual members of thealliance partners. Relational capital based on mu-tual trust and respect fosters learning by encour-aging and facilitating such contact. It alsoincreases the willingness and ability of partnersto engage in a mutual exchange of informationand know-how to achieve reciprocal learning.Highlighting the role of relational capital, ourresults complement the work of other scholarswho have stressed the role of trust and personalinteraction in interorganizational relationships(Gulati, 1995; Zaheeret al., 1998). We show,however, that relational capital is linked not onlyto alliance success in general, but also to veryspecific and important objectives such as learningand limiting partner opportunism.

Although some alliance research has suggestedthat conflict management is an inherent andimportant part of most alliances, evidence of effec-tive outcomes based on conflict management islimited. Our study, however, highlights its impor-tance in enabling the outcomes of interest discussedhere. We see that managing conflicts integrativelyappears to foster learning in alliances in differentways. The communication and contact-rich mannerof resolving conflicts creates a channel for sharingand learning other useful information and know-how from the alliance partner. More importantly,


it also seems to influence the relational capitalbetween alliance partners, which in turn helps learn-ing. Most theory also suggests that alliances oftenraise the possibility of losing critical informationand competencies to the partner. Such losses mayoccur, either because of deliberate and opportunisticattempts by the partner to absorb such learning, orbecause of unexpected leakage through personalinteraction across the alliance interface. Our resultsshow that integrative conflict management helps inminimizing such occurrence. Close monitoring ofinterorganizational interaction to identify potentialconflict situations also helps detect and preventpartner behavior that might be directed towardssuch goals. Further, we see that conflict man-agement also enhances the relational capitalbetween the alliance partners, which acts as aninformal mechanism to check the leakage or steal-ing of core proprietary information or know-howacross the alliance interface. It reduces the moti-vation of each partner to engage in opportunisticacquisition and internalization of its partners’ skills.

Our findings are consistent with the relationalview of competitive advantage offered by Dyerand Singh (1998), who suggest that trust-basedgovernance is an important source of interorgani-zational rents, because it provides alliance partnerswith appropriate incentives to share valuableknowledge with each other. Such rents are sus-tainable because the relational safeguards are resist-ant to easy imitation by competition owing to thesocially complex and idiosyncratic nature of theexchange relationship. Dyer and Singh (1998) andDyer and Nobeoka (2000) have also argued thatthe existence of trust and relational capital betweenpartners encourages firms to set up idiosyncraticknowledge-sharing routines to further facilitate thelearning of specified and agreed-upon informationand know-how between them. In fact, we feel thatinclusion of variables that represent such knowl-edge-sharing routines will empirically improveoverall model fit quite substantially.

Besides having an impact at the dyadic levelbetween alliance partners, we believe thatrelational capital can also play a significant roleat the network level. Certain scholars have arguedthat strong interpersonal ties among existing part-ners create a basis for larger alliance networksto evolve (Gulati and Gargiulo, 1999). Relationalcapital that rests upon such ties engenders greatertrust between partners, thereby inducing them toform more alliances with each other in the future.


It also facilitates each partner to form allianceswith other companies, based on the referrals oftrustworthiness that each partner is ready to givefor its current partners owing to the strongrelational capital between them. Thus relationalcapital opens up greater opportunities for thefirms concerned to form new linkages and collab-orations with each other and with other companiesand thereby increase the network of alliances inwhich they are embedded (Ahuja, 2000; Gulatiand Gargiulo, 1999).

Relational capital can also influence the per-formance of individual firms embedded in alliancenetworks in several ways. Afuah (2000) arguesthat a firm’s competitive advantage is often ham-pered when technological change renders thecapabilities of its network of co-opetitors obso-lete. He goes on to suggest that companies canpotentially minimize damage if they have a meansof detecting such changes in a timely and correctmanner. We have seen that strong relational capi-tal between partners can foster timely and accur-ate exchange of information across the interface.It should thus enable individual firms to identify,better and faster, technological discontinuities thatmight hamper their co-opetitors and act accord-ingly to minimize the damage that may result.Similarly in the context of startups, Baum, Cala-brese, and Silverman (2000) have suggested thattheir performance is significantly influenced bythe size, composition, and diversity of theiralliance network. Startups with larger and morediverse networks are expected to enjoy superiorearly performance, because of greater and richeraccess to relevant information and capabilities ofsuch a network. We would like to suggest thatwhile network size and diversity provide a greaterpotential for accessing and learning importantinformation and capabilities, it is the quality ofthe relationship between network partners thatenables true and full realization of this potential.Startups that build a stronger relational capitalwith their network partners would exhibit higherperformance, all other aspects of the networkremaining the same. Essentially relational capitalat the dyadic level acts as a lubricant for poten-tially useful and important information to travelquickly and accurately through the network. Thus,having empirically established the role ofrelational capital with respect to learning in adyadic exchange, we believe that future researchcan further examine its role and impact in net-


work situations, as described in some of theabove situations.

