the role of incentives and communication in strategic ...terpconnect.umd.edu/~rajshree/research/36...

Strategic Management JournalStrat. Mgmt. J., 31: 413–437 (2010)

Published online EarlyView in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/smj.818

Received 24 October 2008; Final revision received 14 September 2009

THE ROLE OF INCENTIVES AND COMMUNICATIONIN STRATEGIC ALLIANCES: AN EXPERIMENTALINVESTIGATION

RAJSHREE AGARWAL,1* RACHEL CROSON,2 and JOSEPH T. MAHONEY1

1 Department of Business Administration, College of Business, University of Illinois,Champaign, Illinois, U.S.A.2 School of Management, and School of Economics, Political and Policy Sciences,University of Texas-Dallas, Richardson, Texas, U.S.A.

This paper experimentally examines the determinants of the deviation between potential andrealized value creation in strategic alliances. To better understand how decision making inalliances may influence success, we use an experimental design that juxtaposes two importantfactors that affect alliance members’ decisions: economic incentives and communication. Theevidence from our experiment sheds light on the relative impact of each, and more importantly,how both factors interact to explain successful outcomes. Copyright 2009 John Wiley & Sons,Ltd.

INTRODUCTION

Strategic alliances are ongoing cooperative rela-tionships and represent an important organizationalform for governing transactions (Reuer, Zollo,and Singh, 2002; Zaheer and Bell, 2005). Strate-gic alliances have the potential to create eco-nomic value (Gulati and Singh, 1998; McEvilyand Zaheer, 1999) and, on average, empirical evi-dence corroborates this view (Chan et al., 1997;Sarkar, Echambadi, and Harrison, 2001). However,approximately half of all strategic alliances fail(Kale, Dyer, and Singh, 2002). Indeed, the largegap between potential economic value creationand realized economic value creation in strategicalliances suggests that there are formidable imped-iments to successful alliance outcomes (Anand andKhanna, 2000; Gottschlag and Zollo, 2007).

Keywords: strategic alliances; economic incentives; com-munication; social dilemma; game theory; experiments*Correspondence to: Rajshree Agarwal, Department of BusinessAdministration, College of Business, University of Illinois, 350AWohlers Hall, 1206 South Sixth Street, Champaign, IL 61820,U.S.A. E-mail: [email protected]

The challenges in achieving successful outcomesin strategic alliances relate to the inherent ten-sion between cooperation and competition (Hamel,1991). To realize potential value, alliance partnersmust invest resources, share knowledge, and buildsynergies through cooperation (Dyer, 1997; Dyerand Singh, 1998). However, given that the ben-efits of alliance activity are commonly availableto all alliance partners, there arises the potentialfor ‘free-riding,’ or engaging in learning races inthe pursuit of private benefit at the expense of thetotal value creation (Khanna, Gulati, and Nohria,1998). Thus, decision makers in strategic alliancesmust deal with substantial uncertainty and coordi-nation failures, which can lead to real or perceivedopportunism, miscalculation, and low performance(Dyer and Hatch, 2006; Shan, Walker, and Kogut,1994). Managerial solutions to such impedimentsare often learned over time, albeit in a less thansystematic fashion, through hard-earned experi-ence with strategic alliance partners (Ahuja, 2000;Madhavan, Koka, and Prescott, 1998).

Researchers have begun to systematically exam-ine factors that facilitate successful outcomes (Kaleand Singh, 2007; Luo, 2008, Tiwana, 2008). Within

Copyright 2009 John Wiley & Sons, Ltd.

414 R. Agarwal, R. Croson, and J. T. Mahoney

the context of alliance dynamics specifically,Khanna et al., (1998) use an economics game the-oretic lens and develop theoretical propositionsrelating economic incentives to success in strategicalliances. This research underscores the potentialfor learning races when strategic alliance partnersmay benefit more through noncooperation than bypursuing a common goal in the absence of incen-tive alignment. Zeng and Chen, in contrast to such‘structural’ solutions, emphasize, using a socialpsychology lens, the role of ‘motivational’ solu-tions (Zeng and Chen, 2003: 591) (e.g., commu-nication) as a potential way to increase coopera-tive rather than competitive outcomes. This workthus highlights the role of organizational designand deliberate efforts to develop a common under-standing and trust among strategic alliance part-ners.

While both research literature streams provideimportant insights regarding each mechanism inisolation, several fundamental questions remainunanswered. How important are incentive align-ment and communication to achieving successin cooperative alliances? What conditions mayimpact their efficacy in achieving success? Arethere synergies between the two underlying mech-anisms, or do these mechanisms substitute for eachother?

We examine the interplay of both mechanismsin this paper, and employ experimental methodol-ogy that permits the disentangling of the effects ofthese mechanisms on decision making in strategicalliances. Widely employed in economics, psychol-ogy, and to an increasing extent in strategic man-agement (Davis 2003; Kagel and Roth 1995; Song,Calantone, and Di Benedetto, 2002), our empiri-cal approach complements extant research methodsused for examining success in strategic alliances(e.g., stock market returns, survey-based post-alliance perceptions of success). Despite increasedrecognition of endogenous selection in strategyresearch and adoption of new empirical methods,there are limits to the extent to which one canadequately control for endogenous selection giventhe dearth of valid instruments in most empiricalcontexts relevant to strategy (Hamilton and Nick-erson, 2003). In contrast, the laboratory setting inexperimental methodology allows the creation ofa simulated environment that controls for selectioneffects by random assignment of strategic alliancepartners. It also enables a direct and clean mea-surement of both the dependent variable (success

in the strategic alliance), and the independent vari-ables (economic incentives and communication)through the creation of independent ‘treatments’that represent each underlying causal mechanism.A salient feature of this methodology is that bysimulating treatments that may not occur in thefield, it enables us to identify the independent andcombined effects of these variables (Friedman andSunder, 1994). This type of evidence is criticalfor advancing our understanding of the theory ofeconomic organization. The premise of our paper,thus, is in line with the recommendation of Ireland,Hitt, and Vaidyanath (2002) that multiple theo-ries and methodologies be used for studying howalliances can be effectively managed for achievingcompetitive advantage.

We posit and show that alignment of economicincentives is a necessary, but not a sufficient, con-dition for achieving successful alliance outcomes.Contrary to economic theorizing that talk is cheap(Ledyard, 1995), we find strong empirical evidenceof additional benefits of communication. Our paperthus shows that economic incentives (as empha-sized, for example, in property rights theory) andinterpersonal (communication) processes to over-come bounded rationality problems (as empha-sized in social psychology and classic organiza-tion theory) are both important to decision mak-ing in strategic alliances. In the next section, wedevelop our theoretical framework and hypothesesfor the mechanisms that impact decision makingin strategic alliances. We then describe the exper-imental methodology that simulates participantsin an alliance setting and provide the empiricalresults. The concluding section includes a discus-sion, limitations of our study, and avenues forfuture research.

THEORETICAL FRAMEWORK

Strategic alliances are an interorganizational formwhere multiple exchange partners agree to investresources, share knowledge, and engage in eco-nomic value-creating activities that build on syn-ergies between the resources and capabilities thateach of the exchange partner firms bring to thealliance. While strategic alliances are formed withthe intent that all exchange partners will gainfrom cooperation through economic value creation,there is nonetheless a competitive element; strate-gic alliance partners have an incentive to compete

Copyright 2009 John Wiley & Sons, Ltd. Strat. Mgmt. J., 31: 413–437 (2010)DOI: 10.1002/smj

Incentives and Communication in Strategic Alliances 415

for the largest share of the economic benefits. Inaddition, exchange partners can pursue their owninterest over the strategic alliance by engaging ineconomic holdup and/or in learning races (Doz,1996; Khanna et al., 1998).

