￠0530￡When the Social Structure Overshadows Competitive Incentives

WHEN THE SOCIAL STRUCTURE OVERSHADOWS COMPETITIVE INCENTIVES:THE EFFECTS OF NETWORK EMBEDDEDNESS ON JOINT VENTURE DISSOLUTION

FRANCISCO POLIDORO J R .University of Texas at Austin GAUTAM AHUJA University of Michigan WILL MITCHELL Duke University

The embeddedness of interfirm relationships in a social structure can engender order in new tie formation,but competitive incentives may undermine the order that firms seek to achieve and lead to tie dissolution.We examine how relational embeddedness (history of interactions),positional embeddedness (network centrality),and structural embeddedness (common partners)influence tie stability,focusing on unplanned joint venture dissolution.Prior work suggests that relational embeddedness facilitates alli-ance stability.This study shows that positional embeddedness does not promote stability,but structural embeddedness does help sustain alliances,particularly when partners have strong incentives to pursue self-interest at the expense of joint benefits.

The network structure that encompasses firms’commercial alliances creates both potential strate-gic benefits and potential risks for collaborative activities.Extant literature shows that the embed-ding of alliance activity in a preexisting network structure when selecting partners produces infor-mational,reputational,and social monitoring ben-efits that help firms mitigate collaboration hazards (Gulati &Gargiulo,1999;Hitt,Dacin,Levitas,Ar-regle,&Borza,2000;Walker,Kogut,&Shan,1997).However,firms face obstacles to effective collabo-ration even after they create seemingly desirable linkages (Brass,Galaskiewicz,Greve,&Tsai,2004;Gulati,Nohria,&Zaheer,2000).Network embed-dedness is relevant to alliance success because,in addition to assisting partner selection,embedded-ness promotes cohesion between partners during the course of collaboration (Kogut,1989;Levinthal &Fichman,1988;Park &Russo,1996).In many cases,though,the frictions that occur during alli-ance life are so severe that the cohesive effects of embeddedness are not sufficient to ensure alliance stability (Greve,Baum,Mitsuhashi,&Rowley,2010).Despite researchers’growing understanding of the cohesive and frictional forces affecting alli-ance longevity,the literature remains silent about

the reasons that many alliances experience fric-tions if embeddedness helps firms avoid forming such alliances in the first place.Further,research has not examined the possibility that the cohesive effects of embeddedness are stronger in relation-ships that are more vulnerable to collaboration problems.

To further understanding of how network em-beddedness influences alliance instability,we examine unplanned joint venture dissolution.Al-though partners invest significant relationship-spe-cific resources in the formation of a joint venture with the goal of reaping joint benefits (Gulati,1995a;Rowley,Behrens,&Krackhardt,2000),firms have incentives to compete with their partners,which can eclipse the benefits of collaboration and lead to joint venture dissolution (Kogut,1989;Park &Russo,1996;Park &Ungson,1997).Partners may dissolve a joint venture for a variety of reasons,including the achievement of their goals or the demands of antitrust authorities.We focus explic-itly on unplanned joint venture dissolutions that occur because of problems arising during the course of the alliances.Such problems are particu-larly likely to occur when partners are close com-petitors (Park &Russo,1996;Park &Ungson,1997),when they perceive inequity in contributions to their venture (Arino &Torre,1998),or when changes in the market make the venture more at-tractive for one of the partners (Kogut,1991;Li,

We gratefully acknowledge the valuable comments and suggestions by Wenpin Tsai and three anonymous reviewers.

?Academy of Management Journal 2011,Vol.54,No.1,203–223.

203

Copyright of the Academy of Management,all rights reserved.Contents may not be copied,emailed,posted to a listserv,or otherwise transmitted without the copyright holder’s express written https://www.wendangku.net/doc/7a4384519.html,ers may print,download or email articles for individual use only.

Dhanaraj,&Shockley,2008).In our empirical methods section,we detail the procedure we used to identify unplanned joint venture dissolution.To date,the literature on alliance dissolution has focused primarily on relational embeddedness 1—that is,on how the history of interactions between two partners influences the quality and strength of their direct ties (Moran,2005;Nahapiet &Ghoshal,1998).Direct ties provide partners with superior information about each other,engender knowl-edge-based trust (Gulati,1995b;Uzzi,1997;Zaheer &Venkatraman,1995)and,as a result,mitigate tie instability (Kogut,1989;Park &Russo,1996;Uzzi,1996).In this study,in contrast,we consider the influences of the broader network of interfirm alli-ances and investigate how the informational and reputational benefits of positional embeddedness (partners’positions in the network structure [e.g.,Gulati &Gargiulo,1999;Powell,Koput,&Smith-Doerr,1996])and the social monitoring advantages of structural embeddedness (partners that collabo-rating firms have in common [e.g.,Coleman,1988;Gulati,1995b;Walker et al.,1997])affect the hazard of unplanned joint venture dissolution.

When examining how network embeddedness af-fects unplanned joint venture dissolution,this study highlights the trade-off that firms face be-tween incentives to form an alliance and problems that can arise during the life of the alliance.In general,before forming an alliance,a firm can be expected to balance problems that might arise dur-ing collaboration against potential alliance bene-fits and to only form the partnership if the antic-ipated problems are not severe enough to offset the projected benefits.However,this interpreta-tion ignores the possibility that new and only partially predictable circumstances arising dur-ing the course of collaboration may shift the bal-ance of partners’incentives,thereby engendering unexpected instability.

This study shows that many alliances turn out to be unstable because the expectation of deriving substantial benefits sometimes leads firms to create alliances that also have a substantial chance to be unstable.For instance,although the reputational imbalance inherent in asymmetric relationships is a source of instability,firms sometimes form asym-metric partnerships with the expectation of access-ing complementary resources (Ahuja,2000;Gulati &Gargiulo,1999;Mitchell &Singh,1996).Like-

wise,although firms that exhibit greater resource similarity and market commonality have greater propensity to engage in interfirm rivalry (Chen,1996),which is a source of instability,firms some-times ally with partners that have a high level of competitive overlap in order to manage mutual competitive interdependence or enjoy scale bene-fits (Gimeno,2004;Kogut,1988;Park &Ungson,2001;Pfeffer &Nowak,1976).Network embedded-ness becomes relevant during the course of collab-oration because it mitigates collaboration hazards,especially in alliances that have high risk of expe-riencing implementation problems.

This study furthers network research by showing that the effects of network embeddedness on alli-ance dissolution are not simply the mirror image of its effects on alliance formation.Typically,network embeddedness affects alliance formation by indi-cating partnerships that are less likely to experi-ence collaboration problems.In contrast,some forms of network embeddedness influence alliance dissolution because they provide stability to rela-tionships that,although expected to be attractive,are prone to instability.Hence,although examina-tion of alliance formation shows that network embeddedness helps firms minimize potential collaboration problems,examination of alliance dissolution reveals that some types of network em-beddedness enable firms to pursue alliances that have attractive potential benefits despite the pres-ence of obstacles to collaboration.

With respect to research on joint ventures,this study shows how the aggregate network of relation-ships among a set of firms affects joint venture survival,moving this literature beyond the tradi-tional dyadic level of analysis.According to the dyadic perspective taken in prior work,the key social mechanisms for maintaining order in an al-liance are the mutual trust that partners develop and “the shadow of the future”(Axelrod,1984:124),which reflects the benefits of continuing the relationship (Gulati,1995b;Heide &Miner,1992;Parkhe,1993b).By adopting a broader network per-spective,we emphasize the “shadow of others”that arises from structural embeddedness as an addi-tional social mechanism for maintaining order in joint ventures.The shadow of others arises because partners have incentives to behave fairly with allies because of concern for sanctions that common part-ners may impose.Thus,although prior work has shown that competitive tensions increase the haz-ard of joint venture dissolution,this study shows that it is precisely in such high-risk relationships that embedding alliances in a network of com-mon partners most contributes to mitigating this

1

Greve et al.(2010),who examined the influences of positional and structural embeddedness on member withdrawal from interfirm alliances,have provided a notable exception.

204February

Academy of Management Journal

hazard,a finding of both theoretical and practical importance.

NETWORK EMBEDDEDNESS AND JOINT

VENTURE DISSOLUTION

Extant literature has indicated that relational,po-sitional,and structural embeddedness influence al-liance formation(Gulati&Gargiulo,1999)and has also demonstrated that relational embeddedness mitigates tie instability(Broschak,2004;Kogut, 1989;Levinthal&Fichman,1988;Park&Russo, 1996;Uzzi,1996).Hence,though controlling em-pirically for the influence of relational embedded-ness,we focus on how the informational and repu-tational benefits of positional embeddedness(e.g., Gulati&Gargiulo,1999;Powell et al.,1996)and the social monitoring advantages of structural em-beddedness(e.g.,Coleman,1988;Gulati,1995b; Walker et al.,1997)affect the hazard of unplanned joint venture dissolution.

Informational and Reputational Benefits of Positional Embeddedness

Positional embeddedness relates to the extent to which organizations occupy a central position in a network structure(Gulati&Gargiulo,1999).Firms in central network positions have greater scope than noncentral firms for collecting and dissemi-nating information.The information that flows to central firms helps them identify alliance opportu-nities and select partners that possess appropriate resources and demonstrate reliable collaborative behavior(Ahuja,2000;Gulati,1995b).The greater the combined centrality of two firms,the fewer informational constraints they face when searching for partners,which,in turn,suggests that if these firms ally,it is because they knowingly regard each other as appropriate partners.Further,the informa-tional benefits accruing to central firms also en-hance trust between partners(Gulati et al.,2000; Gulati&Singh,1998)and can be thought of as providing an initial stock of trust that enables part-ners to collaborate more effectively during the course of an alliance.