In the network context, Gulati and Lawrence(1999) have examined value chain alliances(VCAs), which are vertical links between inde-pendent firms operating at successive stages inthe production chain. The authors argue thatVCAs are superior to arm’s-length arrangementsor even to vertically integrated firms because oftheir ability to provide high levels of differen-tiation and integration simultaneously, especiallyunder conditions of high task and environmentaluncertainty. Integration, which refers to unity ofeffort and information exchange between VCApartners (Gulati and Lawrence, 1999), enablesthem to leverage their differentiated and special-ized capabilities more effectively. Based on ourresearch, we believe that relational capital andintegrative conflict management can play a keyrole in enhancing such integration in VCAs. Mu-tual interaction and trust that engender relationalcapital not only will enable VCA partners towork more unitedly, but also facilitate easier flowof information and skills between them.

Before we conclude we would like to highlightseveral important limitations of this paper. Owingto practical considerations, like most large-samplesurvey research on alliances we too haveresponses, for both the dependent and independentvariables, from just one of the alliance partners.Ideally it would be beneficial to get an assessmentfrom all/both partners on aspects like relationalcapital or conflict management since they relateto aspects concerning both/all partners. It wouldbe equally interesting to examine how these vari-ables impact learning opportunities and successof both partners. While strong relational capitalcan enhance the learning potential for both part-ners, actual learning may perhaps differ becauseof differential learning abilities of the concernedpartners. Our data do not allow us to examinethis issue. Second, in this research, we have reliedonly on perceptual measures to assess learningand protection of core assets. It would be usefulto develop alternative measures for these variablesusing more objective data and examine how theyrelate to their corresponding perceptual measuresas well as to the explanatory variables. Forexample, Moweryet al. (1996) have used patentcross-citations to assess learning in alliances.Future research could benefit by combining suchobjective measures of learning with survey-based


perceptual measures to investigate the importanttension that we have highlighted in this paper.Third, it is important to recognize that aspectslike relational capital and learning evolve overtime; so might the relationship between them.Yet, in the current study, we have only cross-sectional data on these aspects, which limits ourability to understand the full richness of theirdynamic nature and interaction and to infer strongcausal links between them.

We also believe that there is scope to improveupon and refine some of the measures that havebeen used. This study is one of the few that triesto examine and measure postformation alliancemanagement aspects like relational capital andconflict management, using survey-basedresearch, and there was little empirical precedentto develop most of the measures that were used.Future research can also work towards includingother important variables that might have an impactupon the dependent variables examined here. Forexample, research suggests that learning in allianceswould also be influenced by the learning/absorptivecapacity of the firms concerned (Cohen and Levin-thal, 1990), or the extent to which the partnersestablish knowledge-sharing or learning routines(Dyer and Singh, 1998) between them. Inclusion ofthese variables will improve model fit considerablycompared to some of the models estimated here.Besides including additional variables, futureresearch could also examine other theoreticalrelationships that might exist between some of theconstructs considered here and estimate whethermodels corresponding to alternate theories providebetter explanation of our data.

CONCLUSION

This paper is one of the few empirical studiesthat explores alliance dynamics at a cross-industrylevel and provides some empirical evidence tohighlight the significance of alliance managementpractices such as managing conflicts integrativelyand building relational capital. It shows that thesepractices can, in fact, help firms simultaneouslyachieve alliance objectives that are often believedto be mutually exclusive, i.e., learning criticalskills, capabilities, or information from the partnerand at the same time protecting one from losingcore proprietary assets to the partner.