Since our primary objective is to move beyondpotential value creation and analyze economic andstrategic management issues concerning realizedvalue creation in alliances, Olson’s (1965) sem-inal research The Logic of Collective Action isespecially salient. Olson (1965) combines aspectsof property rights theory (e.g., the tragedy of thecommons) with game theoretic insights (Camerer,1991; Saloner, 1991) in which social dilemmasituations can result in persistent severe under-performance of economic value creation potential(Arend and Seale, 2005). The key idea is thatstrategic alliances typically create economic valuethat have a ‘common pool’ component, and thislack of well-defined property rights invites poten-tial opportunistic behavior and free-riding (Mow-ery, Oxley, and Silverman, 1996; Oxley, 1997).The ‘common-pool problem’ has many applica-tions in economics and sociology. For exam-ple, an extensive literature on depletable naturalresources, such as oil fields and fisheries, describeshow inefficiency arises due to a lack of well-defined property rights that causes individuals orfirms to overharvest resources (Libecap, 1989).Therefore, a few researchers have underscoredthe usefulness of examining the conflict betweencompetition and cooperation through the lens ofgame theory and social psychology, where strate-gic alliances can be represented as a public goodor a social dilemma problem (Gulati, Khanna, andNohria, 1994; Zeng and Chen, 2003). The socialdilemma arises because exchange partners mustdecide on whether to pursue a higher individualpayoff through competitive choices, even thoughthe collective payoff is larger with cooperation.Furthermore, the common sharing of the economicvalue created in the alliance introduces the pub-lic good element: it is difficult to exclude alliancepartners from sharing in the gains, regardless ofwhether or not they contributed toward the eco-nomic value creation.

When framed as a social dilemma problem,strategic alliances can be modeled as either a pris-oner’s dilemma or an assurance/coordination game(Gulati et al., 1994). While the prisoner’s dilemmagame is commonly discussed in the research lit-erature, a few remarks about the assurance game

are perhaps in order. The assurance game is alsocalled the Stag Hunt game that was derived from astory of social contract originally told by Rousseau(1754/1984): Two individuals go on a huntingexpedition together. Each person can individuallychoose to hunt a stag or a hare. Each individualmust choose an action without knowing the choiceof the other hunter. If an individual hunts a stag, hemust have the cooperation of his hunting partnerin order to succeed. An individual can get a hareby himself, but the hare is worth less than the stag.In this stag-hunt game, the rational choice for oneperson depends on what the other individual willdo. There are two equilibria to this game: (1) bothpersons choose to hunt the stag (which is the largermutual payoff) and (2) both persons choose to huntthe hare. The individual decision to hunt the stagentails both the potential for greater mutual benefitand the potential for greater personal risk (Skyrms,2003).

In both the prisoner’s dilemma game and assur-ance game settings, strategic alliance partnersdecide how much to invest toward the joint allianceactivity. Their investments are crucial for eco-nomic value creation, but the rewards of theirinvestments are common to all alliance partners,and are contingent on how much the other alliancepartners contribute. Gulati et al. (1994) note thatwhile strategic alliances are often viewed as pris-oner’s dilemma games, they should be more appro-priately characterized as assurance games. Toexplicate this logic, we provide a simple illustra-tion of the different payoff matrices in two-persongames in Table 1. In Panel 1, we start with the‘invisible hand’ game where there is no socialdilemma and the dominant strategy (Nash) equilib-rium is for both players to cooperate. Each playeris led, by their individual rational self-interest asdescribed by Adam Smith (1776/1937), to thedesirable Pareto optimal outcome of (170, 170). Incontrast, the other three panels of Table 1 demon-strate social dilemma problems.

In the prisoner’s dilemma game of Panel B,the dominant strategy (Nash) equilibrium is to notcooperate. Each player chooses their individuallyrational (self-interested) move, which results in aPareto inferior outcome of (120, 120), rather thanthe collectively rational outcome of (170, 170).While there is a dominant strategy of no cooper-ation in prisoner’s dilemma games, the assurancegame, depicted in Panels C and D (for homoge-nous and heterogeneous payoffs respectively) has



Table 1. Payoff matrices for social or public dilemma games

Panel A: Invisible hand game

Player 2

Cooperate Do not cooperate

Player 1 Cooperate (170, 170) (130, 140)Do not cooperate (140, 130) (120, 120)

Panel B: Prisoner’s dilemma game

Player 2



Panel C: Assurance game

Player 2



Panel D: Assurance game with heterogeneous payoffs

Player 2



two potential Nash equilibria outcomes—a payoffdominant strategy, and a risk dominant strategy(Harsanyi and Selten, 1988). The optimal decisionfor each strategic alliance partner is dependent ontheir alliance partner’s decisions. If the strategicalliance partner cooperates, then the individual’sbest response is also to cooperate, while if thestrategic alliance partner does not cooperate, thebest response is to not cooperate as well. Thecooperative equilibrium is payoff dominant (earn-ing 170 for each player in Panel C). The do notcooperate equilibrium is risk dominant. This equi-librium outcome earns less for each player (110 inPanel C), but it is also less risky. Each player gains10 if his strategic alliance partner chooses the otheraction. In contrast, in the cooperative equilibrium,each player loses 70 if his strategic alliance partnerchooses the other action.

Successful strategic alliance outcomes thus relyon exchange partners choosing the payoff dom-inant strategy where they all cooperate towardthe joint goal of economic value creation, ratherthan the risk dominant strategy of not investingin the joint alliance activity. Researchers in botheconomics and social psychology have empha-sized different mechanisms through which a higherlikelihood of successful coordination on the pay-off dominant equilibrium can be achieved. Notsurprisingly, economists assume perfectly ratio-nal decision makers and have tended to focus onstructural solutions such as economic incentivesalignment, while social psychologists assume thatpeople act with limited rationality and empha-size motivational solutions such as communication(Zeng and Chen, 2003). We turn to the role ofeconomic incentives and communication among



strategic alliance partners as factors representingeach dominant paradigm below.

Economic incentives

Property rights theory emphasizes the sanctionedbehavioral relations among decision makers in theuse of potentially valuable resources (Barzel, 1989;Libecap, 1989). Coase (1960) introduced propertyrights into the economics of organization and ques-tioned why firms, formal alliance structures, andother institutions existed at all if the price systemwere perfectly efficient. Coase (1937, 1960) notedthat in a world of positive transaction costs, organi-zational forms matter for achieving economic effi-ciency. Property rights theory has much to offer indeveloping a more systematic approach for under-standing strategic alliances (Chi, 1994; Foss andFoss, 2005; Oxley, 1999).

Other related theories to property rights the-ory include agency theory and transactions coststheory. All three organizational economics theo-ries hypothesize that it is necessary for the eco-nomic incentives to be right in order to attainefficient outcomes (Argyres and Liebeskind, 1998;Kim and Mahoney, 2005). Since our frameworklater also examines the role of bounded rational-ity, it is within the camp of incomplete contracting(e.g., transaction costs and property rights theory)(Coase, 1960; Libecap, 1989; Williamson, 1996)and not the complete contracting principal-agentmodel (Holmstrom, 1982). Thus, the current paperis more precisely about property rights theory thanagency theory.