Simply extending the argument proposed in the alliance formation literature to its logical conclu-sion,we argue that the informational benefits that network-central firms possess enhance the stability of the alliance that they form.If two central firms choose each other as partners,they can ex ante expect to have made better partnering choices that will lead to a more stable alliance than partnering with noncentral firms.In other words,if firms’ex ante expectations about the value of their network in helping them choose partners are true,we should expect centrally positioned firms to form more stable alliances than noncentral firms.

The informational benefits of centrality are most relevant when firms are searching for partners and forming joint https://www.wendangku.net/doc/7a4384519.html,rmation gathered after the start of an alliance has less value because it is likely to overlap with information partners al-ready possess.Moreover,even if a firm needs further information about its ally after joint ven-ture founding,its own tie with the partner can provide this information,without reference to network positions.

Hypothesis1a.The greater the combined net-work centrality of two partnering firms at joint venture founding(positional embeddedness), the lower the hazard of unplanned joint ven-ture dissolution at any subsequent point. Relational embeddedness refers to the presence of prior direct ties between two firms,which enable the exchange of information about each other’s ca-pabilities and reliability,enhance mutual trust and, in turn,mitigate the hazard of joint venture disso-lution(Kogut,1989;Park&Russo,1996;Park& Ungson,1997).Although controlling for the direct effect of relational embeddedness,we argue that because relational embeddedness(like positional embeddedness)produces informational benefits, it is likely to moderate the impact of positional embeddedness.

To understand the moderating effect of relational embeddedness,note that in the absence of a prior direct tie between two firms,their current network positions function as partial surrogates for the missing relationship and help them learn about each other through their allies.When two firms have collaborated prior to the formation of an ad-ditional alliance,by contrast,their current network positions convey less novel information because their prior ties allowed them to observe each oth-er’s competencies and behavior directly. Although the information that a firm collects di-rectly from its partner may not fully overlap with the information that stems from positional embed-dedness(Burt,1992;Gulati,1995b),it reduces the marginal value of the information the firm obtains through its network position.Moreover,the knowl-edge-based trust resulting from superior informa-tion that two central partners have prior to forming a joint venture partly overlaps with the knowledge-based trust stemming from prior direct ties.Hence, partners’centrality in the network structure will have less influence on joint venture dissolution when the partners have prior direct ties.

2011205

Polidoro,Ahuja,and Mitchell

Hypothesis1b.The combined network central-ity of two partnering firms at joint venture founding(positional embeddedness)contrib-utes less to mitigating the hazard of unplanned joint venture dissolution when they are con-nected by prior direct ties(relational embeddedness).

In addition to informational benefits,positional embeddedness also offers reputational advantages. Central firms are perceived to be more capable and reliable partners than firms that occupy peripheral network positions(Powell et al.,1996).Centrally positioned firms generally prefer to ally with organ-izations that occupy similar positions to mitigate the hazards of collaboration and to avoid diminish-ing their reputations(Chung,Singh,&Lee,2000; Gulati&Gargiulo,1999;Podolny,1994).Further, the positional imbalance between a central and a peripheral actor creates power imbalance between them(Mizruchi,1982).Although the reputational imbalance inherent in asymmetric relationships is a source of instability,firms sometimes form asym-metric partnerships with the expectation of access-ing complementary resources(Ahuja,2000;Gulati &Gargiulo,1999;Mitchell&Singh,1996).Periph-eral firms,in turn,have incentives to ally with central partners because such collaboration can vouch for their reliability and enhance their repu-tations(Podolny,1994).

Again,a strict interpretation of the instrumental goals associated with reputational imbalance is that if positionally asymmetric partners form an alli-ance,it is because they expect benefits to outweigh potential collaboration problems,yet the fact re-mains that over time new and only partially pre-dictable circumstances can shift the balance of ben-efits that partners can derive from their alliance, thus creating tension between the partners.The more central firm may derive greater private bene-fits than its partner expected because its wider portfolio of relationships,skills,and absorptive ca-pacity allows it to leverage the benefits from the information it receives(Cohen&Levinthal,1990; Zollo,Reuer,&Singh,2002),well beyond what its partner anticipated.Alternatively,the asymmetry might well favor the less embedded partner,be-cause the affiliation ends up providing a reputation enhancement that goes beyond what the central partner predicted,or the less embedded partner uses the affiliation to obtain relationships with ad-ditional relatively central partners(Ahuja,Poli-doro,&Mitchell,2009).

By shifting the balance of benefits that position-ally asymmetric partners can derive from an alli-ance,partners’reputational imbalance at the outset of collaboration increases their tendency to revisit the terms negotiated for the alliance in an attempt to restore exchange balance(Cook,1977;Emerson, 1972)and affects the amount of resources they al-locate to their joint venture(Khanna,Gulati,& Nohria,1998).These factors,along with shifts in bargaining power during the course of the alliance, create alliance instability(Das&Teng,2000;Kogut, 1989).

Hypothesis2.Joint ventures formed between a high-centrality firm and a low-centrality firm (positional asymmetry)experience greater haz-ard of unplanned dissolution than joint ven-tures between firms with similar centrality(po-sitional symmetry).

Social Monitoring Benefits of

Structural Embeddedness

We now turn to structural embeddedness,which refers to partners that collaborating firms have in common(Coleman,1988).The partners common to two organizations encompass the set of direct ties common to both of them.Prior studies on alliance formation have shown that firms that have partners in common exhibit increased propensity to ally with each other(Gulati,1995b;Gulati&Gargiulo,1999). Besides referral advantages,structural embed-dedness also produces social monitoring benefits that help organizations monitor and enforce collab-orative behavior(Coleman,1988).The presence of common partners engenders deterrence-based trust by giving greater visibility to norm-breaking behav-ior.If,in a relationship between two firms,one partner exploits the vulnerabilities of the other,the occurrence of such behavior can be revealed to common partners and,through them,reach a larger number of firms in the network of interfirm collab-oration.The presence of common partners pro-duces social monitoring benefits because it ampli-fies opportunities to sanction norm breaking.Even if the firm that faces opportunism is unable to im-pose sanctions on the ally,common partners may inflict penalties on the norm breaker.The deter-rence-based trust stemming from commonality of partners helps mitigate the incidence of collabora-tion problems between two firms.

An alternative way in which partners can deter opportunistic behavior is by resorting to formal governance structures,such as joint ventures,to govern collaboration.Firms commonly resort to joint ventures to organize collaboration when they perceive high interorganizational hazards at the time they create their alliance(Gulati&Singh, 1998).Because of the joint commitment of re-

206February

Academy of Management Journal

sources and the administrative apparatus required for their operation,joint ventures help align incen-tives between partners(Kogut,1988).The“mutual hostage”positions implied by the shared equity in a joint venture engender deterrence-based trust be-tween partners.

The social monitoring benefits of structural em-beddedness reinforce the deterrence-based trust in-herent in the governance structure of joint ven-tures.Even if partners form a joint venture under the presumption that the benefits will supersede potential collaboration problems,new circum-stances arising during the course of the alliance can shift the balance of partners’incentives and,as a result,engender instability.Structural embedded-ness,however,helps firms deter opportunistic be-havior.If a firm behaves opportunistically in the context of a joint venture,it faces the risk of pun-ishment not only by the venture partner but also by all other partners they have in common. Additionally,common partners may impose reputational damage on the norm-breaking firm and refrain from forming future alliances with that firm(Park&Ungson,1997).The fear of loss of reputation deters firms linked by common partners from behaving opportunistically(Gulati&Gar-giulo,1999;Gulati&Westphal,1999).Hence,com-mon partners increase the cost of opportunism by affecting not only the current ties but also the po-tential future ties of the norm breaker.The joint venture between Kuwait’s Petrochemicals Indus-tries Co.and Dow Chemical illustrates that firms, when responding to collaboration problems,con-sider the effect of such response on their current and future ties.In December2008,Petrochemicals Industries decided to pull out of its joint venture with Dow.Although the contract specified that in such an event Dow Chemical should receive$2.5 billion from the partner,industry observers noted that Dow was reluctant to launch litigation proce-dures because of concern that such behavior could jeopardize its other joint ventures and its attempts to strengthen ties with the Kuwaiti investment au-thorities(National Post,2009;Times,2008). Whereas informational and reputational benefits of positional embeddedness are greatest at the time of joint venture founding,by ensuring that firms ally with the right partners,the social monitoring benefits of structural embeddedness can accrue at any point of network evolution.When two firms create a joint venture,the partners they have in common confer deterrent effects.During the course of the alliance,as the number of common partners linking the two firms increases,the social monitor-ing effects become stronger.Additional common partners enhance firms’embeddedness in local net-works(Gulati&Gargiulo,1999),reinforce firms’concern for local reputation and,as a consequence, further mitigate interorganizational hazards. Hypothesis3.The more common partners that two firms forming a joint venture possess at any point of network evolution(structural em-beddedness),the less the hazard of unplanned joint venture dissolution.

Because the presence of common partners in-creases the costs of opportunism,we expect the deterrent benefits that common partners provide to be more pronounced when the edge of incentives to behave opportunistically is especially sharp.Firms sometimes form alliances that are prone to instabil-ity because the formation incentives go hand in hand with factors that increase the chance of shifts in the balance of partners’incentives,which is a source of instability.For instance,as we argued earlier,although firms form positionally asym-metric ties with the expectation of accessing complementary resources,partners’reputational imbalance increases their propensity to behave opportunistically after alliance formation.Like-wise,although firms have incentives to ally with partners exhibiting competitive overlap in order to manage mutual competitive interdependence(Gi-meno,2004;Kogut,1988;Park&Ungson,2001; Pfeffer&Nowak,1976),partners exhibiting com-petitive overlap are more likely to let competitive incentives overrule the benefits of collaboration (Chen,1996;Kogut,1989;Park&Ungson,1997). We argue that the deterrent effects of common part-ners mitigate collaboration hazards especially in these types of alliances that,despite the expected benefits,have high risk of experiencing collabora-tion problems.

