ALLIANCE PORTFOLIOS AND FIRM PERFORMANCE: A ......Alliance Portfolios and Firm Performance 1189 endowed partners may in fact undermine the mar-ket performance of ﬁrms. By distinguishing

Strategic Management JournalStrat. Mgmt. J., 28: 1187–1212 (2007)

Published online 24 July 2007 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.637

Received 29 April 2004; Final revision received 6 June 2007

ALLIANCE PORTFOLIOS AND FIRM PERFORMANCE:A STUDY OF VALUE CREATION ANDAPPROPRIATION IN THE U.S. SOFTWARE INDUSTRY

DOVEV LAVIE*Faculty of Industrial Engineering and Management, Technion—Israel Institute ofTechnology, Haifa, Israel; McCombs School of Business, University of Texas atAustin, Austin, Texas U.S.A.

This study reveals the multifaceted contribution of alliance portfolios to firms’ market perfor-mance. Extending prior research that has stressed the value-creation effect of network resources,it uncovers how prominent partners may undermine a firm’s capacity to appropriate value fromits alliance portfolio. Analysis of a comprehensive panel dataset of 367 software firms and their20,779 alliances suggests that the contribution of network resources to value creation varies withthe complementarity of those resources. Furthermore, the relative bargaining power of partnersin the alliance portfolio constrains the firm’s appropriation capacity, especially when manyof these partners compete in the focal firm’s industry. In turn, the firm’s market performanceimproves with the intensity of competition among partners in its alliance portfolio. These findingsadvance network research by highlighting the trade-offs that alliance portfolios impose on firmsthat seek to manage and leverage their alliances. Copyright 2007 John Wiley & Sons, Ltd.

INTRODUCTION

How does an alliance portfolio contribute to firmperformance? Despite the recent surge in thenumber of interfirm alliances (henceforth termedalliances), the accumulated alliance research offersonly limited insights into this phenomenon. Priorstudies either investigate the performance impli-cations of individual alliances or concentrate onrelational and structural properties of alliance net-works. More recently, scholars studying allianceportfolios have begun to recognize the role ofpartners’ resources in value creation. Neverthe-less, they have tended to neglect the appropriationhazards that partners impose. The current study

Keywords: alliance; portfolio; network; performance;value creation; value appropriation*Correspondence to: Dovev Lavie, Faculty of Industrial Engi-neering and Management, Technion—Israel Institute of Tech-nology, Haifa 32000, Israel. E-mail: [email protected]

bridges these gaps in the literature by simulta-neously incorporating value creation and appro-priation mechanisms and taking into account theresources and competitive positions of partners inthe alliance portfolio. Thus, it sheds light on themeans by which an alliance portfolio can influencefirm performance.

An alliance is a voluntary arrangement amongindependent firms to exchange or share resourcesand engage in the co-development or provision ofproducts, services, or technologies (Gulati, 1998).Alliances take different forms, including joint ven-tures, collaborative R&D, and joint marketing.Traditionally, alliances had been conceived ofas ad hoc arrangements serving specific needs,but more recently firms have begun to engageextensively in multiple simultaneous alliances.For example, in the U.S. software industry, thepercentage of publicly traded firms that engagein alliances has increased from 32 percent to 95percent, and the average number of alliances per

Copyright 2007 John Wiley & Sons, Ltd.

1188 D. Lavie

0

5

10

15

20

25

30

35

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Year

Num

ber

of A

llian

ces

per

Firm

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Perc

enta

ge o

f Fi

rms

Eng

agin

g in

Alli

ance

s

Average Portfolio Size % Firms Engaging in Alliances

Figure 1. The evolution of alliance portfolios in the U.S. software industry, 1990–2001.This figure presents the percentage of publicly traded firms in the U.S. software industry that engage in

alliances and their corresponding alliance portfolio size, measured in terms of number of alliances

firm rose from four to more than 30 during the1990s (see Figure 1). Not only has the numberof alliances increased, but their scope has alsobeen extended. Whereas firms historically formedalliances to perform relatively simple peripheralactivities, alliances are now employed at variousstages of the value chain (Powell, Koput, andSmith-Doerr, 1996). An alliance portfolio refers toa firm’s collection of direct alliances with partners.It is akin to the notion of the egocentric network,which encompasses the focal firm (ego), its set ofpartners (alters), and their connecting ties (Wasser-man and Faust, 1994). The challenge of coordinat-ing multiple simultaneous alliances has promptedfirms to establish dedicated alliance functions andformalize their alliance programs (Kale, Dyer, andSingh, 2002). However, scholars and practition-ers still debate the desirable properties of allianceportfolios.

Considering the potential advantages thatalliances may provide, such as cost sharing, riskreduction, and flexibility (Hagedoorn, 1993; Har-rigan, 1988; Kogut and Kulatilaka, 1993; Ohmae,1989; Powell et al., 1996), practitioners and schol-ars often assume that alliance portfolios willenhance corporate performance. Surprisingly, thisassumption received only limited support in earlyresearch. A multi-industry study by Berg, Duncan,and Friedman (1982) that examined the impact ofa firm’s joint venturing activity on its profitability

found short-term negative effects in the chemi-cals and mechanical engineering industries, but nolong-term performance effects. Similarly, Hage-doorn and Schakenraad (1994) found no directeffect of technology alliance portfolios on firmprofitability.

The limited evidence on the contribution ofalliance portfolios to firm performance is under-standable given the empirical challenges inherentin measuring such a complex phenomenon (Gulati,1998). For example, firms rarely report formalquantitative measures of alliance operations andoutcomes in their financial statements. In addi-tion, the performance contribution of alliances isoften confounded with that of the firm’s inter-nal operations. Recent research has made someprogress toward overcoming these challenges byshifting from the study of dyadic alliances to theexamination of network structures, the nature ofalliance relationships, and the attributes of part-ner firms (Baum, Calabrese, and Silverman, 2000;Rothaermel, 2001; Stuart, 2000; Stuart, Hoang, andHybels, 1999; Zaheer and Bell, 2005). This worksuggests that ‘large firms and those that possessleading-edge technological resources are positedto be the most valuable associates’ (Stuart, 2000:791).

In the current study, I qualify this assertion bynoting that although some complementary networkresources create value, alliances with well-

Copyright 2007 John Wiley & Sons, Ltd. Strat. Mgmt. J., 28: 1187–1212 (2007)DOI: 10.1002/smj

Alliance Portfolios and Firm Performance 1189

endowed partners may in fact undermine the mar-ket performance of firms. By distinguishing valuecreation from value appropriation mechanisms,this study elucidates why the nurturing of allianceportfolios rich in prominent partners serves asa double-edged sword. It sheds further light onthe implications of alliance portfolios by takinginto account the nature of relationships betweenthe focal firm and its partners, as well as amongpartners in its alliance portfolio. These ideas aretested empirically with a pooled time-series anal-ysis of the alliance portfolios of 367 U.S. soft-ware firms during the period 1990–2001. The find-ings confirm that enhancements of a firm’s marketvalue are related to the availability of both mar-keting and financial network resources, as wellas to the prominence of partners in its allianceportfolio, but not to the availability of technol-ogy and human network resources. To the extent,however, that partners in a firm’s alliance portfo-lio are more profitable and gain access to morealliances than the focal firm, the firm’s marketvalue will decline. Finally, bilateral competitionbetween the firm and its partners exacerbates someof these negative effects, while multilateral com-petition among partners in the alliance portfolioattenuates them. Thus, this study advances allianceportfolio research by juxtaposing value creationand appropriation mechanisms and accounting forinterdependencies across alliances in the firm’salliance portfolio.

LITERATURE REVIEW

Alliance research has evolved in four streams:(1) the strategic alliance literature; (2) studies ofstock market returns following alliance announce-ments; (3) social network theory applications; and(4) the emerging research on strategic networks.These research streams enhance our understandingof alliances and networks but fall short of fullyaccounting for the contribution of alliance portfo-lios to firm performance.1

1 Since my focus is limited to relevant implications for firmperformance, I do not review the general alliance literature,except as it relates to performance implications. More thoroughreviews of this literature can be found elsewhere (Gulati, 1998;Koza and Lewin, 1998; Osborn and Hagedoorn, 1997).

When considering performance implications, adistinction should be made between alliance per-formance and the impact of alliances on firm per-formance. Alliance performance has been associ-ated with interfirm trust, strategic and organiza-tional compatibility, knowledge exchange, adap-tive governance, conflict resolution mechanisms,and dedicated alliance personnel (Dussauge, Gar-rette, and Mitchell, 2000; Dyer and Nobeoka,2000; Dyer and Singh, 1998; Gulati, 1995a; Inkpenand Beamish, 1997; Kale et al., 2002; Kale, Singh,and Perlmutter, 2000; Khanna, Gulati, and Nohria,1998; Kumar and Nti, 1998; Madhok and Tall-man, 1998; Parkhe, 1993; Saxton, 1997; Zaheer,McEvily, and Perrone, 1998). Hence, the strate-gic alliance literature has offered a rich perspec-tive on why some alliances are more successfulthan others. However, this perspective may notfully explain the impact of alliances on firm per-formance, for several reasons. First, the analysishas often been conducted at the dyadic level, witheach alliance considered as an isolated event ratherthan as an interdependent element in a portfolioof alliances. Second, there is no immediate wayto infer alliance contributions to firm performancefrom alliance success because of asymmetry inthe value appropriated by the partners involved inthe alliance (Khanna et al., 1998). Finally, mostalliance studies have relied on alliance stabil-ity, alliance longevity, or qualitative managerialassessments, rather than on financial measures ofcorporate performance.

Originating in the finance discipline, researchon abnormal stock market returns overcomes thelast limitation by studying residual changes infirms’ stock prices during various event win-dows surrounding their alliance announcements(McConnell and Nantell, 1985). This researchstream has provided solid evidence regarding theexpected performance implications of individualalliances. According to this research, firms typ-ically enjoy significant positive abnormal stockmarket returns following alliance announcements.The heterogeneity of market returns has beenascribed, for instance, to the type of alliance, therelative size of partners, their partnering expe-rience, and the relatedness of their businesses(Anand and Khanna, 2000a; Balakrishnan andKoza, 1993; Beck, Larson, and Pinegar, 1996;Chan et al., 1997; Das, Sen, and Sengupta, 1998;Koh and Venkatraman, 1991; Lee and Wyatt,1990; McConnell and Nantell, 1985; Merchant and


1190 D. Lavie

Schendel, 2000; Park and Kim, 1997; Reuer andKoza, 2000; Woolridge and Snow, 1990). Somestudies have found significant correlation betweenthese abnormal stock market returns and subjectivemanagerial assessments of alliance performance(Kale et al., 2002; Koh and Venkatraman, 1991).Nevertheless, this research offers only limited evi-dence of the performance implications of allianceportfolios, because the majority of findings referto joint ventures rather than to the more com-mon non-equity alliances. More importantly, theanalysis of residual market returns considers onlythe marginal contribution of individual alliancesand neglects the simultaneous and interdependenteffects of other alliances in the focal firm’s portfo-lio. Finally, whereas these studies typically reportsignificant positive market returns, such returns arerelatively small and are recorded for only half ofthe alliance participants, thus leaving considerableunexplained heterogeneity in the contribution ofalliance portfolios to a given firm’s market perfor-mance.

The social network literature, in turn, overcomesthe alliance independence assumption by focusingon enduring patterns of relationships among inter-acting social actors. Social network perspectivesconceive of actors (such as firms) as interdepen-dent, and place emphasis on the social, economic,or political network structures of ties that provideactors with opportunities and constraints. Actorscan enjoy certain benefits, referred to as socialcapital, by virtue of their membership in social net-works (Portes, 1998). These benefits derive fromclosed network structures that bond actors (Cole-man, 1988) or from brokerage positions that bridgedisconnected actors (Burt, 1992). Social capitalmay contribute to firm performance by enhancinginnovation (Ahuja, 2000; Tsai and Ghoshal, 1998),knowledge transfer (Burt, 1992; Inkpen and Tsang,2005; Koka and Prescott, 2002), intellectual cap-ital (Nahapiet and Ghoshal, 1998), and efficiency(Baker, 1990; Burt, 2000). In addition, social net-work studies emphasize either structural or rela-tional aspects of networks (Granovetter, 1985).Structural embeddedness studies highlight actors’positions in the overall network structure by con-sidering properties such as structural holes (Burt,1992), centrality (Bonacich, 1987; Freeman, 1979;Ibarra, 1993; Podolny, 1993), structural equiva-lence (Burt, 1987), and density (Coleman, 1988).Relational embeddedness studies, in turn, examinethe nature of dyadic relationships, considering, for

instance, the strength of ties and the evolving trustbetween actors (Granovetter, 1973; Podolny, 1994;Powell, 1990; Uzzi, 1996). Nevertheless, the socialnetwork literature offers only a partial account ofthe performance implications of alliance portfoliosbecause it focuses on network ties while assum-ing away differences in the inherent attributes ofactors. In addition, it partially relies on an inter-personal reasoning involving emotional intensityand intimacy, which is not applicable for inter-firm alliances (Rowley, Behrens, and Krackhardt,2000).

