DISENTANGLING THE INFLUENCES OF LEADERS’ …willm/bio/cv/papers/AMJ2007... · RELATIONAL EMBEDDEDNESS ON INTERORGANIZATIONAL EXCHANGE ... creased the influences of those experiences

DISENTANGLING THE INFLUENCES OF LEADERS’RELATIONAL EMBEDDEDNESS ON

INTERORGANIZATIONAL EXCHANGE

JEFFREY Q. BARDENUniversity of Washington

WILL MITCHELLDuke University

Drawing on the concept of relational embeddedness and the associated mechanisms ofmutual understanding, trust, and commitment, we examine how leaders’ prior ex-change experiences influence the likelihood of subsequent interorganizational ex-change. We begin to develop a microlevel model of organization-level relations thataccounts for nodal multiplexity. In data on baseball player trades, we found thatindividual leaders’ ties affected exchanges less than did an organization’s other ties.The sharing of exchange experiences by organizations and their current leaders in-creased the influences of those experiences on exchange behavior. Thus, leaders havemore influence within their organizational contexts than in isolation.

Organizations commonly undertake exchangeswith other organizations to gain access to neededresources. In doing so, they often face multipleoptions in deciding which organizations to engageand which to avoid. Traditionally, research on eco-nomic exchange has highlighted the defining roleof incentives and complementarities within an ide-alized, atomistic market (Aiken & Hage, 1968;North, 1990). Although this perspective has beenadapted to account for market imperfections, suchas “bounded rationality” and “small numbers bar-gaining,” it does not adequately explain the rela-tional prerequisites of exchange, such as mutualawareness and trust.

To help account for the role of relational mech-anisms in shaping exchange behavior, researchershave increasingly focused on social embeddedness(Adler & Kwon, 2002; Inkpen & Tsang, 2005; Pow-ell, 1990; Uzzi, 1996, 1997). Social embeddednessrefers to the influences that prior relations amongactors have on their subsequent economic behav-ior. By facilitating the spread of social informationthat promotes mutual understanding, trust, andcommitment, prior relations determine the relativeimpacts of different actors and, in turn, shape eco-nomic behavior (Granovetter, 1985). Thus, to theextent that prior relations facilitate access to re-

sources, social embeddedness is the fundamentalmechanism underlying social capital.

Social embeddedness includes both relationaland structural components (Granovetter, 1992). Re-lational embeddedness refers to the influences ofthe content of direct, dyadic relations. Structuralembeddedness refers to the influences of the over-all pattern of direct and indirect relations in a set ofactors. Because the influence of structural embed-dedness is contingent upon the content of dyadicrelations (Podolny & Baron, 1997; Rowley, Behrens,& Krackhardt, 2000), relational embeddedness is animportant starting point for understanding the in-fluences of network structures on economic behav-ior. However, research on the role of relationalcontent remains underdeveloped (Burt, 1997).Thus, this study focuses on relational embedded-ness, while controlling for key aspects of structuralembeddedness.

Theoretical development and testing of the con-cept of relational embeddedness initially were uti-lized to examine relations among individuals, butmore recent research shows that relational embed-dedness predicts several forms of interorganization-al exchange, including strategic alliances (Gulati,1995), buyer-supplier relations (Uzzi, 1997), andgovernment relations (Hitt, Bierman, Uhlenbruck,& Shimizu, 2006). However, translating ideas con-cerning relational embeddedness from the individ-ual to the organization level can lead to misspeci-fication, owing to the “nodal multiplexity” oforganizational ties. The problem stems from thefact that interorganizational relationships are

For their helpful advice on this project, we thankJames Cook, Henrich Greve, Joe Labianca, Marjorie Lyles,Sim Sitkin, Ken Spenner, Kevin Steensma, KatherineStovel, Mike Hitt, and our anonymous reviewers. Weappreciate thoughtful copyediting by Tamzin Mitchell.

� Academy of Management Journal2007, Vol. 50, No. 6, 1440–1461.

1440

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 expresswritten permission. Users may print, download or email articles for individual use only.

inherently multilevel (Brass, 2001; Klein, Palmer, &Conn, 2001). Relationships among organizationsand the subgroups that compose them originatewith individuals. Thus, whereas conventional mul-tiplexity refers to variation in the content of ties(e.g., friendship versus business ties) (Granovetter,1973), nodal multiplexity refers to variation in re-lational experiences within pluralistic actors (e.g.,teams, organizations, collectivities).

Nodal multiplexity is relevant when assessinghow individuals’ and organizations’ prior exchangeexperiences influence their subsequent interorgan-izational exchange behavior. Individual membersof the same organization often do not share identi-cal sets of exchange experiences, because of em-ployment turnover and variation in exchange deci-sion processes. Hence, it is not always clear whoseexperiences shape an organization’s behavior.Moreover, it is also not clear whether exchangeexperiences that members of an organizationshared have the same effects as prior exchangesthat organization members experienced separately.

This theoretical problem manifests itself in theresearch on how relational embeddedness affectsinterorganizational exchange. Researchers have ei-ther constrained analyses to the interorganizationallevel without considering the roles of interpersonalties (e.g., Gulati, 1995) or relied heavily on dataprovided by organizational leaders in a context inwhich individual-level and organization-level in-fluences cannot easily be disentangled (e.g., Uzzi,1997). In the first case, researchers risk overattrib-uting interorganizational exchange behavior to or-ganizations’ embeddedness when, in fact, individ-ual ties could be the determinants. Indeed, to theextent that influential individuals change their or-ganizational allegiances over time, it is possiblethat individual-level ties formed before the exis-tence of an interorganizational relationship lead tonew interorganizational exchange relations. In thesecond case, the influences of leaders’ exchangeexperiences could be overemphasized when, infact, the experiences of other members and parts oforganizations are more influential. Moreover, nei-ther of these approaches entertains the possibilitythat the sharing of an exchange experience by or-ganization members could have influence aboveand beyond the influences of distinct, parallel tiesthat an organization and a leader have with thesame exchange partner.

Each of these approaches embodies a distinctrationale. The rationale for explaining interorgani-zational exchange behavior on the basis of leaders’experiences is that leaders have the authorityneeded to inject their views into organizations’ de-cisions (Brass, Galaskiewicz, Greve, & Tsai, 2004).

By contrast, other researchers suggest that leaders’exchange experiences might have only limited in-fluence on organizations’ relational behavior. Otherboundary-spanning members of an organization,such as buying agents and sales personnel, possessrelevant interorganizational experience that mayinfluence organizations’ decisions (Burt, 1999). Be-liefs about the interorganizational exchange envi-ronment may also become embedded in organiza-tion-level repositories such as routines, cultures,and protocols (Fiol & Lyles, 1985). Such beliefsoften become institutionalized before the tenures ofcurrent leaders or lie beyond the control of evenlong-tenured leaders. Thus, a more comprehensivemodel of relational embeddedness and interorgan-izational exchange must accommodate multipleexplanations.

Given this theoretical imperative, we tested amodel that assesses both the influences of leaders’individual-level experiences and those of the aggre-gate of other prior organizational experiences.Moreover, our approach went beyond simply iso-lating the direct effects of leader-to-leader ties onsubsequent interorganizational exchange behaviorfrom the effects of organization-to-organizationties. It included examination of the influences ofleader-to-organization ties, which could reinforceor substitute for organization-to-organization orleader-to-leader ties. In addition, the model iso-lated the influences of shared and unshared ex-change experiences among organizations and theirleaders. We examined 1,657 player trades among30 clubs in Major League Baseball (6,771 club dy-ads), using data from 1978 to 2003.

CONCEPTUAL BACKGROUND

Embeddedness and Interorganizational Exchange

Interorganizational exchange is a means bywhich firms respond to dynamic and unpredictablebusiness environments (Eisenhardt & Martin,2000). Researchers have used several labels to de-scribe organizations that rely on exchange relation-ships, such as “network organizations” (Jones, Hes-terly, & Borgatti, 1997; Miles & Snow, 1986, 1992),“virtual organizations” (Chesbrough & Teece,1996), and “modular organizations” (Lei, Hitt, &Goldhar, 1996; Sanchez & Mahoney, 1996; Schill-ing & Steensma, 2001). Interorganizational ex-change has multiple forms that vary by temporalduration, recurrence, and governance structure(Ring & Van de Ven, 1992). Examples of interorgan-izational exchange include business acquisitions,buyer-supplier relations, and alliances. This studyfocuses on repeated, arm’s-length exchange

2007 1441Barden and Mitchell

relations among horizontally related industry ac-tors. Examples of this kind of interorganizationalexchange include initial public offering syndica-tions, horizontal acquisitions and divestitures,patent and product licensing arrangements amongcompetitors, standard-setting alliances, and sharedproduction arrangements, in addition to the playertrades among sports clubs that this study examines.

The common element among forms of interorgan-izational exchange is the bilateral transfer of re-sources triggered by mutual decision of the ex-change partners (Cook & Emerson, 1978; Jones etal., 1997; Ring & Van de Ven, 1994). Actors inexchange processes serve in dual roles, as both themakers of the exchange decisions and referents ofothers’ decisions. This implies that embeddednessincludes decisional and relational components. In-deed, the mechanisms of embeddedness—knowl-edge, trust, and commitment—facilitate both an ac-tor’s exchange decisions and others’ decisions toexchange with the actor.

Early explanations of exchange highlighted theroles of complementarity (Thibaut & Kelley, 1959;Levine & White, 1961; Pfeffer & Salancik, 1978;Schermerhorn, 1975) and calculative judgmentsabout the utility of a proposed exchange (Cook &Emerson, 1978; North, 1990). However, economiccalculations are not the sole determinant of ex-change. Many dyads of actors with complementaryresources do not exchange those resources. Stillother dyads exchange resources even when actorsknow that superior economic deals exist elsewherein the market. In such cases, the structure andcontent of prior relations that define embeddednessoften play important roles in influencing subse-quent exchange relations by spreading marketknowledge, reducing risks associated with oppor-tunism, and increasing the noneconomic rewardsof exchange (Gulati, 1995; Powell, 1990; Uzzi,1996, 1997). In practice, virtually all relations pos-sess both economic and social elements (Adler,2001; Cook & Emerson, 1978), making this studyapplicable to many forms of exchange.

Relational information acquired through priorexchange experiences promotes subsequent interor-ganizational exchange in two basic ways. First, ex-change ties support the economic basis of exchangeby facilitating calculative judgments. Actors’ priorexchange experiences establish mutual awarenessof each other’s resource portfolios. Because thesearch for exchange opportunities is costly (Geertz,1978), a model of exchange must account for whatmarket actors know and how they learn about eachother (Stigler, 1968). Prior exchange partners rep-resent current options for future social and eco-nomic resources (McGrath, Ferrier, & Mendelow,

2004). Prior exchanges provide information aboutothers’ incentives to behave reliably and can helpdeter opportunism (Gulati, 1995). Prior exchangesalso allow partners to develop relationship-specificexchange routines and structures (Amburgey &Miner, 1992) that reduce the transaction costs ofsubsequent negotiation and resource transfer.