To the extent that building relational capital


and managing conflicts in an integrative mannerare important to the success of alliances, com-panies can benefit substantially by possessing asuperior capability of managing these aspects ofalliance management. Research shows, however,that firms are quite heterogeneous with respect totheir alliance capabilities and that this heterogen-eity is linked both to the amount of prior allianceexperience they have had (Anand and Khanna,2000) and how they learn and leverage from thatexperience (Kale and Singh, 1999). It has beenobserved that the prior alliance experience of thefirm is important in being able to build or utilizeappropriate routines and mechanisms to buildrelational capital and manage conflicts. Futureresearch needs to explore these important researchquestions in greater detail.

REFERENCES

Afuah, A. (2000). ‘How much do your co-opetitors’capabilities matter in the face of technologicalchange?’, Strategic Management Journal, SpecialIssue,21, pp. 387–404.

Ahuja, G. (1996). ‘Collaboration and innovation: Alongitudinal study of inter-firm linkages and firmpatenting in the global advanced materials industry’,Ph.D. dissertation, University of Michigan.

Ahuja, G. (2000). ‘The duality of collaboration: Induce-ments and opportunities in the formation of interfirmalliances’, Strategic Management Journal, SpecialIssue,21, pp. 317–343.

Anand, B. and T. Khanna (2000). ‘Do firms learn tocreate value? The case of alliances’,Strategic Manage-ment Journal, Special Issue,21, pp. 295–315.

Anderson, E. 1990. Two firms, one frontier: Onassessing joint venture performance.Sloan Manage-ment Review, 31(2), pp. 19–30.

Anderson, J. C. and D. W. Gerbing (1988). ‘Structuralmodeling in practice: A review and recommendedtwo-step approach’,Psychological Bulletin, 103(3),pp. 411–423.

Arino, A. and J. de la Torre (1998). ‘Learning fromfailure: Towards an evolutionary model of collabo-rative ventures’, Organization Science, 9(3),pp. 306–325.

Arrow, K. J. (1974).The Limits of Organization. Nor-ton, New York.

Badaracco, J. L. (1988).General Motor’s Asian Alliances.Harvard Business School Press, Boston, MA.

Badaracco, J. L. (1991).The Knowledge Link. HarvardBusiness School Press, Boston, MA.

Bazerman, M. and M. Neal (1984).Negotiating Ration-ally. Free Press, New York.

Baum, A. C., T. Calabrese and B. Silverman (2000).‘Don’t go it alone: Alliance network compositionand startups’ performance in Canadian biotechnolo-


gy’, Strategic Management Journal, Special Issue,21, pp. 267–294.

Beamish, P. J. (1987). ‘Joint ventures in LDCs: Partnerselection and performance’,Management Inter-national Review, 27, pp. 23–37.

Blau, P. (1977). ‘A macrosociological theory of socialstructure’, American Journal of Sociology, 83,pp. 26–54.

Bleeke, J. and D. Ernst (1991). ‘The way to win incross-border alliances’,Harvard Business Review,69(6), pp. 127–135.

Borys, B. and D. Jemison (1989). ‘Hybrid arrangementsas strategic alliances: Theoretical issues in organi-zational combinations’,Academy of ManagementReview, 14, pp. 234–249.

Bradach, J. L. and R. G. Eccles (1989). ‘Marketsversus hierarchies: From ideal types to plural forms’,Annual Review of Sociology, 15, pp. 97–118.

Carmines E. and J. McIver (1981). ‘Analyzing modelswith unobserved variables: Analysis of covariancestructures’. In G. Bohrnstedt and E. Borgatta (eds.),Social Measurement: Current Issues. Sage, BeverlyHills, CA, pp. 61–73.

Cohen, W. and D. Levinthal (1990). ‘Absorptive capacity:A new perspective on learning and innovation’,Admin-istrative Science Quarterly, 35, pp. 128–152.

Culpan, R. and K. A. Eugene (1993). ‘Cross-nationalcorporate partnerships: Trends in alliance formation’.In R. Culpan (ed.), Multinational StrategicAlliances. Hayworth Press, New York, pp. 153–171.

Cummings, T. (1984). ‘Trans-organizational develop-ment’, Research in Organizational Behavior, 6,pp. 367–422.

De la Sierra, C. M. (1995).Managing Global Alliances:Key Steps for Successful Collaboration. Addison-Wesley, New York.

Deutsch, M. (1969). ‘Conflicts: Productive and destruc-tive’, Journal of Social Issues, 25, pp. 7–41.

Dore, R. (1983). ‘Goodwill and the spirit of marketcapitalism’, British Journal of Sociology, 34, p. 4.