Recently, scholars have begun to apply socialnetwork theories in studies of the performanceimplications of strategic networks (Gulati, Nohria,and Zaheer, 2000). Several studies have exam-ined how the number of alliances, and networkproperties such as density and structural holes,affect firms’ innovation output, new product devel-opment, revenue growth, market share, marketvalue, or profitability (Ahuja, 2000; Baum et al.,2000; Rothaermel, 2001; Rowley et al., 2000; Stu-art et al., 1999; Zaheer and Bell, 2005). Some stud-ies have considered the composition of partnersin the network (Baum et al., 2000; Goerzen andBeamish, 2005), their innovativeness, and promi-nence (Gulati and Higgins, 2003; Rothaermel,2001; Saxton, 1997; Stuart, 2000; Stuart et al.,1999; Zaheer and Bell, 2005). In particular, Saxton(1997) found that firms benefit from their partners’reputation. Stuart et al. (1999) found that the tech-nological and commercial prominence of partnersaffect the IPO performance of startup firms. Simi-larly, Gulati and Higgins (2003) demonstrated thatrelationships with prestigious venture capital firmsand underwriters contribute to IPO performance.Rothaermel (2001) showed that biopharmaceuticalfirms facilitate new product development by lever-aging their partners’ assets. Finally, Stuart (2000)found that the technological innovativeness of part-ners contributes to the sales growth and innova-tion rates of semiconductor firms. This stream ofresearch deviates from traditional social networkanalysis by focusing on the attributes of partners,rather than on the quality of ties or the propertiesof the network structure. It adheres to the notionthat alliance portfolios provide the focal firm withaccess to network resources (Gulati, 1999) thatextend its opportunity set and potentially enhanceits performance.

Although this research advances an understand-ing of the role of alliance portfolios, the empirical



evidence on the contribution of partners’ resourceshas been generally limited to startup firms andintermediate outcomes, such as innovation outputand revenue growth. Moreover, this research hasconcentrated on the contributions of partners tovalue creation, concluding that firms would benefitfrom alliances with prominent and well-endowedpartners that endorse or otherwise support the focalfirm. For example, Stuart concluded that ‘bothfrom a resource access and reputation standpoint,large and innovative firms are likely to be the mostvaluable associates’ (Stuart, 2000: 808). The cur-rent study cautions, however, that firms that followthis advice may suffer a decline in market perfor-mance insofar as dominant partners that contributeto joint value creation may also be well positionedto appropriate a larger share of this value at thefocal firm’s expense. It extends recent research thathas suggested that the costs of forming allianceswith critical resource-rich partners may outweighthe benefits of this partnering strategy (Bae andGargiulo, 2004) by distinguishing between the net-work resources that the alliance portfolio providesand the bargaining power of partners in that portfo-lio. By simultaneously considering value creationand appropriation mechanisms, this study offers amore nuanced account of the impact of allianceportfolios on firms’ market performance.

THEORY AND HYPOTHESES

To untangle the impact of alliance portfolios onfirms’ market performance, this study movesbeyond prior research that has emphasized rela-tional or structural aspects of network ties, andunderscores instead the contribution of networkresources (Gulati, 2007; Lavie, 2006), which moredirectly captures the characteristics and endow-ments of partners in the firm’s alliance port-folio. The emphasis shifts from the ties thatserve as channels for transferring resources amongfirms to the resources themselves. In addition, thisstudy concurrently incorporates aspects of valuecreation and appropriation mechanisms. Value-creation mechanisms enhance the focal firm’s abil-ity to generate value from its relationships withpartners as they collectively pursue shared objec-tives and extend the range of value chain activ-ities that contribute to the overall value of thealliance portfolio. These mechanisms produce rela-tional rents that cannot be generated independently

by individual participants in alliances (Dyer andSingh, 1998). In turn, value appropriation mecha-nisms do not create new value but instead deter-mine the relative share of relational rents thatthe focal firm can appropriate (Gulati and Wang,2003; Hamel, 1991; Khanna et al., 1998). Valueappropriation mechanisms often elicit co-opetition(Brandenburger and Nalebuff, 1996), in whichpartners competitively pursue self-interested objec-tives in an attempt to increase their share of appro-priated relational rent. The disparity between valuecreation and value appropriation is akin to thedistinction between common and private benefits(Khanna et al., 1998). Value-creation mechanismsare collective processes that generate common ben-efits that are shared by all partners in an alliance,whereas value appropriation mechanisms deter-mine the distribution of these common benefitsto individual partners as well as the capacity ofpartners to unilaterally extract private benefits thatare unavailable to other partners. The combinationof value creation and appropriation mechanismsaccounts for the contribution of the alliance port-folio to firm performance.

Value creation in alliance portfolios

The resource-based view (Barney, 1991; Penrose,1959; Peteraf, 1993; Wernerfelt, 1984) has tradi-tionally analyzed the value-creating contributionsof a firm’s internal resources and capabilities.When the firm becomes embedded in a networkof alliances, a distinction can be made betweenthe internal resources that it owns or controls(Barney, 1991), and the network resources thatare owned by its partners yet can be potentiallyaccessed by the focal firm through its ties to thesepartners (Gnyawali and Madhavan, 2001; Gulati,1999; Lavie, 2006). Network resources encompasspartners’ tangible and intangible assets, includingtheir human resources, financial assets, market-ing efforts, R&D investments, and reputation. Afirm may enhance its performance by leveragingthese resources through three value-creation mech-anisms.

First, a focal firm can use network resourcesto directly extend and enrich its value-creationopportunities. In particular, alliance partners canprovide immediate access to complementary assetsthat support the commercialization of the focalfirm’s products or enhance its service offerings(Koh and Venkatraman, 1991; Mowery, Oxley,


1192 D. Lavie

and Silverman, 1998; Pisano, 1990; Rothaermel,2001). The firm can rely on its partners’ sales-people as a secondary marketing channel, gainaccess to new customers by leveraging its partners’marketing investments, facilitate product introduc-tions by incorporating partners’ technologies, oruse partners’ funding to support its own R&Dactivities. In this sense, network resources con-tribute directly to firm performance by supplement-ing the firm’s resources with resources that may beotherwise unavailable internally, and by makingavailable a wider range of strategic opportunities(Barney, 1991). Although the firm may invest ininternal development of resources or acquire exter-nal resources in the factor market, the assump-tion is that some network resources are special-ized (Teece, 1986) and, because of their scarcityand the presence of isolating mechanisms suchas causal ambiguity (Barney, 1991; Dierickx andCool, 1989), more costly to develop internally oracquire. For instance, the firm may flexibly targetspecific resources of partners instead of engagingin corporate acquisitions and paying a premiumto own redundant resources that do not directlycontribute to value creation.

Second, a focal firm can generate value fromresource combinations. According to the relationalview (Dyer and Singh, 1998), the firm can gen-erate synergies by combining network resourceswith its internal resources. The bundles of net-work resources and the firm’s internal resourcesmay be richer relative to bundles comprising inter-nal resources exclusively. Besides the increase inthe range of available resource combinations, thefocal firm enjoys the flexibility of drawing networkresources as needed while limiting its investmentsto the maintenance of relationships. Moreover,the firm may create value by combining networkresources of distinct partners, and thus enjoy syn-ergies that are unavailable to individual partners inits alliance portfolio. For example, a firm that spe-cializes in systems integration may combine thehardware platforms of one partner with the soft-ware development expertise of another partner inthe course of system implementation projects.

Finally, the focal firm can indirectly benefitfrom network resources that enhance the value ofits internal resources or provide it with oppor-tunities to internalize external resources (Lavie,2006). In particular, resource-rich partners offerintangible assets that enhance the firm’s legiti-macy and capacity to acquire additional resources

(Weigelt and Camerer, 1988). The alliance portfo-lio can also facilitate the accumulation of exter-nal knowledge and new skills (Balakrishnan andKoza, 1993; Kogut, 2000) through imitation, learn-ing, and acquisition of network resources (Hamel,1991; Kale et al., 2000). Thus, the focal firm canbenefit from network resources that enrich its inter-nal set of resources or enable it to develop newresources and capabilities. This is illustrated, forexample, by the demonstrated link between startupfirms’ revenue growth and innovation rates and theinnovativeness and prominence of partners in theiralliance portfolios (Stuart, 2000).

In sum, alliances with well-endowed partnersthat can endorse the firm and provide technologi-cal, financial, marketing, and human resources areexpected to contribute to value creation, and thusenhance the market performance of the focal firm.The richer the firm’s alliance portfolio in valuablenetwork resources, the better its expected perfor-mance will be.

Hypothesis 1: The focal firm’s market perfor-mance will be positively associated with thenetwork resources possessed by partners in itsalliance portfolio.

Value appropriation in alliance portfolios

The proliferating research on strategic networkshas focused almost exclusively on the value-creation effects of alliances while overlookingvalue appropriation considerations. Scholars havelong acknowledged that interfirm collaborationcreates value by generating relational rents (Dyerand Singh, 1998), but only recently have they con-sidered how these relational rents are appropriatedby individual alliance partners (Dyer, Kale, andSingh, forthcoming; Lavie, 2006). Prior researchthat analyzed the asymmetric distribution of com-mon and private benefits in alliances (Khannaet al., 1998) has underscored the incentives thatsuch benefits provide for continued collaboration,but paid less attention to the antecedents of valueappropriation. Similarly, empirical research thatreveals asymmetry in the market returns of part-ners following alliance announcements has beenconducted at the dyad level only (e.g., Gulati andWang, 2003), and thus falls short of fully explain-ing value appropriation in alliance portfolios. Thecapacity of the focal firm to appropriate value



from its alliance portfolio depends not only on theattributes of its partners, but also on the nature ofthe firm’s relationships with these partners and thecharacteristics of the alliance portfolio as a whole.A firm whose alliance portfolio comprises a setof powerful partners may be unable to appropri-ate the benefits that its alliances produce (Bae andGargiulo, 2004). The paradox is that prominentpartners that stand to make a significant contri-bution to joint value creation may also be ableto restrict the appropriation capacity of the focalfirm, and consequently undermine its market per-formance.

The appropriation capacity of the focal firmdepends in part on its relative bargaining powervis-a-vis partners in its alliance portfolio. Bargain-ing power is defined as the ability to favorablychange the terms of agreements, to obtain accom-modations from partners, and to influence theoutcomes of negotiations (Yan and Gray, 1994).Because of the quasi-formal nature of alliancesand the inherent incompleteness of alliance agree-ments, bargaining power matters not only at thealliance formation stage but also throughout thealliance life cycle, when market conditions changeor underspecified agreement terms need to be rene-gotiated. Asymmetric rent appropriation can there-fore be ascribed to partners’ attempts to lever-age their bargaining power vis-a-vis the focal firmand to extract a disproportional share of relationalrents. The relative bargaining power of partnersthus affects the distribution of rents in alliances(Hamel, 1991) and can reduce the focal firm’sshare of common benefits relative to its sharein joint investments (Khanna et al., 1998). Analliance portfolio composed of partners with strongrelative bargaining power increases the focal firm’sinvestments while reducing its share of relationalrents, thus negatively affecting its market perfor-mance.

Bargaining power derives from the interdepen-dence between the focal firm and its partners(Pfeffer and Salancik, 1978). According to bar-gaining power theory (Bacharach and Lawler,1984), bargaining power is based on two dimen-sions of interdependence: the stakes of the par-ties involved in the negotiation; and the availabil-ity of alternatives. In alliances, relative bargain-ing power derives from partners’ stakes in thealliance portfolio and the availability of alternativealliances.

First, to the extent that the outcomes of alliancesare more critical to the performance of the focalfirm than to the performance of its partners, thepartners enjoy strong bargaining positions. Whenthe focal firm has a weaker bargaining positionrelative to partners in its alliance portfolio, thisincreases the stakes for the focal firm, whichbecomes more dependent on its alliances and theiroutcomes (Yan and Gray, 1994). According toPorter, ‘the bargaining relationship with buyersand suppliers is reflected in both the configura-tion of a firm’s value chain and how margins aredivided with buyers, suppliers, and coalition part-ners’ (Porter, 1985: 58). Consequently, the relativebargaining position of partners is manifested intheir profit margins. Hence, high-margin partnerscan leverage their flexibility to negotiate favorableterms with the focal firm and attenuate its appro-priation capacity, leading to negative performanceimplications for the focal firm.

Hypothesis 2a: The focal firm’s market perfor-mance will be negatively associated with therelative profitability of partners in its allianceportfolio.

Second, the relative bargaining power of part-ners in the alliance portfolio derives from the setof alternatives available to them. To the extent thatpartners enjoy better access to alternative alliancesthan the focal firm does, they can obtain greaterbenefits outside the scope of their joint allianceswith the firm. The option of reaching alternativearrangements for pursuing similar objectives withother firms improves partners’ positions in negoti-ations with the focal firm. According to Yan andGray (1994), the lower switching costs of suchpartners reduce their dependence on the firm andenhance their relative bargaining power vis-a-visthe firm. This idea is akin to Burt’s (1983) notionof structural autonomy, wherein firms that faceintense competition in their industry and transactwith partners that operate in more concentratedindustries are subject to constraints when pursuingtheir interests. This logic, while originally referringto the dynamics of bargaining among exchangepartners, may also apply in the case of alliances,where the relative availability of alternatives deter-mines the extent to which the firm depends onits partners. Hence, a firm with fewer alternativealliances relative to partners in its alliance portfo-lio enjoys less bargaining power and consequently


1194 D. Lavie

weaker appropriation capacity, resulting in nega-tive performance implications.