Common partners and positional asymmetry. As Hypothesis2predicts,partners’reputational imbalance at joint venture founding increases the chances that they will experience collaboration problems during the course of their alliance,which in turn exacerbate the importance of common part-ners as a stabilizing mechanism.When positionally symmetric firms create a tie,their structural simi-larity and relatively congruent attendant routines, processes,and objectives provide a degree of sta-bility to the alliance.But when asymmetric part-ners form a tie,fewer of these stability-enhancing factors are present,which in turn enhances the value of common partners as mechanisms of stabil-ity.Hence,in joint ventures created between posi-tionally asymmetric partners,the norms and repu-tational pressures associated with the presence of common partners in the ongoing network structure become more relevant to mitigating collaboration

2011207

Polidoro,Ahuja,and Mitchell

problems than they are in joint ventures formed between positionally symmetric partners,which are less likely to experience collaboration hazards in the first place.

Hypothesis 4.The number of common part-ners between two joint venture partners (structural embeddedness)contributes more to mitigating the hazard of unplanned joint venture dissolution when one firm has high centrality and the other has low centrality (positional asymmetry)than when firms have similar centrality (positional symmetry).Common partners and competitive overlap.Re-searchers have argued that interfirm ties enhance the likelihood of coordination,tacit or otherwise,and has shown that firms with high levels of com-petitive overlap form joint ventures in an attempt to manage their competitive interdependence (Kogut,1988;Pfeffer &Nowak,1976).However,the attempt to manage mutual competitive intensity is both an incentive to form joint ventures and a source of instability (Kogut,1989;Park &Russo,1996).Firms that exhibit greater resource similarity and market commonality have high propensity to engage in interfirm rivalry (Chen,1996)and represent a com-petitive threat to each other (Pfeffer &Nowak,1976).They can,for example,enter or reinforce their presence in markets in which competitors

operate and,by doing so,affect each other’s perfor-mance and viability (Barnett,1997).

Some scholars have argued that firms with high levels of competitive overlap,fearing retaliation,may curb competitive actions and engage in less vigorous competition (Gimeno &Woo,1996;Kar-nani &Wernerfelt,1985).However,the stability of such mutual forbearance is fragile in the face of competitive incentives (Baum &Korn,1999;Sti-gler,1964).When partners exhibit high competitive overlap,there is a high probability that one will learn about the other’s vulnerabilities and encoun-ter opportunities to exploit those weaknesses.Di-rect retaliation between firms with high competi-tive overlap may lead to escalation of mutual rivalry and become either ineffective or overly costly.As a result,enforcement through common partners becomes even more important in relation-ships between firms that have high competitive overlap.

Hypothesis 5.The number of common partners between two joint venture partners (structural embeddedness)contributes more to mitigating the hazard of unplanned joint venture dissolu-tion when the firms have high competitive overlap than when they have low competitive overlap.

Figure 1summarizes our hypotheses about the

FIGURE 1

Influences of Network Embeddedness on Unplanned Joint Venture Dissolution

Relationsh ips discussed in prior work Relationships examined in this study

208February

Academy of Management Journal

influence of different aspects of network embed-dedness on unplanned joint venture dissolution.

METHODS

We used data on the duration of technology-related joint ventures formed among the leading firms in the global chemicals industry.Although joint ventures constitute formal interfirm ties (Contractor&Lorange,1988;Gulati&Singh, 1998)that require significant relationship-spe-cific investments and imply high exit costs(Gu-lati,1995a;Rowley et al.,2000),partners often dissolve alliances before reaping the joint bene-fits(Kogut,1989;Park&Russo,1996;Park& Ungson,1997).Joint ventures involving technol-ogy are especially vulnerable to opportunism (Gulati,1995a;Gulati&Singh,1998)and,hence, constitute an appropriate context for examining the influence of network embeddedness on un-planned alliance dissolution.

Using trade journals such as Chemical Week and C&E News,we identified a primary sample of107 firms from Western Europe,Japan,and the United States—the core of the global chemicals industry. Because collecting reliable,comprehensive data on joint venture formation and dissolution among smaller firms is difficult,past network studies have used a similar strategy of focusing on the leading firms in an industry(Gulati,1995b;Rosenkopf,Me-tiu,&George,2001).Reliable financial and other data were not available for10firms,limiting the analysis to97firms.

We collected data on the joint venture activity of the97companies between1979and1991.The period of analysis started in1979because that is when news media databases began to systemati-cally report data on interfirm collaboration(Gulati, 1995b;Hagedoorn&Schakenraad,1989).Prior studies have suggested that alliance activity was low until the early1980s(Gulati&Gargiulo,1999; Hergert&Morris,1988).

We adopted a two-stage data collection proce-dure.In the first stage,we identified joint ventures between firms in the sample that included the cre-ation of new technologies or the sharing of an ex-isting technology or manufacturing process by the partners.To do so,we used(1)general business news media electronic databases,such as Dow Jones News Retrieval Text Index and Lexis-Nexis, and sector-specific electronic databases,such as METADEX,(2)general business print media,such as the Frost and Sullivan Predicasts index,as well as industry-specific publications such as Chemical Week and C&E News,and(3)government publica-tions and consultants’reports for the industry over the relevant period.In the second stage,we searched the above media to collect all relevant news stories about each joint venture.We analyzed news stories dating until2006to ascertain whether the joint ventures survived beyond1992.These two stages involved studying over130,000news stories and took several years.

Although we had data on joint ventures formed since1979,we needed to use data about the first few years to construct the network emerging from prior joint ventures to capture the network struc-ture at joint venture founding.We used data on joint ventures formed between1979and1982to construct a baseline network.We then identified 201dyadic ties resulting from168joint ventures formed between1983and1991and examined the hazard of unplanned dissolution of these ties be-tween1984and1992.

From the search described above,we unambigu-ously identified the dissolution of36dyads and the survival beyond1992of128dyads.To establish either the date of dissolution or survival beyond 1992for the other37dyads,we contacted the com-panies and obtained information on13additional dyads,all of which survived beyond1992.For the remaining24dyads,we were unable to unambigu-ously determine survival or dissolution.Given our relatively high rate of success in identifying disso-lution,we believe that the most accurate assump-tion is that these24dyads survived beyond1992; otherwise,their dissolution would have been re-ported.As we report later,we obtained robust re-sults when we tested the sensitivity of the analysis to this assumption.

A joint venture ceases to exist when partners liquidate it,when one firm acquires the stake of the partner,or when partners sell their stake to a third party(Dussauge,Garrette,&Mitchell,2000;Kogut, 1989;Park&Russo,1996).Although dissolution represents only one form of alliance performance,it usually reflects a business failure or irresolvable conflict between partners(Gehringer&Herbert, 1989;Kogut,1989).To ascertain whether the cases of dissolution in the sample reflected collaboration problems,we examined the news stories about each dissolved joint venture.In a few cases,disso-lution reflected goal attainment or changes in the business landscape unrelated to partners’opportu-nistic behavior.We did not consider these cases to be unplanned dissolution,as they do not reflect the types of interfirm hazards we discuss above.In the vast majority of cases,unplanned joint venture dis-solution reflected either partners’difficulty in col-laborating or an attempt by one of the partners to exploit a particular business opportunity alone

2011209

Polidoro,Ahuja,and Mitchell

rather than share the benefits.Appendix A presents examples of joint venture dissolution.Dependent Variable

The dependent variable in our study is the haz-ard of unplanned dissolution of a joint venture by two firms in a given year.We created a dummy variable for each dyad in each year until the year of dissolution or until 1992,whichever occurred first;this variable was coded 1if the firms dissolved the joint venture in that year and 0otherwise.Our analysis of the reasons leading to joint venture dis-solution enabled us to identify 31cases of un-planned dissolution.2The average duration of joint venture dyads that experienced unplanned disso-lution was 3.5years,with a standard deviation of 2.2years.

Independent Variables

Combined centrality at founding (positional embeddedness).In line with prior studies,we measured each firm’s position in the network struc-ture in which it was embedded (i.e.,the aggregate network of joint ventures among a set of partnering firms)using Bonacich’s (1987)eigenvector central-ity scores (e.g.,Chung et al.,2000;Gulati &Gar-giulo,1999;Podolny,1994).This measure suits our analysis because it takes into account that a firm has greater scope for collecting and disseminating information when it is linked to a greater number of partners and when partners are themselves well connected in a network structure—that is,this mea-sure results in higher centrality scores for firms that are linked to many firms,which are in turn linked to many other firms.The combined centrality of the two firms in a dyad is the geometric mean of the two firms’scores in a given joint venture founding year (Mizruchi,1993),each firm’s score being nor-malized relative to the most central firm in that year (Gulati &Gargiulo,1999).Our findings are robust to using the arithmetic mean of the central-ity scores.We computed the centrality scores using UCINET 5(Borgatti,Everett,&Freeman,1999).To test Hypothesis 1b,we split observations into two groups,one with,and the other without,prior direct ties.

Positionally asymmetric dyads at founding (positional asymmetry).Interorganizational net-works display a pattern of core-periphery struc-tures (Gulati &Gargiulo,1999),and firms’behavior is affected by whether they occupy central or pe-ripheral network positions (Mizruchi,1993).Repu-tational imbalance arises when partners occupy different status brackets—that is,when one partner belongs to a high-status group while the other is a member of a low-status group (Podolny,1993).We adopted Mizruchi’s (1992)procedure to identify dyads between a central and a peripheral firm.For each dyad we coded a dummy variable as 1if one firm in the dyad had a centrality score lower than the mean and the other had centrality equal to or greater than the mean observed in the founding year,and as 0otherwise.We obtained similar re-sults when defining central firms as those with centrality scores in the top quartile.