Second, exchange ties enhance the social attrac-tiveness of potential partners by building trust andcommitment. Even when repeated exchanges be-tween partners are unlikely to continue indefi-nitely, partners can build trust by developing reci-procity (Blau, 1964), shared norms of efficiencyand fairness (Ring & Van de Ven, 1994), and otherdispositional similarities (McPherson, Smith-Lovin, & Cook, 2001). Prior exchange relations alsogenerate social outcomes such as friendship thathave value beyond economic calculations (Lawler& Yoon, 1998). Moreover, actors may have interestsin supporting other actors with similar values orpolitical views (McPherson et al., 2001).

Exchange Ties at the Leader and OrganizationalLevels of Analysis

Social influences shape both interpersonal (Blau,1964; Emerson, 1967; Homans, 1974; Thibaut &Kelley, 1959) and interorganizational (Burt, 1976;Coleman, 1973; Cook, 1975; Van de Ven, 1976)exchange behavior. The social determinants of in-terpersonal and interorganizational exchange sharemany characteristics (Emerson, 1976). The similar-ities show that interorganizational researchers haveborrowed from social theories of interpersonal re-lations (Klein, Palmer, & Conn, 2001). Indeed,many studies have extended social structural mod-els of exchange to interorganizational settings (e.g.,Ahuja, 2000; Baker, 1990; Gulati, 1995; Gulati &Westphal, 1999; Kogut, Shan, & Walker, 1992; Uzzi,1997).

Nonetheless, differences arise between ex-changes at the interpersonal and interorganization-al levels, owing to the nodal multiplexity of organ-ization ties. Whereas interpersonal exchangeinvolves atomistic mutual decisions, the multilevelnature of organizations makes interorganizationalexchange more complex (Klein et al., 2001). Inter-actions among microlevel elements can have dis-tinct influences on interorganizational exchangedecisions. Typically, microlevel elements are indi-vidual organization members, but they can also berepositories of organizational schemata such asdocumentation and routines. Each of these microelements is shaped by a distinct set of social expe-riences, and each mediates the influence of socialcontext on organization-level behavior. The

1442 DecemberAcademy of Management Journal

theoretical challenge is determining which mic-rolevel social experiences guide organization-levelbehavior (Adler & Kwon, 2002; Brass, 2001; Oh,Labianca, & Chung, 2005). How one understandswhose exchange ties, or what combinations of ties,shape interorganizational exchange remains anopen question.

In helping answer this question, social structuralmodels offer tools for linking micro- and mac-rolevel structures (Granovetter, 1973). Indeed,some empirical research links microlevel ties tointerorganizational exchange. For example, Gulatiand Westphal (1999) examined how corporateboard interlocks influence alliance formation. Suchresearch focuses on causal links between qualita-tively different types of exchange, however, anddoes not disentangle hierarchically nested ex-change experiences (e.g., organizations’ and lead-ers’ experiences).

Most existing empirical research that links em-beddedness to subsequent interorganizational ex-change constrains analyses to a single level, byfocusing on the influences of organizations’ ties. Aswe have shown, this approach risks theoreticalmisspecification and misleading empirical conclu-sions because individuals and organizations ex-change ties overlap. If organization members spendtheir entire careers with a single organization, theissue is less important because the individuals’ andorganizations’ experiences overlap. However, indynamic labor markets in which individuals tendto change their organizational allegiances, theseeds of an interorganizational exchange relation-ship can arise from individual-level ties that ex-isted before the interorganizational relationship.Studies that constrain analyses to the organization-al level without simultaneously testing the influ-ences of individuals’ exchange ties risk producingresults that overstate the influences of organiza-tion-level relationships.

The complexity of organizations requires a step-by-step approach to a multilevel empirical study ofinterorganizational exchange (Emerson, 1976;Klein et al., 2001). This study takes the first steptoward the development of a microlevel model ofinterorganizational exchange by focusing on isolat-ing the influences of leaders’ ties from the influ-ences of other ties within an organization. Leaders’ties represent a valuable theoretical starting pointbecause leaders’ instrumental and symbolic rolesin organizations suggest that their social experi-ences are likely to be highly influential (Brass et al.,2004).

In this study, organization ties refers to the col-lective exchange experiences of all structures andindividuals that define an organization, excluding

the experiences of its senior leaders. Thus, the pre-dictions and results of this study related to organ-ization ties should be interpreted with care, be-cause organization ties are more complex than asimple leader-organization dichotomy implies.However, this bilevel approach offers a conceptu-ally meaningful and empirically tractable startingpoint for a multilevel theory.

This seemingly simple dichotomy leads to sixcombinations of ties that can influence interorgan-izational exchange behavior. These combinationsinclude prior exchange ties between two organiza-tions (organization-organization, henceforth de-noted as “OO”), prior ties between leaders (leader-leader, “LL”), and prior exchange ties between anorganization and the leader of the other organiza-tion (organization-leader, “OL”). Because organiza-tions and their current leaders often share exchangeexperiences, three additional types of ties can existat any time within a given dyad. Such shared tiesinclude exchange ties shared by both organizationsand both current leaders (organization-organiza-tion-leader-leader, “OOLL”), ties shared by bothorganizations and only one current leader (organi-zation-organization-leader, “OOL”), and ties sharedby only one organization but both current leaders(organization-leader-leader, “OLL”). Table 1a listsexamples of these six types of ties.

Each of these types of ties affects the level atwhich prior relations influence subsequent interor-ganizational exchange. If social influences operateprimarily at the interpersonal level between lead-ers, then exchanges in which both leaders werepresent (LL, OLL, OOLL) will affect subsequentexchange. If social influences operate at other or-ganizational levels, then previous exchanges in-volving both organizations will affect exchange(OO, OOL, OOLL). If social influences operateacross leaders and other organizational elements,then exchanges involving both an organization andthe leader of the other organization (OL, OOL, OLL,OOLL) will affect future exchange. Furthermore,shared exchange experiences of organizations andtheir leaders may reinforce each other, leading tosynergistic influence going beyond the effects ofindividual- or organization-level ties alone. Thechallenge is to reconcile these different kinds ofties with the mechanisms underlying social expla-nations of interorganizational exchange.

THEORY AND HYPOTHESES

Given the dual roles of an actor in an exchangeprocess, the question of whose relational embed-dedness is most influential has two components:First, who shapes a decision to exchange? Before an


interorganizational exchange decision can bemade and executed, organizations’ membersmust reconcile intraorganizational differences inunderstanding the resources, processes, and par-ties involved. It is often unclear whose beliefsand, ultimately, whose prior exchange experi-ences guide subsequent organization-level ex-change decisions.

Second, who or what serves as the referent forothers’ judgments when they make exchange de-cisions? Actors have limited capacity to judge thevalues and reliabilities of all potential transac-tions and partners (Simon, 1955). Actors oftenlack complete information about potential part-

ners and their resource portfolios and/or the cog-nitive agility to process all information that theydo have about a potential partner. As a result,actors must rely on partial information and/orsimplify their cognitive judgments by using heu-ristics (Tversky & Kahneman, 1974). This is im-portant because an actor’s exchange ties to a po-tential partner can vary depend on the way thatpotential organizational partner has changedsince prior exchange experiences. Thus, deter-mining which ties influence subsequent interor-ganizational exchange requires understandingwhich subcomponents of a potential partnercommand the decision maker’s attention.

TABLE 1aPossible Organization and Leader Exchange Ties between Two Organizations

Tie Type Generalized IllustrationaMajor League Baseball Illustration for

Potential Trades in 2002

OO Prior exchanges that involved both organizations, butnot their current leaders: The Star Company andMoon Inc. dealt with each other when the two or-ganizations had different leaders.

In 1999, the Boston Red Sox traded players with the DetroitTigers when neither Theo Epstein nor David Dombrowskiwas general manager of their respective 2002 team.

LL Prior exchanges that involved both current leaders ofthe organizations but not the organizationsthemselves: Norma and Chandra dealt with eachother when they led other organizations.

In 2001, the 2002 Baltimore Orioles’ general manager, JimBeattie, and the 2002 Milwaukee Brewers’ generalmanager, Doug Melvin, traded players when they weregeneral managers of different teams.

OL Prior exchanges that involved one organization andthe current leader of the other organization but notthe focal organization’s current leader and not theother organization: Star dealt with Chandra beforeNorma’s tenure at the Star Company and whenChandra led an organization other than Moon Inc.;symmetrically, Moon Inc. dealt with Norma beforeChandra’s tenure at Moon and when Norma led anorganization other than the Star Company.

In 1999, the 2002 Anaheim Angels’ general manager, BillStoneman, traded players with the Seattle Mariners whenStoneman was the general manager of a different teamand before the 2002 Seattle Mariners’ general manager,Pat Gillick, joined the Mariners’ organization.

OOL Prior exchanges that involved both organizations andthe current leader of one of those organizations butnot the other current leader: The Star Companydealt with Moon Inc. while Norma led Star butChandra did not lead Moon; symmetrically, theStar Company dealt with Moon Inc. while Chandraled Moon but Norma did not lead Star.

In 2000, the Boston Red Sox traded with the CincinnatiReds when Jim Bowden was general manager of the Reds,but Theo Epstein was not general manager of the RedSox.

OLL Prior exchanges that involved the current leaders ofboth organizations and one of the organizations butnot the other organization: Norma dealt withChandra while Norma led the Star Company butChandra did not lead Moon Inc.; symmetrically,Norma dealt with Chandra while Chandra ledMoon Inc. but Norma did not lead the StarCompany.

In 2000, the Cincinnati Reds’ general manager, Jim Bowden,traded with David Dombrowski when Dombrowski wasgeneral manager of a team other than the Detroit Tigers.

OOLL Prior exchanges that involved both organizations andtheir current leaders: The Star Company and MoonInc. dealt with each other while Norma andChandra led the organizations

In 1999, the Atlanta Braves traded with the San DiegoPadres when John Schuerholz and Kevin Towers weregeneral managers of the two teams.

a The examples focus on two leaders and their organizations: Norma is the current leader of the Star Company; Chandra is the currentleader of Moon, Inc.


Interleader Ties

Both the decisional and referential aspects ofleaders’ relational embeddedness influence interor-ganizational exchange. First, the knowledge, trust,and commitments that leaders garner from theirpersonal exchange experiences are influentialbecause leaders bear the ultimate responsibility fororganizational decisions, including those related tointerorganizational exchange. This responsibility isparticularly strong in highly centralized, hierarchi-cal organizations where, by definition, power tendsto be concentrated in senior leaders (Crozier, 1964;Mintzberg, 1979). Behavioral decision making re-search also suggests that leaders’ experiences areparticularly influential because decision makerstend to limit their searches for information andalternatives to their own experience, given searchcosts and cognitive limits (Simon, 1955). Further-more, decision making research suggests that whenselecting among alternatives, leaders tend to be bi-ased toward their own ideas (Beach, 1993).

Second, a leader can also shape the way othersin a marketplace view the attractiveness of theleader’s organization. If one views an organiza-tion as a resource delivery system for which aleader bears responsibility for effective function-ing, then others must make some judgment aboutthe reliability and value that the leader can pro-vide. As an organization’s key boundary spanner,a leader must be trusted to negotiate honestly andcompetently. As an architect of organizationalaction, a leader must be trusted to direct ex-change processes competently. Even when aleader has less than complete influence over theorganization’s resource delivery system, judg-ments about the leader can be important criteriaduring the exchange decision. Indeed, a leader’sreputation can provide a heuristic for the reliabil-ity of a referent organization in others’ exchangedecisions. Finally, a leader can be a direct sourceof value during an exchange because of the lead-er’s knowledge, skills, legitimacy, and friend-ships. Thus, potential partners’ exchange experi-ences with an organization’s leader constituteimportant bases for subsequent exchange deci-sions, even if those experiences occurred whenthe leader was not a member of the focalorganization.