In particular, friction in establishing propertyrights helps to explain and predict why there canbe large and persistent economic gaps betweenpotential and realized value creation (Kim andMahoney, 2005; Mahoney 2005). Property rightstheory emphasizes ‘getting the economic incen-tives right’ (Kim and Mahoney, 2005: 233). Absentsome mutual resource commitments from alliancemembers to align their economic incentives in astrategic alliance environment, the alliances willfail to achieve synergies and sustained economicvalue creation (Gulati et al., 1994). FollowingKhanna et al.’s (1998) theoretical model that isbased on economic reasoning pertaining to prop-erty rights, we analyze the payoff structures ofstrategic alliances in terms of their private andcommon benefits. Khanna et al. define ‘private’benefits as those accruing to individual firms from

activities not governed by the alliance, and ‘com-mon’ benefits as those accruing to all partici-pants in the alliance (Khanna et al., 1998: 195).More specifically, in the context of the assurancegame, private benefits occur when exchange part-ners ‘take’ from others in the form of unilaterallearning of skills and knowledge and applicationin areas unrelated to the alliance’s activities, whilecommon benefits are realized by collective ‘giv-ing’ or sharing of information and application ofthe learning in areas related to the alliance (Khannaet al., 1998).

Since strategic alliances typically result in bothkinds of benefits, the decision makers in a strategicalliance face an inherent tension between compe-tition and cooperation, as exemplified by learningraces where an alliance partner can privately ben-efit at the expense of the others’ in the alliance(Hamel, 1991; Khanna et al., 1998). Therefore, theprobability of strategic alliance success dependson the extent to which the decision makers per-ceive common benefits to be greater than privatebenefits. From a property rights perspective, astrategic alliance has elements of the ‘tragedy ofthe commons’ (Hardin, 1968). If the benefits forcontributing to the strategic alliance (and main-taining the economic value of the common pool)are less than the private benefits from ‘raiding’the pool, then the strategic alliance is less likelyto result in cooperative behavior among the deci-sion makers. Consistent with property rights theory(Coase, 1960), we predict that aligning of eco-nomic incentives is critical to ensuring strategicalliance success. Accordingly, and consistent withKhanna et al. (1998), we posit:

Hypothesis 1: Alliances wherein decision mak-ers have a higher ratio of common to privatebenefits are more likely to achieve success thanalliances wherein decision makers have a lowerratio of common to private benefits.

The above hypothesis is consistent with the earlyproperty rights research literature and the opti-mistic view that efficiency will be readily achieved(Demsetz, 1967). However, more recently, boththeorists and property rights historians have chal-lenged this optimistic view (Eggertsson, 1990;North, 1990). A key idea is that ‘macro’ (prop-erty rights) problems have ‘micro’ foundationsin terms of communication and incentive prob-lems. Recent property rights research literature,



which works from an incomplete contracting per-spective (Hart, 1995), emphasizes that the coor-dination procedures by which one can obtain thecorrect economic incentives are exceedingly dif-ficult. For example, Libecap (1989) notes thatasymmetric information and distributional conflictsoften lead to persistent suboptimization of eco-nomic outcomes. Further, the more heterogeneousthe contractual bargaining parties are, the greaterthe impediments to achieving the full potential ofeconomic value.

These contributions in property rights theory areconsistent with the theoretical reasoning in thestrategic management research literature, whichfocuses on the impact of heterogeneity in part-ner scope and resultant differences in economicincentives on strategic alliance success and failure(Khanna et al., 1998). Given the need for coordina-tion among strategic alliance partners, a key factorthat can impact alliance success is whether deci-sion makers are similar or different to each other interms of perceived benefits of the strategic alliance.An increase in the heterogeneity of decision mak-ers, as modeled in Panel D of Table 1, increasesthe difficulty of reaching the efficient economicoutcome (Libecap, 1989). Khanna et al. (1998)discuss differences in relative scope of alliancepartners and predict that asymmetric common ben-efits can cause problems in achieving coordina-tion or cooperation in strategic alliances. Thereare several reasons why one may expect increasesin heterogeneity to cause a decrease in successrates of strategic alliances. First, the need for coor-dination is greater when exchange partners mustdetermine the optimal allocation of effort, givendifferences in relative common benefits from thealliance activities. Second, heterogeneity amongalliance partners increases the perception of oppor-tunistic behavior by exchange partners, even whennone may be present. Heterogeneity increases thepotential for misunderstanding and creating diver-gent expectations among strategic alliance partners(Goerzen and Beamish, 2005). Accordingly, wehypothesize:

Hypothesis 2: Alliances in which there is het-erogeneity in strategic alliance partners’ ratioof common to private benefits will have a lowerlikelihood of success than alliances whereexchange partners are relatively homogenous intheir ratio of common to private benefits.

Communication

The two hypotheses above relate to factors influ-encing economic incentive alignment and hetero-geneity, and focus on what social psychologiststerm structural solutions to a social dilemma prob-lem (Komorita and Parks, 1994). Importantly, bothhypotheses, and Hypothesis 1 in particular, reston the implicit assumption made by economists:that strategic alliance partners act rationally, andwhen provided with full information about appro-priate economic incentive alignment, gain a com-plete understanding of the coordination problemthat they face.

However, as social psychologists and more mod-ern property rights economists have noted, thestrong form property rights theory abstracts awayfrom considerations arising from coordination fail-ure, miscalculation, free-riding behavior, and dis-tributional conflicts (Libecap, 1989; North, 1990;Olson, 1965). The early strong form view hadrelied on optimization based on economic incen-tive alignment alone, and made many behavioralassumptions that may not hold in reality. Indeed, inan assurance game context, there is a strong likeli-hood that the risk dominant strategy overshadowsthe payoff dominant strategy (Harsanyi and Selten,1988). Even in the absence of actual opportunisticbehavior by any of the strategic alliance partners,the risk and associated fear that others may notcontribute toward joint alliance interests may pre-vent individual decision makers from undertakingactions that will result in alliance success. Suspi-cion breeds distrust and reversion to competitiverather than cooperative actions. Such coordinationcosts are further exacerbated by decision-makingbiases caused by uncertainty (Zajac and Bazer-man, 1991), the anchoring and/or framing problem(Kahneman, Slovic, and Tversky, 1982), or by dif-ferences in the considerations of fairness acrossstrategic alliance partners (Messick, 1991).

In sum, even in strategic alliances wherein deci-sion makers perceive a higher ratio of common toprivate benefits ratio, lack of coordination due toinsufficient common knowledge, differential per-ceptions of other decision makers’ actions, and thebounded rationality of the participants to clearlysee what actions are in their best interests (Simon,1947) can result in the realized value creation froma strategic alliance falling short of the potentialvalue creation. Indeed, Simon’s (1957) definitionof bounded rationality suggests why a substantial



gap between realized value creation and poten-tial value creation may occur: ‘The capacity ofthe human mind for formulating and solving com-plex problems is very small compared with thesize of the problems whose solution is required forobjectively rational behavior in the real world—oreven for a reasonable approximation to such objec-tive rationality’ (Simon, 1957: 198) and ‘it is onlybecause individual human beings are limited inknowledge, foresight, skill and time that organi-zations are useful instruments for the achievementof human purpose’ (Simon, 1957: 199).