Common partners in previous year (structural embeddedness).We measured the number of com-mon partners two firms shared in the evolving al-liance network as the number of common partners they had in the year before the observation year (Gulati &Gargiulo,1999;Mizruchi,1992).To test Hypotheses 4and 5,we interacted this variable with the measures capturing positionally asymmet-ric dyads at founding and competitive overlap,respectively.

Control Variables

Prior research has shown that relational embed-dedness between partners reduces the hazard of joint venture dissolution (Kogut,1989;Park &Russo,1996;Park &Ungson,1997).We controlled for this influence with a variable that contains the cumulative number of prior joint ventures formed between two partners until the year preceding a focal joint venture’s founding.We also added a variable with the number of ties that partners formed since the founding of the focal joint venture to control for the possibility that firms may need to dissolve the focal joint venture to free up re-sources and use them to seize emerging alliance opportunities.

Because firms exhibit high propensity to engage in mutual competitive behavior when they have a high degree of similarity in both the resources that they possess and the markets where they operate (Chen,1996),we measured competitive overlap as a dummy variable set to 1if the firms in the dyad had high technical resource overlap and high mar-

2

Additionally,in 5cases we were unable to identify whether the dissolution was planned or unplanned.Given that the vast majority of dissolutions were un-planned,we assumed these cases to reflect unplanned dissolutions.Sensitivity analyses considering these 5cases as planned dissolutions produced similar results.

210February

Academy of Management Journal

ket overlap,and to 0otherwise.3We considered overlap along each dimension to be high if the respective score for firms in the dyad was above the sample mean.Sensitivity analysis yielded robust results using above-the-median and top-quartile measures of high overlap.

To measure firms’overlap along each dimension in a given year,we used the Euclidean distance between them (Sohn,2001),which takes the value (?k ?1,n [P ikt –P jkt ]2)1/2,where P ikt (P jkt )captures the proportion of firm i ’s (firm j ’s)resources in each of n areas where resources might overlap.We used patent data to measure firms’technical resources (Griliches,1990;Hall,Jaffe,&Trajtenberg,2001).To measure technical resource overlap,we used the proportion of firms’patent applications in each of the 80technological classes that chemical compa-nies use.To measure overlap of market resources,we considered the average of firms’overlap in both geographic and product markets.For geographic overlap,we used the proportion of subsidiaries that firms owned in each of 156countries,and for prod-uct-market overlap,we used the proportion of sales they obtained in each of 120market segments de-fined at the level of four-digit SIC codes.These measures vary between 0(complete overlap)and the square root of 2(no overlap).To facilitate the interpretation of results,we linearly transformed these measures and subtracted them from the square root of 2.

Several variables address other possible influ-ences.We controlled for the possibility that part-nering firms’ages affect the hazard of joint venture dissolution (Kim,Oh,&Swaminathan,2006)by including a variable with the difference between the firms’founding years (Park &Ungson,1997).To control for the possibility that the total amount of firms’technical resources affects the stability of their relationship,we added controls for firms’combined R&D expenditures and patents in the chemical industry.Further,we included measures to account for the possibility that differences in firms’financial performance or financial resources affect their propensity to dissolve a joint venture.We calculated performance asymmetry by subtract-ing the lesser from the greater value of return on assets (ROA)within each dyad.To measure asym-metry in liquidity (current assets/current liabili-ties)and in debt-equity (long term debt/share-holder equity),we used the ratio of the lesser to the greater value within a dyad.We included a dummy variable to control for cultural differences (Barkema,Shenkar,Vermeulen,&Bell,1997;Park &Ungson,1997;Parkhe,1993a).This variable equals 1when parent firms are based in different regions (U.S.,Europe,or Japan)and 0otherwise.We lagged all control variables.

Table 1reports descriptive statistics and the cor-relation matrix for the variables.

Model Estimation and Econometric Issues Because the time elapsed since tie formation may influence the hazard of tie dissolution (Levinthal &Fichman,1988),we chose a parametric survival function that enables the baseline hazard function to vary as a function of time and then estimates the influences of the covariates as multiplicative ef-fects that either increase or decrease the baseline hazard.We used the Weibull distribution,which accounts for a monotonic effect of time (Barkema et al.,1997;Dussauge et al.,2000;Park &Russo,1996).Our results are robust to the lognormal and gamma distributions that prior work has used as alternatives (e.g.,Dussauge et al.,2000;Park &Russo,1996).We used robust standard errors ad-justed for clustering at the joint venture level to allow for nonindependence of observations and correct for groupwise heteroskedasticity (Greene,2003).

RESULTS

Table 2presents the results of the regression analyses.The natural logarithm of the shape pa-rameter of the Weibull distribution is statistically significant in all models,showing that the hazard of dissolution increases monotonically with time elapsed since joint venture founding.Model 1con-tains the control variables;model 2introduces the main effects;and model 3adds the interactions.Hypothesis 1a predicts that firms’combined cen-trality at joint venture founding reduces the hazard of unplanned dissolution,and Hypothesis 1b pre-dicts that this effect is smaller when firms have prior direct ties.The logic drew on arguments about the informational benefits of positional em-beddedness.In contrast to the predictions,how-ever,model 3shows that firms’centrality does not affect unplanned joint venture dissolution when partners are forming their first direct tie (??1.47,n.s.)but increases the hazard of dissolution when partners have prior direct ties (??12.58,p ?.001).

3

Results are robust to an alternative measure for com-petitive overlap considering the interaction between technical resource overlap and market overlap.A con-cern with the interpretation of these results is that the addition of several interaction terms may create multi-collinearity problems.The simplified variable,although it implies some loss of information,allows for easier interpretation.

2011211

Polidoro,Ahuja,and Mitchell

T A B L E 1D e s c r i p t i v e S t a t i s t i c s a n d C o r r e l a t i o n M a t r i x

V a r i a b l e s M e a n

s .d .

12345678910111213141516171819

1.J o i n t v e n t u r e d i s s o l u t i o n 0.030.18

2.C o m b i n e d c e n t r a l i t y a t f o u n d i n g 0.24

0.25?.08

3.C o m b i n e d c e n t r a l i t y a t f o u n d i n g i n d y a d s w i t h n o d i r e c t t i e s 0.17

0.22

?.08.644.C o m b i n e d c e n t r a l i t y a t f o u n d i n g i n d y a d s w i t h d i r e c t t i e s 0.070.20?.02.55?.28

5.P o s i t i o n a l l y a s y m m e t r i c d y a d s a t f o u n d i n g 0.33

0.47.11?.38?.29?.16

6.C o m m o n p a r t n e r s i n p r e v i o u s y e a r 0.71

1.16

?.11.66.38.40?.21

7.C o m m o n p a r t n e r s i n p r e v i o u s y e a r ?p o s i t i o n a l l y a s y m m e t r i c d y a d s a t f o u n d i n g 0.12

0.45?.05?.12?.08?.06.39.258.C o m m o n p a r t n e r s i n p r e v i o u s y e a r ?c o m p e t i t i v e o v e r l a p 0.361.01?.07.43.21.32?.14.76.16

9.D i r e c t t i e s a t f o u n d i n g 0.200.45?.02.38?.34.84?.03.23?.03.1610.T i e s f o r m e d s i n c e f o u n d i n g 7.31

5.29?.0

6.44.43.09?.11.50.19.34

.01

11.C o m p e t i t i v e o v e r l a p 0.330.47.00.11.02.12?.12.25.08.52.04.0712.T e c h n i c a l r e s o u r c e o v e r l a p 0.84

0.13

?.04.19.14.08?.08.13.07

.25

.06

.22

.41

13.M a r k e t o v e r l a p 0.770.16?.06.22?.03.30?.07.34.13.36.23.19.58.1414.A g e a s y m m e t r y 35.5434.02?.01.08.15?.06.01?.18?.13?.17?.05.03?.05.18?.1815.C o m b i n e d c h e m i c a l R &D 430.84409.14.01.01.08?.08.06?.13?.03?.08?.02.13.12.37.06.5416.C o m b i n e d c h e m i c a l p a t e n t s 187.30

166.93

?.03.05.09?.04.05?.08

?.01

?.02

.01

.17

.19.44

.10

.57

.90

17.P e r f o r m a n c e a s y m m e t r y 0.030.03.12?.11?.03?.10?.06?.25?.12?.19?.02?.16?.11.07?.33.17.09.0818.D e b t -e q u i t y a s y m m e t r y 0.550.23?.05.06.01.07?.02.03.03.02.01?.05.03.05.05.16.11.14?.2319.L i q u i d i t y a s y m m e t r y 0.750.18?.05.19.13.09.01.19.11.10.03.15.02?.01.10?.05?.12?.16?.11.3120.T r a n s r e g i o n a l d u m m y

0.500.50.10?.21?.08?.17.08

?.41

?.15

?.33

?.10

?.17

?.40.07

?.68

.37.20

.17

.40

.02

?.14

We note that caution is in order in interpreting the latter effect,because of the substantial correlation with direct ties at founding,so that the negative impact of direct ties at founding partially offsets the positive impact of the combined centrality measure for firms with direct ties.Nonetheless,the findings do not support our predictions.Although posi-tional embeddedness may facilitate tie formation,premised on informational advantages that enable firms to select better partners,it does not decrease the hazard of unplanned joint venture dissolution.Instead,positional embeddedness can actually in-crease the hazard of unplanned dissolution when firms have prior direct ties.This unexpected result might stem from the formation of redundant alli-ances that generate fewer benefits to partners,which makes collaboration vulnerable to the emer-gence of more attractive opportunities.