We state a general proposition, followed by ahypothesis that refers to specific ties:

Proposition 1. Prior exchange ties between twoorganizations’ leaders increase the likelihoodof subsequent exchange between the twoorganizations.

Hypothesis 1. LL, OLL, and OOLL ties all in-crease the likelihood of subsequent interorgan-izational exchange.

Organization-Level Ties

Several elements of organizations other thantheir leaders can play decisional and referentialroles in interorganizational exchange. First, otherorganization members can play boundary-spanningroles. Because leaders often do not have the capac-ity to gather and process comprehensive informa-tion about the marketplace, support staff such assales and purchasing personnel frequently mustformulate exchange alternatives. Such diffusion ofdecision-making responsibilities defines the decen-tralized organization (Crozier, 1964) and allowsnonleaders to inject their own social-experience-based beliefs into interorganizational exchange de-cisions. Boundary-spanning organization memberswith little decision-making authority can alsoshape organization-level exchange decisions whenleaders see value in their alternative perspectives(Burt, 1999). When a leader draws on the experi-ences of other boundary spanners to make interor-ganizational exchange decisions, the social basis ofthe exchange shifts from the leader’s exchange ex-periences to those of the broader organization. Inaddition, other boundary-spanning organizationmembers, such as salespeople, frequently serve asheuristics for reliability and value in others’ interor-ganizational exchange decisions.

Second, social information can be processedwithin organizational cognitive structures. Organi-zational learning research suggests that organiza-tions have cognitive capabilities that are distinctfrom the cognitive capabilities of their individualmembers (Fiol & Lyles, 1985). Organizations canstore knowledge and beliefs in locations other thantheir individual members’ memories. Such reposi-tories include archives, routines, culture, and tech-nologies (Hedberg, 1981). For example, onlinesocial networking technologies, such as Meeting-Maker™ and LinkedIn®, can facilitate interorgani-zational links. The interactivity of collectiveknowledge acquisition and usage can also influ-ence organizations’ social behavior (Thompson, Le-vine, & Messick, 1999). This view of organizationsas distinct, interactive cognitive systems is consis-tent with the idea that trust operates at the interor-ganizational level (Zaheer, McEvily, & Perrone,1999). When organization-level social cognitionprocesses exist, organization-level ties will influ-ence interorganizational exchange behavior.

Third, prior exchange experiences can determinethe expected value of future potential exchange


with previous organization-level partners. Struc-tures and routines developed during past exchangerelationships (Gulati, 1995) enhance the referentvalue of potential organization-level partners. Pastexperiences not only reduce the costs of replicatingexchange routines, but also build trust in an organ-ization as a resource delivery system. The referentvalue of potential organization-level exchangepartners also arises because organizations can takeon distinct psychological meaning (Selznick,1957). Such meaning causes others to project hu-man qualities onto organizations and helps developpsychologically valuable relational commitment(Levinson, 1965). In sum, because the cognitive andreferent elements of organizations influence interor-ganizational exchange processes, interorganization-al exchange ties will influence patterns of subse-quent interorganizational exchange.

Proposition 2. Prior exchange ties between adyad of organizations increase the likelihoodof subsequent exchange between the twoorganizations.

Hypothesis 2. OO, OOL, and OOLL all increasethe likelihood of subsequent interorganization-al exchange.

Exchange ties between a leader and an organiza-tion can also be influential. Such ties take multipleforms. Interpersonal ties can exist between a leaderfrom one organization and nonleaders from anotherorganization. If a nonleader is sufficiently influen-tial, that interpersonal relationship can serve as abasis for future interorganizational exchange. Thus,the logic of interpersonal relational embeddednessis applicable.

The mechanisms of relational embeddedness canalso operate over levels between an individual andan organization. Exchange ties between individualsand organizations have been discussed in the con-text of employment relationships (Levinson, 1965).The existence of individual-organization relationsis a basic assumption underlying concepts such asorganizational commitment (Mowday, Porter, &Steers, 1982) and organizational citizenship (Organ& Konovsky, 1989). Indeed, research shows thatemployee-leader relationshipsandemployee-organ-ization relationships are distinct (Wayne, Shore, &Liden, 1997).

Although some scholars have suggested that theonly social links that can exist between individualsand collectivities is membership (e.g., Goffman,1971), others have cited alternative links (Breiger,1974). Research on interactions between patientsand hospitals (Reider, 1953) and on consumer atti-tudes toward product suppliers (Lessig, 1973) sup-

ports the view that individual-to-organization rela-tionships can span organizational boundaries.Indeed, concepts such as brand equity and brandloyalty imply that trust and commitment cross lev-els of analysis. Thus, when an influential leaderuses prior experiences with another organizationwhile making subsequent interorganizational ex-change decisions, leader-to-organization ties arerelevant. Alternatively, when an organization usesprior experiences with another organization’sleader while making subsequent interorganization-al exchange decisions, organization-to-leader tiesbecome relevant.

Proposition 3. Prior exchange ties between anorganization and the leader of another organ-ization increase the likelihood of subsequentexchange between the two organizations.

Hypothesis 3. OL, OLL, OOL, and OOLL allincrease the likelihood of subsequent interor-ganizational exchange.

Shared Ties

When an organization’s leader and other actorsshare prior exchange experience, two sets of factorsinteract to increase the degree to which those ex-change ties influence subsequent interorganization-al behavior. First, for several reasons, knowledge,trust, and commitments developed through ex-change experiences shared by both an organizationand its leader are more likely to be available forsubsequent exchange decisions. On the most basiclevel, this shared experience creates a richer organ-izational memory system that increases the likeli-hood that leaders and other organizational ele-ments will recall and use the resulting relationalinformation when weighing subsequent exchangealternatives (Walsh & Ungson, 1991). In addition,relational information garnered through prior ex-change experience shared by a leader and otherorganizational elements is also more likely to be-come institutionalized within the organization’sstructures and culture, increasing the legitimacy ofits subsequent use (Douglas, 1986). Furthermore,relational information acquired during shared ex-change experiences is more likely to create salientsymbols of the intraorganizational relationshipsthat produced the exchange. For example, sharedexchange experiences may affirm and streamlinethe subsequent interactions between leaders andthe other organization members. This relational in-teraction inside the organization may institutional-ize information and beliefs about the interorganiza-tional environment in cultural structures, such asthe organization’s rules, mythology, and written


records. Thus, organizations are more likely to userelational information garnered through exchangeexperiences organizational leaders and otherboundary-spanning elements share.

Second, external market actors are likely to bemore confident in the reliability of their own judg-ments about a potential exchange partner whentheir prior exchange experiences involved both thepartner organization and its current leader. A po-tential partner cannot comprehensively judge itspartner’s systemic reliability by observing the or-ganization’s subcomponents independently. Ex-change experiences with an organization and itscurrent leader provide opportunities to observe theinteractions between the two. Decision makers willhave greater confidence in exchange decisionswhen those decisions are based on prior experi-ences with both a potential organizational partnerand its current leader.

Proposition 4. The influence of prior exchangeties on the likelihood of subsequent interorgan-izational exchange is stronger when leadersand other organizational elements share theprior experiences.

Because there are three types of unshared ties(OO, LL, and OL) and one type of fully shared tie(OOLL), the logic underlying Proposition 4 leads tofour comparative hypotheses:

Hypothesis 4a. OLL and OOLL both havestronger positive influences than LL on thelikelihood of subsequent interorganizationalexchange.

Hypothesis 4b. OOL and OOLL both havestronger positive influences than OO on thelikelihood of subsequent interorganizationalexchange.

Hypothesis 4c. OOL, OLL, and OOLL all havestronger positive influences than OL on thelikelihood of subsequent interorganizationalexchange.

Hypothesis 4d. OOLL has a stronger positiveinfluence than OLL and OOL on the likelihoodof subsequent interorganizational exchange.

The hypotheses define a partially specified rankordering of the relationships between the types ofexchange ties and subsequent interorganizationalexchange, while also posing empirical questionsabout the relative importance of organization tiesand leader ties. Table 1b summarizes the ordering:9 of the 15 pairwise comparisons have a clear order,but the other 6 are uncertain. The clearest rankedposition in the table is that OOLL (fully sharedorganization-leader ties) will have the strongest in-fluence. OLL (one organization and both leaders)dominates OL (one leader and one organization)and LL (both leaders but neither organization), butmight not dominate OO (both organizations with-out current leaders) if organization ties are veryinfluential. In parallel, OOL (one leader and bothorganizations) dominates OO (both organizations)and OL (one leader and one organization), butmight not dominate LL (both leaders in other or-ganizations) if leader ties are very influential. Sim-ilarly, the rank order of OOL and OLL depends onwhether organization ties or leader ties have themost influence. The rank order of the three simplercombinations—OO, LL, and OL—also depends onwhether organization ties or leader ties have moreinfluence. We used this empirical ordering to helpdevelop a more nuanced understanding of whenorganization experience is more important andwhen leader experience dominates in a decision toexchange.

METHODS

Sample, Data, and Statistical Method

We examined the determinants of Major LeagueBaseball (MLB) player trades. Player trading is animportant part of many professional sports indus-tries. Because teams own the contractual rights to

TABLE 1bPredicted Rank-Order Influence of Combinations of Leader-Organization Exchange Ties

on Subsequent Interorganizational Exchangea

Type of Tie OL OO LL OOL OLL

OOLL �� OLL �� ? �� ?OOL �� ?LL ? ?OO ?

a Each cell reports the expected relationship of a type of tie to the type in the top row. Cells containing a question mark (“?”) haveambiguous relative influence.


specific players, they are able to trade these rightswith other teams to adjust financial liabilities andstocks of player skills.

MLB player trades offered several advantages as aresearch context. The relational mechanisms of em-beddedness (Uzzi, 1997)—information, trust, andcommitment—arise in this setting. Prior exchangesoften provide information about the values andavailability of the players who are contractuallybound to major league teams at any given time.Although a variety of player statistics and otherfacts are publicly available, the availability andusefulness of this public information are limited(Schuerholz & Guest, 2006). For example, it is moredifficult to get complete information about the4,000–5,000 minor league players who are contrac-tually bound to major league organizations than itis to get information about major league players.Making this informational task even more challeng-ing is the fact that the market values of these play-ers constantly change over time owing to injuries,player development, and the spread of knowledgeabout players’ weaknesses. Indeed, teams strugglewith the challenges of player information overload(Lewis, 2003), making exchange processes impor-tant means by which teams refresh their knowledgeabout other teams’ players.