When viewed through classic organization the-ory or social psychology lenses, strategic alliancesmay also benefit from incorporation of motiva-tional/design solutions, chief among which is com-munication (Ledyard, 1995; Simon 1947). Thefundamental insight from classical organizationtheory is that effective coordination requires notonly monetary incentives, but also nonmonetaryrewards, and that both formal and informal man-agerial communication can increase the likelihoodof cooperation and coordination (Barnard, 1938).What sets organization theory apart from muchof economics is this emphasis on the nonmate-rial, informal, interpersonal, and moral basis ofbehavior (Scott, 1987). Contemporary organizationtheory concerning social capital—which can bedefined as resources embedded in a social structurethat are accessed and/or mobilized in purposiveactions—is in many ways connected to classicalorganization theory (Koka and Prescott, 2002).

Within the context of the strategic managementliterature, institutional design factors such as therole of both formal and informal communicationin achieving cooperation and coordination has beenemphasized since Barnard (1938). A key differencebetween structural solutions (such as economicincentive alignment) and motivational/design solu-tions is that the former aims to change the under-lying structure, such as the economic payoffs asso-ciated with the problem, while the latter solu-tions are more ‘intangible,’ and attempt to addressthe institutional design to elicit greater coopera-tion (Komorita and Parks, 1994; Ledyard, 1995).Importantly, Zeng and Chen (2003) note that com-munication among alliance partners can poten-tially offer more cost-effective solutions than struc-tural modifications in economic payoffs, giventhe high costs associated with altering monetaryreward structure, monitoring, and potential restruc-turing/consolidation of partner businesses.

Communication matters because it can helpchange strategic alliance partner perceptions of theproblem from competitive to cooperative in twodistinct ways. First, communication can reducecoordination costs and address management issuesrelated to bounded rationality and decision biases.Strategic alliance research has recognized the roleof personal communications among decision mak-ers as a means to achieving cooperation and coor-dination (Rodan and Galunic, 2004; Zaheer andVenkatraman, 1995). Investing time and effortin improving communication improves economicreturns (Adner and Helfat, 2003) by facilitatinga flow of information, which can clarify expecta-tions and causal connections between individualactions and group outcomes (Kogut, 2000). Whendecision makers are aware of each other’s incen-tives and orientation toward the strategic alliance,there is an alleviation of fears related to poten-tial exchange partner misconduct. By reducingthe possibility of surprises, communication canprovide convergent expectations that enhance thecoordination and cohesion of the group (Malm-gren, 1961; Williamson, 1975). Through com-munication, managers can minimize the boundedrationality problem through joint problem solv-ing (McEvily and Marcus, 2005). Indeed, prioralliance related research has found a significantpositive relationship between inter-partner com-munication and superior strategic alliance perfor-mance (Doz, 1996).

Secondly, communication can engender cooper-ation through moral suasion, development of groupidentity, and trust (Komorita and Parks, 1994;Zeng and Chen, 2003). Indeed, Orbell, Dawes,and van de Kragt (1990) underscore the use ofmultilateral promises in increasing cooperation.These communication efforts may exert influenceon decision makers since appeals for cooperationmay influence the decision, which may be rein-forced by identification with the cooperative sys-tem (Barnard, 1938; Simon 1947). Communicationcan thus permit the development of social capitaland trust among strategic alliance partners (Gulati,1998, 1999).

Economists, however, have responded to thesecritiques by highlighting the possibility of ‘cheaptalk’ (Farrell and Rabin, 1996). In game theo-retic language, preplay communication carries nopayoff relevant information, it is simply cheaptalk and should have no effect on the equilib-rium outcomes, particularly if there is a single



dominant strategy (Ledyard, 1995). Indeed, in thesingle-period prisoner’s dilemma context, there issubstantial evidence that communication does notincrease the probability of cooperation (Crawford,1998; Farrell and Rabin, 1996). However, the evi-dence is mixed for game theoretic settings suchas assurance games with multiple Nash equilibriathat better represent the strategic alliance context;and several researchers have found significant sup-port for communication enhancing the probabilityof successful outcomes (Crawford, 1998; Ledyard,1995).

We posit that communication should increase thelikelihood that partners choose the payoff dom-inant strategy over the risk dominant strategy(Harsanyi and Selten, 1988) when the underly-ing economic incentives are appropriately aligned.More specifically, we maintain that due to factorshighlighted in this section, economic incentives arenot by themselves sufficient, and that the additionof communication will significantly increase therate of successful alliances. Thus:

Hypothesis 3: The effect of incentive alignmenton the probability of success in alliances ishigher in the presence of communication thanin its absence.

METHODOLOGY

We test the above hypotheses using experimen-tal methodology (Kahneman, Knetsch, and Thaler,1990; Plott, 1982; Smith, 2000). Well establishedin the social psychology and economics of organi-zation as a fruitful approach for the study of issuespertaining to social dilemma problems (Hazlett,1997; Poppe and Utens, 1986), and consideredcommonplace within the economics and socialpsychology discipline (Kagel and Roth, 1995;Komorita and Parks, 1994; Samuelson, 2005),experiments have also begun to be utilized in thestrategic management field (Knez and Camerer,1994; Song et al., 2002). The use of experimentalmethodology enables us to directly test the theo-ries proposed by implementing different treatmentscorresponding to each while controlling for factorsthat may confound with these mechanisms in thereal world.

As indicated in the theory section, we model thestrategic alliance context as an assurance game.Specifically, we model decision making within a

strategic alliance as a threshold ‘take some orgive some’ game, where decision makers eithercontribute to the alliance for common economicbenefit, or use the alliance for private economicadvantage. Each participant in the experiment rep-resents a firm making a decision about the extentto which to engage in cooperative activities withintheir strategic alliance. Each alliance partner hasdifferent monetary benefits from the success of thealliance, which affects their decisions concerninghow much of the resources to contribute (give) orextract (take) from the strategic alliance. In a seriesof experiments, we examine the behavior of partic-ipants under different assumptions of the ratio ofcommon to private economic benefits accruing asa result of the strategic alliance. We also examinethe impact of communication on strategic allianceperformance by implementing communication pro-tocols to examine their impacts.

Experimental design

Our experimental design is developed for bothexternal and internal validity. The experimentswere designed so that strategic alliance issues groworganically out of the hypotheses that they aredesigned to distinguish (Kagel and Roth, 1995).Moreover, our experiments involve induced valu-ation of participants (Smith, 1976, 2000)—theyare paid for their participation in the experimentin a way that is responsive to the choices theymade—to ensure that participants are motivatedby the same factors they would encounter in thereal world.

For internal validity, we designed a settingwhere theories about strategic alliance behaviorcould be tested directly. Five treatments weredeveloped for the experiment, each representingan interaction between economic incentives (ratioof common to private benefit) and communication.(Details of the implementation are provided in thenext section.) The first treatment of low commonbenefit represents a scenario where the ratio ofcommon to private benefits is low, that is, none ofthe decision makers have an economic incentivefor the alliance to succeed. Second, we consider ascenario of high common benefit of alliance activ-ity, where all of the decision makers have a highcommon to private benefit ratio, that is, the eco-nomic incentives are aligned so that the payoffswhen the alliance succeeds are significantly higherfor every decision maker than when the alliance is



not successful. The third treatment of mixed com-mon benefit allows for heterogeneity in the ratioof common to private benefit among the alliancepartners. For some decision makers, the ratio ofcommon to private benefit is very high, while forothers, the ratio is very low.