Hypothesis 2b: The focal firm’s market perfor-mance will be negatively associated with therelative availability of alternative alliances topartners in its alliance portfolio.

The focal firm’s appropriation capacity isexpected to diminish with partners’ attempts toleverage their relative bargaining power in nego-tiations with the firm. Hence, the negative perfor-mance implications of these attempts depend notonly on the extent to which partners enjoy superiorbargaining positions but also on their motivationto increase their relative share of relational rents.The motivation of partners to compete away thefocal firm’s share of relational rents derives fromthe level of bilateral competition in the allianceportfolio, that is, the extent to which partners canbe considered competitors of the focal firm. Analliance portfolio featuring a high proportion of thefirm’s competitors is characterized by opportunis-tic and co-opetitive behavior (Brandenburger andNalebuff, 1996), disputes, and considerable risk ofundesirable leakage of resources, which may leadto spillover rents (Lavie, 2006). Bilateral competi-tion encourages partners to increase their residualclaims for relational rents instead of focusing oncollaborative rent generation. Accordingly, priorresearch notes that competition between the focalfirm and its partners increases the ratio of uni-lateral private benefits to collaborative commonbenefits (Khanna et al., 1998) and prompts part-ners to internalize the focal firm’s intangible assetsand improve their competitive positions vis-a-visthe firm (Hamel, 1991). Thus, bilateral competi-tion stimulates partners’ attempts to compete awaythe focal firm’s share of relational rents.

Nevertheless, studies that have analyzed theperformance implications of bilateral competitionreport mixed findings. Some note that bilateralcompetition increases alliance failure rates (Parkand Russo, 1996) and reduces partners’ revenues(Baum et al., 2000) and market returns (Balakr-ishnan and Koza, 1993; Chang and Chen, 2002),while others report positive effects on marketreturns (Koh and Venkatraman, 1991; Park andKim, 1997). This perplexity can be resolved byacknowledging that bilateral competition providesthe motivation, but not the capacity, for dispro-portional rent appropriation. An increase in the

likelihood of disproportional rent appropriation inalliances does not reveal which party has the upperhand, and thus can extract a larger share of rela-tional rent. This notion follows the fundamen-tal psychology of relations (Heider, 1958), whichrecognizes that in addition to motivation (‘try’),power and ability (‘can’) are required for action.Hence, even when partners in the alliance port-folio have incentives to compete away the focalfirm’s rents, they may lack the capacity to do so.If, however, they enjoy a strong relative bargain-ing power by virtue of their relative profitabilityor available alternatives, bilateral competition mayimpel them to leverage their bargaining positionsand attenuate the firm’s appropriation capacity.Alliance partners may have an inherent inclinationto maximize their payoffs irrespective of bilateralcompetition (Khanna et al., 1998), but bilateralcompetition induces the motivation to restrict thefirm’s share of relational rents because it impelspartners to view the bargaining situation as a zero-sum game rather than a positive-sum game. Thus,bilateral competition intensifies the negative per-formance implications of relative profitability andrelative availability of alternatives to partners inthe alliance portfolio.

Hypothesis 3a: The negative associationbetween the focal firm’s market performanceand the relative profitability of partners in itsalliance portfolio will intensify with increases inthe level of bilateral competition.

Hypothesis 3b: The negative associationbetween the focal firm’s market performanceand the relative availability of alternativealliances to partners in its alliance portfolio willintensify with increases in the level of bilateralcompetition.

Notwithstanding the above, competition in thealliance portfolio can enhance the focal firm’sappropriation capacity insofar as it emerges amongpartners rather than between the firm and its part-ners. This is illustrated in the following statementof an alliance manager whom I interviewed:

I am working with IBM, BEA and Microsoft. Eachone of them has a lead product in this interactivedevelopment environment. All compete head-to-head with each other, and we work with all threeof them as strategic partners. This brings greatervalue to us.



The extent to which partners in the alliance port-folio are considered direct competitors of eachother can be referred to as multilateral competi-tion.2 Multilateral competition evolves when thefocal firm allies with multiple partners that oper-ate in the same product markets or offer similarservices, and thus play a similar role in its allianceportfolio. Prior research has indicated that analliance portfolio that provides access to redundantpartners is inefficient and may therefore limit thefocal firm’s growth prospects (Baum et al., 2000).This approach, however, is strictly concerned withvalue creation. While multilateral competition mayintroduce redundancies to the alliance portfolio,it can also enhance the firm’s value appropria-tion capacity by producing brokerage benefits andattenuating opportunistic behavior of partners.

Specifically, multilateral competition leads part-ners to compete for the focal firm’s business,resources, and attention, enabling the firm to arbi-trate among partners and control the flow of assetsand information in its alliance portfolio. This issomewhat akin to Burt’s (1992) notion of broker-age benefits, except that the broker in this casedoes not foster mediated transactions between dis-connected partners, but instead can play partnersagainst each other. Multilateral competition allowsthe focal firm discretion in allocating resources tocompeting alliances and in choosing partners forpursuing particular collaborative initiatives. Theflexibility to juggle among competing partnersenables the firm to exercise control and focus onthose alliances that provide more favorable terms,and consequently increase its relative share of rela-tional rents.

Additionally, multilateral competition reducesthe likelihood that partners will behave oppor-tunistically. Prior research has shown that oppor-tunistic behavior by partners negatively affectsalliance performance (Parkhe, 1993). However,the presence of multiple competing partners in

2 The notion of multilateral competition differs from relativebargaining power. Relative bargaining power derives from thedyadic bargaining situation between the focal firm and each ofits partners. In contrast, multilateral competition does not entaildirect negotiations with the firm and is conceptualized insteadbased on the similarity in industry focus of partners, irrespectiveof the firm’s position. In addition, multilateral competition differsfrom the notion of structural equivalence (Burt, 1976), whichrefers to similarity in partners’ market transactions with actors inother industry sectors. Whereas structural equivalence considerswhom partners are tied to, multilateral competition takes intoaccount the industries in which partners compete, regardless oftheir ties to other partners.

the focal firm’s alliance portfolio establishes alarge-numbers exchange condition that attenuatesthe hazards of opportunistic behavior (Williamson,1975). This condition ensures that partners’ oppor-tunistic behavior would not pay off, since the focalfirm’s other partners can offer similar benefits,thus creating viable alliance exit options for thefirm. Taken together, the brokerage benefits and thefavorable exchange conditions entailed by multi-lateral competition enable the focal firm to appro-priate a larger share of relational rents from itsalliance portfolio. Hence, at higher levels of mul-tilateral competition in the alliance portfolio, thefocal firm enjoys a stronger appropriation capacitythat may contribute to its market performance.

Hypothesis 4: The focal firm’s market perfor-mance will be positively associated with thelevel of multilateral competition among partnersin its alliance portfolio.

Nevertheless, the firm’s favorable position inan alliance portfolio rich in competing partnersdoes not necessarily warrant that the firm willleverage this competitive situation. For example,a vice president of corporate alliances, whom Iinterviewed, noted:

For us to position ourselves to play multiple part-ners against each other where we are not the marketleader would cause unrecallable damage to our rep-utation as a partner. We cannot afford to do that justto win tactical engagements.

Hence the firm’s tendency to leverage multi-lateral competition depends, at least in part, onits status and relative bargaining power. In thesame vein, the negative performance implicationsof partners’ relative bargaining power can be atten-uated by leveraging multilateral competition.

As previously noted, to the extent that partnersare more profitable than the focal firm, they areless dependent on their joint alliances with thatfirm (Yan and Gray, 1994). Relative profitabilityprovides partners with favorable bargaining posi-tions in negotiations with the firm, yet when thefirm allies with multiple partners in the same indus-try the competition among these partners increasestheir dependence on their alliances with the focalfirm. Thus multilateral competition can attenuatethe negative performance implications of the firm’sdependence on relatively profitable partners.


1196 D. Lavie

In addition, multilateral competition influencesthe value of alternative alliances. By definition,the relative availability of alternative alliances toa partner is negatively related to the number ofother partners in the focal firm’s alliance portfo-lio. However, when these other partners operatein industries different from that of the partner inquestion, the firm’s capacity to leverage its multi-ple alliances in negotiations with that partner islimited. The question of interest here concernsthe extent to which the available alliances areindeed substitutes. The various alliances in thefocal firm’s portfolio are more likely to substi-tute for each other when they are formed withpartners that operate in the same industries. Thus,from the perspective of the focal firm, multilat-eral competition increases the opportunity valueof its other alliances. The substitution among thefocal firm’s alliances, which derives from multilat-eral competition, can therefore compensate for thelimited number of alternative alliances availableto the firm relative to the number of alternativealliances available to its partners. In sum, multilat-eral competition can restrict the decline in marketperformance attributed to the relative bargainingpower of partners in the firm’s alliance portfolio.

Hypothesis 5a: The negative associationbetween the focal firm’s market performanceand the relative profitability of partners in itsalliance portfolio will be weakened withincreases in the level of multilateral competition.

Hypothesis 5b: The negative associationbetween the focal firm’s market performanceand the relative availability of alternativealliances to partners in its alliance portfolio willbe weakened with increases in the level of mul-tilateral competition.

RESEARCH METHODS

Research setting and sample

The theoretical framework is summarized inFigure 2. The hypotheses were tested with com-prehensive panel data on the alliance portfoliosof firms operating in the U.S. software industry(SICs 7371 through 7374). I selected this settingfor several reasons. First, the software industryhas experienced dynamic and extensive allianceactivity (see Figure 1). The practice of engagingin multiple simultaneous alliances in this indus-try enhances the meaningfulness, reliability, andvariance of alliance portfolio variables. Second,this industry features a high proportion of publiclytraded firms with accessible financial data, thusincreasing the representativeness of the sample andreducing potential size- and age-related biases thatcan emerge when young and small firms are under-represented. Finally, the sample is highly represen-tative, since U.S. firms account for nearly half ofthe worldwide software market (Rudy, 2000).

RelativePartner

Profitability

RelativePartner

Alternatives

+ (H1)

– (H2a)

– (H2b)

+ (H3a)

+ (H3b) – (H5b)

– (H5a) + (H4)

MultilateralCompetition

BilateralCompetition

Focal FirmMarket

Performance

NetworkResources

Figure 2. Theoretical framework and hypotheses



I designed the study as a pooled time-series anal-ysis spanning the years 1990–2001, yet trackedhistorical alliances back to the year 1985 in orderto ensure the coverage of active alliances. This5-year window followed conventional assumptionsin alliance network research (Stuart, 2000). Thesample included the 367 U.S.-based software firmsthat were listed as publicly traded in 2001 and thathad at least five annual records in the Compus-tat database. The exclusion of firms with fewerthan five records during the study’s time frame isjustified by the longitudinal nature of the analy-sis. Specifically, the inclusion of firm fixed effectsin this analysis is essential to control for unob-served heterogeneity and potential endogeneity inthe panel data, but it can produce biased esti-mates when the sample period is too short (Gulati,1995b; Hsiao, 1986). Therefore, a minimum num-ber of observations per firm is required. Never-theless, I found no significant differences betweenthe 367 sampled firms and the remaining 297 pub-licly traded firms in this industry that had less thanfive years of records, were inactive in 2001, orwere headquartered in foreign countries.3 In addi-tion, since five firms had no alliances during thestudy’s time frame and certain firms had no activealliances in certain years, I tested the implicationsof self-selection bias in firms’ decisions to engagein alliances with a two-stage Heckman procedure,which produced results that were consistent withthe reported findings. Furthermore, survivor bias isunlikely given the poor performance of the sam-pled firms. For example, their return on assetsmonotonically declined by 35 percent annually onaverage between 1990 and 2001. In addition, 52percent of the sampled firms divested parts oftheir business or were acquired between 2001 and2005. Hence, there is no reason to believe that thesampled firms were better performing or differedsignificantly from other firms in their industry.

Data collection

Following Anand and Khanna (2000b), I first com-piled records of alliances formed by each focal firmbetween 1985 and 2001 from the SDC Platinum

3 For example, no significant differences were found in termsof total assets (t = 1.430, p = 0.152), net sales (t = 0.525,p = 0.600), number of employees (t = 0.274, p = 0.785), netincome (t = 1.481, p = 0.139), cash (t = 1.505, p = 0.133),long-term debt (t = 0.066, p = 0.947), stock price (t = 1.273,p = 0.204), and other indicators.

database. In order to ensure complete coverageof publicly announced alliances, I complementedand corrected the SDC data by searching allianceannouncements and status reports in the Factivadatabase (over 8,000 news sources worldwide),press releases, and alliance listings posted on cor-porate web sites, as well as SEC filings accessedvia the Edgar database. Most alliance announce-ments were cross-validated by at least two inde-pendent sources. Following this corroboration, Iadded, eliminated, or corrected alliance recordsand made additional corrections based on a cor-porate history search that tracked name changes,mergers, acquisitions, and spin-offs involving eachfocal firm and its identified partners. In total, Iidentified 20,779 alliances, of which only 5,135were reported in SDC. The identified alliancesinvolved 8,801 unique partners, with 2,884 pub-licly traded partners accounting for 66 percent ofthe alliances.4 A focal firm participated in 56.70alliances on average during the time frame of thestudy.