Trust developed through prior exchanges is im-portant because player trades involve substantialuncertainty (Schuerholz & Guest, 2006; Thrift &Shapiro, 1990). Many injuries and other problems(e.g., hairline bone fractures, bone spurs, back prob-lems, and illegal drug use) are not obvious duringstandard physical exams and sometimes spur pub-lic disputes between teams. For example, the NewYork Yankees’ general manager, Bob Watson, pub-licly suggested that his trust in the MilwaukeeBrewers’ general manager, Sal Bando, was shakenafter a 1996 trade dispute involving an injuredplayer. The Kansas City Royals’ general manager,John Schuerholz, kept a San Francisco Giants bat-ting helmet displayed in his office for three years asa reminder not to deal with the Giants organizationbecause he believed that they knowingly traded aplayer with a cocaine addiction to the Royals in1982 (Schuerholz & Guest, 2006: 239). Generalmanagers must also be trusted not to waste eachothers’ time because the values of different poten-tial trades are often contingent on each other (Thrift& Shapiro, 1990).

Finally, there is anecdotal evidence that commit-ments developed through prior trades influence ex-change decisions (Schuerholz & Guest, 2006; Thrift& Shapiro, 1990). Indeed, John Schuerholz statedthat he began considering trades with the Giantsagain after his friend and former trading partner, Al

Rosen, left the Houston Astros to become the Gi-ants’ general manager in 1985 (Schuerholz & Guest,2006: 240).

MLB trades offered three other methodologicaladvantages as a study setting. First, the definedboundaries of the industry allowed us to fully spec-ify the set of potential trading partners in any givenperiod. Defined scope was particularly valuablegiven the longitudinal nature of the study. In 2002,there were 30 MLB team organizations. In thisstudy, each team organization is defined as a con-tinuous legal entity, which is bound neither byteam name nor by the franchise’s host city. Forexample, the Seattle Pilots moved to Milwaukee in1970, with a change of name to the Brewers. Weconsidered the two teams to be the same organiza-tion. We treated the Seattle Mariners club, whichwas created in 1978 as an expansion team, as a newentity. We used the dyad-year as the base unit ofanalysis (e.g., there were 435 club dyads in 2002).Study years began on April 1, at the beginning ofthe baseball season, and ended on March 31 of thefollowing calendar year. Because of incrementalexpansion in the number of club franchises overthe years, not all baseball clubs included in thestudy existed throughout the entire period of thestudy, April 1, 1985, to March 31, 2003. We used1978 as the starting point for observing prior ex-change experiences. There were 6,771 dyad-yearsat risk of exchange.

Second, longitudinal data were available for allplayer trades, which helped us assess causality.Trades involved the exchange of exclusive contrac-tual rights to players’ services. Between 1985 and2002, the number of trades per year ranged from 41in the strike-shortened 1994 season to 123 in 2000.A total of 1,657 distinct trades between dyads ofteams occurred over the 17-year period. We col-lected data on trades and team management fromthe Sporting News Official Baseball Guide. Therewere no missing trade data.

Third, baseball organizations possess well-defined repositories of knowledge and beliefs thatdefine the mechanisms of relational embedded-ness. The focus of this study was the role of ageneral manager as one of these repositories. “Gen-eral manager” (GM) is the title usually used inbaseball to identify the leader of a team’s baseballoperations. Although GMs often do not presideover nonbaseball functions, such as marketing andpark maintenance, they are directly responsible forengineering player trades and other important base-ball-related decisions (Thrift & Shapiro, 1990).Since the retirement of Minnesota Twins’ ownerCalvin Griffith in 1984, no team owner has held thedual role of owner and general manager. Active


team owners, such as George Steinbrenner of theNew York Yankees, sometimes make suggestionsabout the kinds of human resources needed andassess the financial viability of potential trades, butthey are not directly responsible for formulatingand executing player trades. Although proactiveowners sometimes shape teams’ complementarities(e.g., refusals to trade certain players), owners haveless direct influence on the information, trust, andcommitment mechanisms of relational embedded-ness that are at the heart of this study.

The leadership role of general managers in for-mulating and executing player trades is clear; theactual influence of a particular general manager’sexchange experiences on subsequent exchange be-havior is less clear. Indeed, the dichotomy of aca-demic viewpoints about the influence of leaders oninterorganizational exchanges extends into popularpress coverage of baseball. With some evidencepointing to GMs as the dominant force in interor-ganizational exchange relations, other evidencesuggests that the influence of general managers isless direct.

A number of authors have articulated the viewthat baseball general managers are the dominantforce in teams’ exchange relations (e.g., Brown &Eisenhardt, 1998; Lewis, 2003; Thrift & Shapiro,1990). Belief in the direct influence of general man-agers is also evident in the high accountability towhich team owners, popular press analysts, andfans hold GMs. This intense accountability contrib-utes to frequent movement of general managers be-tween teams. There is intense competition for suc-cessful general managers, who are often lured awayby other teams. Moreover, general managers whoare fired for poor team performance are often re-hired by other teams. Of the people who took gen-eral manager positions during the observation pe-riod and served in that capacity for longer than oneyear, 40 percent were employed as general managerby more than one team. This frequent job switchinghad the empirical advantage of creating a diver-gence between the exchange ties of the organiza-tions we studied (the baseball clubs) and the ex-change ties of the individual leaders (the generalmanagers).

In addition, the relatively small size and hierar-chical nature of Major League Baseball organiza-tions suggested that general managers and theirexchange experiences should be influential for ourprediction of subsequent interorganizational ex-change. Decision-making authority is generallymost concentrated in leaders in small, hierarchicalorganizations, so that the influence of leader em-beddedness should be strongest in such a context(Uzzi, 1997). Therefore, Major League Baseball as a

context permits a conservative test of whether lead-ers are as influential as the popular press and base-ball’s other constituencies imply.

General managers also face constraints. Some au-thors have highlighted the influences that organi-zational support staff members, such as talentscouts and medical personnel, have in generalmanagers’ exchange decisions (Schuerholz &Guest, 2006; Shanks, 2005; Thrift & Shapiro, 1990).If such influences are strong, then general manag-ers’ ties may be less influential than those of therest of the organization.

Thus, there are varying views of the relative rolesthat leaders’ and aggregate organizations’ social re-lations play in shaping interorganizational ex-change behavior. Overall, Major League Baseballprovided a definable context that allowed us tostudy processes that arise in many business andsocial contexts.

We used Cox regression analysis to predict thelikelihood (log hazard rate) of a player trade be-tween the members of a dyad of teams. The depen-dent variable was a dichotomous measure signify-ing whether a trade event occurred between twoteams in a given year in the 17-year observationwindow (1 indicated a trade; 0 indicated no trade).Although a few team dyads traded more than onceduring a given year, dichotomization caused littleinformation loss because less than 3 percent of thedyad-years at risk in the study experienced morethan one trade (removing these cases did not sub-stantially change the results). Moreover, dichoto-mization enhanced the clarity of the analyses andprevented problems created by contagion withinthe dependent variable (e.g., a trade in July couldmotivate a trade in November of the same yearunder observation).

The longitudinal nature of the study made theuse of Cox regression particularly apt. Cox regres-sion allows covariates to change after the last oc-currence of the dependent variable (Tuma & Han-nan, 1984). This variability is especially usefulbecause complementarities tend to ebb and flowover time as a result of actors’ exchange behaviorsand resource portfolio redevelopments. Cox regres-sion also accounts for the statistical informationassociated with “right–censored” cases, or spells oftime between dyad exchange events that are cut offby the end of a study’s observation window (thepresent study had no “left-censored” cases).

Independent Variables

Disaggregated exchange ties. The strength ofties between actors at different levels captures thesocial structure of interorganizational exchange.


These include interorganizational ties, interper-sonal ties, and person-organization ties. Becauseoverlap of these ties at different levels creates mul-ticollinearity, we separated shared ties from un-shared ties to isolate the effects of each level as abasis for a type of ties. The tendency of MLB gen-eral managers to move from organization to organ-ization makes this disaggregation meaningful. Theanalysis modeled the probability of an exchange ina dyad of teams in a given year as a function of theorthogonal set of six leader and organization tiecombinations:

P(Exchange) � OOLL � OOL � OLL � OO � LL

� OL � e.

We weighted the exchange ties measures to ac-count for both the number of players involved inprevious trades and the decay of prior experienceover time. First, we expected prior exchanges in-volving greater complexity and risk to have greaterinfluence on the value of experience. Thus, weweighted each prior exchange by the number ofplayers involved. Second, we weighted prior ex-changes by a decay function that captured memoryloss, resource portfolio changes (to teams’ playerrosters), changes in organizations’ personnel, andother structural changes. Because rates of decay aredifficult to determine a priori, we tested the explan-atory values of exchange ties using one, two, three,four, five, and seven-year linearly decayed and un-decayed sliding windows. We found two relevanthistorical decays, which reflected intriguing differ-ences in the decay rate of organization and leaderexperience. A linearly decayed three-year windowprovided the best statistical fit for the interorgani-zational tie measures (OO, OOL, and OOLL), andan undecayed five-year sliding window providedthe best fit for the other exchange tie measures. Insum, coefficients on the exchange tie variables canbe interpreted as the degree to which a change inthe decayed number of previously traded playerschanges the likelihood that a dyad of teams willtrade players again at a given future time.

Control variables. We used several control vari-ables and sensitivity analyses to assess the robust-ness of the results. This section describes variablesthat appear in the reported analysis. The Appendixdescribes other efforts to assess sources of variance.

The analyses include a dummy variable, “dyadteams in same division,” indicating whether a dy-ad’s members operated in the same division, as acontrol for the fact that interdivisional rivals com-pete against each other for playoff spots. In 1993,team owners also approved the “wild card playoffrule,” which increased rivalry among teams in the

same league. Therefore, we included anotherdummy variable that indicated whether dyad teamswere in the same league after 1992 (“same league �wild card years”).

Other sources of organizational embeddednessmay also be relevant. Teams from the same leagueplay substantially more games against each otherthan against teams from different leagues. Thispractice allows teams from the same league togather more information about each others’ players(Thrift & Shapiro, 1990). Therefore, the analysesincluded a dummy variable indicating whetherdyad teams were from the same league for all yearsof the study (“dyad teams in same league”). Inaddition, because each member of an organizationis a potential interorganizational contact pointthrough which relational information can travel,we included the total employment of team dyads asa control variable (“dyad total employment”). Thismeasure included all executives, scouts, and ad-ministrative personnel, but not players. The em-ployment histories of team leaders can also shapeorganizational embeddedness. General managerswho have been employed by other teams couldpossess more information about previous employ-ers or feel trust and commitment to former employ-ers. Thus, we included a variable, “cumulativeleader employment ties,” that denotes the numberof general managers in a given dyad who have beenemployed by their potential trading partners asplayers or administrators (decaying employmentties did not improve the statistical fit of the model).

Other factors may influence the sense of urgencywith which organizations engage in exchange rela-tions. Research on “problemistic search” (Cyert &March, 1963) suggests that performance can influ-ence exchange behavior. We controlled for dyadteams’ average winning percentage with the vari-able “dyad average team performance.” Becausequalitative evidence suggests that new generalmanagers tend to reshape teams, we included avariable denoting the cumulative tenures of generalmanagers with their respective teams, “dyad totalleader tenure.”