The results support Hypothesis 2,which predicts that joint ventures between a high-centrality and a low-centrality firm experience greater hazard of un-planned dissolution than those between firms with

TABLE 2

Weibull Estimates of Influences on the Hazard of Unplanned Joint Venture Dissolution a

Variables

Model 1

Model 2

Model 3

Robustness Tests

Gamma Frailty

Dropping Dyads without Data on Dissolution

Model 4

Model 5

Combined centrality at founding (H1a:?0)

3.03*(1.84)Combined centrality at founding in dyads with no direct ties at founding (H1a:?0)

1.47(

2.23) 1.31(2.93)0.94(2.23)

Combined centrality at founding in dyads with direct ties at founding (H1b:?H1a)

12.58***(3.79)12.81***(3.75)10.10***(3.33)

Positionally asymmetric dyads at founding (H2:?0)

1.76***(0.54)

2.05***(0.58) 2.07***(0.61) 2.03***(0.60)Common partners in previous year (H3:?0)

?1.52*(0.88)?1.39?

(0.96)

?1.39?

(0.93)

?1.36?

(0.91)

Common partners in previous year ?positionally asymmetric dyads at founding (H4:?0)

?14.43***(1.20)?20.93***(1.31)?16.86***(1.34)Common partners in previous year ?competitive overlap (H5:?0)?12.23***(0.97)?18.71***(0.89)?14.65***(0.89)Direct ties at founding 0.21(0.62)?0.35(0.68)?2.95**(1.09)?3.05*(1.34)?2.17*(0.88)Ties formed since founding ?0.06(0.05)?0.03(0.06)?0.04(0.08)?0.04(0.08)?0.03(0.07)Competitive overlap 1.07(0.74) 1.52*(0.79) 1.32?(0.75) 1.34?(0.81) 1.03(0.72)Technical resource overlap ?3.04(2.17)?3.70?(2.31)?4.11?(2.20)?4.17?(2.28)?3.87?(2.16)Market overlap ?0.63(2.16)0.08(2.36)0.23(2.37)0.16(2.61)0.57(2.46)Age asymmetry ?0.01(0.01)?0.01(0.01)?0.01(0.01)?0.01(0.01)?0.005(0.01)Combined chemical R&D 0.002?(0.001)0.002*(0.001)0.003**(0.001)0.003?(0.002)0.002*(0.001)Combined chemical patents ?0.005(0.005)?0.006(0.004)?0.007(0.004)?0.007(0.005)?0.006(0.004)Performance asymmetry 7.95(5.17)10.14*(5.30)14.41**(4.58)14.71**(4.75)15.10**(5.32)Debt-equity asymmetry ?0.20(1.01)0.09(1.28)0.34(1.28)0.32(1.29)0.26(1.31)Liquidity asymmetry ?0.84(1.18)?2.06(1.27)

?2.17?(1.21)?2.17?(1.24)?1.82(1.24)Transregional dummy 1.32?

(0.75) 1.47*(0.72) 1.14?(0.66) 1.18(0.76)0.84(0.65)Constant ?1.81(2.17)?2.70(2.45)?2.64(2.32)?2.64(2.37)?3.08(2.50)ln(Weibull shape parameter (?)0.53***(0.15)0.70***(0.18)0.75***(0.20)0.78*(0.33)0.79***(0.20)ln(Overdispersion parameter frailty model (?)

?2.08(9.45)

Log pseudo-likelihood ?65.14?55.79?51.71?51.70?47.58?log pseudo-likelihood 9.35 4.08Observations 692692692

692

585

a

Robust,heteroskedasticity-adjusted standard errors are in parentheses.

?

p ?.10*p ?.05**p ?.01***p ?.001

One-tailed tests for hypotheses.

2011213

Polidoro,Ahuja,and Mitchell

similar centrality.This prediction built on arguments about the reputational benefits of positional embed-dedness.As expected,results show that the partners’positional asymmetry at joint venture founding in-creases the dissolution hazard (??2.05,p ?.001).The results support Hypothesis 3,which predicts that the number of common partners connecting two firms reduces the hazard of unplanned joint venture dissolution.This prediction drew on argu-ments concerning social monitoring benefits of structural embeddedness.As expected,the pres-ence of common partners shared by two firms in the evolving network structure mitigates the hazard of alliance dissolution (???1.52,p ?.05,in model 2,before adding interactions;???1.39,p ?.10,in model 3,after adding interactions).The findings support Hypotheses 4and 5,which predict that the benefits of having common partners are increased in the relationships most vulnerable to instability,including joint ventures between posi-tionally asymmetric partners and between partners with competitive overlap.Supporting the predic-tions,the coefficient on common partners is signifi-cantly more negative in positionally asymmetric dy-ads (???14.43,p ?.001)and in dyads with high competitive overlap (???12.23,p ?.001).More-over,the statistical significance of the main effects of both common partners and competitive overlap de-clines once the analysis adds the interactions,high-

lighting the importance of the contingent effects (al-though the main effect of positional asymmetry remains highly significant in model 3).

Figures 2and 3show the multiplicative effect of social asymmetry and competitive overlap relative to a baseline group in which all covariates are 0.As the slopes in Figure 2indicate,common partners contribute to reducing collaboration hazards more dramatically for positionally asymmetric firms.Similarly,the slopes in Figure 3show that the stabilizing effects of common partners are more notable when firms exhibit competitive overlap.Several control variables in model 3affect alli-ance dissolution.The presence of direct ties at founding (relational embeddedness),technical re-source overlap,and liquidity asymmetry at least moderately reduce dissolution risk.Firms’com-bined R&D expenditures,performance asymmetry,and alliance transregional span at least moderately increase dissolution hazards.Sensitivity Tests

To rule out the possibility that unobserved het-erogeneity across observations affects our findings,we ran a model with a frailty distribution (Hou-gaard,1986),based on a gamma distribution.Model 4shows that the findings we reported earlier are robust to this specification.Moreover,the parame-

FIGURE 2

Effects of Common Partners Relative to Baseline Group in Positionally Symmetric and Asymmetric

Dyads

?30

?25

?20

?15

?10?5

5

ln (Hazard Ratio/Hazard Ratio Baseline Group)

214February

Academy of Management Journal

ter of the frailty model is not statistically signifi-cant,suggesting that little unobserved heterogene-ity is present.

As noted earlier,we were unable to unambigu-ously establish dissolution or survival for 24dyads in the sample and assumed that they survived be-yond 1992.We obtained similar results in models assuming that these dyads lasted only three,four,or five years after the year of formation.Further,to rule out the possibility that particular assumptions about the duration of these dyads affected the re-sults,in model 5we drop them from the sample and obtain robust results.

Because of multicollinearity among the variables capturing firms’age asymmetry,combined chemi-cal R&D,and combined chemical patents,we ran additional models with orthogonal measures for these variables (Rhee &Haunschild,2006;Sine,Shane,&DiGregorio,2003).This procedure purges variables from information contained in the other measures.These models produced similar results.Results were robust when we dropped these vari-ables in additional models,ruling out the possibil-ity that multicollinearity affected the findings.We argue that the informational and reputational benefits of positional embeddedness accrue to partners at joint venture founding,whereas the social monitoring benefits of structural embed-dedness that accrue at alliance founding are re-inforced through the life of the partnership.In sensitivity analyses,we added time-varying co-variates with partners’combined centrality and

positional asymmetry,as well as the number of common partners at joint venture founding.Con-sistently with our arguments,we found that part-ners’combined centrality and positional asymme-try in the evolving network structure do not affect the hazard of alliance dissolution.

Further,we conducted a post hoc analysis to examine whether results were robust to models in which the influences of embeddedness could vary with time elapsed since alliance founding (Greve et al.,2010).We advise caution in the interpretation of such an analysis,since it entails the addition of interaction terms that compound multicollinear-ity.4The analysis showed robust results.Moreover,it did not show that the effects of partners’com-bined centrality and positional asymmetry varied significantly over time.Addition of the interaction between common partners and time since found-ing,on the other hand,indicated that the influence of common partners on tie dissolution became weaker over time.Nonetheless,although this inter-action was statistically significant,the main effect for common partners became more robust in both magnitude and statistical significance (p ?.001).Finally,we conducted sensitivity tests with sample selection analysis.If firms decide to form an alliance on the basis of the factors expected to influence dis-

4

The fact that the analysis uses information on the duration of joint ventures to estimate dissolution hazards increases the concern with potential multicollinearity.

FIGURE 3

Effects of Common Partners Relative to Baseline Group in Dyads with High and Low Competitive Overlap

?30

?25

?20?15

?10?5

5

2

1

ln (Hazard Ra tio/ Hazard Ratio Baseline Group)

Common Partners in Previous Year

Competitive overlap = 0 (low)

Competitive overlap = 1 (high)

2011215

Polidoro,Ahuja,and Mitchell

solution,empirical models of tie dissolution that do not account for this decision may be biased.To our knowledge,no previous empirical examination of joint venture dissolution has addressed this issue.For our sensitivity analysis,therefore,we adopted Heck-man’s(1979)method to estimate firms’propensity to create joint ventures and used a self-selection correc-tion term in the estimation of the hazard of un-planned dissolution.As Appendix B reports,this pro-cedure showed robust results.

DISCUSSION AND CONCLUSION

This study reveals that the network structure in which firms are embedded,besides affecting which ties the firms create,as prior research has shown,also affects the hazard of unplanned tie dissolution.Per-haps most strikingly,we find that positional embed-dedness and structural embeddedness have distinct effects on alliance stability.Positional embed-dedness arising from combined network central-ity does not stabilize alliances and may even contribute to instability.By contrast,structural embeddedness via having partners in common stabilizes alliances,especially joint ventures be-tween positionally asymmetric firms and between firms with high competitive overlap.Thus,com-mon partners constitute a social mechanism for maintaining order,especially in risky interfirm activities.