Doz, Y. L. (1988). ‘Technology partnerships betweenlarger and smaller firms: Some critical issues’. InF. K. Contractor and P. Lorange (eds.),CooperativeStrategies in International Business. LexingtonBooks, Lexington, MA, pp. 169–186.

Doz, Y. (1996). ‘The evolution of cooperation in stra-tegic alliances: Initial conditions or learning proc-esses’,Strategic Management Journal, Summer Spe-cial Issue,17, pp. 55–84.

Doz, Y., and G. Hamel (1998).Alliance Advantage:The Art of Creating Value through Partnering. Har-vard Business School Press, Boston, MA.

Dyer, J. (1996). ‘Specialized supplier networks as asource of competitive advantage: Evidence from theauto industry’, Strategic Management Journal,17(4), pp. 271–292.

Dyer, J., and K. Nobeoka (2000). ‘Creating and manag-ing a high performance knowledge-sharing network:The Toyota case’,Strategic Management Journal,Special Issue,21, pp. 345–367.

Dyer, J. and H. Singh (1998). ‘The relational view:Cooperative strategy and sources of interorgani-


zational competitive advantage’,Academy of Man-agement Review, 23(4), pp. 660–679.

Gerbing, D. W. and J. C. Anderson (1988). ‘Anupdated paradigm for scale development incorporat-ing unidimensionality and its assessment’,Journalof Marketing Research, 25, pp. 186–192.

Geringer, J. M. (1988). ‘Partner selection criteria fordeveloped country joint ventures’,Business Quar-terly, 53, pp. 55–62.

Granovetter, M. (1985). ‘Economic action and socialstructure: A theory of embeddedness’,AmericanJournal of Sociology, 91, pp. 481–510.

Gulati, R. (1995). ‘Does familiarity breed trust? Theimplications of repeated ties for contractual choicein alliances’,Academy of Management Journal, 38,pp. 85–112.

Gulati, R., and M. Gargiulo (1999). ‘Where do inter-organizational networks come from?’,AmericanJournal of Sociology, March, pp. 177–231.

Gulati, R. and P. Lawrence (1999). ‘Organizing verticalnetworks: A design perspective’, paper presented atthe Special Issue Conference on Strategic Networksat the Kellogg School, Northwestern University.

Hair, J., R. Anderson, R. Tatham, and W. Black (1998).Multivariate Data Analysis. Macmillan, New York.

Hamel, G. (1991). ‘Competition for competence andinter-partner learning within international strategicalliances’, Strategic Management Journal, SummerSpecial Issue,12, pp. 83–103.

Harrigan, K. R. (1985).Managing for Joint VentureSuccess. Praeger, New York.

Harrigan, K. R. (1988a). ‘Joint ventures: Linking fora leap forward’,Planning Review, 14, pp. 10–14.

Harrigan, K. R. (1988b). ‘Strategic alliances and partnerasymmetries’, Management International Review,28, pp. 53–72.

Hennart, J. (1988). ‘A transaction costs theory of equityjoint ventures’,Strategic Management Journal, 9(4),pp. 361–374.

Hergert, M. and D. Morris (1987). ‘Internationalcollaborative agreements’,Columbia Journal ofWorld Business, 22(2), pp. 15–21.

Hill, C. W. L. (1995). ‘National institutional structure,transaction cost economizing and competitive advan-tage: The Case of Japan’,Organization Science, 5,pp. 594–607.

Hladik, K. J. (1988). ‘R&D and international jointventures’. In F. J. Contractor and P. Lorange (eds.),Cooperative Strategies in International Business.Lexington Books, Lexington, MA, pp. 15–21.

Inkpen, A. (1994). ‘The characteristics and performanceof Japanese–North American joint ventures in NorthAmerica’. In S. B. Prasad and R. B. Peterson (eds.),Advances in International Comparative Manage-ment, Vol. 9. JAI Press, Greenwich, CT, pp. 83–108.

Jarillo, C. (1988). ‘On strategic networks’,StrategicManagement Journal, 9(1), pp. 31–41.

Joreskog, K. G. (1969). ‘A general approach to con-firmatory maximum likelihood factor analysis’,Psy-chometrika, 34, pp. 183–202.

Joreskog, K. G. (1971). ‘Statistical analysis of coge-neric tests’,Psychometrika, 36, pp. 109–133.