Several contextual forces can influence patternsof exchange. First, because historical period effectssuch as economic cycles can influence the risksassociated with interorganizational exchange, theanalyses included dummy variables for each yearfrom 1985 to 2001. Second, because individualteams may have distinct propensities to undertakeexchanges as the result of inertia or distinct cul-tural characteristics, the analyses included dummyvariables for each team (the Toronto Blue Jays clubserved as the comparison group). Because the pop-ulation was small and the statistical power of the


study was substantial, this fixed-effects solutionwas a feasible method of controlling for the non-independence of the dyad-years.

Major League Baseball created four new teamsduring the study period (the Colorado Rockies andFlorida Marlins in 1993; the Arizona Diamond-backs and Tampa Bay Devil Rays in 1998). The newteams faced different regulatory and economiccontexts than long-standing teams; for example, ex-pansion drafts were conducted in which expansionteams could select players from other teams. Theanalyses included two control variables to capturethe numbers of teams in each dyad that were ex-pansion teams in the year after the expansiondrafts.

The overall network structure also might affectdyadic exchange relations. We tested the effects ofseveral structural measures at both the interorgan-izational and interleader levels. The final analysisincluded the only measure that had a significanteffect on interorganizational exchange, dyad teams’“flow betweenness.” Flow betweenness indicatesan actor’s power within an exchange network be-cause it measures the degree to which the actoroccupies valued paths between other actors. (Free-man, Borgatti, & White, 1991). (The Appendix de-scribes other structural measures not used in ourfinal analysis.)

Anecdotal evidence suggests that teams are morelikely to trade players who are in the final year oftheir contracts (Thrift & Shapiro, 1990). Once aplayer’s contract expires, the player becomes a“free agent” and is allowed to sign a contract withany team. The prospect of losing contractual con-trol of a player reduces a team’s incentive to keephim on the roster, especially if he is highly paidand the team has little chance of making the play-offs. Therefore, we controlled for the number ofplayers in the final year of their contracts on eachteam’s major league roster at the beginning of aseason (“dyad total free agents”).

Finally, controlling for asset complementarity isessential in any study of exchange behavior (Gulati,1995). We tested four measures of complementarity(skill differences, winning percentages, marketsize, and payrolls). When the regression analysisincluded all four measures, the only significantvariable was “dyad teams’ skill difference,” whichwe therefore retained in the final analysis (the dif-ference in dyad teams’ winning percentages wassignificant when models omitted the skill differ-ence measure). The Appendix describes thecomplementarity measures.

The skill difference variable was needed becauseteams with lower performance often trade experi-enced and more expensive veteran players to teams

with stronger performance during each season’srace to qualify for the playoffs (Lewis, 2003;Schuerholz & Guest, 2006; Shanks, 2005). Suchtrades allow the poorer performers to lower payrollcosts and lets the better performers make rosteradjustments to maximize chances of succeeding inthe playoffs. To control for dyad teams’ skill differ-ences, we created a variable measured by summingthe dyad teams’ differences in normalized pitchingand batting statistics each season:

SDAB, t � Zt(RPIA, t) � Zt(RPIB, t)� � �Zt(ERAA, t)

� Zt(ERAB, t)�.

This measure of skill differences between teamsA and B in year t (SDAB, t) is calculated as a functionof each team’s earned-run average (ERA; a measureof the strength of a team’s pitching staff) and runsscored per inning (RPI; a measure of the strength ofa team’s offense), standardized in relation to year t(Zt).

Table 2 reports the descriptive statistics of thevariables in the final analysis.

RESULTS

Table 3 reports the results of the regression anal-yses. Model 1 illustrates how the control variablesinfluence the likelihood that the two teams in adyad will trade players in a given year. Model 2,which explains significantly more variance thanmodel 1, reports the effects of the exchangetie measures.

The results in model 2 partially support Hypoth-esis 1, which predicts that exchange ties betweenorganizational leaders (LL, OLL, and OOLL) in-crease the likelihood that those leaders’ organiza-tions subsequently exchange. On the one hand,prior exchanges in which both general managerswere involved but only one was leading his/hercurrent organization (OLL) significantly predictedthe likelihood of subsequent exchange. On theother hand, ties formed between general managerswhen they were not leading their current organiza-tions (LL) did not significantly predict subsequentexchange between their current organizations. Thisresult would not be possible if leaders’ exchangeexperiences had no influence above and beyondorganization exchange experiences (OO). In addi-tion, prior exchanges that involved both currentleaders and both of the organizations being ob-served in the dependent variable (OOLL) signifi-cantly influenced the likelihood of subsequent ex-change. One must interpret these results withcaution, because it is possible that leader-to-organ-


TA

BL

E2

Des

crip

tive

Sta

tist

ics

and

Cor

rela

tion

s

Var

iabl

es1

23

45

67

89

1011

1213

1415

1617

18

Cor

rela

tion

s1.

Dya

dex

chan

ge2.

Dya

dT

eam

s’to

tal

flow

betw

een

nes

s.0

83.

Dya

dT

eam

s’sk

ill

dif

fere

nce

.06

.02

4.D

yad

aver

age

team

per

form

ance

�.0

2.1

8.0

95.

Dya

dto

tal

emp

loym

ent/

10.0

2.0

5.0

8.2

56.

Dya

dte

ams

insa

me

div

isio

n�

.03

.03

.03

.02

�.0

17.

Dya

dte

ams

insa

me

leag

ue

0.0

1.0

3.0

6.0

3.4

68.

Cu

mu

lati

vele

ader

emp

loym

ent

ties

.01

.01

.01

�.0

20

.02

.02

9.D

yad

tota

lle

ader

ten

ure

�.0

7�

.18

0.1

8.0

9.0

1.0

2�

.01

10.1

993

exp

ansi

onte

ams

�.0

3�

.19

�.0

5�

.54

�.3

0�

.04

�.1

.01

�.1

111

.199

8ex

pan

sion

team

s�

.01

�.1

5.0

5�

.06

.11

0.0

10

.02

�.0

212

.Dya

dto

tal

free

agen

ts0

�.0

60

.26

.31

�.0

2.0

10

.08

�.1

4�

.07

13.B

oth

orga

niz

atio

ns,

both

lead

ers

(OO

LL

).0

4.1

8.0

2.0

8.0

5�

.04

0�

.01

.11

�.0

6�

.01

.02

14.B

oth

orga

niz

atio

ns,

one

lead

er(O

OL

)0

.2.0

3.0

5.0

50

.03

.01

�.1

3�

.06

�.0

5�

.02

.03

15.O

ne

orga

niz

atio

n,

both

lead

ers

(OL

L)

.03

.03

.01

�.0

6�

.04

�.0

1�

.04

.02

0.0

9�

.03

�.0

40

.04

16.B

oth

orga

niz

atio

ns

(OO

).0

7.1

40

�.0

30

0.0

4�

.01

�.2

5�

.03

�.0

3�

.06

.01

.09

�.0

217

.Bot

hle

ader

s(L

L)

0.0

3.0

4�

.02

.01

.05

.02

0�

.06

.01

�.0

1�

.01

00

�.0

1.0

118

.On

eor

gan

izat

ion

,on

ele

ader

(OL

).0

4.0

80

�.0

4�

.03

0�

.03

0�

.21

.02

�.0

3�

.04

.02

.01

.02

.11

.16

Des

crip

tive

stat

isti

csM

inim

um

00

00.

343.

60

00

0.5

00

00

00

00

0M

axim

um

120

.75

6.43

0.68

221

12

292

222

1510

.715

.55

1618

15M

ean

s0.

217.

671.

520.

513

.10.

160.

480.

129.

140.

020.

028.

270.

660.

40.

230.

120.

030.

1S

tan

dar

dd

evia

tion

s0.

412.

631.

120.

062.

20.

370.

50.

334.

380.

110.

123.

771.

40.

950.

980.

550.

390.

65

ization ties in the OLL and OOLL combinations,rather than the leader-to-leader ties, could havedriven the significant results.

The results support Hypothesis 2, which predictsthat exchange ties between organizations in a dyad(OO, OOL, and OOLL) increase the likelihood thatthe dyad members will subsequently exchange. Ex-change ties between organizations when they werenot led by their current leaders (OO) significantlypredicted subsequent interorganizational ex-change. Furthermore, prior exchanges involvingboth dyad organizations with at least one leaderalso involved (OOL and OOLL) also significantlypredicted the likelihood of subsequent exchange.

The results partially support Hypothesis 3,which predicts that exchange ties between an or-ganization and the leader of another organizationincrease the likelihood that the organizations willexchange (OL, OLL, OOL, OOLL). Three of the fourties involving an organization and the currentleader of another organization (OOL, OLL, andOOLL) significantly predicted the likelihood of

subsequent exchange. However, the coefficient thatcould isolate the influence of leader-organizationties (OL) is not statistically significant. Thus, theseresults cannot directly distinguish the influences ofleader-to-organization ties from the influences ofinterorganizational ties or interleader ties. We dis-cuss this point in more detail below.

Proposition 4 generates four hypotheses focusingon prior exchange experiences shared by leadersand other members of their organizations. Mostresults support the predictions.

The results in model 2 of Table 3 partially sup-port Hypothesis 4a, which predicts that interleaderexchange experiences shared with one or more ofthe organizations under observation (OLL, OOLL)have stronger influence than unshared interleaderties (LL). Coefficient comparison tests (Table 4)suggest that the positive influences of OOLL (bothorganizations and both leaders) are larger than theinfluences of LL (both leaders outside their currentorganizations). In turn, the positive influences ofOLL (both leaders and one organization) are mod-

TABLE 3Results of Cox Regression Analysis Predicting Interorganizational Exchange

Variables

Model 1 Model 2

b s.e. Exp(b) b s.e. Exp(b)

Exchange ties betweenBoth organizations, both leaders (OOLL) 0.34*** 0.02 1.40Both organizations, one leader (OOL) 0.33*** 0.03 1.40One organization, both leaders (OLL) 0.07** 0.03 1.08Both organizations (OO) 0.19*** 0.03 1.21One organization, one leader (OL) 0.02 0.02 1.02Both leaders (LL) �0.06 0.08 0.94

Control variablesDyad teams’ total flow betweenness 0.06*** 0.01 1.06 0.03* 0.01 1.03Dyad teams’ skill difference 0.13*** 0.02 1.14 0.13*** 0.02 1.14Dyad Average team performance �1.60** 0.63 0.20 �1.66** 0.64 0.19Dyad total employment / 10 0.05** 0.02 1.05 0.04* 0.02 1.04Dyad teams in same division �0.39*** 0.09 0.68 �0.30*** 0.09 0.74Dyad teams in same league 0.31*** 0.09 1.36 0.16* 0.10 1.18Same league � wild card years �0.24* 0.11 0.79 �0.13 0.11 0.87Cumulative leader employment ties 0.13 0.08 1.09 0.12† 0.08 1.14Dyad total leader tenure �0.04*** 0.01 0.96 �0.03*** 0.01 0.971993 expansion teams 0.39 0.42 1.47 0.83* 0.42 2.281998 expansion teams 0.95** 0.39 2.59 1.26*** 0.40 3.53Dyad total free agents 0.02** 0.01 1.02 0.03** 0.01 1.03Year and team dummy variables Included Included

�2 (df) 480.3*** (58) 1,577.4*** (64)��2 (df) 1,097.1*** (6)

† p � .10* p � .05

** p � .01*** p � .001One-sided tests.


erately significantly larger than the influences of LL(p � .09).