For each alliance, I coded the date of announce-ment, pre-specified duration or termination date,5

number of participating partners, partners’ names,public status and countries of origin, indication ofthe strategic significance of the alliance, whether

4 Whereas the lack of financial information for privately heldpartners did not affect variables that measured the type ofalliance, the size of the alliance portfolio, and network structure,variables that relied on financial measures could be calculatedmore accurately for firms with higher proportions of publiclytraded partners in their alliance portfolios. However, this poten-tial effect was controlled for by including a measure of thepercentage of publicly traded partners in the portfolio. In orderto assess the differences between the public partners and privatepartners, I empirically contrasted these two subsamples, findingthat, compared with private partners, public partners engaged inmore strategic long-term alliances and favored joint ventures andtechnology alliances rather than marketing alliances.5 Alliance termination dates were unavailable for many alliances,since firms rarely announce alliance termination and occasionallymaintain inactive alliances. To the extent that the date of alliancetermination was unavailable from archival sources, it was calcu-lated based on alliance extension announcements and reports ofactive alliance status in a given year. Remaining alliances wereassumed to have a 3-year duration based on the average specifiedduration of other alliances in the sample as well as assessmentsof industry experts whom I interviewed. Alliance terminationdates were imputed for 67 percent of the alliances. The impu-tation of alliance termination dates is a conventional practicein alliance research. For example, Stuart (2000) imputed 5-yearduration for all alliances using a linear depreciating weightingfor alliances with an earlier date of formation. In this study, I wasable to reduce the use of imputation by searching alliance statusreports and recording alliance termination dates when available.The impact of the age of alliances in the portfolio was measuredseparately with a control variable.


1198 D. Lavie

the alliance was a joint venture, and its classifica-tion to agreement types: licensing, manufacturing,sales and service, OEM and value-added resale,R&D, royalties, and supply. A given alliance couldinvolve more than one type of agreement.

I extracted firm-specific and partner-specificdata, such as historical SIC code, number ofemployees, total assets, long-term debt, R&Dexpenses, and net income, annually from Compu-stat for the years 1990–2001. I converted all thefinancial data to billion U.S. dollar units. I furtherextracted data on common shares outstanding andstock prices from the joint CRSP database and dataon S&P 500 stock market value from the CRSPIndices database.

Since the firm-year was considered the opera-tional unit of analysis, I pooled the data on the20,779 alliances across all alliances in each focalfirm’s portfolio in a given year, producing 2,595firm-year observations. This sample excluded pre-1990 and year-2001 records, which were elimi-nated because of the time frame setting and the lag-ging of independent variables by one year relativeto the dependent variable. This one-year lag wasnecessary in order to allow for causal interpreta-tion of the findings (Stuart, 2000). The final sampleincluded 1,822 observations after listwise deletionwas applied to treat missing values. Missing val-ues occurred, for instance, because firms were notrequired by SEC regulations to report R&D andadvertising investments and because stock marketprices were available only for the years in whichfirms were publicly traded.

Most of the firms in this sample operated inthe pre-packaged software segment (SIC 7372,68.7%) or the computer integrated system designsegment (SIC 7373, 23.5%), with a few firmsengaging in computer processing and data prepa-ration services (SIC 7374, 4.7%) or computer pro-gramming services (SIC 7374, 3.1%). On aver-age, a focal firm owned $412 million in assets,had $304 million in annual sales, employed 1,730employees, and spent $38.4 million annually onR&D and $8.5 million on advertising. In addi-tion, it had $58 million in cash, and achieved a−24 percent return on assets with a $1,068 millionmarket value. On average, focal firms were in busi-ness for 14 years and maintained a network of15.7 partners. Partners came primarily from thebusiness services industry (SIC 73, 52.3%), theindustrial, commercial, and computer equipmentindustry (SIC 35, 18.1%), electronics (SIC 36,

9.4%), or communications (SIC 48, 4.1%), or from56 other industries. On average, a partner owned$4,749 million in assets, had $2,252 million inannual sales, employed 8,500 employees, spent$60 million annually on R&D and $36 million onadvertising, and had $210 million in cash.

Dependent variable: Market performance

I used market performance as the dependent vari-able to overcome the recognized limitations ofaccounting measures of performance. These mea-sures often discount firms’ intangible assets (Brush,Bromiley, and Hendrickx, 2000), which are per-vasive in the software industry. Indeed, the meanTobin’s q —a proxy for intangible assets—of thesampled firms reached 2.9, with values higher than1 indicating a high proportion of intangible assets.In addition, accounting performance measures aresomewhat distant from alliance activities becauseof potential intervening and confounding firm-level factors, which may account for the insignifi-cant direct effects of alliances on firm profitabilityobserved in some prior studies (Berg et al., 1982;Hagedoorn, 1993). Finally, my auxiliary analy-ses revealed that market performance models havestronger explanatory power than alternative mod-els based on dependent variables such as return onsales and Tobin’s q.

I calculated the market performance measure foreach focal firm, capturing the annual change in themarket value of the firm’s common shares. Thisis a financial measure based on investors’ expec-tations about the future performance of the firm.The market value is typically calculated by mul-tiplying the firm’s stock price by the number ofcommon shares outstanding. Because of the highvolatility of this measure, I calculated the annualmarket value by averaging the 12 end-of-monthdaily values of the relevant calendar year. In orderto control for stock market fluctuations and tempo-ral trends (see Figure 3), I adjusted the measure bydividing it by the ratio of the compound S&P 500market value at year t to the compound S&P 500market value at the year 1990.6 Thus, the adjustedmarket value of firm i’s common shares at timet + 1 took the following form:

6 An alternative model that incorporated the compound softwareindustry market value index (SICs 7371–7374) instead of thecompound S&P 500 index generated similar results.



0

200

400

600

800

1000

1200

1400

1600

1800

2000

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Year

Mar

ket V

alue

[$

Mill

ion]

-60%

-40%

-20%

0%

20%

40%

60%

80%

Mar

ket V

alue

Gro

wth

Average Market Cap Average Market Cap Growth

Average S&P500 Market Cap Growth

Figure 3. Market performance of firms in the U.S. software industry, 1990–2001

Market valuei,t+1 = (S&P 500 market value1990/

S&P 500 market valuet+1) × (

12∑m=1

Stock pricei,t+1,m × Outstanding sharesi,t+1,m/12)

I calculated the annual change in market valueby dividing the adjusted market value at year t + 1by the adjusted market value at year t . Follow-ing prior research (Brush et al., 2000; Podolny,Stuart, and Hannan, 1996; Stuart, 2000), I log-transformed this ratio to achieve the desirable sta-tistical properties under the linearity, homoskedas-ticity, and independence assumptions. Hence, if xi,t

is a covariate matrix, then the change in marketvalue can be expressed by the following powerfunction:

ln(Market valuei,t+1) = α ln(Market valuei,t )

+ π ′xi,t + ei,t+1

In this model, all variables are annually updatedand lagged by one year relative to the dependentvariable. When estimated with OLS regression,this model produces efficient and unbiased esti-mates. Another advantage of this specification isthe embedded control for past performance, which

enables the interpretation of causal effects of theindependent variables on market performance.

Independent variables and moderators

Network resources

A set of five variables measured the networkresources of partners in the alliance portfolio. Tech-nology network resources were measured as themean value of R&D investments of partners in thefocal firm’s alliance portfolio. Similarly, marketingnetwork resources were proxied by the mean valueof advertising investments of partners, financialnetwork resources were calculated based on themean value of cash funds available to partners, andhuman network resources indicated the mean num-ber of employees of partner organizations (reportedin 100Ks). The underlying assumption is that evenif these resources are not fully shared with the focalfirm, they may be accessible through the firm’sties to its partners (Lavie, 2006). The incorpora-tion of the number of employees as an independentvariable also controls for the size of partners inthe firm’s alliance portfolio. Finally, I measurednetwork prominence as the percentage of publiclytraded partners in the focal firm’s alliance portfo-lio. Publicly traded partners typically play the roleof prominent endorsers that are well known to, andfrequently appreciated by, the investor community,suppliers, and customers.


1200 D. Lavie

Relative profitability

The relative profitability variable measured themean difference between the return on assets(ROA) of partners in the alliance portfolio andthe focal firm’s ROA in a given year, using theformula:

[Ki,t∑j=1

(ROAj,t /Ki,t )] − ROAi,t

where Ki,t is the number of partners in firm i’sportfolio in year t . This variable represents therelative bargaining power of partners based on theprofitability of partners relative to the profitabilityof the focal firm. Higher values of this variablesuggest that partners in firm i’s alliance portfolioare less dependent on the outcomes of alliancesrelative to that firm.7

Relative alternatives

The relative alternatives variable captures the struc-tural dependence of the focal firm on its allianceportfolio, under the assumption that partners withmany alternative ties to other firms in the softwareindustry hold strong bargaining positions. I calcu-lated this variable as a function of the ratio of eachpartner’s number of alliances with other focal firmsto the number of alliances that each focal firm hadwith other partners, averaged across partners in thefirm’s alliance portfolio in a given year:

Ki,t∑j=1

ln(

Kj,t − Ki,t,j

Ki,t − Ki,t,j

)/Ki,t

7 An alternative measure of bargaining power based on the rela-tive size of the focal firm and its partners was abandoned sinceit cannot discriminate between the value-creation effect of net-work resources and the value-appropriation effect of bargainingpower. Moreover, software firms are typically specialized andrely on intangible assets that are not captured in conventionalorganizational size measures (as indicated by the high value ofTobin’s q). This may introduce a systematic bias when relativesize is used as a proxy for bargaining power. For example, themean value of the total assets of partners was more than tentimes that of the focal firms, and clearly does not reflect the rel-atively strong bargaining position of software firms. Moreover,according to Porter (1980: 24, 1985: 6), bargaining power inexchange transactions derives from a variety of factors, one ofwhich is the volume of the transaction relative to the overall salesof the seller or the overall purchases of the buyer. This is notequivalent to the relative size of the partners involved in the bar-gaining situation, but instead captures their mutual dependence,which more directly derives from the perceived importance ofthe transaction and the availability of alternatives.

where Kj,t is the total number of alliances betweenpartner j and all the focal firms in year t , and Ki,t,j

is the number of alliances between focal firm i andpartner j in year t . Higher values of this variablesuggest that partners in the alliance portfolio haverelatively more alternatives than the focal firm.

Bilateral competition

I calculated the level of competition between thefocal firm and partners in its alliance portfolioas the percentage of matches between the firm’sprimary industry segment (four-digit SIC) and theprimary industry segments of its partners, taking

the formKi,t∑j=1

mi,t,j/Ki,t , where mi,t,j is a dummy

receiving a value of 1 when the SIC of firm i

matches that of partner j in year t .

Multilateral competition

Assuming that competition among partners inten-sifies with the degree of similarity in their busi-ness portfolios, this variable was operationalizedas the sum of squared proportions of partners’sales in each industry segment. This is a con-centration measure that resembles the inversedBerry–Herfindahl diversification index (Mont-gomery, 1982), except that it is based on the salesdistributions of partners rather than on those of thefocal firm. It takes the form:

Ci,t∑c=1

p2i,t,c, with pi,t,c =

Ki,t,∑j=1

sj,t,c/

Ki,t,∑j=1

∑c

sj,t,c

where sj,t,c measures partner j ’s sales in four-digitprimary SIC code c in year t . For instance, if all ofthe firm’s partners generate their sales in the sameindustry, this measure receives a value of 1, andif each partner specializes and therefore extractsequivalent revenues from a unique industry, thismeasure becomes proportional to the inverse of thenumber of industries in which partners compete.

Control variables

I controlled for interindustry variation by focus-ing on the analysis of a single industry and forinter-temporal trends and shocks by standardizingthe dependent variable by the S&P 500 stock mar-ket index. The remaining controls included firm-level and portfolio-level variables that I lagged by



one year relative to the dependent variable. Annu-ally updated, firm-level controls included firm size,measured as the focal firm’s total asset value, andfirm R&D intensity, measured as the firm’s R&Dinvestments divided by its net sales. I controlledfor all remaining interfirm heterogeneity by includ-ing firm fixed effects, thus separating the impactof alliance portfolios from other firm activities thatmay influence market performance.

I controlled for the size of the alliance port-folio, which may positively affect firm perfor-mance (Ahuja, 2000; Baum et al., 2000; Stuartet al., 1999). Because of the correlation betweenthe number of alliances and firm size (r = 0.73,p < 0.001) and the inherent association betweenthese two variables (Hagedoorn and Schakenraad,1994), I adjusted the portfolio size variable bytaking the logarithm of the number of alliancesdivided by the total assets of the focal firm. Addi-tional controls included portfolio age, measured asthe average age of alliances in the firm’s allianceportfolio, which I introduced in order to controlfor the alliance life cycle. Multi-partner alliancemeasured the average number of partners involvedin each of the firm’s alliances, assuming thatmulti-partner alliances entail more complex man-agement. Portfolio internationalization measuredthe percentage of foreign partners in the allianceportfolio, assuming that high proportions of for-eign partners may be more difficult to managebecause of geographical and cultural distance andlanguage and time zone differences, or because ofthe increased possibility of government interven-tion.