Joreskog, K. G. and D. Sorbom (1988).LISREL 7.SPSS, Chicago, IL.

Kale, P. and H. Singh. (1999). ‘Building alliance capa-bilities: A knowledge-based approach’, Academy ofManagement Best Paper Proceedings, Chicago, IL.

Khanna, T., R. Gulati and N. Nohria (1998). ‘Thedynamics of learning alliances: Competition,cooperation and relative scope’,Strategic Manage-ment Journal, 19(3), pp. 193–210.

Killing, J. P. (1982). ‘How to make a global jointventure work’, Harvard Business Review, 60(3),pp. 120–127.

Kim, W. C. and R. Mauborgne (1998). ‘Proceduraljustice, strategic decision making and the knowledgeeconomy’, Strategic Management Journal, 19(4),pp. 323–338.

Kogut, B. (1988). ‘A study of the life cycle of jointventures’. In F. K. Contractor and P. Lorange (eds.),Cooperative Strategies in International Business.Lexington Books, Lexington, MA, pp. 169–186.

Kogut, B. and U. Zander (1992). ‘Knowledge of thefirm, combinative capabilities and the replication oftechnology’,Organization Science, 3, pp. 383–397.

Koh, J. and N. Venkatraman (1991). ‘Joint ventureformations and stock market reactions: An assess-ment in the information technology sector’,Academyof Management Journal, 34, pp. 869–892.

Macaulay, S. (1963). ‘Non-contractual relations in busi-ness: A preliminary study’,American SociologicalReview, 28, pp. 55–67.

MacNeil, I. (1981). ‘Economic analysis of contractualrelations’, Northwestern University Law Review, 75,pp. 1018–1063.

Madhok, A. (1995). ‘Opportunism and trust in jointventure relationships: An exploratory study and amodel’, Scandinavian Journal of Management,11(1), pp. 57–74.

Marsden, P. V. (1990). ‘Network data and measurement’,Annual Review of Sociology, 16, pp. 435–463.

Merchant, H. (1997). ‘International joint venture per-formance of firms in the non-manufacturing sector’.In P. Beamish and A. Inkpen (eds.),CooperativeStrategies. Lexington Books, Lexington, MA,pp. 428–456.

Mohr, J. and R. Spekman (1994). ‘Characteristics ofpartnership success’,Strategic Management Journal,15(2), pp. 135–152.

Mowery, D. C., J. E. Oxley and B. S. Silverman(1996). ‘Strategic alliances and interfirm knowledgetransfer’, Strategic Management Journal, WinterSpecial Issue,17, pp. 77–92.

Nelson, R. and S. Winter (1982).An EvolutionaryTheory of Economic Change. Belknap Press, Cam-bridge, MA.

Nunnally, J. C. (1978).Psychometric Theory(2nd ed.).McGraw-Hill, New York.

Palay, T. (1985). ‘Avoiding regulatory constraints: Con-tractual safeguards and the role of informal agree-ments’, Journal of Law, Economics and Organiza-tions, 1, pp. 155–175.

Parkhe, A. (1993). ‘Strategic alliance structuring: Agame theoretic and transaction cost examination of


interfirm cooperation’, Academy of ManagementJournal, 36(4), pp. 794–829.

Piore, M. J. and C. F. Sabel (1984).The SecondIndustrial Divide. Basic Books, New York.

Pisano, G. P. (1989). ‘Using equity participation tosupport exchange: Evidence from the biotechnologyindustry’, Journal of Law, Economics, and Organi-zation, 5, pp. 109–126.

Pisano, G. P., M. V. Russo and D. Teece (1988). ‘Jointventures and collaborative agreements in the telecom-munications equipment industry’. In D. Mowery (ed.),International Collaborative Ventures in US Manufac-turing. Ballinger, Cambridge, MA, pp. 23–70.

Powell, W. (1987). ‘Hybrid organizational arrange-ments: New form or transitional development’,Cali-fornia Management Review, 30, pp. 67–87.

Prahalad C. K. and G. Hamel (1990). The core com-petence of the corporation,’Harvard BusinessReview, 68(3), pp. 79–91.

Reich, R. B. and E. D. Mankin (1986). ‘Joint ventureswith Japan give away our future’,Harvard BusinessReview, 64(2), pp. 78–86.