The results support Hypothesis 4b, which pre-dicts that the influences of shared interorganiza-tional ties (OOL and OOLL) are both stronger thanthe influence of OO. Coefficients on OOLL andOOL (both organizations and one leader) are allsignificantly larger than the coefficient on OO (bothorganizations without their current leaders).

The results partially support Hypothesis 4c,which predicts that the influences of OOL, OLL,and OOLL all dominate the influence of OL. Asexpected, the coefficients on OOLL and OOL aresignificantly greater than the coefficient on OL. Bycontrast, though, the coefficients on OLL (bothleaders and one organization) are not.

Finally, the results support Hypothesis 4d,which predicts that fully shared ties (OOLL) havemore influence on future exchange than partiallyshared ties (OLL and OOL). The positive coefficienton OOLL is significantly larger than the coefficientson both OOL and OLL.

We used multiple sensitivity analyses to assessthe robustness of the results; the Appendix reports

these analyses in greater detail. The sensitivity testsassess across-team variation in the power of generalmanagers and other organizational elements, exec-utive history (negative employment history, gen-eral manager success, and executive turnover),structural embeddedness, player power, othercomplementarity effects (winning percentage, mar-ket size, differential skills), and dyadic disaggregat-ing (asymmetric measures of OL, OOL, and OLL).The reported results did not change materially inany of these analyses. We believe that any remain-ing unwanted variance does not create a seriousbias because the observed pattern of results re-mains consistent throughout the many combina-tions of control variables that we assessed in sen-sitivity analyses.

Although we discuss the implications of the re-sults in greater detail below, we note that our corefinding is that the greatest impact arises from or-ganization-to-organization experience, especially ifat least one leader in an organization dyad sharesthat experience (i.e., OOLL and OOL have the great-est impact, with smaller but still substantial impactfrom OO). Figure 1 illustrates the relative influenceof each kind of tie, given an increase in that mea-sure equal to one standard deviation of allexchange ties.

DISCUSSION

This study contributes to the literature that ex-tends the concepts of embeddedness and socialcapital to interorganizational settings. Prior re-search linking exchange ties to subsequent interor-ganizational exchange offers limited insight overlevels because it constrains analyses to a singlelevel or conflates multiple levels. The exchange tiemeasures of prior research have reflected eitherinterorganizational relations or relations among or-

TABLE 4Coefficient Comparison Tests

Model �2 Change p <

Unconstrained 1,577.0OOL � OO 1,552.4 24.6 0.0001

OOLL � OO 1,570.3 6.7 0.01OLL � LL 1,574.2 2.8 0.09

OOLL � LL 1,420.2 156.8 0.0001OOL � OL 1,441.4 132.6 0.0001OLL � OL 1,576.1 0.9 0.34

OOLL � OL 933.5 643.5 0.0001OOLL � OOL 1,559.6 17.4 0.0001OOLL � OLL 1,248.7 328.3 0.0001

FIGURE 1Cumulative Hazard Function of Interorganizational Exchange


ganizational leaders. In contrast, we examined howboth leaders’ and organizations’ exchange ties in-fluence subsequent interorganizational exchangebehavior. By simultaneously examining organiza-tional and leader exchange ties, the present studybegins to isolate the influences of nodal multiplex-ity on interorganizational exchange and the forma-tion of larger network structures.

The disaggregation of leaders’ and organizations’exchange ties allows us to address two related is-sues. First, debates in the academic and practition-er literatures about the relative influences of lead-ers on organizations’ economic behavior areemblematic of the underlying theoretical questionof the role of nodal multiplexity. Some authorshave suggested that the exchange experiences ofleaders with decision-making authority play dom-inant roles in shaping organizations’ economic be-havior (Brass et al., 2004), and others have sug-gested that the influences of other organizationalrepositories of exchange experiences condition andpossibly supplant the influence of leaders’ experi-ences (Burt, 1999). Second, we consider an array ofties in which organizations and their current lead-ers share prior exchange experiences. Disaggregat-ing exchange ties and then examining the patternsof results allow us to isolate the effects of exchangeties at different levels of analysis.

We found strong evidence that interorganization-al exchange ties have influences on interorganiza-tional exchange behavior that go above and beyondthe influences of leaders’ exchange ties. These re-sults suggest that social influences on interorgani-zational exchange arise, in large part, outside theexperiences of individual leaders. As the resultsillustrate, we found no support for the direct effectsof unshared leader-to-leader and leader-to-organi-zation exchange ties. In addition, the positive in-fluence of shared exchange experiences involvingboth leaders and only one current organization issignificantly weaker than the influences of all otherkinds of exchange ties involving both current or-ganizations. Thus, contrary to the intuition that thesocial experiences of leaders with decision-makingauthority should have the strongest effect, the ac-tual influence of leaders’ exchange ties appearslimited.

The finding that leader ties are not as influentialas other organizational ties is particularly intrigu-ing, given received wisdom about the empiricalcontext of this study and other more general organ-izational settings. Writers in the popular press com-monly hold that general managers are highly in-strumental in formulating interorganizationalexchanges among Major League Baseball teams.Furthermore, Major League Baseball clubs are rel-

atively small and simple hierarchies. Thus, com-mon management theory would suggest that lead-ers’ schemata heavily influence organizationalbehavior (e.g., Mintzberg, 1979). Given this positedcentral role of senior leadership, one might expectgeneral managers’ exchange experiences to influ-ence interorganizational exchange behaviorstrongly (Uzzi, 1997). The fact that individual lead-ers’ exchange ties had little direct influence helpsreshape such seemingly intuitive conclusions bothin this empirical setting and more broadly.

Although this study does not identify specificreasons for this counterintuitive finding, at leastthree causes are possible. First, key knowledgeabout other teams used in exchange decisions mayoriginate from individuals other than a team’s GM.Teams generally employ scouts, advisors, and as-sistant general managers who help shape generalmanagers’ trading decisions through personal ad-vice and player scouting reports. The complexityand dynamism of teams’ player rosters could rein-force general managers’ reliance on support staff.Second, the employing organization may be theprimary referent for others’ exchange decisions,rather than the GM, because players are legallybound to the organization and not to the generalmanager. Third, individual-level negative ex-change experiences may moderate others’ reliabil-ity judgments more than organization-level experi-ences. That is, as attribution error research hasshown (Ross, 1977), exchange decision makersmight be more likely to attribute negative priorexchange experiences to general managers’ actionsthan to other aspects of their organizational con-text. If so, then the negative effects of bad exchangeexperiences could counterbalance the positive ef-fects at the level of individual leaders’ ties, but notat the level of organization ties.

Although we found little direct effect of individ-ual leaders’ ties, we did find that leaders had indi-rect influence via their reinforcement of organiza-tion-level exchange ties. The results suggest thatleaders’ exchange experiences are influential whenthey occur within one or more organizations in anexchange dyad. That is, leaders matter, but theymatter more in their broader organizational contextthan they do as isolated individuals.

Anecdotal evidence from John Schuerholz, gen-eral manager of the Atlanta Braves, reinforces thisresult. Scheuerholz and Guest (2006: 211) both lik-ened the responsibilities of a major league generalmanager to those of an orchestra conductor andcalled the GM a “final filter” in decision making.This description suggests that leadership, even insmall organizations, often does not involve directlyinjecting one’s own knowledge and beliefs into or-


ganizational decisions. Instead, leadership in organ-ization exchange decisions involves gathering in-sights from others’ experiences before reconcilingthose insights with one’s own schema.

We believe that such inferences about the mic-rolevel foundations of organization-level embed-dedness generalize to a broad range of organization-al settings. Nonetheless, as with any industrystudy, the question of boundaries on generalizationof the results arises. The results apply most directlyto situations in which organizations of at least mod-erate complexity undertake repeated transactionalexchanges (e.g., initial public offering syndications,patent- or product-licensing arrangements, stan-dard-setting alliances, and shared production ar-rangements, as mentioned earlier). The results maybe weaker for very simple organizations, in whichan individual executive is able to identify resourceneeds, assess sources, and negotiate terms of ex-change. In addition, the results may be less appli-cable to idiosyncratic exchanges for which priorexperience—whether individual or organization-al—will be less relevant. Moreover, temporally ex-tended forms of prior exchange, such as ongoingalliances, could foster stronger, more complex de-velopment of interpersonal relations than arm’s-length exchanges. If so, then individual leaders’prior experiences might have stronger effects on thelikelihood of subsequent interorganizational ex-change than the more transaction oriented experi-ences in this study. The study’s results showingthat the leaders’ employment ties could have someinfluence are evidence of that. Future researchcould explore these boundaries.

The natures of organizations’ cognitive processesalso raise questions for future research. Organiza-tion-level processes, such as collective interpreta-tion of past experiences, collective storage ofknowledge and beliefs, and collective decisionmaking, involve many individuals in and aspects ofan organization other than its leader. Variation inthese processes is likely to moderate the influencesof specific individuals, subgroups, and other socialstructures. A comprehensive model of organiza-tion-level economic and social behavior requires acomplete understanding of contextual forces aswell as the dynamic, underlying processes.

The processes that define organization-level em-beddedness are likely to vary as a result of severalcontextual factors. Organizational characteristicssuch as size, structure, culture, and stability arelikely to influence how capabilities and responsi-bilities for interorganizational exchange processesare distributed (Uzzi, 1997). For example, in a mul-tidivisional organization, divisional leaders mayhave more responsibility for interorganizational ex-

change decisions than the corporation’s CEO. Thisexample highlights the value of distinguishing in-fluences at the group and organization levels infuture research.

This discussion calls attention to the importanceof identifying variations in the roles and character-istics of leaders. An operational leader’s exchangeexperiences might have a different level of influ-ence on an organization’s embeddedness thanthose of a more symbolic leader. Numerous charac-teristics of leaders, including their capabilities andlegitimacy, could determine the degree to whichtheir experiences influence organizational ex-change decisions. In the Major League Baseballcontext, for instance, we cannot fully rule out thepossibility that team owners’ exchange experienceshave greater influence than those of general man-agers, although we attempted to control for thispossibility.

In addition to the organizational context, struc-tural aspects of interorganizational and interper-sonal environments are intriguing. Although thisstudy assesses network density, actors’ closenesscentralities, actors’ betweenness centralities, andcommon third-party ties, more comprehensive ex-amination of larger network structures is war-ranted. For example, different kinds of structuralequivalence and network density might affect ex-change. Structure-based studies of organiza-tion-level behavior also could consider cross-levelrelationships and the degrees to which organiza-tion members share prior experiences.

Other characteristics of an exchange process arelikely to influence which individuals and struc-tures are responsible for key exchange-related ac-tivities. For example, if the level at which an or-ganization’s relational embeddedness operatesdepends on which individuals’ knowledge and be-liefs are incorporated into interorganizational ex-change decisions, then theories of informationsearch (Feldman & March, 1981; Marschak, 1968;Ocasio, 1995) will offer insights. Alternatively, ifthe level at which organization-level relational em-beddedness operates depends on who or whatserves as the referent focus of others’ interorgani-zational exchange decisions, then theories of atten-tion (Ocasio, 1997; Simon, 1955) and attribution(Jones & Nisbett, 1972) will be helpful. The vastwealth of research on decision making and judg-ment could offer insights into the microlevel foun-dations of organizational embeddedness.