Another set of variables controlled for rela-tional aspects of the firm’s ties with partners in itsalliance portfolio (Granovetter, 1985; Uzzi, 1996).Technology agreements measured the proportion ofR&D and manufacturing agreements with partners,while marketing agreements captured the propor-tion of joint sales and service, OEM, value-addedresale, and supply agreements with partners. Theassumption is that the former type of agreementsentail greater interdependence and interaction withpartners than the latter (Rowley et al., 2000). Tiemultiplicity controlled for another relational aspectby measuring the average standardized number ofsimultaneous categories of agreements signed in agiven alliance. Joint ventures measured the pro-portion of equity-based joint ventures out of thetotal number of alliances in the firm’s portfolio,in order to control for the governance mode of

alliances. Finally, the strategic status control mea-sured the proportion of alliances that were referredto as ‘strategic’ in press releases, and to which thefirm may be inclined to commit more resources.

A final set of variables controlled for structuralaspects of the alliance portfolio. Since informa-tion about possible ties among the 8,479 non-focalpartners was unavailable, I considered the sub-set network of ties among the 367 focal firmswhen calculating these measures. Structural holescontrolled for the resource access and brokeragebenefits derived from a sparse network structure,in which the focal firm bridges otherwise discon-nected partners (Burt, 1992). This measure wascomputed as the inverse of the network constraintmeasure using the structural holes procedure inUCINET (Borgatti, Everett, and Freeman, 2002;Zaheer and Bell, 2005). Network closure controlledfor the social norms and sanctions that may enforcecollaboration in dense network structures (Cole-man, 1988, 1990). I computed this measure usingthe ego network density procedure in UCINET(Bae and Gargiulo, 2004; Borgatti et al., 2002),which provides the ratio of observed to maximumalliances possible among firms in the network.

Analysis

I report descriptive statistics in Table 1, whichdetails the means, standard deviations, and pair-wise correlations of the variables. Variance infla-tion factors were considerably lower than the crit-ical value, ruling out potential multicollinearity.Analysis of the unbalanced panel data was carriedout using cross-section time-series regressions withfirm fixed effects. Fixed-effects models incorporatesuperior controls for time-invariant variables andare thus preferred to random-effects models thatmay produce biased estimates (Mundlak, 1978).The inclusion of fixed effects suggests that thereported models explain within-firm variation infirm performance over time, rather than interfirmvariation in performance. I treated potential auto-correlation by incorporating first-order autoregres-sive errors in the models, assuming correlation oferrors across adjacent years. I also tested contem-poraneous correlation across firms in the panel dataand ruled it out because of insignificance. Thus, themodels took the form:

yi,t = α + βxi,t + ui + εi,t , with εi,t

= ρεi,t−1 + µi,t and − 1 < ρ < 1


1202 D. Lavie

Tabl

e1.

Sam

ple

stat

istic

san

dco

rrel

atio

nm

atri

x

Var

iabl

ena

me

1.2.

3.4.

5.6.

7.8.

9.10

.11

.12

.13

.14

.15

.16

.17

.18

.19

.20

.21

.22

.23

.24

.

1.L

ogad

j.m

arke

tva

lue t

+1

1.00

2.L

ogad

j.m

arke

tva

lue t

0.94

1.00

3.Fi

rmsi

ze0.

410.

451.

004.

Firm

R&

Din

tens

ity−0

.06

−0.0

5−0

.02

1.00

5.Po

rtfo

liosi

ze−0

.56

−0.6

4−0

.29

0.11

1.00

6.Po

rtfo

lioag

e−0

.07

−0.0

70.

03−0

.05

−0.0

91.

007.

Mul

ti-pa

rtne

ral

lianc

e0.

260.

270.

13−0

.02

−0.0

7−0

.02

1.00

8.Po

rtfo

lioin

tern

atio

naliz

atio

n−0

.02

−0.0

10.

00−0

.01

−0.0

20.

030.

021.

00

9.Te

chno

logy

agre

emen

ts0.

160.

160.

020.

030.

01−0

.02

0.10

−0.0

41.

00

10.

Mar

ketin

gag

reem

ents

−0.1

1−0

.12

−0.0

4−0

.05

−0.0

1−0

.03

−0.0

70.

03−0

.71

1.00

11.

Tie

mul

tiplic

ity0.

060.

05−0

.01

0.01

−0.0

40.

030.

030.

090.

390.

121.

0012

.Jo

int

vent

ures

0.03

0.03

0.06

−0.0

2−0

.07

0.12

0.11

0.17

0.00

0.06

0.04

1.00

13.

Stra

tegi

cst

atus

0.25

0.26

0.11

−0.0

1−0

.11

−0.0

10.

15−0

.01

0.18

−0.0

70.

28−0

.08

1.00

14.

Stru

ctur

alho

les

0.02

0.02

−0.0

10.

040.

12−0

.05

−0.0

6−0

.03

−0.0

10.

00−0

.06

−0.0

7−0

.11

1.00

15.

Net

wor

kcl

osur

e0.

010.

010.

000.

050.

04−0

.03

−0.0

4−0

.02

0.00

0.01

−0.0

2−0

.02

−0.0

70.

421.

0016

.Te

chno

logy

netw

ork

reso

urce

s−0

.08

−0.0

9−0

.04

0.01

0.03

0.11

−0.0

20.

070.

01−0

.05

−0.0

4−0

.04

−0.0

9−0

.02

0.00

1.00

17.

Mar

ketin

gne

twor

kre

sour

ces

−0.1

0−0

.10

−0.0

3−0

.01

−0.1

00.

21−0

.03

0.18

0.01

−0.0

40.

04−0

.06

−0.0

9−0

.03

0.00

0.43

1.00

18.

Fina

ncia

lne

twor

kre

sour

ces

0.00

−0.0

20.

02−0

.01

−0.0

20.

120.

000.

140.

00−0

.02

0.00

−0.0

2−0

.08

−0.0

2−0

.01

0.79

0.42

1.00

19.

Hum

anne

twor

kre

sour

ces

−0.0

3−0

.05

−0.0

1−0

.01

0.04

0.07

0.01

0.11

0.02

−0.0

30.

00−0

.02

0.00

−0.0

4−0

.01

0.87

0.38

0.70

1.00

20.

Net

wor

kpr

omin

ence

0.05

0.03

0.01

−0.0

10.

06−0

.03

0.02

−0.1

00.

11−0

.04

0.14

0.06

0.08

−0.0

20.

000.

08−0

.01

0.05

0.13

1.00

21.

Rel

ativ

epr

ofita

bilit

y−0

.34

−0.3

2−0

.09

0.19

0.43

−0.0

3−0

.06

0.04

−0.0

50.

03−0

.02

0.00

−0.0

50.

050.

010.

010.

000.

02−0

.03

−0.0

51.

0022

.R

elat

ive

alte

rnat

ives

−0.1

2−0

.15

−0.1

0−0

.01

0.12

−0.0

5−0

.07

−0.1

90.

07−0

.04

0.02

−0.0

50.

030.

010.

030.

300.

060.

220.

290.

59−0

.03

1.00

23.

Bila

tera

lco

mpe

titio

n0.

090.

11−0

.02

0.02

0.08

−0.0

6−0

.01

−0.1

10.

14−0

.08

0.03

−0.1

20.

010.

040.

03−0

.22

−0.1

2−0

.15

−0.2

4−0

.10

−0.0

6−0

.01

1.00

24.

Mul

tilat

eral

com

petit

ion

−0.3

3−0

.38

−0.2

30.

030.

19−0

.09

−0.1

9−0

.15

−0.0

50.

06−0

.01

0.02

−0.0

30.

010.

000.

01−0

.08

−0.0

30.

040.

150.

060.

340.

011.

00

N22

2620

5623

1020

6825

9525

9525

9525

9325

9525

9525

9525

9525

9525

9525

9522

9118

2923

8723

6025

9521

3325

9524

4523

89M

ean

3.74

3.85

0.41

0.39

−0.8

31.

822.

150.

170.

480.

700.

060.

040.

310.

260.

110.

970.

451.

190.

540.

750.

18−2

.09

0.25

0.63

Stan

dard

devi

atio

n2.

101.

992.

011.

893.

290.

630.

590.

210.

320.

280.

060.

140.

30.

340.

251.

010.

511.

300.

630.

210.

663.

800.

300.

34

Coe

ffici

ents

larg

erth

an0.

045

inab

solu

teva

lue

are

sign

ifica

ntat

the

5%le

vel.



where ui represents the firm fixed effects and ρ

is the autoregressive AR(1) parameter that has azero mean, homoskedastic and serially uncorre-lated error term µi,t . I tested the models using theMIXED procedure in SAS, which relies on maxi-mum likelihood estimation. The lag structure, thecontrol for past performance, and the incorporationof firm fixed effects and a first-order autoregressivecoefficient in the model increase confidence in thecausal interpretation of the findings.

The analysis followed a hierarchical approach inwhich I introduced independent variables and mod-erators in subsequent models. I based hypothesistesting on the results of the full model (Model 7).Missing values were treated with listwise deletion,which accounts for the variation in sample sizesacross models. I used log-likelihood ratio tests toassess the improvement in model fit by comparingeach model to the adjusted baseline model, whichincluded only the controls and fixed effects, dis-carding the observations missing in the partial orfull models.

RESULTS

I report the findings in Table 2. Based on thefull model (Model 7), the firm’s market per-formance declines with increases in the propor-tion of foreign partners in the alliance portfolioas indicated by the portfolio internationalizationeffect (β = −0.55, p < 0.001). This finding maybe ascribed to cultural and organizational differ-ences, geographical distance, or communicationand learning challenges that are typical of cross-border alliances (Barkema et al., 1997; Parkhe,1991; Simonin, 1999) and may impede value cre-ation and eventually impair the firm’s marketperformance. Additionally, the firm’s market per-formance is positively related to tie multiplicity(β = 2.14, p < 0.01), which captures the numberof distinctive types of agreements signed in thecourse of a given alliance. This finding is consis-tent with the relational embeddedness perspective(Granovetter, 1985; Powell, 1990; Uzzi, 1996),wherein strong or rich ties to partners promoteinterfirm trust and norms of reciprocity that mayfacilitate the exchange of valuable knowledge andinformation between the firm and its partners. Thestructural holes and network closure controls pro-duced no significant effects. According to recentresearch, however, it is possible that the effects

of network structure depend on firm- or partner-specific factors (Bae and Gargiulo, 2004; Shipilov,2006; Zaheer and Bell, 2005).8

Hypothesis 1 is partially supported by the resultsof Model 7. In accordance with this hypothe-sis, the marketing and financial network resourcespossessed by the firm’s partners enhance its mar-ket performance (β = 0.15, β = 0.09 respectively,p < 0.01). Similarly, the prominence of partnersin the alliance portfolio is positively related tomarket performance (β = 0.48, p < 0.01). How-ever, human network resources do not enhancemarket performance, whereas technology networkresources produce a marginally significant negativeeffect (β = −0.09, p < 0.1).

In support of Hypothesis 2, Model 7 reveals neg-ative effects of relative partner profitability (β =−0.40, p < 0.001) and relative partner alternatives(β = −0.09, p < 0.05) on market performance.Hence, alliance portfolios that feature partners thatare less dependent on the outcomes of alliancesand enjoy better access to other alliances relativeto the focal firm are detrimental to the firm’s mar-ket performance. The negative effect of relativepartner profitability intensifies with the level ofcompetition between the focal firm and its partnersas indicated by the significant moderating effectof bilateral competition (β = −0.30, p < 0.05).However, the moderating effect of bilateral com-petition on the association between relative partneralternatives and market performance is insignifi-cant. Thus, Hypothesis 3a is supported by the data,whereas Hypothesis 3b is not.

Consistent with Hypothesis 4, the firm’s marketperformance is enhanced with the level of multi-lateral competition among partners in its allianceportfolio (β = 0.42, p < 0.01). In addition, multi-lateral competition weakens the negative effects ofrelative partner profitability (β = 0.29, p < 0.05)and relative partner alternatives (β = 0.07, p <

0.05). Hence, in support of Hypotheses 5a and 5b,

8 In auxiliary analysis I incorporated alternative structural holesand network closure measures corresponding to the global net-work of ties with both focal and non-focal partners. In thisanalysis, structural holes negatively affected market performance(β = −0.66, p < 0.01), whereas network closure had a marginalpositive effect (β = 0.44, p < 0.1). All of the main effectsand interactions remained significant in the full model with theexception of multilateral competition and its interaction withrelative partner alternatives. In addition, I examined models thatcontrol for either structural holes or network closure, excludingthe other measure in turn. The effects of these controls wereinsignificant, with all the main effects and interactions retainingtheir levels of significance.