The fourth and fifth treatments permit strategicalliance partners to communicate with each other.In the fourth treatment of high common benefit withcommunication, decision makers have a high ratioof common to private benefits and can communi-cate with each other, while in the fifth treatment ofmixed common benefit with communication deci-sion makers have heterogeneous ratios of commonto private benefits, and the ability to communi-cate with each other. Thus, comparisons betweenthe different treatments permit an assessment ofthe individual and interaction effects of economicincentives and communication on strategic allianceoutcomes.

Experimental procedure

Our experiment involved 405 participants who par-ticipated as decision makers in strategic alliances.All participants were business students at aresearch one institution in the United States, withthe majority of the students enrolled in the MBA(regular and executive) program. Specifically, therewere 60 executive MBA students, 300 MBA stu-dents, and 45 senior-level undergraduate students.Since the undergraduate pool of students was rel-atively small, convenience in scheduling consid-erations caused us to mix these students with theMBA students. The executive MBA students hadan average work experience of 12 years, whilethe regular MBA students had an average workexperience of four years. The experiments wereconducted in the context of coursework in cor-porate strategy, and the participants were familiarwith both the concept of strategic alliances and thekey management challenges within this context. Asdescribed below in greater detail, the participantswere also provided with a realistic description ofa learning alliance, to simulate their participationas managers in an alliance setting. Importantly,almost all of these students had been involvedin social dilemma settings in their workplace—ifnot directly in an interfirm alliance setting, in set-tings that required team synergies and a similartension between competition and cooperation. Sub-sequent to the alliance experiments conducted for

research, there were additional experimental sim-ulations conducted for pedagogical purposes in anexecutive education setting (not included in thedata to be compliant with Institutional ReviewBoard guidelines). The outcomes in these exper-iments were consistent with the results reported inthe empirical section.

Participation was strictly voluntary, and in accor-dance with the principle of induced valuation, par-ticipants were paid in cash based on their per-formance. Participants were randomly assigned toone of the five treatments and to their role withinthe treatment. When they arrived at the labora-tory, the participants were seated at a computerterminal. In compliance with Institutional ReviewBoard guidelines, they read and signed a consentform. Participants were provided with a copy oftheir role-specific instructions, and prior to thestart of the experiment, a composite version of theinstructions was read aloud. To ensure that partic-ipants had understood the instructions, the neces-sary decisions they were being asked to make, andthe resulting payoffs, each participant filled a pre-experiment questionnaire, and the answers to thequestions were discussed until there was a consen-sus on the understanding of the experiment. In par-ticular, the pre-experiment questionnaire explicitlyasked the participants to calculate and report pay-offs under scenarios where the alliance was suc-cessful and where the alliance was not success-ful. The entire experiment was computer-aided,and implemented using a Web-based Java appli-cation; participants input their decisions and weregiven feedback electronically at the end of eachperiod of decision making. After the experimentended, participants completed an exit questionnairedescribing their experiences.

Participants were informed that they would roleplay managers who were responsible for allocatingresources to their own firm or to an existing, five-firm alliance to which they belonged. Per Zeng andChen (2003), we model a multiparty alliance ratherthan a two-person alliance. We designed the studyso that participants were (privately) motivated viacash received at the end of the experiment. Theexperiment involved no deception and thus con-tamination effects are not a major concern. Further,participants were asked not to discuss the experi-ment with others.

The experimental context represented a thresh-old social dilemma game, wherein the common



benefits are realized if strategic alliance part-ners meet a certain threshold level of collec-tive contributions (Ledyard, 1995). At the begin-ning of the simulation, the alliance common poolwas endowed with 100 resource units.1 As mem-bers representing decision makers of a five-firmalliance, each participant received 20 resourceunits (created by their research and development[R&D] staff) in every period. The primary decisionconcerned how many resource units they chose togive to or take from the alliance common pool.At the end of every period, each alliance memberreceived $1,000 experimental dollars for resourcesheld within their own firm (private benefit).

Furthermore, if the alliance common poolhad at least 150 resource units, the strategicalliance achieved a successful outcome and eachalliance member received a bonus representingbenefits accruing from the strategic allianceactivity (common benefit). The common benefits(bonus amounts) were calibrated for the differenttreatments described above (low, high, andmixed common benefit). While each alliancemember knew his or her own bonus, noneknew with certainty the bonuses of the otherstrategic alliance members. However, in eachtreatment, the alliance members were informedabout (a) whether the other strategic alliancemembers received similar (homogeneous) ordifferent (heterogeneous) bonuses, and (b) therange of the bonuses across all alliance members:whether the bonuses were all high, all low, or amix of the two. This information was sufficient todetermine the type of game the participants faced.

At the end of each period, after the decisions hadbeen made regarding resource transfer to or fromthe alliance common pool and the members hadreceived their experimental earnings, two eventsoccurred. First, the resources in the alliance poolwere depreciated by 33 percent. The depreciationwas implemented to ensure that if the strategicalliance threshold of 150 was met, the start ofthe next period would replicate the conditions forthe first period (100 units of resources). Second,a random draw indicated whether the game wouldcontinue (80% likely) or end (20% likely) at thatperiod. The random draw enabled us to implement

1 Given the focus on learning alliances, we used the term ‘infor-mation units’ rather than ‘resource units’ in the experimentalsimulation. The two terms, however, are interchangeable giventhe experimental context.

an infinitely repeated game in the lab (with a dis-count rate of 0.2) avoiding endgame effects (Fried-man and Sunder 1994). To ensure compatibilityacross treatments and to reduce variance, the real-ization of the continuation probability draws wasdetermined in advance and applied to all five treat-ments, as recommended in Friedman and Sunder(1994). This random draw resulted in 11 periods,which we used in all treatments.

Implementation of experimental treatmentsand empirical model

As indicated above, the differences in the ratio ofcommon to private benefit treatments were imple-mented by differences in bonus structure across thealliance simulations. The details of the implemen-tation of the payoff matrices, the stage game equi-libria, and the repeated game equilibria for eachtreatment are provided in the Technical Appendix.In the high common benefit treatment, the bonuses(in experimental dollars) were $35,000 for two ofthe firms and $40,000 for three of the firms. Thisbonus ensured that economic incentives made itworthwhile for each of the decision makers to con-tribute to the strategic alliance; in any one period,it collectively cost the firms in the strategic alliance$50,000 to contribute to the alliance common pool,and they collectively received $190,000 in theform of common benefits (bonuses). These param-eters are chosen so as to make it unattractive forany player to take the alliances’ resources in anygiven period as well. By taking, they earn at most$150,000, but they lose their bonuses in this andall future rounds, which occur with an 80 percentchance. Thus, the expected discounted value ofthe losses is at a minimum ($35,000 +0.8∗$35,000+ 0.82∗$35,000, . . .) = $175,000 > $150,000.

Of course, there is still an economic incentiveproblem, since each firm would prefer that theother firms do the contributing while they freeride (or even worse, take resources), as long as thestrategic alliance remained successful. Note thatno one firm has an economic incentive (or theresources) to unilaterally contribute 50 resourceunits to receive the bonuses (costing $50,000 andgaining at most $40,000). Nonetheless, the pay-offs are consistent with economic incentives beingaligned toward success of the alliance. As in theassurance game, if others are contributing, it is inthe best interests of a target firm to contribute aswell.