The outcomes of prior exchanges also will shapeperceptions of reliability during subsequent interor-ganizational exchange decisions, raising a complexset of questions about how norms of fairness andefficiency moderate organizational embeddedness


(Ring & Van de Ven, 1994). For two reasons, how-ever, we believe that accounting for exchange out-comes would not fundamentally alter the results ofthis study, or of studies in similar contexts. First,even moderately negative exchange experiencesshort of egregious opportunism provide relationalinformation that can facilitate subsequent exchangebetween parties. Second, although partner oppor-tunism presents a meaningful risk, participatingactors in the context of this study viewed the vastmajority of player trades as being reasonably fair.Such satisfaction arises, in part, because the publicnature of the exchanges and the ability of teams toappeal perceived injustices to their league govern-ing body offer protections against egregious oppor-tunism, as do legal and arbitration processes incommercial exchanges. Moreover, to the extent thatexchange outcomes introduce unwanted varianceinto the model, these analyses represent conserva-tive tests of the hypotheses. Nonetheless, the im-pact of exchange outcomes warrants greater study.

Several other issues merit study. Other character-istics of prior exchange ties (e.g., duration, risk,conventional content multiplexity) could shapehow prior exchange ties influence subsequent inter-organizational exchange because different types ofexchange produce and require different combina-tions of mutual knowledge, trust, and commitment.Indeed, this study suggests that a general manager’sprior employment relationships directly influencesubsequent player trading behavior. It would beuseful to unpack the influences of knowledge,trust, and commitment and determine how the na-ture of exchanges moderates the influences of thosedifferent mechanisms of embeddedness.

Finally, future researchers should entertain thepossibility that embeddedness at different levelshas different implications for organization perfor-mance. This article begins to disaggregate how dif-ferent forms of established exchange relations pro-vide access to resources. In doing so, the study putsthe impact of leaders’ social capital in perspective.The results suggest caution in assessing the socialsignificance of even industry-experienced leaders.In contrast, the research emphasizes the impor-tance of broader organizational characteristics inshaping interorganizational social relationships.

REFERENCES

Adler, P. S. 2001. Market, hierarchy, and trust: Theknowledge economy and the future of capitalism.Organization Science, 12: 215–234.

Adler, P. S., & Kwon, S. 2002. Social capital: Prospectsfor a new concept. Academy of Management Re-view, 27: 17–40.

Ahuja, G. 2000. The duality of collaboration: Induce-ments and opportunities in the formation of inter-firm linkages. Strategic Management Journal, 21:317–343.

Aiken, M., & Hage, J. 1968. Organizational interdepen-dence and interorganizational structure. AmericanSociological Review, 33: 912–930.

Amburgey, T., & Miner, A. 1992. Strategic momentum:The effects of history, strategic position and struc-ture on merger activity. Strategic ManagementJournal, 13: 335–348.

Baker, W. 1990. Market networks and corporate behavior.American Journal of Sociology. 96: 589–625.

Beach, L. 1993. Broadening the definition of decision-making: The role of prechoice screening of options.Psychological Science, 4: 215–220.

Blau. P. 1964. Exchange and power in social life. NewBrunswick, NJ: Wiley.

Bonacich, P. 1972. Factoring and weighting approachesto status scores and clique identification. Journal ofMathematical Sociology, 2: 113–120.

Brass, D. 2001. Networks and frog ponds: Trends in mul-tilevel research. In K. Klein & S. Kozlowski (Eds.),Multilevel theory, research, and methods in organ-izations: 557–571. San Francisco: Jossey-Bass.

Brass, D., Galaskiewicz, J., Greve, H., & Tsai, W. 2004.Taking stock of networks and organizations: A mul-tilevel perspective. Academy of Management Jour-nal, 47: 795–819.

Breiger, R. L. 1974. The duality of persons and groups.Social Forces, 53: 181–190.

Brown, S., & Eisenhardt, K. 1998. Competing on theedge: Strategy as structured chaos. Boston: Har-vard Business School Press.

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

Burt, R. 1997. A note on social capital and networkcontent. Social Networks, 19: 355–373.

Burt, R. 1999. The social capital of opinion leaders. An-nals of the American Academy of Political andSocial Science, 566: 37–54.

Chesbrough, H., & Teece, D. 1996. When is virtual virtu-ous? Organizing for innovation. Harvard BusinessReview, 74(1): 65–73.

Coleman, J. 1973. The mathematics of collective action.Chicago: Aldine Atherton.

Cook, K. 1975. Towards a theory of interorganizationalrelations: An extension of the exchange model.Seattle: Institute of Sociological Research, Univer-sity of Washington.

Cook, K., & Emerson, R. E. 1978. Power, equity andcommitment in exchange networks. American So-ciological Review, 43: 721–739.


Crozier, M. 1964. The bureaucratic phenomenon. Chi-cago: University of Chicago Press.

Cyert, R. M., & March, J. G. 1963. A behavioral theory ofthe firm. Englewood Cliffs, NJ.: Prentice-Hall.

Douglas, M. 1986. How institutions think. Syracuse: Syr-acuse University Press.

Eisenhardt, K. M., & Martin, J. A. 2000. Dynamic capa-bilities: What are they? Strategic Management Jour-nal, 21: 1105–1121.

Emerson, R. 1967. Exchange theory. In J. Berger, J.Zelditch, & B. Anderson (Eds.), Sociological theo-ries in progress, vol. 2: 38–57. Boston: Houghton-Mifflin.

Emerson, R. 1976. Social exchange theory. In A. Inkeles(Ed.), Annual review of sociology, vol. 2: 335–362.Greenwich, CT: JAI Press.

Feldman, M. S., & March, J. G. 1981. Information inorganizations as signal and symbol. AdministrativeScience Quarterly, 26: 171–186.

Fiol, M. C., & Lyles, M. A. 1985. Organizational learning.Academy of Management Review, 10: 803–813.

Freeman, L. 1979. Centrality in social networks: Concep-tual clarification. Social Networks, 1: 215–239.

Freeman, L., Borgatti, S., & White, D. 1991. Centrality invalued graphs: A measure of betweenness based onnetwork flow. Social Networks, 13: 141–154.

Geertz, C. 1978. The bazaar economy: Information andsearch in peasant marketing. American EconomicReview, 68: 28–32.

Goffman, E. 1971. Relations in public. New York: BasicBooks.

Granovetter, M. S. 1973. The strength of weak ties. Amer-ican Journal of Sociology, 78: 1360–1380.

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

Granovetter, M. S. 1992. Problems of explanation in eco-nomic sociology. In N. Nohria & R. G. Eccles (Eds.),Networks and organizations: Structure, form andaction: 25–56. Boston: Harvard Business SchoolPress.

Gulati, R. 1995. Social structure and alliance formation:A longitudinal analysis. Administrative ScienceQuarterly, 40: 619–642.

Gulati, R., & Westphal, P. 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.

Hedberg, B. 1981. How organizations learn and unlearn.In P. C. Nystrom & W. H. Starbuck (Eds.), Handbookof organization design: 8–27. London: Oxford Uni-versity Press.

Hitt, M., Bierman, L., Uhlenbruck, K., & Shimizu, K.2006. The importance of resources in the interna-

tionalization of professional service firms: The good,the bad, and the ugly. Academy of ManagementJournal, 49: 1137–1157.

Homans, G. C. 1974. Social behavior. New York: Har-court, Brace.

Inkpen, A., & Tsang, E. 2005. Networks, social capital,and learning. Academy of Management Review,30: 146–165.

Jones, C., Hesterly, W. S., & Borgatti, S. P. 1997. A generaltheory of network governance: Exchange conditionsand social mechanisms. Academy of ManagementReview, 22: 911–945.

Jones, E., & Nisbett, R. 1972. The actor and the observer:Divergent perceptions of the causes of the behavior.In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E.Nisbett, S. Valins, & B. Weiner (Eds.), Attribution:Perceiving the causes of behavior: 79–94. Morris-town, NJ: General Learning Press.

Klein, K., Palmer, S., & Conn, A. 2001. Inter-organization-al relationships: A multilevel perspective. In K.Klein & S. Kozlowski (Eds.), Multilevel theory, re-search, and methods in organizations: 267–308.San Francisco: Jossey-Bass.

Kogut, B., Shan, W., & Walker, G. 1992. Competitivecooperation in biotechnology: Learning through net-works? In N. Nohria & R. Eccles (Eds.), Networksand organizations: Structure, form and action:348–365. Boston: Harvard Business School Press.

Lawler, E., & Yoon, J. 1998. Network structure and emo-tion in exchange relations. American SociologicalReview, 63: 871–894.

Lei, D., Hitt, M. A., & Goldhar, J. D. 1996. Advancedmanufacturing technology: Organizational designand strategic flexibility. Organization Studies, 17:501–523.

Lessig, V. P. 1973. Consumer store images and storeloyalties. Journal of Marketing, 37: 72–74.

Levine, S., & White, P. E. 1961. Exchange as a conceptualframework for the study of interorganizational rela-tionships. Administrative Science Quarterly, 5:583–601.

Levinson, H. 1965. Reciprocation: The relationship be-tween man and organization. Administrative Sci-ence Quarterly, 9: 370–390.

Lewis, M. 2003. Moneyball. New York: Norton.

Marschak, J. 1968. Economics of inquiring, communicat-ing, deciding. American Economic Review, 58:1–18.

McGrath, R., Ferrier, W., & Mendelow, A. 2004. Realoptions as engines of choice and heterogeneity.Academy of Management Review, 29: 86–101.

McPherson, M., Smith-Lovin, L., & Cook, J. M. 2001.Birds of a feather: Homophily in social networks. InK. Cook & J. Hagan (Eds.), Annual review of sociol-ogy, vol. 27: 415–44. Greenwich, CT: JAI Press.


Miles, R., & Snow, C. 1986. Organizations: New conceptsfor new forms. California Management Review,28(2): 62–73.

Miles, R., & Snow, C. 1992. Causes of failure in networkorganizations. California Management Review,34(4): 53–72.

Mintzberg, H. 1979. The structuring of organizations.Englewood Cliffs, NJ: Prentice-Hall.

Mowday, R. T., Porter, L. W., & Steers, R. M. 1982. Em-ployee-organization linkages: The psychology ofcommitment, absenteeism, and turnover. NewYork: Academic Press.

North, D. 1990. Institutions, institutional change, andeconomic performance. New York: Cambridge Uni-versity Press.

Ocasio, W. 1995. The enactment of economic diversity: Areconciliation of theories of failure-induced changeand threat-rigidity. In L. Cummings & B. Staw (Eds.),Research in organizational behavior, vol. 17: 287–331. Greenwich, CT: JAI Press.

Ocasio, W. 1997. Towards an attention-based view of thefirm. Strategic Management Journal, 18: 187–206.

Oh, H., Labianca, G., & Chung, M.-H. 2005. A multilevelmodel of group social capital. Academy of Manage-ment Review, 31: 569–582.

Organ, D., & Konovsky, M. 1989. Cognitive versus affec-tive determinants of organizational citizenship be-havior. Journal of Applied Psychology, 74: 157–164.

Pfeffer, J., & Salancik, G. R. 1978. The external control oforganizations. Stanford, CA: Stanford UniversityPress.