1204 D. Lavie

Table 2. Fixed effects panel AR(1) models for market performance

Dependent variable:Log adj. market valuet+1

Model1

Model2

Model3

Model4

Model5

Model6

Model7

Model8

Intercept 1.76 0.39 0.32 0.35 0.37 0.23 0.04 0.51Firm fixed effects Included Included Included Included Included Included Included IncludedLog adj. market valuet 0.61∗∗∗ 0.60∗∗∗ 0.59∗∗∗ 0.59∗∗∗ 0.58∗∗∗ 0.59∗∗∗ 0.59∗∗∗ 0.58∗∗∗

Firm size −0.04† −0.03 −0.03 −0.03 −0.03 −0.02 −0.02 −0.02Firm R&D intensity −0.03 −0.04∗ −0.02 −0.02 −0.02 −0.02 −0.02 −0.02Portfolio size −0.09∗∗∗ −0.06∗ −0.02 −0.02 −0.02 −0.00 −0.00 0.01Portfolio age −0.04 −0.07† −0.08† −0.08† −0.07† −0.05 −0.04 −0.02Multi-partner alliance 0.03 0.02 0.02 0.02 0.02 0.02 0.02 0.05Portfolio internationalization −0.29∗∗ −0.51∗∗∗ −0.54∗∗∗ −0.54∗∗∗ −0.53∗∗∗ −0.51∗∗∗ −0.55∗∗∗ −0.67∗∗∗

Technology agreements −0.08 −0.34† −0.24 −0.24 −0.27 −0.26 −0.22 −0.07Marketing agreements −0.11 −0.40† −0.33 −0.33 −0.33 −0.34 −0.30 −0.23Tie multiplicity 1.16∗ 2.42∗∗ 2.28∗∗ 2.28∗∗ 2.21∗∗ 2.25∗∗ 2.14∗∗ 1.68∗

Joint ventures 0.24 0.41 0.31 0.30 0.35 0.28 0.38 0.13Strategic status 0.04 0.05 0.04 0.04 0.07 0.04 0.05 −0.03Structural holes 0.07 0.06 0.06 0.06 0.05 0.05 0.05 0.06Network closure 0.02 0.04 0.05 0.05 0.04 0.04 0.05 0.03Technology network resources −0.12∗ −0.11† −0.11† −0.11† −0.10† −0.09† −0.06Marketing network resources 0.10∗ 0.10∗ 0.10∗ 0.11∗ 0.13∗∗ 0.15∗∗ 0.12∗

Financial network resources 0.09∗∗ 0.09∗∗ 0.09∗∗ 0.09∗∗ 0.09∗∗ 0.09∗∗ 0.08∗

Human network resources 0.11 0.13 0.12 0.12 0.10 0.12 0.10Network prominence 0.15 0.35∗ 0.35∗ 0.38∗ 0.39∗ 0.48∗∗ 0.46∗∗

Relative partner profitability −0.29∗∗∗ −0.29∗∗∗ −0.22∗∗∗ −0.22∗∗∗ −0.40∗∗∗ −0.40∗∗∗

Relative partner alternatives −0.02† −0.02† −0.02 −0.02† −0.09∗ −0.10∗∗

Bilateral competition 0.04 −0.09 −0.09 −0.06 −0.05Bilateral competition ×

Relative partner profitability−0.33∗ −0.33∗ −0.30∗ −0.29†

Bilateral competition ×Relative partner alternatives

−0.04 −0.03 −0.03 −0.04

Multilateral competition 0.22∗∗ 0.42∗∗ 0.38∗∗

Multilateral competition ×Relative partner profitability

0.29∗ 0.30∗∗

Multilateral competition ×Relative partner alternatives

0.07∗ 0.08∗

Probability of non-missingvalues

−0.68∗∗∗

Number of firm-years 1822 1346 1343 1343 1343 1343 1343 1343Number of firms 331 301 300 300 300 300 300 300Autocorrelation coefficient

AR(1)−0.06 −0.141 −0.13 −0.13 −0.13 −0.13 −0.13 −0.11

−2 Log-likelihood 3251.6 2388.0 2330.1 2330.0 2323.9 2315.0 2304.6 2293.2Adjusted � − 2LL (n = 1343) 21.3∗∗∗ 72.4∗∗∗ 72.5∗∗∗ 78.6∗∗∗ 87.5∗∗∗ 97.9∗∗∗ 109.3∗∗∗

Pseudo R2 0.92 0.92 0.93 0.93 0.93 0.93 0.93 0.93

†p < 0.1; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.

the negative performance implications of partners’sources of bargaining power are attenuated by theextent to which partners in the firm’s alliance port-folio compete against each other.

Overall, the explanatory power of the full modelreached 93 percent, which is typical of fixed-effects models. The contribution of value-creation

variables to a firm’s market performance basedon the improvement in goodness of fit of Model2 relative to Model 1 after adjusting for samplesize suggests that some network resources sig-nificantly contribute to a firm’s market perfor-mance (χ 2

df=5 = 21.3, p < 0.001). However, thecontribution of value appropriation variables to the



explanatory power of the model is even greater(χ 2

df=8 = 83.4, p < 0.001) as revealed by thecomparison between Model 7 and Model 2. Basedon Model 7, an auxiliary analysis revealed thatwith respect to the value-creation variables a onestandard deviation increase in marketing networkresources improves the firm’s adjusted marketvalue by 7.8 percent, while a similar increase infinancial network resources leads to a 10.7 percentincrease in adjusted market value. In addition,a one standard deviation increase in networkprominence increases the adjusted market valueby 9.2 percent. With respect to value appropria-tion variables, a one standard deviation increasein relative partner profitability reduces the firm’sadjusted market value by 21.6 percent, and a simi-lar increase in relative partner alternatives reducesthe adjusted market value by 21.0 percent. Finally,an increase of one standard deviation in multi-lateral competition increases the adjusted marketvalue by 14.7 percent. The moderating effects ofbilateral and multilateral competition are depicted

in Figure 4, in which the variables of interest arepresented at their minimum and maximum values,while all remaining variables are held at their meanlevels.

To test the robustness of the findings, I con-sidered alternative specifications. For example, Itested for selection bias that may arise if values inthe full model are not missing at random (Alli-son, 2001). For this purpose, I ran a two-stageregression analysis, where the first stage predictsthe probability of non-missing values using pro-bit analysis with robust estimates, accounting forthe panel structure of the data. The results ofthis model (see Appendix) suggest that missingvalues are more likely in the years 1995–1998,and for strategic R&D agreements, but not forOEM/VAR agreements or for firms operating inSICs 7372–7373. Finally, the occurrence of miss-ing values is minimized with increases in the num-ber of partners in the alliance portfolio. Whencontrolling for the probability of non-missing val-ues (see Table 2, Model 8), the results remain

0

10

20

30

40

50

60

70

80

Relative Partner Alternatives

Adj

uste

d M

arke

t V

alue

t+1

0

20

40

60

80

100

120

140

-2 -1

Relative Partner Profitability

Adj

uste

d M

arke

t V

alue

t+1

Mean

Mean

Max BilateralCompetition

Min BilateralCompetition

0

20

40

60

80

100

120

140

Relative Partner Profitability

Adj

uste

d M

arke

t V

alue

t+1

Mean

Mean Max MultilateralCompetition

Min MultilateralCompetition

0

10

20

30

40

50

60

70

80

-8 -6 -4 -2

Relative Partner Alternatives

Adj

uste

d M

arke

t V

alue

t+1

Mean

Mean

Max BilateralCompetition

Min BilateralCompetition

Mean

Mean

Max MultilateralCompetition

Min MultilateralCompetition

0 1 2 3 4 5 6 -2 -1 0 1 2 3 4 5 6

0 2 4 6 8 10 12 -8 -6 -4 -2 0 2 4 6 8 10 12

Figure 4. Predicted market value by relative network profitability/alternatives and bilateral/multilateral competition(Model 7)


1206 D. Lavie

fully consistent with Model 7, although the interac-tion between relative partner profitability and bilat-eral competition becomes marginally significant(β = −0.29, p = 0.056).9 Additionally, when dif-ferent lag structures are considered, the good-ness of fit of alternative models diminishes withincreases in the lag period. For example, whenthe change in market value is measured over a 2-year period, R2 declines from 0.93 to 0.89, but theeffects of network prominence, multilateral com-petition, relative partner alternatives, and the inter-actions of the last with bilateral and multilateralcompetition remain significant.

DISCUSSION AND CONCLUSIONS

This study offers rich evidence on the contributionof alliance portfolios to firm performance. It com-plements the traditional focus on the implicationsof structural and relational embeddedness in net-works with a framework of network resources thathighlights the attributes of partners with whom thefirm collaborates. It further advances research onalliances and networks by juxtaposing value cre-ation and appropriation mechanisms, and therebyrevealing the dual role of dominant partners inthe alliance portfolio. On the one hand, prominentpartners that can endorse the focal firm and endowvaluable resources may enhance its market per-formance. On the other hand, powerful partnersmay leverage their bargaining power and impairthe focal firm’s capacity to appropriate rents fromits alliance portfolio. Moreover, this study revealshow the firm’s appropriation capacity is contingenton the levels of bilateral and multilateral competi-tion in its alliance portfolio.

9 In addition, when the power function of growth in market valueis replaced with a dependent variable based on a linear ratio ofthe market value at time t + 1 to the market value at time t , theeffects of network marketing resources, relative partner prof-itability, and multilateral competition remain significant. Whenthe absolute market value at time t + 1 is considered as thedependent variable, the effects of network prominence, relativepartner profitability, relative partner alternatives, and the inter-actions involving multilateral competition remain significant.These specifications, however, are inferior to the reported modelfor the reasons discussed in the Methods section. Some findingsare also robust to the use of revenue growth instead of marketvalue growth as a dependent variable (Stuart, 2000). Specifically,the effects of marketing and financial network resources, rela-tive partner alternatives, and the interactions between relativepartner profitability and bilateral competition as well as betweenrelative partner alternatives and multilateral competition remainsignificant.

The findings demonstrate how the network re-sources furnished by partners in the alliance port-folio enhance the focal firm’s market perfor-mance. This value-creation effect is ascribed tothe ability of the firm to leverage these exter-nal resources, create synergies by combining themwith its internal resources, and eventually inter-nalize them through learning and imitation (Lavie,2006). Nevertheless, not all types of networkresources create value. Whereas ties to prominentpartners with abundant marketing and financialresources enhance market performance, technol-ogy and human network resources fall short ofcreating value. This result is surprising given priorindications that partners’ size and innovativenesslead to revenue growth (Stuart, 2000). This contra-diction can be resolved, however, by consideringthe complementary nature of network resources.Specifically, Stuart studied horizontal alliances inthe semiconductor industry, where the ‘surest pathto commercial success has been to develop newtechnologies’ (Stuart, 2000: 796). In addition, hor-izontal alliances in that industry create economiesof scale by sharing production facilities. Theseindustry characteristics increase the complemen-tary value of partners’ sales volume and inno-vativeness. In contrast, software firms specializein the development of intellectual property basedon their proprietary technology and human assets.In the software industry, alliances enable firmsto integrate complementary software components,embed them in systems, resell and implement solu-tions, and engage in joint marketing activities.Hence, the marketing resources of partners, theirindustry prominence, and financial endorsementare essential for value creation. Generalizing fromthese findings, the relational rents that firms canderive from their network resources depend onthe complementary value of these resources fromthe perspective of the focal firm (Dyer and Singh,1998). Similar alliance portfolios may contributedifferently to value creation based on the extentto which partners’ resources are complementary.Nevertheless, this argument should be scrutinizedby future research that may directly measure thedegree of complementarity of network resources.

The main contribution of this study is in distin-guishing value creation from value appropriationmechanisms in alliance portfolios. Prior researchhas typically considered either one mechanismor the other, while focusing on dyadic alliances.



The current findings reveal that the firm’s depen-dence on its partners constrains its appropriationcapacity, which in turn results in declining mar-ket performance. In this sense, I incorporate theprivate vs. common benefits logic (Khanna et al.,1998) and extend the relational view (Dyer andSingh, 1998), which has focused on the role oftrust and knowledge sharing in joint value creation,by highlighting some of the unconstructive inter-dependencies between the firm and its partners.Specifically, partners in the alliance portfolio maycapture stronger bargaining positions by virtue oftheir higher profit margins or their access to morealternative alliances relative to the focal firm. Bysimultaneously accounting for the nature of inter-dependencies in alliances and the contributions ofnetwork resources, this study demonstrates howdominant partners can facilitate joint value cre-ation while adversely affecting firm performanceas a result of excessive appropriation of that value.Hence, partners that possess the most valuableresources are not necessarily the finest associates,as suggested in prior research (e.g., Stuart, 2000).Instead, powerful partners may be well positionedto appropriate the lion’s share of relational rents.For this reason, perhaps, some prior studies thatfocused on the value-creation effects of partners’endowments without accounting for their relativebargaining positions report insignificant contribu-tions of network resources (e.g., Zaheer and Bell,2005). Similarly, studies that consider the powerstructure of alliance portfolios without accountingfor the resource endowments of partners (e.g., Baeand Gargiulo, 2004) cannot discern adverse appro-priation effects from the value-creation effect ofnetwork resources. The current study juxtaposesvalue creation and appropriation mechanisms, thusavoiding such underspecification and clarifying themixed implications of strong alliance portfolios forfirm performance.

In addition, this study advances an alliance port-folio perspective by considering both the bilat-eral relationships between the focal firm and itspartners and the multilateral relationships acrosspartners in the portfolio. In particular, the find-ings suggest that multilateral competition amongpartners enhances the firm’s market performance.This outcome is ascribed to the firm’s ability toarbitrate among competing partners and controlresource allocation decisions, which improve itsappropriation capacity. Hence, by distinguishing

partners’ resource endowments from their indus-try affiliation, this study differentiates the negativeconsequences of resource redundancy and networkinefficiency (Baum et al., 2000) from the posi-tive implications of multilateral competition for thefirm’s appropriation capacity. Although the net-work resources accessible from similar partnersmay add limited value, multilateral competitionincreases the firm’s relative share of that addedvalue.