In the low common benefit treatment, thebonuses were $4,000 for two firms and $5,000 forthree firms. This bonus made it economically inef-ficient for any of the firms to contribute to thecommon alliance activity; in any one period, itcollectively cost the firms $50,000 to contributeto the alliance common pool, but their collectivecommon benefit was only $23,000. Thus, the pay-offs in this treatment are consistent with economicincentives not being aligned for strategic alliancesuccess.

Finally, the mixed common benefit treatmentinvolved heterogeneity among the bonusesreceived by the alliance members; while three ofthe firms received high bonuses ($35,000, $35,000,and $40,000), two firms received low bonuses($5,000 each). Importantly, the bonus values arestill consistent with incentive alignment for thestrategic alliance to succeed ($120,000 of benefitsversus $50,000 of costs); however the heterogene-ity in bonuses creates a problem. For instance,were the alliance only among the three firms withhigh bonuses they could contribute enough to makethe threshold and collect their bonuses (the alliancepool shrinks by 50 units each period, and thesethree members together control 60 units). How-ever, there is an incentive for the low commonbenefit firms to take the resources in the alliancepool (earning together $150,000 and losing onlytheir total bonuses of $10,000), and thus prevent-ing the strategic alliance from succeeding. A suc-cessful outcome requires that the members of thestrategic alliance configure the optimal amount ofcontributions so that the low common benefit firmsare also better off from alliance success.

The two treatments that permitted communica-tion were implemented via a free-form chat box.For both the high common benefit with commu-nication treatment and the mixed common ben-efit with communication treatment, the strategicalliance members had the ability to chat with allof the other members in their strategic alliance, orsend private messages to any one alliance member.The ‘chat’ feature was implemented using a ‘chatbox’ resembling instant messaging, and a record ofall prior messages was available for each memberas the strategic alliance progressed across periods.

Our primary dependent variable, alliance suc-cess, is coded as one if the alliance common poolresource units exceeded the threshold of 150 in aparticular period, and zero otherwise. Our second

dependent variable, transfer of resources, is mea-sured as the net amount of resources transferred tothe alliance common pool (total giving minus totaltaking by all alliance members in each period).Our last dependent variable, resources in allianceis the total amount of resources in the alliance poolat the end of each period, which is the sum ofthe residual resources from the prior period (afterdepreciation) and the net transfer of resources inthe current period. As expected, these measures arehighly correlated but they capture multiple aspectsof the same question: to what extent did strategicalliance members create value?

We note that our unit of analysis is at thealliance rather than the firm level. However thereis consistency across the two levels, since thepayoffs are higher for each firm if they cooperaterather than compete where there are high or mixedcommon benefits, and vice versa for low commonbenefits. Therefore, while there may be relativegreater benefits to some firms than others, thereare no clear winners or losers if they all play theoptimal strategy (i.e., either they all collectivelywin by cooperating in a high common benefitstrategy, or they ‘compete’ in the form of learningraces, and others, in the low common benefitscenario).

The main independent variables in the modelinclude the indicator variables for each of thefive treatments described above (e.g. high com-mon benefit = one if the observation was drawnfrom that treatment, and zero otherwise), and theperiod in which the decisions were made. In addi-tion, since the experiments represent hierarchicaldata (alliance members are grouped together), weinclude group fixed effects to control for unob-served heterogeneity due to idiosyncratic, groupspecific factors. We use a logistic regression anal-ysis for our first dependent variable, and multi-variate regression analysis for the other dependentvariables.

RESULTS

Table 2a provides the distribution of the observa-tions across each treatment type, the number ofgroups per treatment, and the number of periodsin each treatment. We conducted one-way ANOVAtests for homogeneity across the executive MBAand MBA/undergraduate participant pool. As can



Table 2a. Sample statistics across treatments

Lowcommonbenefit

Highcommonbenefit

High commonbenefit, with

communication

Mixedcommonbenefit

Mixed commonbenefit, with

communication

Number of observations 187 165 176 176 187Number of groups 17 15 16 16 17Number of periods 11 11 11 11 11

Table 2b. Tests of homogeneity across participant pool

Lowcommonbenefit

Highcommonbenefit

High commonbenefit, with

communication

Mixedcommonbenefit

Mixed commonbenefit, with

communication

Executive MBA 0.00 0.19 0.59 0.11 0.22(0.00) (0.15) (0.24) (0.11) (0.18)

MBAa 0.00 0.27 0.58 0.10 0.21(0.00) (0.15) (0.24) (0.09) (0.16)

F-statisticb n.a 2.45 0.07 0.08 0.14

a Included in this pool are 300 MBA students and 45 senior level undergraduate students. Since the undergraduate pool of studentswas relatively very small, convenience in scheduling considerations caused us to mix these students with the MBA students.b The critical F-statistic for a 5% level of significance is 3.85, thus the hypothesis that the two samples are drawn from the samepopulation cannot be rejected.

be seen in Table 2b, there is no statistically sig-nificant difference in the success rates across theexecutive MBA and the MBA/undergraduate par-ticipants in any of the treatments, justifying ourpooling of these two samples.

Tables 3a and 3b provide information regardingthe differences in the dependent variables acrossthe treatments, and the results from the formalhypotheses testing are reported in Table 4. In par-ticular, the regressions control for unobserved het-erogeneity by including an indicator variable foreach group and for the period in which the deci-sions are made. Consistent with Hypothesis 1,economic incentives do matter, and Panel A ofTable 3b shows the difference between the low andhigh common benefit treatments. In the low com-mon benefit treatment where there is no economicincentive to participate in an alliance, alliancesnever succeeded. On average, less than one unitof resources is transferred to the alliance, andvery little of the initial resource stock remains. Instark contrast, in treatments where there is an eco-nomic incentive for the alliance to succeed, thealliance success rates are positive. For example,27 percent of the strategic alliances in the highcommon benefit succeed. The average amounts ofresources transferred and the total of resources in

the common pool is significantly higher as well.More formally, based on the results in Panel Aof Table 4, Hypothesis 1 is supported. Economicsincentive alignment is a necessary condition foralliances to succeed, and the coefficient of highcommon benefit is significant and positive for allthree dependent variables.

Furthermore, consistent with Hypothesis 2, het-erogeneity in the ratio of common to private ben-efits reduces the likelihood of success; only 10percent of the strategic alliances in mixed com-mon benefit treatment result in successful out-comes relative to 27 percent in the high com-mon benefit treatment (Panel B of Table 3b), andthe success rates with communication are 22 per-cent rather than 59 percent, respectively (Panel Cof Table 3b). The formal test of Hypothesis 2 isreported in Panels B and C of Table 4: regard-less of whether there is communication or not, thehomogeneous high common benefit treatment hassignificantly higher performance than the hetero-geneous mixed common benefit treatment for allthree of the dependent variables in our analysis.

Also, contrary to the optimistic predictions ofproperty rights theory, economic incentives aloneare not sufficient. The success rate of 27 percent inthe high common benefit treatment is a far cry from



Table 3a. Success rates across treatments

Without communication With communication

Low common benefit 0% n.a.High common benefit 27% 59%Mixed common benefit 10% 22%

Table 3b. ANOVA tests across treatments

Panel A. High vs. low common benefit, no communication

High, no communication Low, no communication Chi-squared/F-ratios

Observations 165 187Alliance success 27.27% 0.00% χ 2 = 75.75∗∗

Transfer of resources 13.59 0.47 F = 9.61∗∗

(3.09) (2.90)Resources in alliance 47.00 24.47 F = 37.37∗∗

(3.00) (2.52)

Panel B. High vs. mixed common benefit, no communication

High, no communication Mixed, no communication Chi-squared/F-ratios


Transfer of resources 13.59 3.01 F = 5.28∗


(3.00) (2.98)

Panel C. High vs. mixed common benefit, with communication

High, with communication Mixed, with communication Chi-squared/F-ratios




(2.90) (2.91)

Panel D. High common benefit, with vs. no communication

High, no communication High, with communication Chi-squared/F-ratios




(3.00) (2.90)



Table 3b. (Continued )

Panel E. Mixed common benefit, with vs. no communication

Mixed, no communication Mixed, with communication Chi-squared/F-ratios




(2.98) (2.91)

Standard errors in parentheses; ∗ significant at the 5% level; ∗∗ significant at the 1% level.