Theoretical Implications and Potential for Future Research

By showing how different aspects of network embeddedness influence joint venture dissolution, this study furthers both network research and the literature on joint ventures.

Network research.Our work highlights and em-pirically demonstrates several ways in which the effects of network embeddedness on tie dissolution differ from its effects on tie formation.First,prior research has shown that firms tend to form re-peated ties.Our study shows that the benefits of relational embeddedness(repeated ties)partly overlap with the informational benefits of posi-tional embeddedness(network centrality)and,as a result,repeated ties between central partners can in some cases result in increased hazard of unplanned dissolution.

Second,this study reveals that the informational benefits of positional embeddedness,although they facilitate tie formation,fail to provide stability dur-ing alliance life.Superior information stemming from firms’positions in the alliance network does not help partners manage the tensions that occur during the life of a collaboration.

Third,previous research has also shown that, despite firms’overall preferences to form position-ally symmetric relationships,they sometimes have incentives to form ties with partners in asymmetric network positions.This study demonstrates that, despite the presence of instrumental goals motivat-ing such alliances,the reputational imbalance in-herent in those ties makes the resulting alliances more vulnerable to unplanned dissolution. Fourth,the literature on tie formation indicates that the presence of common partners allays firms’concerns about the reliability of potential partners and,as a result,generally has a positive impact on tie formation.Research has also shown that alli-ance governance mode is an alternative mechanism for mitigating collaboration hazards:firms prefer to structure alliances as joint ventures when they per-ceive greater collaboration hazards.Our study, however,shows that firms benefit from overlap-ping these two mechanisms:joint venture partners that have common partners enjoy more stable col-laboration than allies lacking common partners. This study also demonstrates a previously uniden-tified contingency,showing that the sociological ben-efits of embeddedness are more valuable when the incentives for competitive or opportunistic behavior are greater.Our findings indicate that social monitor-ing through common partners has a greater effect as a stabilizing factor precisely in situations in which partners have otherwise strong incentives to behave opportunistically,such as in joint ventures between positionally asymmetric partners and between part-ners with high competitive overlap.

These findings,summarized in Appendix C,high-light the tension that firms face between alliance for-mation incentives and collaboration challenges. Sometimes the potential benefits of a partnership go hand in hand with factors that are conducive to col-laboration problems.Whereas network embedded-ness affects tie formation by indicating which alli-ances have lower collaboration hazards,it affects tie dissolution by providing stability to alliances that, despite the expected benefits,are more vulnerable to collaboration challenges.Examining how different forms of embeddedness jointly affect alliance stabil-ity furthers understanding of embeddedness by ques-tioning the implicit assumption in prior research that different types of embeddedness are additive and reinforce each other.Our study shows that the bene-fits of relational and positional embeddedness un-dercut each other and that structural embeddedness can partly substitute for asymmetries in positional embeddedness.

216February

Academy of Management Journal

Further,our study,when juxtaposed with prior work,draws attention to the fact that structural embeddedness provides two related but distinct types of benefits and that these benefits map differ-ently onto the outcomes of tie formation and tie dissolution.The first benefit is that of referrals, whereby firms are knowledgeable about their part-ners and refer them to a focal firm(Burt,1992; Gulati,1995b).The second benefit is that of social restraint(Coleman,1988;Dore,1983).Alliance for-mation mainly reflects the referral benefits of hav-ing common partners,perhaps combined with some ex ante expectation of social restraint.Alli-ance dissolution,on the other hand,offers an op-portunity for a closer test of the social restraint mechanism.At the time that firms consider the decision to terminate a relationship with a partner, the referral advantage is no longer relevant;the social restraint is,meanwhile,a valid consider-ation.Forming a tie with its partner’s partner does not require the firm to subordinate its immediate economic interests.However,systematically stay-ing in a relationship longer,even in the face of competitive incentives to break the relationship, reflects the suppression of an atomistic,transac-tional calculus in favor of a longer-term,relational perspective.Thus,tie formation and tie dissolution are not mirror images of each other;rather,they reflect distinct decision processes and offer oppor-tunities to evaluate the efficacy of different embed-dedness mechanisms.

Finally,this study draws attention to the dynam-ics leading to tie instability(Yan&Zeng,1999)and sheds light on the appropriate time scales for social network research(Granovetter,1992;Soda,Usai,& Zaheer,2004).The informational and reputational benefits of network embeddedness are more prev-alent at the time of tie formation,by helping firms choose partners who are more likely to engage in responsible collaborative behavior.By contrast,so-cial monitoring benefits accrue even after alliance formation,by providing deterrents to opportunism. Future research on how changes in network structure over time accentuate or diminish the ben-efits of network embeddedness can further expand scholars’understanding of network dynamics.An interesting opportunity in this line of inquiry is the investigation of how tie dissolution,in turn,affects subsequent tie formation.Another fruitful opportu-nity for further inquiry would be to examine the circumstances under which structural embedded-ness may lead firms to jointly behave in ways that appear to undermine network stability,in light of the unexpected finding in the study by Greve et al. (2010)showing that firms connected by common partners are more likely to jointly withdraw from liner shipping alliances.The stabilizing effects of structural embeddedness are centered on the prem-ise that firms connected by common partners re-frain from behaving opportunistically owing to concern with their other ties.In keeping with that logic,a firm’s decision to engage in opportunistic behavior despite the potential for sanctions by common partners may induce those partners to concomitantly discontinue ties with the norm-breaking firm.Hence,what appears as firms’joint withdrawal from alliances may indeed reflect com-mon partners’decisions to inflict penalties on norm-breaking firms.

Research on joint ventures.Prior studies have approached joint venture dissolution from a dyadic perspective and highlighted the effects of partners’prior ties,cultural differences,and competitive ri-valry(e.g.,Hennart&Zeng,2002;Kogut,1991;Park &Russo,1996).This study,in contrast,approaches joint venture instability from a broader network perspective and shows that the aggregate network of ties among a set of firms affects joint venture dissolution.

Complementing the dyadic perspective with a network approach has important implications for the management of joint ventures.The normative implication of the literature on alliance formation is that managers,if they can,should embed their search for alliance partners in their social context to avoid forming alliances with considerable poten-tial to turn out badly.But,as this study highlights, managers can only partially foresee shifts in indus-try dynamics that can engender alliance instability. Prior research has suggested that managers can re-sort to the administrative apparatus of joint ven-tures,coupled with contractual safeguards,to align partners’incentives and deter opportunism.How-ever,because managers are unable to fully antici-pate partners’future behavior,the formality of joint venture governance and the shadow of the future between partners are insufficient mechanisms for stability.By shifting from a dyadic approach to a network approach,this study highlights an addi-tional mechanism through which managers can mitigate joint venture instability:common partners make the shadow of the future more dramatic by casting their own shadow over the relationship be-tween two firms.The deterrence-based trust that common partners promote overshadows the incen-tives that partners have to pursue self-interest at the expense of joint benefits and,as a result,reduces joint venture instability.

Future research can expand the literature on joint ventures by examining how network embeddedness, besides affecting tie dissolution,also affects other col-laboration outcomes,such as resource exchange and

2011217

Polidoro,Ahuja,and Mitchell

knowledge flow between partners during the course of an alliance.The informational,reputational,and social monitoring benefits associated with network embeddedness may affect not only the patterns of joint venture formation and dissolution,but also the thickness of resource flow through network ties.So-cial considerations can affect network dynamics by influencing what flows through a given network at different points in time even if the ties in that struc-ture remain relatively stable over time.Whereas this study expands understanding of network dynamics by shifting the focus from tie formation to tie disso-lution,one next step is to consider the dynamics in the content of network ties.

Limitations

Similarly to prior researchers looking at joint venture dissolution,we were unable to control for joint venture characteristics such as size or perfor-mance(Barkema et al.,1997;Dussauge et al.,2000; Kogut,1989;Park&Russo,1996;Park&Ungson, 1997).Obtaining reliable data on specific joint ven-ture attributes such as size and financial perfor-mance is extremely difficult,especially in a longi-tudinal study.However,our analysis does account for joint venture age,because the Weibull model enabled the baseline hazard function to vary with time elapsed since founding.

To the extent that longitudinal data on specific characteristics of alliances become more easily available,researchers will have the opportunity to expand research on network embeddedness in new ways.Availability of these data would enable the investigation of how embeddedness affects other alliance outcomes,such as innovations that part-ners jointly create or alliances’financial perfor-mance.Additionally,such data would enable fur-ther examination of the argument that network embeddedness may prevent partners from discon-tinuing poorly performing alliances(Kim et al., 2006;Seabright,Levinthal,&Fichman,1992). Overall,the study highlights the challenges that even intendedly rational decision makers face in cre-ating and managing successful alliances.Firms have many incentives to form alliances.However,the same social and economic forces that create forma-tion incentives sometimes undermine the subsequent performance of the alliances and lead to their un-planned dissolution.We believe that our study pro-vides a base for further work exploring these tensions.

REFERENCES

Ahuja,G.2000.The duality of collaboration:Inducements and opportunities in the formation of interfirm link-ages.Strategic Management Journal,21:317–343.Ahuja,G.,Polidoro,F.,&Mitchell,W.2009.Structural homophily or social asymmetry?The formation of alliances by poorly embedded firms.Strategic Man-agement Journal,30:941–958.

Arino,A.,&Torre,J.1998.Learning from failure:To-wards an evolutionary model of collaborative https://www.wendangku.net/doc/7a4384519.html,anization Science,9:306.