Therefore, the firm’s appropriation capacitydepends not only on its relative bargaining powervis-a-vis partners but also on the competitive ten-sion with its partners and among partners in itsalliance portfolio. Specifically, bilateral competi-tion motivates appropriation contests whereby thefirm and its partners struggle to maximize theirown share of relational rent. However, unlike priorresearch that has predicted a direct effect of bilat-eral competition on firm performance and pro-duced conflicting findings (Balakrishnan and Koza,1993; Chang and Chen, 2002; Koh and Venkatra-man, 1991; Park and Kim, 1997), this study sug-gests that the outcomes of these contests dependon relative bargaining power. Partners that com-pete with the focal firm in the same industry areencouraged to leverage their bargaining positions,which restrict the benefits that the firm can extractfrom its alliance portfolio. This assertion, however,receives more support with respect to the relativeprofitability of partners than with respect to the rel-ative availability of alternative alliances. Perhapsrelative profitability is a more immediate sourceof bargaining power, since relative availability ofalternatives entails negotiation with other partnersand thus offers weaker leverage to the firm.

Finally, the findings reveal how multilateralcompetition attenuates some of the negative effectsof partners’ bargaining power in the alliance port-folio. Although the tension among the firm’s part-ners may rise and perhaps even limit their com-mitments to alliances in the presence of competi-tors in the alliance portfolio, the firm can lever-age the competition among its partners to restricttheir ability to exploit their relative bargainingpositions in bilateral engagements. It can reduceits dependence on any given partner by form-ing alliances with that partner’s competitors. Inturn, the firm may face hazards when its partnersengage in alliances with its competitors. Thus, thisstudy complements prior research that noted howcompetitors that develop alliance portfolios can


1208 D. Lavie

undermine the focal firm’s competitive positionirrespective of the firm’s own alliances (Silver-man and Baum, 2002). This study also extendsthe countervailing alliance perspective (Gimeno,2005), which has focused on the tendency offirms to ally with the partners of their rivals, byproviding a rationale for the counterpart strategywhereby firms attempt to ally with the rivals oftheir partners. For example, as Sony increases thenumber of competing video game developers forits PlayStation platform in its alliance portfolio,some game developers in that portfolio may tryto balance Sony’s enhanced bargaining positionby allying with Sony’s competitors and migrat-ing their software to Microsoft’s Xbox and Nin-tendo’s GameCube platforms (Venkatraman andLee, 2004). According to the current study, sucha strategy may enable these game developers toimprove their market performance. Overall, thisstudy stresses the merits of considering the over-all configuration of the alliance portfolio and theinterdependencies across partners instead of envi-sioning alliance portfolios as mere aggregations ofdyadic relationships.

Future research may extend this study in severalways. First, it may test the findings in differentindustry contexts. Specifically, alliances may servedifferent roles in industries that are less dynamicthan the U.S. software industry, in which a dif-ferent balance may prevail between explorationand exploitation alliances (Lavie and Rosenkopf,2006; Rothaermel and Deeds, 2004; Rowley et al.,2000). Second, researchers may study the impli-cations of alliance portfolios for various perfor-mance measures that encompass not only forward-looking measures such as growth in market value,but also accounting measures such as profitabil-ity. Because of the inherent trade-off betweenlong-term and short-term performance objectives(Meyer, 2002), alliance portfolios may produceinconsistent effects across different performancemeasures. Third, there is merit in studying theevolution of alliance portfolios and consideringhow, for instance, the relative bargaining powerof partners and the level of multilateral competi-tion change over time (Hite and Hesterly, 2001;Koza and Lewin, 1998; Lavie, 2004). Fourth,field studies may offer more refined measures ofvalue creation and appropriation based on primarydata sources, distinguishing, for example, betweenpartners’ shared and non-shared resources, andaccounting for isolating mechanisms that may

potentially limit the flow of network resources(Lavie, 2006). In addition, future research mayexamine the potential availability of new alliancesin addition to the availability of existing alterna-tives when assessing relative bargaining power.Finally, researchers may study additional facets ofalliance portfolios and their performance implica-tions, for example, by considering the implicationsof the variance in partners’ network resources andbargaining positions.

In conclusion, this study contributes to theemerging research on alliance portfolios andinforms the broader literature on alliance networks.Most prior research in this field of inquiry hasemphasized structural properties of networks (e.g.,Ahuja, 2000; Bae and Gargiulo, 2004; Rowleyet al., 2000; Shipilov, 2006) with less attention tothe role of partner attributes. The current studybrings these attributes to the foreground, suggest-ing that the question of whom to partner with maybe as critical as the question of which networkstructure is desirable. In fact, the number of part-ners, the density of the network, or the prevalenceof structural holes may be less important than thenature of network resources that partners furnishto the focal firm or the nature of bilateral and mul-tilateral interdependencies in the alliance portfolio.A careful examination of the reasoning behind thetraditional focus on network structure would revealthat structural properties serve as mere proxiesfor more fundamental concepts. For example, thenotion of structural holes (Burt, 1992) is intendedto capture aspects of a firm’s bargaining powervis-a-vis its network partners, whereas structuralequivalence (Burt, 1987) implicitly assumes com-petition between equivalent partners. In a similarvein, the notion of embeddedness (Granovetter,1985; Uzzi, 1996) highlights the conditions thatfacilitate the flow of resources in the network, butit does not shed light on the nature and availabil-ity of such network resources. Hence the currentstudy offers theoretical arguments and evidencethat invoke more direct mechanisms that drive thecontribution of alliance portfolios to firm perfor-mance. It goes beyond exclusive concern with thevalue-creation impact of network resources (Stu-art, 2000), noting that resource-rich partners aredesirable as long as they do not leverage theirsuperior bargaining positions to appropriate valueat the firm’s expense. The firm may accommodatesuch prominent partners in its alliance portfolioby avoiding direct competition with them or by



attracting competing partners to balance its disad-vantageous position and restore its appropriationcapacity.

ACKNOWLEDGEMENTS

This paper is based on my dissertation, titled‘The Interconnected Firm: Evolution, Strategy, andPerformance’ (University of Pennsylvania, 2004),which won the 2005 TMS INFORMS Best Dis-sertation Award. An abbreviated version of thispaper was published in the Best Paper Proceed-ings of the 2007 Academy of Management Con-ference in Philadelphia, where it was a finalistfor the best paper award of the BPS division. Iwould like to thank my advisor, Harbir Singh,for his guidance and support, and extend gratitudeto my dissertation committee members—RanjayGulati, Lori Rosenkopf, Nicolaj Siggelkow, andSid Winter—for their helpful comments. I amgrateful for the feedback received from SMJ edi-tor Ed Zajac and two anonymous SMJ review-ers. Additional feedback was received from JimFredrickson, Pam Haunschild, Andy Henderson,Mike Hendron, Anita McGahan, and Jim West-phal. Earlier versions of this paper were presentedat the 2004 Academy of Management Conferencein New Orleans and seminars held at the Univer-sity of Texas at Austin, the University of Maryland,the University of California at Irvine, the Tech-nion, the Hebrew University of Jerusalem, and TelAviv University. I am grateful for the financial sup-port received from the Alfred P. Sloan IndustryStudies Fellowship, the Mack Center for Techno-logical Innovation and the Penn Lauder CIBER.This research was also supported by The IsraelScience Foundation (Grant No. 917/07).

REFERENCES

Ahuja G. 2000. Collaboration networks, structural holes,and innovation: a longitudinal study. AdministrativeScience Quarterly 45(3): 425–455.

Allison PD. 2001. Missing Data . Sage: Thousand Oaks,CA.

Anand BN, Khanna T. 2000a. Do firms learn to createvalue? The case of alliances. Strategic ManagementJournal , March Special Issue 21: 295–317.

Anand BN, Khanna T. 2000b. The structure of licensingcontracts. Journal of Industrial Economics 48(1):103–135.

Bacharach SB, Lawler JL. 1984. Bargaining: Power,Tactics, and Outcomes . Jossey-Bass: San Francisco,CA.

Bae J, Gargiulo M. 2004. Partner substitutability, alliancenetwork structure, and firm profitability in the telecom-munications industry. Academy of Management Jour-nal 47(6): 843–859.

Baker WE. 1990. Market networks and corporate behav-ior. American Journal of Sociology 96: 589–625.

Balakrishnan S, Koza MP. 1993. Information asymmetry,adverse selection and joint-ventures. Journal ofEconomic Behavior and Organization 20(1): 99–117.

Barkema H, Shenkar O, Vermeulen F, Bell J. 1997.Working abroad, working with others: how firms learnto operate international joint ventures. Academy ofManagement Journal 40: 426–442.

Barney J. 1991. Firm resources and sustained competitiveadvantage. Journal of Management 17(1): 99–120.

Baum JAC, Calabrese T, Silverman BS. 2000. Don’tgo it alone: alliance network composition andstartups’ performance in Canadian biotechnology.Strategic Management Journal , March Special Issue21: 267–294.

Beck JC, Larson AB, Pinegar JM. 1996. The wealtheffects of non-equity alliances: the U.S.–Japaneselicensing experience. Pacific-Basin Finance Journal 4:393–408.

Berg SV, Duncan J, Friedman P. 1982. Joint VentureStrategies and Corporate Innovation . Oelgeschlager,Gunn & Hain: Cambridge, MA.

Bonacich P. 1987. Power and centrality: a familyof measures. American Journal of Sociology 92:1170–1182.

Borgatti SP, Everett MG, Freeman LC. 2002. UCINET6 for Windows: Software for Social Network Analysis(6.24 edn). Analytic Technologies: Needham, MA.

Brandenburger A, Nalebuff BJ. 1996. Co-opetition .Doubleday: New York.

Brush TH, Bromiley P, Hendrickx M. 2000. The freecash flow hypothesis for sales growth and firmperformance. Strategic Management Journal 21(4):455–472.

Burt RS. 1976. Positions in networks. Social Forces 55:93–122.

Burt RS. 1983. Corporate Profits and Cooptation:Networks of Market Constraints and Directorate Tiesin the American Economy . Academic Press: NewYork.

Burt RS. 1987. Social contagion and innovation: cohesionversus structural equivalence. American Journal ofSociology 92: 1287–1335.

Burt RS. 1992. Structural Holes: The Social Structure ofCompetition . Harvard University Press: Cambridge,MA.

Burt RS. 2000. The network structure of social capital.Research in Organizational Behavior 22: 345–423.

Chan SH, Kensinger JW, Keown AJ, Martin JD. 1997.Do strategic alliances create value? Journal ofFinancial Economics 46(2): 199–221.

Chang S-C, Chen S-S. 2002. The wealth effect ofdomestic joint ventures: evidence from Taiwan.Journal of Business Finance and Accounting 29(1–2):201–222.

Coleman JS. 1988. Social capital in the creation of humancapital. American Journal of Sociology 94: S95–S120.


1210 D. Lavie

Coleman JS. 1990. Foundations of Social Theory .Harvard University Press: Cambridge, MA.

Das S, Sen PK, Sengupta S. 1998. Impact of strategicalliances on firm valuation. Academy of ManagementJournal 41(1): 27–41.

Dierickx I, Cool K. 1989. Asset stock accumulation andsustainability of competitive advantage. ManagementScience 35: 1504–1513.

Dussauge P, Garrette B, Mitchell W. 2000. Learningfrom competing partners: outcomes and durations ofscale and link alliances in Europe, North America andAsia. Strategic Management Journal 21(2): 99–126.

Dyer JH, Kale P, Singh H. forthcoming. Splitting the pie:rent distribution in alliances and networks. Managerialand Decision Economics .

Dyer JH, Nobeoka K. 2000. Creating and managinga high-performance knowledge-sharing network: theToyota case. Strategic Management Journal , MarchSpecial Issue 21: 345–367.

Dyer JH, Singh H. 1998. The relational view: coopera-tive strategies and sources of interorganizational com-petitive advantage. Academy of Management Review23(4): 660–679.

Freeman LC. 1979. Centrality in social networks:conceptual clarification. Social Networks 1: 215–239.

Gimeno J. 2005. Competition within and betweennetworks: the contingent effect of competitiveembeddedness on alliance formation. Academy ofManagement Journal 47(6): 820–842.

Gnyawali DR, Madhavan R. 2001. Cooperative networksand competitive dynamics: a structural embeddednessperspective. Academy of Management Review 26(3):431–445.

Goerzen A, Beamish PW. 2005. The effect of alliancenetwork diversity on multinational enterprise per-formance. Strategic Management Journal 26(4):333–354.

Granovetter M. 1973. The strength of weak ties.American Journal of Sociology 78: 1360–1380.

Granovetter M. 1985. Economic action and socialstructure: the problem of embeddedness. AmericanJournal of Sociology 91: 481–510.

Gulati R. 1995a. Does familiarity breed trust? Theimplications of repeated ties for contractual choices.Academy of Management Journal 35(4): 85–112.