Table 4. Fixed effects regression

Panel A. High vs. low common benefit, no communication

Independent variable Alliance success Transfer of resources Resources in alliance

Intercept — 3.60 55.32∗∗

(4.18) (2.21)High common benefit 1.23∗∗ 6.56∗∗ 11.27∗∗

(0.19) (1.95) (1.03)Period 0.35∗∗ 0.57 −3.26∗∗

(0.04) (0.62) (0.33)

Observations 352 352 352Log-likelihood −134.56R2 0.08 0.25 0.74χ2/F χ 2 = 22.35∗∗ F = 3.25∗∗ F = 28.55∗∗

Panel B. High vs. mixed common benefit, no communication


Intercept — 7.58 61.56∗∗

(4.94) (4.27)High common benefit 0.60∗∗ 5.29∗ 7.49∗∗

(0.15) (2.31) (1.99)Period 0.05 0.12 −3.67∗∗

(0.05) (0.73) (0.63)

Observations 343 343 343Log-likelihood −163.58R2 0.06 0.02 0.13χ2/F χ 2 = 18.29∗∗ F = 5.27∗ F = 14.07∗∗

Panel C. High vs. mixed common benefit, with communication


Intercept — 25.47∗∗ 73.08∗∗

(4.37) (2.84)High with communication 0.79∗∗ 7.42∗∗ 11.35∗∗

(0.11) (2.04) (1.33)Period 0.05∗∗ 0.42 −0.97∗

(0.02) (0.64) (0.42)




Table 4. (Continued )

Panel D. High common benefit, with vs. no communication


Intercept — 22.82∗∗ 74.55∗∗

(3.99) (2.19)High with communication 0.66∗∗ 10.90∗∗ 15.82∗∗

(0.12) (1.86) (1.02)Period 0.04∗ 0.28 −1.96∗∗

(0.02) (0.59) (0.32)


Panel E. Mixed common benefit, with vs. no communication


Intercept — 10.18∗ 59.87∗∗

(4.75) (3.05)Mixed with communication 0.35∗∗ 8.76∗∗ 12.11∗∗

(0.13) (2.22) (1.42)Period 0.22∗∗ 0.27 −2.68∗∗

(0.02) (0.70) (0.45)


Standard errors in parentheses; group dummies included (not reported); ∗ significant at the 5% level; ∗∗ significant at the 1% level.

the theoretical optimum of 100 percent. Even whenpayoffs would be uniformly higher for everyone ifthey all cooperated, some individuals choose the

risk dominant strategy of not cooperating in almost75 percent of the cases. Tables 3a and 3b showthe importance of communication for successful

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

1 3 4 6 7 9 10 11 1 3 4 5 7 8 10 11

No communication With communication

Suc

cess

rat

e

Period Period

2 5 8 2 6 9

Figure 1. Effect of communication on success rate in high common benefit treatments. This figure is available incolor online at www.interscience.wiley.com/journal/smj



alliance performance. As seen in Panels D and E ofTable 3b, when economic incentives are in place,adding communication in the high common benefittreatment more than doubles the success rate (from27% to 59%) and the amount of resources trans-ferred to the strategic alliance (from 13 units to 35units), and increases the amount of resources con-tained in the alliance (from 47 units to 78 units).The same effect occurs when communication isincluded in the mixed common benefit treatment.This result is consistent with Hypothesis 3 andis further empirical evidence supporting organi-zational theory—that is, economic incentives arenot, by themselves, sufficient, and the addition ofcommunication significantly increases the rate ofsuccessful alliances. The formal tests of Hypothe-sis 3 reported in Panels D and E of Table 4 showstrong support for all three dependent variables inthe study.

DISCUSSION AND CONCLUSIONS

Strategic alliances are an important mode of capa-bility development in the face of environmen-tal changes and increases in competitive inten-sity. However, successful outcomes from coopera-tive alliances are contingent on exchange partners’decisions to contribute to the strategic alliance. Inan attempt to better understand decision makingin strategic alliances, our research brings togethercomplementary streams of literature that empha-size the role of economic incentives (as in eco-nomic property rights theory) and the role ofcommunication (as in social psychology and clas-sic organizational theory) and tests their relativeeffects by using an experimental design that per-mits the isolation of the underlying causal mecha-nisms. Consistent with property rights theory, wefind that aligning economic incentives is necessaryfor success (Barzel, 1989), but is not sufficient assome would predict (Demsetz, 1967). This findingsuggests that a myriad of factors like coordina-tion costs, bounded rationality, and lack of trustin the absence of shared knowledge can createendogenous uncertainty regarding partner actions,and cause realized value creation to be less thanpotential value creation. While designing the rightpayoff structure is a necessary condition for strate-gic alliance success (Khanna et al., 1998), it doesnot seem to be sufficient.

In this context, the ability to communicate sig-nificantly increases the probability of success.The results are both economically and statisticallysignificant; communication approximately doublesthe rate of strategic alliance success. As depicted inFigure 1, the success rates in the absence of com-munication are low and relatively constant acrossperiods. If the decision makers did not ‘get it right’the first time, the probability of success was verylow for the strategic alliance across time. In con-trast, communication increases success rates notonly in the first period, but also causes successrates to increase as decision makers interact witheach other in subsequent periods.

An informal inspection of the communicationcontent of the decision makers revealed that com-munication allowed strategic alliance members torecover from mistakes and coordinate. The fol-lowing excerpts from an alliance communicationrepresent the various effects of communication,including but not limited to, creating a sharedunderstanding of the rules, explanation of behav-ior, and development of trust.

Company 5: ‘Alliance partners let’s aim to max-imize profits.’

Company 2: ‘I think it is beneficial if we all worktogether.’

Company 4: ‘If everyone enters 10 every time,there will always be 100 at the beginning ofthe quarter. Therefore, we can easily reach 150every time [to make our bonus]. What doeseveryone think?

Company 1: ‘There are many [computer] win-dows to manage; hard to keep an eye on theclock . . . You guys . . . woops, I made a mistake!I am putting 15 now to win back your trust. [Oth-erwise] revenge will destroy us.’

Company 3: ‘Yeah, please do not do that again. . . You’ll make me paranoid.’

Thus, in line with classical organizational the-ory and social psychology, in our setting webelieve that communication enables managers toset goals, to coordinate, and to provide initial com-munications and subsequent feedback. Communi-cation can reduce defection from cooperative out-comes, mitigate problems of bounded rationality



(Simon, 1947), lessen fears of opportunism (Hen-nart, 1988), and enable the group to recover frommistakes. Communication also enables leaders ofthe group to make appeals not to be selfish andto cooperate (Barnard, 1938; Miller, 1992), whichseem to work for some of the participants for someof the time.