Ring, P. S. and A. Van de Ven (1989). ‘Formal andinformal dimensions of transactions’. In A. Van deVen, H. Angel, and M. S. Poole (eds.),Research onthe Management of Innovation: The Minnesota Studies.Ballinger/Harper Row, New York, pp. 171–192.

Ring, P. S. and A. Van de Ven (1992). ‘Structuringcooperative relationships between organizations’,Strategic Management Journal, 13(7), pp. 483–498.

Sabel, C. (1993). ‘Studied trust: Building new form ofcooperation in a volatile economy’,HumanRelations, 46(9), pp. 1133–1170.

Sako, M. (1991). ‘The role of trust in Japanese buyer–supplier relationships’,Ricerche Economiche, 45,pp. 449–474.

Simonin, B. L. (1997). ‘The importance of collaborativeknow-how: An empirical test of the learning organi-zation’, Academy of Management Journal, 40(5),pp. 1150–1174.

Szulanski, G. (1996). ‘Exploring internal stickiness:Impediments to the transfer of best practice withinthe firm’, Strategic Management Journal, WinterSpecial Issue,17, pp. 27–43.

Tucchi, C. L. (1996). ‘Firm heterogeneity and perform-ance of international strategic technology alliances’,working paper, MIT Sloan School of Management.

von Hippel, E. (1988).The Sources of Innovation.Oxford University Press, New York.

Williamson, O. E. (1985).The Economic Institutionsof Capitalism. Free Press, New York.

Williamson, O. E. (1991). ‘Comparative economicorganization: The analysis of discrete structural alter-natives’, Administrative Science Quarterly, 36,pp. 269–296.

Yoshino, M. and U. S. Rangan (1995).StrategicAlliances: An Entrepreneurial Approach to Globali-zation. Harvard Business School Press, Boston, MA.

Zaheer, A., B. McEvily and V. Perrone (1998). ‘Doestrust matter? Exploring the effects of interorgani-zational and interpersonal trust on performance’,Organization Science, 9, pp. 1–20.


APPENDIX: List of items used to measure each theoretical construct

Note: Respondents used a 7-point Likert scale to provide responses on each item, such that‘1 = Strongly Disagree and 7= Strongly Agree’

Reference source

(a) Independent variablesRelational capital (RC)1. There is close, personal interaction between the partners at multiple levels Dyer and Singh (1998);2. The alliance is characterized by mutual respect between the partners at Madhok (1995); Dyer

multiple levels (1996); Gulati (1995);3. The alliance is characterized by mutual trust between the partners at multiple Inkpen (1994); Badaracco

levels (1991); Mohr and Spekman4. The alliance is characterized by personal friendship between the partners at (1994)

mulitple levels5. The alliance is characterized by high reciprocity among the partners

Conflict management (CM)1. An explicit mechanism has been established and used to address or resolve Parkhe (1993); MacNeil

conflicts (1981); Mohr and Spekman2. Managerial interaction between partners is closely monitored for identifying (1994)

potential conflicts3. There is strong two-way communication while resolving conflicts4. Great emphasis is placed on dealing with cultural obstatcles while resolving

conflicts5. The partners engage in joint problem solving while resolving conflicts6. Top management from both sides is involved in resolving conflicts

(B) ControlsPartner fit: Complementarity and compatibility (PP)1. There is high Complementarity between the resources/capabilities of the two Beamish (1987); Harrigan

partners (1988b); Tucchi (1996);2. There is high similarity/overlap between the core capabilities of each partner Geringer (1988); Parkhe3. The organizational cultures of the two partners are compatible with each (1993); Dyer and Singh

other (1998); De la Sierra (1995)4. The management and operating styles of the partners are compatible with

each other

Other controls (these three items were not measured with a Likert scale)1. What was the structure of this alliance (choose one): equity or nonequity?2. Did the partners have other alliances between them, prior to this relationship

(choose one): Yes/No?3. Do the partners belong to the same nationality (choose one): Yes/No?

(C) Dependent variablesLearning (L)1. Your company learnt or acquired some new or important information from

the partner2. Your company learnt or acquired some critical capability or skill from the

partner3. This alliance has helped your company to enhance its existing

capabilities/skills

Protection of proprietary assets (PC)1. Your company has been able to protect its core capabilities or skills from the

partner2. Your company has been successful in protecting its crown jewels from being

appropriated by the partner


learning and protection of proprietary assets in strategic alliances: building relational capital

Documents