Axelrod,R.1984.The evolution of cooperation.New York:Basic Books.

Bae,H.,&Gargiulo,M.2004.Partner substitutability, alliance network structure,and firm profitability in the telecommunications industry.Academy of Man-agement Journal,47:843–859.

Barkema,H.G.,Shenkar,O.,Vermeulen,F.,&Bell,J.

1997.Working abroad,working with others:How firms learn to operate international joint ventures.

Academy of Management Journal,40:426–442.

Barnett,W.P.1997.The dynamics of competitive intensity.

Administrative Science Quarterly,42:128–160.

Baum,J.A.C.,&Korn,H.J.1999.Dynamics of dyadic competitive interaction.Strategic Management Journal,20:251–278.

Bonacich,P.1987.Power and centrality:A family of measures.American Journal of Sociology,92: 1170–1182.

Borgatti,S.P.,Everett,M.G.,&Freeman,L.C.1999.

UCINET5.0(version1.00).Natick,MA:Analytic Technologies.

Brass,D.J.,Galaskiewicz,J.,Greve,H.R.,&Tsai,W.

2004.Taking stock of networks and organizations:A multilevel perspective.Academy of Management Journal,47:795–817.

Broschak,J.P.2004.Managers’mobility and market in-terface:The effect of managers’career mobility on the dissolution of market ties.Administrative Sci-ence Quarterly,49:608–640.

Burt,R.S.1992.Structural holes:The social structure of competition.Cambridge,MA:Harvard University Press.

Chen,https://www.wendangku.net/doc/7a4384519.html,petitor analysis and interfirm rivalry: Toward a theoretical integration.Academy of Man-agement Review,21:100–134.

Chung,S.,Singh,H.,&Lee,https://www.wendangku.net/doc/7a4384519.html,plementarity, status similarity and social capital as drivers of alliance formation.Strategic Management Journal,21:1–22. Cohen,W.M.,&Levinthal,D.A.1990.Absorptive capacity:

A new perspective on learning and innovation.Ad-

ministrative Science Quarterly,35:128–152.

Coleman,J.S.1988.Social capital in the creation of human capital.American Journal of Sociology,94: S95–S120.

Contractor,F.,&Lorange,P.1988.Cooperative strate-gies in international business.Lexington,MA:Lex-ington Books.

218February

Academy of Management Journal

Cook,K.S.1977.Exchange and power in networks of interorganizational relations.Sociological Quar-terly,18:62–82.

Das,T.K.,&Teng, B.2000.Instabilities of strategic alliances:An internal tensions https://www.wendangku.net/doc/7a4384519.html,ani-zation Science,11:77–101.

Dore,R.1983.Goodwill and the spirit of market capital-ism.British Journal of Sociology,34:459–482. Dussauge,P.,Garrette,B.,&Mitchell,W.2000.Learning from competing partners:Outcomes and durations of scale and link alliances in Europe,North America and Asia.Strategic Management Journal,21:99–126. Emerson,R.M.1972.Exchange theory part II:Exchange relations and networks.In J.Berger,M.Zelditch Jr.,&

B.Anderson(Eds.),Sociological theories in progress,

vol.7:58–87.Boston:Houghton Mifflin. Gehringer,J.M.,&Herbert,L.1989.Control and perfor-mance of international joint ventures.Journal of International Business Studies,20:235–254. Gimeno,https://www.wendangku.net/doc/7a4384519.html,petition within and between net-works:The contingent effect of competitive embed-dedness on alliance formation.Academy of Man-agement Journal,47:820–842.

Gimeno,J.,&Woo,C.Y.1996.Hypercompetition in a multimarket environment:The role of strategic sim-ilarity and multimarket contact in competitive https://www.wendangku.net/doc/7a4384519.html,anization Science,7:322–341. Granovetter,M.1992Problems of explanation in eco-nomic sociology.In N.Nohria&R.G.Eccles(Eds.), Network and organizations:25–56.Boston:Har-vard Business School Press.

Greene,W.H.2003.Econometric analysis.Upper Sad-dle River,NJ:Prentice Hall.

Greve,H.R.,Baum,J.A.C.,Mitsuhashi,H.,&Rowley,T.

2010.Built to last but falling apart:Cohesion,fric-tion and withdrawal from interfirm alliances.Acad-emy of Management Journal,53:302–322. Griliches,Z.1990.Patent statistics as economic indica-tors:A survey.Journal of Economic Literature,28: 1661–1707.

Gulati,R.1995a.Does familiarity breed trust?The implica-tions of repeated ties for contractual choice in alli-ances.Academy of Management Journal,38:85–112. Gulati,R.1995b.Social structure and alliance formation patterns:A longitudinal analysis.Administrative Science Quarterly,40:619–652.

Gulati,R.,&Gargiulo,M.1999.Where do interorganiza-tional networks come from?American Journal of Sociology,104:1439–1493.

Gulati,R.,Nohria,N.,&Zaheer,A.2000.Strategic net-works.Strategic Management Journal,21:203–215. Gulati,R.,&Singh,H.1998.The architecture of cooper-ation:Managing coordination costs and appropria-tion concerns in strategic alliances.Administrative Science Quarterly,43:781–814.Gulati,R.,&Westphal,J.D.1999.Cooperative or control-ling?The effects of CEO-board relations and the con-tent of interlocks on the formation of joint ventures.

Administrative Science Quarterly,44:473–506.

Hagedoorn,J.,&Schakenraad,J.1989.Strategic partner-ing and technological cooperation.In B.Dankbaar,J.

Groenewegen,&H.Scwenk(Eds.),Perspectives in industrial economics:171–191.Dordrecht,The Netherlands:Kluwer.

Hall,B.H.,Jaffe,A.B.,&Trajtenberg,M.2001.The NBER patent citations data file:Lessons,insights,and methodological tools.Working paper no.8498,Na-tional Bureau of Economic Research,Cambridge,MA.

Heckman,J.J.1979.Sample selection bias as a specifi-cation error.Econometrica,47:153–161.

Heide,J.B.,&Miner,A.S.1992.The shadow of the future:Effects of anticipated interaction and fre-quency of contact on buyer-seller cooperation.

Academy of Management Journal,35:265–291.

Hennart,J.,&Zeng,M.2002.Cross-cultural differences and joint venture longevity.Journal of Interna-tional Business Studies,33:699.

Hergert,M.,&Morris,D.1988.Trends in international collaborative agreements.In F.K.Contractor&P.Lor-ange(Eds.),Cooperative strategies in international business:99–110.Lexington,MA:Lexington Books. Hitt,M.A.,Dacin,M.T.,Levitas,E.,Arregle,J.,&Borza,

A.2000.Partner selection in emerging and devel-

oped market contexts:Resource-based and organiza-tional learning perspectives.Academy of Manage-ment Journal,43:449–467.

Hougaard,P.1986.Survival models for heterogeneous populations derived from stable distributions.Bio-metrica,73:387–396.

Karnani,A.,&Wernerfelt,B.1985.Multiple point com-petition.Strategic Management Journal,6:87–96. Khanna,T.,Gulati,R.,&Nohria,N.1998.The dynamics of learning alliances:Competition,cooperation,and relative scope.Strategic Management Journal,19:193–210. Kim,T.,Oh,H.,&Swaminathan,A.2006.Framing interor-ganizational network change:A network inertia perspec-tive.Academy of Management Review,31:704–720. Kogut,B.1988.Joint ventures:Theoretical and empirical per-spectives.Strategic Management Journal,9:319–332.

Kogut,B.1989.The stability of joint ventures:Reciproc-ity and competitive rivalry.Journal of Industrial Economics,38:183–198.

Kogut,B.1991.Joint ventures and the option to expand and acquire.Management Science,37:19–33. Lane,P.J.,Salk,J.E.,&Lyles,M.A.2001.Absorptive capacity,learning,and performance in international joint ventures.Strategic Management Journal,22: 1139–1161

Levinthal,D.,&Fichman,M.1988.Dynamics of interor-

2011219

Polidoro,Ahuja,and Mitchell

ganizational attachments:Audit-client relationships.

Administrative Science Quarterly,33:345–369.

Li,J.,Dhanaraj,C.,&Shockley,R.L.2008.Joint venture evolution:extending the real options approach.

Managerial&Decision Economics,29:317–336.

Li,S.X.,&Rowley T.2002.Inertia and evaluation mech-anisms in interorganizational partner selection:Syn-dicate formation among U.S.investment banks.

Academy of Management Journal,45:1104–1119.

Mitchell,W.,&Singh,K.1996.Survival of business using collaborative relationships to commercialize complex goods.Strategic Management Journal,17:169–195.

Mizruchi,M.S.1982.The American corporate network.

Beverly Hills,CA:Sage.

Mizruchi,M.S.1992.The structure of corporate polit-ical action:Interfirm relationships and their con-sequences.Cambridge,MA:Harvard University Press.

Mizruchi,M.S.1993.Cohesion,equivalence,and simi-larity of behavior:A theoretical and empirical as-sessment.Social Networks,15:275–307.

Moran,P.2005.Structural vs.relational embeddedness: Social capital and managerial performance.Strate-gic Management Journal,26:1129–1151. Nahapiet,J.,&Ghoshal,S.1998.Social capital,intellec-tual capital,and the organizational advantage.Acad-emy of Management Review,23:242–266. National Post.2009.Financial Post&FP Investing:Dow Chemical seeks to settle with Kuwaitis.February21: FP5.

Oxley,J.E.1997.Appropriability hazards and gover-nance in strategic alliances:A transaction cost ap-proach.Journal of Law,Economics&Organiza-tion,13:387–409.