Gulati R. 1995b. Social structure and alliance formationpatterns: a longitudinal analysis. AdministrativeScience Quarterly 40(4): 619–652.

Gulati R. 1998. Alliances and networks. StrategicManagement Journal , April Special Issue 19:293–317.

Gulati R. 1999. Network location and learning: theinfluence of network resources and firm capabilitieson alliance formation. Strategic Management Journal20(5): 397–420.

Gulati R. 2007. Managing Network Resources: Alliances,Affiliations, and Other Relational Assets . OxfordUniversity Press: New York.

Gulati R, Higgins MC. 2003. Which ties matter when?The contingent effects of interorganizational partner-ships on IPO success. Strategic Management Journal24(2): 127–144.

Gulati R, Nohria N, Zaheer A. 2000. Strategic networks.Strategic Management Journal , March Special Issue21: 203–215.

Gulati R, Wang LO. 2003. Size of the pie and shareof the pie: implications of structural embeddednessfor value creation and value appropriation in jointventures. Research in the Sociology of Organizations20: 209–242.

Hagedoorn J. 1993. Understanding the rationale ofstrategic technology partnering: interorganizationalmodes of cooperation and sectoral differences.Strategic Management Journal 14(5): 371–385.

Hagedoorn J, Schakenraad J. 1994. The effect of strategictechnology alliances on company performance.Strategic Management Journal 15(4): 291–309.

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

Harrigan KR. 1988. Joint ventures and competitive strat-egy. Strategic Management Journal 9(2): 141–158.

Heider F. 1958. The Psychology of InterpersonalRelations . Wiley: New York.

Hite JM, Hesterly WS. 2001. The evolution of firmnetworks: from emergence to early growth of the firm.Strategic Management Journal 22(3): 275–286.

Hsiao C. 1986. Analysis of Panel Data . CambridgeUniversity Press: New York.

Ibarra H. 1993. Network centrality, power, and inno-vation involvement: determinants of technical andadministrative roles. Academy of Management Journal36: 471–501.

Inkpen AC, Beamish PW. 1997. Knowledge, bargainingpower, and the instability of international jointventures. Academy of Management Review 22(1):177–202.

Inkpen AC, Tsang EWK. 2005. Social capital, networks,and knowledge transfer. Academy of ManagementReview 30(1): 146–165.

Kale P, Dyer JH, Singh H. 2002. Alliance capability,stock market response, and long-term alliance success:the role of the alliance function. Strategic ManagementJournal 23(8): 747–767.

Kale P, Singh H, Perlmutter H. 2000. Learning andprotection of proprietary assets in strategic alliances:building relational capital. Strategic ManagementJournal , March Special Issue 21: 217–237.

Khanna T, Gulati R, Nohria N. 1998. The dynamicsof learning alliances: competition, cooperation, andrelative scope. Strategic Management Journal 19(3):193–210.

Kogut B. 2000. The network as knowledge: generativerules and the emergence of structure. StrategicManagement Journal , March Special Issue 21:405–425.

Kogut B, Kulatilaka N. 1993. Operating flexibility, globalmanufacturing, and the option value of a multinationalnetwork. Management Science 39(11): 123–139.

Koh J, Venkatraman N. 1991. Joint venture formationsand stock market reactions: an assessment inthe information technology sector. Academy ofManagement Journal 34(4): 869–892.



Koka BR, Prescott JE. 2002. Strategic alliances associal capital: a multidimensional view. StrategicManagement Journal 23(9): 795–816.

Koza MP, Lewin AY. 1998. The co-evolution of strategicalliances. Organization Science 9(3): 255–264.

Kumar R, Nti KO. 1998. Differential learning andinteraction in alliance dynamics: a process andoutcome discrepancy model. Organization Science9(3): 356–367.

Lavie D. 2004. The evolution and strategy of intercon-nected firms: a study of the Unisys alliance net-work. In Proceedings of the 63rd Annual Meeting ofthe Academy of Management, Nagao DH (ed), NewOrleans, LA.

Lavie D. 2006. The competitive advantage of intercon-nected firms: an extension of the resource-based view.Academy of Management Review 31(3): 638–658.

Lavie D, Rosenkopf L. 2006. Balancing explorationand exploitation in alliance formation. Academy ofManagement Journal 49(4): 797–818.

Lee I, Wyatt SB. 1990. The effects of international jointventures on shareholder wealth. Financial Review25(4): 641–649.

Madhok A, Tallman SB. 1998. Resources, transactionsand rents: managing value through interfirm col-laborative relationships. Organization Science 9(3):326–339.

McConnell JJ, Nantell TJ. 1985. Corporate combinationsand common stock returns: the case of joint ventures.Journal of Finance 40(2): 519–536.

Merchant H, Schendel D. 2000. How do internationaljoint ventures create shareholder value? StrategicManagement Journal 21(7): 723–737.

Meyer MM. 2002. Rethinking Performance Measure-ment: Beyond the Balanced Scorecard . CambridgeUniversity Press: Cambridge, U.K.

Montgomery CA. 1982. The measurement of firm diver-sification: some new empirical evidence. Academy ofManagement Journal 25(2): 299–307.

Mowery DC, Oxley JE, Silverman BS. 1998. Technolog-ical overlap and interfirm cooperation: implications forthe resource-based view of the firm. Research Policy27(5): 507–523.

Mundlak Y. 1978. On the pooling of time series and crosssection data. Econometrica 46(1): 69–85.

Nahapiet J, Ghoshal S. 1998. Social capital, intellectualcapital, and the organizational advantage. Academy ofManagement Review 23: 242–266.

Ohmae K. 1989. The global logic of strategic alliances.Harvard Business Review 67(2): 143–157.

Osborn RN, Hagedoorn J. 1997. The institutionalizationand evolutionary dynamics of interorganizationalalliances and networks. Academy of ManagementJournal 40(2): 261–278.

Park SH, Kim D. 1997. Market valuation of jointventures: joint venture characteristics and wealthgains. Journal of Business Venturing 12(2): 83–108.

Park SH, Russo MV. 1996. When competition eclipsescooperation: an event history analysis of joint venturefailure. Management Science 42(6): 875–890.

Parkhe A. 1991. Interfirm diversity, organizationallearning, and longevity in global strategic alliances.

Journal of International Business Studies 22(4):579–601.

Parkhe A. 1993. Strategic alliance structuring: a gametheoretic and transaction cost examination of interfirmcooperation. Academy of Management Journal 36(4):794–829.

Penrose ET. 1959. The Theory of the Growth of the Firm .Oxford University Press: Oxford, U.K.

Peteraf MA. 1993. The cornerstones of competitiveadvantage: a resource-based view. Strategic Manage-ment Journal 14(3): 179–191.

Pfeffer J, Salancik GR. 1978. The External Control ofOrganizations: A Resource Dependence Perspective.Harper & Row: New York.

Pisano G. 1990. The R&D boundaries of the firm: anempirical analysis. Administrative Science Quarterly33: 153–176.

Podolny JM. 1993. A status-based model of marketcompetition. American Journal of Sociology 98:829–872.

Podolny JM. 1994. Market uncertainty and the socialcharacter of economic exchange. AdministrativeScience Quarterly 39: 458–483.

Podolny JM, Stuart TE, Hannan MT. 1996. Networks,knowledge, and niches: competition in the worldwidesemiconductor industry, 1984–1991. American Jour-nal of Sociology 102: 659–689.

Porter ME. 1980. Competitive strategy: Techniques forAnalyzing Industries and Competitors . Free Press:New York.

Porter ME. 1985. Competitive Advantage: Creating andSustaining Superior Performance. Free Press: NewYork.

Portes A. 1998. Social capital: its origins and applicationsin modern sociology. Annual Review of Sociology 24:1–24.

Powell W. 1990. Neither market nor hierarchy: networkforms of organization. Research in OrganizationalBehavior 12: 295–336.

Powell WW, Koput KW, Smith-Doerr L. 1996. Interor-ganizational collaboration and the locus of innovation:networks of learning in biotechnology. AdministrativeScience Quarterly 41(1): 116–145.

Reuer JJ, Koza MP. 2000. Asymmetric information andjoint venture performance: theory and evidence fordomestic and international joint ventures. StrategicManagement Journal 21(1): 81–88.

Rothaermel FT. 2001. Incumbent’s advantage throughexploiting complementary assets via interfirm cooper-ation. Strategic Management Journal , June–July Spe-cial Issue 22: 687–699.

Rothaermel FT, Deeds DL. 2004. Exploration andexploitation alliances in biotechnology: a systemof new product development. Strategic ManagementJournal 25(3): 201–221.

Rowley T, Behrens D, Krackhardt D. 2000. Redundantgovernance structures: an analysis of structural andrelational embeddedness in the steel and semiconduc-tor industries. Strategic Management Journal , MarchSpecial Issue 21: 369–386.


1212 D. Lavie

Rudy J. 2000. Standard & Poor’s Industry Sur-veys—Computers: Software. Standard & Poor’s Cor-poration: New York; 1–38.

Saxton T. 1997. The effects of partner and relationshipcharacteristics on alliance outcomes. Academy ofManagement Journal 40(2): 443–461.

Shipilov A. 2006. Network strategies and performance ofCanadian investment banks. Academy of ManagementJournal 49(3): 590–604.

Silverman BS, Baum JAC. 2002. Alliance-based compet-itive dynamics. Academy of Management Journal 45:791–806.

Simonin BL. 1999. Ambiguity and the process ofknowledge transfer in strategic alliances. StrategicManagement Journal 20(7): 595–623.

Stuart TE. 2000. Interorganizational alliances and theperformance of firms: a study of growth andinnovation rates in a high-technology industry.Strategic Management Journal 21(8): 791–811.

Stuart TE, Hoang H, Hybels R. 1999. Interorgani-zational endorsements and the performance ofentrepreneurial ventures. Administrative Science Quar-terly 44: 315–349.

Teece DJ. 1986. Profiting from technological innovations:implications for integration, collaboration, licensing,and public policy. Research Policy 15: 285–303.

Tsai W, Ghoshal S. 1998. Social capital and valuecreation: the role of intrafirm networks. Academy ofManagement Journal 41(4): 464–476.

Uzzi B. 1996. The sources and consequences of embed-dedness for the economic performance of organi-zations: the network effect. American SociologicalReview 61: 674–698.

Venkatraman N, Lee C-H. 2004. Preferential linkage andnetwork evolution: a conceptual model and empiricaltest in the U.S. video game sector. Academy ofManagement Journal 47(6): 876–892.

Wasserman S, Faust K. 1994. Social Network Analysis:Methods and Applications . Cambridge UniversityPress: Cambridge, U.K.

Weigelt K, Camerer C. 1988. Reputation and corporatestrategy: a review of recent theory and applications.Strategic Management Journal 9(5): 443–454.

Wernerfelt B. 1984. A resource-based view of the firm.Strategic Management Journal 5(2): 171–180.

Williamson OE. 1975. Markets and Hierarchies, Analysisand Antitrust Implications: A Study in the Economicsof Internal Organization . Free Press: New York.

Woolridge JR, Snow CC. 1990. Stock market reaction tostrategic investment decisions. Strategic ManagementJournal 11(5): 353–363.

Yan A, Gray B. 1994. Bargaining power, managementcontrol, and performance in United States–Chinesejoint ventures: a comparative case study. Academy ofManagement Journal 37(6): 1478–1517.

Zaheer A, Bell GG. 2005. Benefiting from networkposition: firm capabilities, structural holes, andperformance. Strategic Management Journal 26(9):809–826.

Zaheer A, McEvily B, Perrone V. 1998. Does trustmatter? Exploring the effects of interorganizationaland interpersonal trust on performance. OrganizationScience 9(2): 141–159.

APPENDIX

First-stage robust probit model for probability ofnon-missing values in model 8

First-stageModel 8

Intercept −1.87∗∗∗

Year 1990 0.30Year 1991 0.29Year 1992 0.31Year 1993 0.17Year 1994 −0.20Year 1995 −0.34∗

Year 1996 −0.29∗

Year 1997 −0.25∗

Year 1998 −0.25∗

Year 1999 −0.09Year 2000Firm SIC 7371 0.58Firm SIC 7372 1.25∗∗∗

Firm SIC 7373 0.77∗

Firm SIC 7374Firm size −0.00Firm age 0.00Firm partnering experience 0.01†

Number of partners 0.03∗∗

Publicly traded partnerss 0.33Portfolio age 0.07Portfolio internationalization −0.36Joint ventures −0.48Strategic alliances −0.04∗

Multi-partner alliances 0.09Licensing agreements −0.07Manufacturing agreements −0.06Sales and service agreements −0.02OEM/VAR agreements 0.48∗∗

R&D agreements −1.12∗

Royalties agreements −0.75Supply agreements 0.48Number of observations 2049Number of observations with non-missing

values1343

Number of firms 360Wald χ 2 (d.f. = 30) 158.49∗∗∗

Pseudo R2 0.23

†p < 0.1; ∗ p < 0.05; ∗∗ p < 0.01; ∗∗∗ p < 0.001.


Documents

ALLIANCE PORTFOLIOS AND FIRM PERFORMANCE: A ......Alliance Portfolios and Firm Performance 1189 endowed partners may in fact undermine the mar-ket performance of ﬁrms. By distinguishing