Podolny, J., & Baron, J. 1997. Relationships and re-sources: Social networks and mobility in the work-place. American Sociological Review, 62: 673–694.

Powell, W. 1990. Neither market nor hierarchy: Networkforms of organization. In B. M. Staw & L. L. Cum-mings (Eds.), Research in organizational behavior,vol. 12: 295–336. Greenwich, CT: JAI Press.

Reider, N. 1953. Transference to institutions. Bulletin ofthe Menninger Clinic, 17: 58–63.

Ring, P., & Van de Ven, A. 1992. Structuring cooperativerelationships between organizations. Strategic Man-agement Journal, 13: 483–498.

Ring, P. S., & Van de Ven, A. H. 1994. Developmentalprocesses of cooperative interorganizational rela-tionships. Academy of Management Review, 19:90–118.

Ross, L. 1977. The intuitive psychologist and his short-comings: Distortions in the attribution process. In L.Berkowitz (Ed.), Advances in experimental socialpsychology, vol. 10: 173–220. New York: AcademicPress.

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.

Sanchez, R., & Mahoney, J. 1996. Modularity, flexibility,and knowledge management in product and organi-zation design. Strategic Management Journal, 17:63–76.

Schermerhorn, J. R. 1975. Determinants of interorganiza-tional cooperation. Academy of Management Jour-nal, 18: 846–856.

Schilling, M., & Steensma, K. 2001. The use of modularorganizational forms: An industry-level analysis.Academy of Management Journal, 44: 1149–1168.

Schuerholz, J., & Guest, L. 2006. Built to win: Insidestories and leadership strategies from baseball’swinningest GM. New York: Warner Books.

Selznick, P. 1957. Leadership in administration. NewYork: Harper & Row.

Shanks, B. 2005. Scout’s honor: The bravest way tobuild a winning team. New York: Sterling & Ross.

Simon, H. 1955. A behavioral model of rational choice.Quarterly Journal of Economics. 69: 99–118.

Stigler, G. 1968. Competition. In D. L. Sills (Ed.), Inter-national encyclopedia of the social sciences, vol. 3:181–185. New York: Macmillan.

Thibaut, J. W., & Kelley, H. H. 1959. The social psychol-ogy of groups. New York: Wiley.

Thompson, L. L., Levine, J. M., & Messick, D. M. (Eds.)1999. Shared cognition in organizations: The man-agement of knowledge. Mahwah, NJ: Erlbaum.

Thrift, S., & Shapiro, B. 1990. The game according toSyd: The theories and teachings of baseball’s lead-ing innovator. New York: Simon & Schuster.

Tuma, N. B., & Hannan, M.,T. 1984. Social dynamics:Models and methods. New York: Academic.

Tversky, A., & Kahneman, D. 1974. Judgment under un-certainty: Heuristics and biases. Science, 185: 1124–1131.

Uzzi, B. 1996. The sources and consequences of embed-dedness for economic performance of organizations: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.

Van de Ven, A. H. 1976. On the nature, formation, andmaintenance of relations among organizations.Academy of Management Review, 1: 24–36.

Walsh, J., & Ungson, G. 1991. Organizational memory.Academy of Management Review, 16: 57–91.

Wayne, S. J., Shore, L. M., & Liden, R. C. 1997. Perceivedorganizational support and leader-member ex-change: A social exchange perspective. Academy ofManagement Journal, 40: 82–111.


Zaheer, A., McEvily, B., & Perrone, V. 1999. Does trustmatter? Exploring the effects of interorganizationaland interpersonal trust on performance. Organiza-tion Science, 9: 141–159.

APPENDIXAdditional Control Variables and

Sensitivity Tests

Power Variations

A few team owners—such as George Steinbrenner ofthe New York Yankees—have reputations for beinghighly involved in baseball operations. This suggestedthat each team’s general manager or organizational tiescould have a different impact on the dependent variable.Because there was no reliable way to know the nature ofthis variance a priori, additional analyses included twosets of 30 controls (one for each team), which we calledacross-team influence variables. In the first set of 30, eachcontrol was equal to the sum of the dyad team’s totalnumber of GM ties. In the second set of 30, each controlwas equal to the sum of the dyad team’s total number oforganization ties. For example, in a dyad consisting ofteam A and team B, team A’s GM ties (GMTA) and teamB’s GM ties (GMTB) would be set according to the follow-ing equations (the other 28 teams’ across-team influencevariables would equal zero):

GMTA�LALB�OALALB�OBLALB�OAOBLALB�OBLA�

OBLALB.

GMTB�LALB�OALALB � OBLALB � OAOBLALB � OALB �

OBLALB.

In parallel, team A’s organizational ties (OTA) andteam B’s organizational ties (OTB) would be set accordingto the following equations (the other 28 teams’ controlvariables would equal zero):

OTA � OAOB � OAOBLA � OAOBLB � OAOBLALB � OALB �

OALALB.

OTB � OAOB � OAOBLA � OAOBLB � OAOBLALB � OBLA �

OBLALB.

Including these variables in the analysis did not ma-terially change the results.

Negative Employment History, General ManagerSuccess, and Turnover

Because negative employment experiences could af-fect exchanges, a supplemental analysis included a vari-able that captured the number of GMs in a dyad who hadever been fired while acting as GMs for the other organ-ization in the dyad. Similarly, because a GM’s overallsuccess could affect trading behavior, an analysis in-cluded variables capturing the teams’ cumulative win-ning percentages during the current GMs’ tenures. An

analysis also added three dummy variables capturingturnover in ownership, field managers, and scouting di-rectors, because executive turnover can disrupt adminis-trative processes and social embeddedness. The reportedresults did not change.

Structural Embeddedness

Supplemental analyses assessed aspects of interorgan-izational and interleader network structure. At eachlevel, we calculated two measures of general networkdensity, four measures of each team’s centrality, and twomeasures of dyad teams’ common ties. The density mea-sures equaled the sum of all network dyad exchanges ina given year divided by the total number of team dyads.The exchanges of one of the density measures wereweighted by the number of players traded and linearlydecayed over three years. None of the density measureswas significant or changed the results of the hypothesistests. The centrality measures for each team and each GMin a given year included Freeman closeness and Freemanbetweenness (Freeman, 1979), the Bonacich eigenvector(Bonacich, 1972), and flow betweenness (Freeman et al.,1991). Inclusion of none of these measures changed thehypothesis tests; the final analysis retained the organiza-tion-level flow betweenness measure because it had asignificant impact on interorganizational exchange). Fi-nally, common ties measures equaled the sum of theproducts of the dyad actors’ ties to the same third-partyactors in a given year. We calculated the common tiesmeasures with and without weighted measures of ex-change ties. Including the common ties measures did notchange the results.

Player Power

Although few professional baseball players can di-rectly influence teams’ trading behavior, a few playershave clauses in their contracts that prevent their teamsfrom trading them without permission. Typically, onlythe very best players have the bargaining power to obtainsuch a no-trade clause; however, some veteran playersqualify for implicit no-trade contracts under the “10-and-5” rule (ten years in MLB; five years with the sameteam). Because historical employment contract informa-tion is difficult to obtain, we could not conduct directanalyses of no-trade clauses over the span of the study.However, an analysis of 2006 contracts suggested thatthese clauses were unlikely to substantially influence ourresults because they are rare. In 2006, 71 players hadexplicit or implicit no-trade clauses in their contracts—about 1.5 percent of all professional baseball players(including minor league players who were contractuallybound to major league organizations). We believe thatthis percentage was even lower in previous years becausecontract negotiations have become significantly more so-phisticated over time (Schuerholz & Guest, 2006). More-over, teams’ 2006 no-trade clauses have a high correla-tion (r � .75) with team payroll, which occurs becausealmost all players with no-trade clauses are highly paid


stars or veterans. Inclusion in a supplemental analysis ofa variable capturing teams’ payrolls did not substantiallychange the results. Moreover, the fixed-effects variablesin the analyses indirectly controlled for no-trade clausesbecause such clauses tend to vary systematically acrossteams. For example, some teams (e.g., the Atlanta Braves)refuse to offer these clauses to any players, includingsuperstars, because doing so reduces their roster flexibil-ity.

Complementarity: Differential Skills

We considered the roles of different skill balances inshaping interorganizational exchange behavior. Teamstend to “[trade] strengths for strengths” (Thrift & Shapiro,1990: 240). In particular, teams with strong pitchingstaffs tend to trade with teams that have strong battingline-ups. The reason is that major league teams have onlya few roster spots available. Talented minor league play-ers who are blocked from playing in the major leaguesbecause their teams are particularly strong in some areahave a diminished marginal value to the teams’ compet-itive capabilities; blocked players often have greatervalue on the open trade market. For example, the AtlantaBraves had a particularly strong pitching staff throughoutthe 1990s. As a result, the Braves used talented pitchersin their minor league system to attract exchange partners(Shanks, 2005). We created a control variable, “relativeskill balance” (RSB) using teams’ relative normalizedpitching and batting statistics each season, as follows:

RSBAB, t � ��Zt(RPIA, t) � Zt(ERAA, t) � �Zt(RPIB, t)

� Zt(ERAB, t)�.

The skill balance between teams A and B in year t(RSB

AB, t) is based on each team’s earned-run average

(ERA) and runs scored per inning (RPI), standardized byyear t (Zt).

Other Complementarity Effects: Winning Percentage,Market Size, and Payrolls

We assessed three other complementarity effects, find-ing no material impact on the reported results. First, we

included a variable equal to the absolute difference in thewinning percentages of each dyad’s teams. Second, weincluded a control for absolute differences in marketpopulation size, because teams in smaller markets (e.g.,Oakland and Kansas City) have an incentive to acquireless expensive players and develop that talent in-house,but teams in larger markets (e.g., New York) can afford toacquire more expensive talent via trades (Lewis, 2003).Third, we controlled for difference in dyad team pay-rolls.

Dyadic Disaggregation

An additional model included nine disaggregated ex-change ties, which treated OL, OOL, and OLL as organi-zation- and leader-specific asymmetric measures ratherthan symmetric measures. The disaggregated analysisdid not explain significantly more variance in the likeli-hood of exchange than the reported effects, and coeffi-cients on all of the disaggregated measures were direc-tionally consistent with the symmetric measuresreported in Table 3.

Jeffrey Q. Barden ([email protected]) is an as-sistant professor in the management and organizationarea of the University of Washington business school. Jeffhas a Ph.D. degree in management from the Fuqua Schoolof Business at Duke University. His research focuses oninterorganizational strategy and applications of theoriesover levels of analysis.

Will Mitchell ([email protected]) is the J. Rex Fu-qua Professor of International Management at Duke Uni-versity’s Fuqua School of Business. He received his Ph.D.from the University of California, Berkeley. Will studiesbusiness dynamics, investigating how businesses changeas the environments in which firms compete change and,in turn, how the business changes contribute to ongoingcorporate success or failure. He teaches courses in busi-ness dynamics, corporate strategy, and pharmaceuticalstrategy.


Documents

DISENTANGLING THE INFLUENCES OF LEADERS’ …willm/bio/cv/papers/AMJ2007... · RELATIONAL EMBEDDEDNESS ON INTERORGANIZATIONAL EXCHANGE ... creased the influences of those experiences