Park,S.H.,&Russo,M.V.1996.When competition eclipses cooperation:An event history analysis of joint venture failure.Management Science,42:875–890. Park,S.H.,&Ungson,G.R.1997.The effect of national culture,organizational complementarity and eco-nomic motivation on joint venture dissolution.

Academy of Management Journal,40:279–307.

Park,S.H.,&Ungson,G.R.2001.Interfirm rivalry and managerial complexity:A conceptual framework of alliance https://www.wendangku.net/doc/7a4384519.html,anization Science,12:37–53. Parkhe,A.1993a.Partner nationality and the structure-performance relationship in strategic alliances.Or-ganization Science,4:301.

Parkhe,A.1993b.Strategic alliance structuring:A game theoretic and transaction cost examination of inter-firm cooperation.Academy of Management Jour-nal,36:794–829.

Pfeffer,J.,&Nowak,P.1976.Joint ventures and interor-ganizational interdependence.Administrative Sci-ence Quarterly,21:398–418.Podolny,J.M.1993.A status-based model of market competition.American Journal of Sociology,98: 829–872.

Podolny,J.M.1994.Market uncertainty and the social character of economic exchange.Administrative Science Quarterly,39:458–483.

Powell,W.W.,Koput,K.W.,&Smith-Doerr,L.1996.

Interorganizational collaboration and the locus of innovation:Networks of learning in biotechnology.

Administrative Science Quarterly,41:116–145. Rhee,M.,&Haunschild,P.R.2006.The liability of good reputation:A study of product recalls in the U.S.auto-mobile https://www.wendangku.net/doc/7a4384519.html,anization Science,17:101–117. Rosenkopf,L.,Metiu,A.,&George,V.P.2001.From the bottom up?Technical committee activity and alli-ance formation.Administrative Science Quarterly, 46:748–772.

Rowley,T.,Behrens,D.,&Krackhardt,D.2000.Redun-dant governance structures:An analysis of structural and relational embeddedness in the steel and semi-conductor industries.Strategic Management Jour-nal,21:369–386.

Seabright,M.A.,Levinthal,D.A.,&Fichman,M.1992.

Role of individual attachments in the dissolution of interorganizational relationships.Administrative Science Quarterly,35:122–160.

Sine,W.D.,Shane,S.,&DiGregorio,D.2003.The halo effect and technology licensing:The influence of institutional prestige on the licensing of university inventions.Management Science,49:478–496. Soda,G.,Usai,A.,&Zaheer,https://www.wendangku.net/doc/7a4384519.html,work memory: The influence of past and current networks on perfor-mance.Academy of Management Journal,47:893–906. Sohn,M.2001.Distance and cosine measure of niche overlap.Social Networks,23:141–165.

Stigler,G.J.1964.A theory of oligopoly.Journal of Political Economy,72:44–61.

Times.2008.Kuwait decision to quit joint venture puts Dow Chemical’s expansion in jeopardy.December 30:Business39.

Uzzi,B.1996.The sources and consequences of embed-dedness for the economic performance of organiza-tions:The network effect.American Sociological Review,61:674–698.

Uzzi,B.1997.Social structure and competition in inter-firm networks:The paradox of embeddedness.Ad-ministrative Science Quarterly,42:35–67. Walker,G.,Kogut,B.,&Shan,W.1997.Social capital, structural holes and the formation of an industry https://www.wendangku.net/doc/7a4384519.html,anization Science,8:109–126. Yan,A.,&Zeng,M.1999.International joint venture insta-bility:A critique of previous research,a reconceptual-ization,and directions for future research.Journal of International Business Studies,30:397–414. Zaheer,A.,&Venkatraman,N.1995.Relational gover-

220February

Academy of Management Journal

nance as an interorganizational strategy:An empiri-cal test of the role of trust in economic exchange.Strategic Management Journal,16:373–392.Zollo,M.,Reuer,J.,&Singh,H.2002.Interorganizational

routines and performance in strategic alliances.Or-ganization Science,13:701–713.

APPENDIX B

Sensitivity Test with Sample Selection Analysis

We used Heckman’s (1979)procedure to account for potential sample selection bias in the analysis of the influences of network embeddedness on joint venture (JV)dissolution.For each pair of firms,we calculated the propensity of JV formation using a probit specification and created the term to correct for self-selection.This term captured the likelihood of a given dyad in our sample of ties being observed to be at risk of dissolution;its inclusion in the analysis of tie dissolution eliminated the potential specification error.Li and Rowley’s (2002)examination of alliance formation by previously con-nected firms and Bae and Gargiulo’s (2004)research on the effects of network participation on organizational profitability used a similar approach.

In the equation estimating the likelihood of JV formation,we included year dummies to account for the effects of temporal unobserved heterogeneity.We used a random-effects probit regression to account for the possibility of random effects across dyads.Further,to ensure that the

selection equation included at least one variable with a nonzero coefficient that did not affect the second equation,we did not include firms’combined size (partners’assets in the chemical industry)in the estimation of JV dissolution,since additional analysis showed that this variable had no significant effect on JV dissolution,thus echoing findings of prior examination (Barkema et al.,1997).As Table B1shows,the results are consistent with prior research.Firms’positions in their network structure and firms’prior direct ties increased the likelihood of JV formation.As do the findings of prior research,the results indicate that firms sometimes ally with positionally asymmetric partners (Gu-lati &Gargiulo,1999).Indirect ties between firms in our sample decreased their propensity to create a joint venture.This result corroborates the argument that firms resort to JV as the preferred alliance governance mode when the haz-ards of collaboration are higher (Gulati &Singh,1998;Ox-ley,1997).The presence of common partners allays part-ners’concern with opportunism and reduces firms’need to resort to a formal governance structure.The findings reveal that resource overlap and market overlap encourage alli-ance formation,which is consistent with the argument that similarity reflects partners’interdependence (Pfeffer &Nowak,1976)and facilitates mutual learning (Cohen &Levinthal,1990;Lane,Salk,&Lyles,2001).Firms possess-ing more resources are more likely to engage in technolog-ical collaboration,and they are less likely to do so when they already invest more in R&D.Table B2shows robust results.

APPENDIX A

TABLE A1

Examples of Reasons for Joint Venture Dissolution

Dissolution Type and

Joint Venture Summary of Archival Stories a

Unplanned

Courtaulds Grafil

A 1988change in US legislation increased the attractiveness of the aerospace market for structural carbon fiber.Following that shift,Courtaulds exited the least attractive technology of the JV formed with Dexter in 1983and later bought the partner’s remaining stake to exploit the venture’s strong position in the carbon fiber business.

Radcure Specialites

In 1989UCB and Rhone Poulenc became partners in a JV dedicated to the development,production,and distribution of radiation-cured resins.Two years later,when demand for the jointly developed resins started to grow,UCB bought Rhone Poulenc’s stake in the JV.

Enimont

In 1989Montedison and Enichem dissolved the JV formed in 1988because they could not agree on an industrial strategy.Sclavo

In 1990Enichem bought Du Pont’s stake in the JV formed in 1987,following Du Pont’s ultimatum that the partner either buy the JV or cede control.Du Pont claimed that the alliance had become blocked.

Planned

BASF-Linde

In 1990BASF and Linde dissolved the JV formed in 1988because the technology they developed jointly reached commercialization stage.

a

“JV”is joint venture.

2011221

Polidoro,Ahuja,and Mitchell

TABLE B1

Random-Effects Probit Estimates of Influences on Joint

Venture Formation

Variables b s.e. Combined centrality at founding 1.60***(0.24) Positionally asymmetric dyads at

founding

0.05(0.06) Common partners in previous year?0.22***(0.07) Direct ties at founding0.20*(0.10) Competitive overlap?0.26(1.43) Technical resource overlap0.86**(0.27) Market overlap0.52?(0.28) Age asymmetry?0.00(0.00) Combined size0.00***(0.00) Combined chemical R&D?0.00*(0.00) Combined chemical patents0.00(0.00) Performance asymmetry?0.99(1.17) Debt-equity asymmetry0.06(0.13) Liquidity asymmetry0.12(0.18) Transregional dummy?0.07(0.08) Constant?4.07***(0.36) Random-effects parameter0.12*(0.06) Log-likelihood?1,022.03 Observations37,824

?p?.10

*p?.05

**p?.01

***p?.001

Two-tailed tests.

TABLE B2

Weibull Estimates of Influences on Joint Venture

Dissolution a

Variables b s.e. Combined centrality at founding in dyads

with no direct ties at founding

0.67(2.60) Combined centrality at founding in dyads

with direct ties at founding

11.55**(4.09) Positionally asymmetric dyads at founding 2.05***(0.59) Common partners in previous year?1.25?(0.99) Common partners in previous year?

positionally asymmetric dyads at

founding

?15.27***(1.26)

Common partners in previous year?

competitive overlap

?12.98***(0.94)

Direct ties at founding?2.88**(1.06) Ties formed since founding?0.05(0.08) Competitive overlap 1.27?(0.73) Technical resource overlap?4.34?(2.24) Market overlap0.07(2.25) Age asymmetry?0.01(0.01) Combined chemical R&D0.003*(0.001) Combined chemical patents?0.007(0.004) Performance asymmetry13.82***(4.34) Debt-equity asymmetry0.33(1.24) Liquidity asymmetry?2.36?(1.33) Transregional dummy 1.21?(0.75) Correction for propensity of JV formation?0.49(1.03) Constant?1.83(2.08) ln(Weibull shape parameter)0.79***(0.23) Log pseudo-likelihood?51.57 Observations692

a Robust,heteroskedasticity-adjusted standard errors are in parentheses.

?p?.10

*p?.05

**p?.01

***p?.001

One-tailed tests for hypotheses.

222February

Academy of Management Journal