Our study thus provides a systematic analy-sis in a laboratory setting of issues encounteredin the real world. Consider, for example, themany strategic alliances forged among automobilemakers across the globe in the late eighties andnineties. In addition to the GM-Toyota learningalliance that has received some scholarly attention(Inkpen, 2000; Khanna, Gulati, and Nohria, 2000),the popular press has identified many other strate-gic alliances among U.S., European and Japanesecorporations (e.g., Treece, Miller, and Melcher,1992). While many of these provided less thanstellar outcomes, an insightful BusinessWeek arti-cle by Treece et al. (1992) provides rich detail onhow the interplay of the mechanisms identified inour study led to the success of the Ford-Mazdaalliance over a 13-year period. Beginning in 1979,when Ford partnered with the ailing Mazda to helpbolster its economic performance, the strategicalliance helped both corporations achieve syner-gies by combining Ford’s international marketingand finance expertise with Mazda’s engineeringand product development know-how. Mutual eco-nomic incentive alignment was ensured, as exem-plified by the following quotes from Treece et al.(1992: 102–107):

In choosing what to work on jointly, the partnersoperate on a project-by-project basis, with ideascoming from people throughout both organiza-tions. The main criterion for approving an ideais that it benefits both companies.

The Navajo project is indicative of how therelationship works. Back in the spring of 1987,Mazda, which didn’t have a sport-utility vehicleof its own, decided it wanted to buy from Ford amodified version of its upcoming Explorer off-road vehicle. Ford was glad to oblige, partlybecause it was a chance to prove it could betrusted to manufacture Mazda’s pickups in 1993.

The [Europe] deal would guarantee Ford extrasales for a new car line while giving Mazdaaccess to a key market where quotas limit Japanese

imports. . . . [but] “the concern is always therethat one party will benefit unfairly from whatyou’re about to do” [David R. Gunderson, Mazdaboard member].

However, much of the article describes why thisunderlying economic alignment was not sufficientin itself. The communication efforts undertakenby both corporations to ensure coordination, rela-tionship management, expectations conformation,and even camaraderie development are illustratedby the following quotes from Treece et al. (1992:102–107):

Ford and Mazda can call on some hard-learnedprinciples for managing a successful strategicalliance, . . . many of which would apply to tiesin any industry. Underlying them all is the ideathat benign neglect is no basis for a partnership.

Says Ford President Philip E. Benton Jr.:“There’s a lot of hard work in making it work.”To help smooth future deals, Ford and Mazdahave even developed a set of Basic BusinessPrinciples. These tenets outline how to price adeal and how to share development costs, forexample.

Communication between the two groups couldbe hard at times, but they did their best tolighten the mood. There was a lot of arm-flailing and miming to get messages across. Andone Saturday, as the Mazda group worked inthe Louisville plant, the lights suddenly wentout and the doors and windows slammed shut.They thought they were locked in for the week-end–until they were introduced to the Americantradition of April Fool’s Day.

What’s the secret? “The most important pointis for people to meet face-to-face and freelytalk,’ explains President Wada [Yoshihiro Wada,president of Mazda Motors Corp.]. Much ofthe credit for keeping everyone talking belongsto the four men who monitor the alliance, W.Wayne Booker, Gunderson, Wada, and ShigeoKasuga.

In highlighting the above mechanisms, we alsocontribute to scholarly work that has built onWilliamson’s (1975) seminal work, aside from therich transactions costs insights regarding the role



of opportunistic behavior, uncertainty, asset speci-ficity, and their interactions in determining thegovernance form choice among feasible alterna-tives (Leiblein and Miller, 2003; Villalonga andMcGahan 2005). Williamson (1975) also identi-fied other factors that are critical determinants ofsuccess, which our study highlights. For instance,Williamson (1975) underscored the presence ofbounded rationality in decision making, and thatheterogeneity among decision makers further exac-erbates difficulties in achieving cooperative out-comes (Williamson, 1975: 239–240). Importantly,Williamson (1975) asserted that the developmentof convergent expectations is a vital managerialrole, and that managerial communication promotesconvergent expectations by attenuating uncertain-ties generated when interdependent parties makeindependent decisions (Malmgren, 1961). Whiletransaction costs theory has emphasized the oppor-tunism problem, our results suggest that greaterattention be given to the problem of boundedrationality in alliance and relational contracting.More generally, our empirical results find strongsupport for the role that economic incentives,bounded rationality, heterogeneity, and communi-cation play in determining successful alliance out-comes, which will hopefully generate new interestand empirical inquiry concerning factors that relateto strategic alliance success. Indeed, the theoreti-cal logic and empirical results presented here canbe applied to a large number of strategic contextsincluding buyer-supplier arrangements, joint ven-tures, franchising, internal corporate ventures, net-works, R&D consortia, technology transfer agree-ments, and intraorganizational teams (Lerner andMerges, 1998; Phene, Fladmoe-Lindquest, andMarsh, 2006; Roethaermel and Deeds, 2004; Reuerand Ragozzino, 2006; Santoro and McGill, 2005).

Our research study has some limitations, whichalso open up avenues for future inquiry. Our lab-oratory setting and use of experimental method-ology allowed us to disentangle the relative andinteraction effects of the causal mechanisms under-lying decision making in strategic alliances, butat some cost of realism incurred by our need toabstract away from the confounding issues thatare clearly relevant in actual strategic alliancesundertaken by corporations in the real world. Forinstance, our experiments controlled for selec-tion effects by randomly assigning participantsto the different treatments, and did not take into

account exogenous uncertainty (either technologi-cal or demand driven). Future research examiningthe role of prior relationships, due diligence in thepre-alliance phase in exchange partner choice, andthe use of contractual safeguards can help shedlight on the effect of these factors on decision mak-ing. In the same vein, our anonymous experimentalsetting did not permit strategic alliance members tocredibly threaten consequences for deviant behav-ior, and research in this area would be beneficial aswell. All of these factors may be important addi-tional criteria for increasing the likelihood of suc-cess. This is a particularly important area of furtherresearch since our empirical results show that evenin the treatments where there were both incentivealignment and communication, there was a 40 per-cent likelihood that the strategic alliance wouldnot succeed. Therefore, our experiment demon-strated that even in the simple game structures thatwe used, strategic alliance partners may place anonzero probability that others will not cooperate,and accordingly choose the risk dominant strat-egy among the multiple Nash equilibria. Exoge-nous uncertainty may interact with the endogenousuncertainty about alliance partner actions to cre-ate additional challenges in the realization of thepotential economic value creation, and researchthat examines such interaction effects would befruitful.

The use of students as subjects is open to thecriticism that students may not emulate the actualdecisions of managers in a strategic alliance set-ting. We think that this concern is somewhat miti-gated in our setting for the following reasons. First,the majority of the participants in our experimentshad at least four years of work experience. Further,there was no significant difference in the resultsobtained for groups in which the alliance mem-bers were executive MBA’s; the average of thiscohort was approximately 40 years of age with aminimum of seven years work experience. Manyof the executives had also participated in strate-gic alliance activity as part of their job descrip-tions. Finally, research in experimental economicshas addressed this issue explicitly. Indeed, Dyer,Kagel, and Levine (1989) and Croson and Dono-hue (2006) find that ‘real world’ decision makersperformed the same or sometimes worse in thelaboratory setting than students. Nonetheless, addi-tional research that uses other sampling frames, notjust MBA/undergraduate students, would clearlyprovide additional value to the field.



We contribute to the extant research literature byexamining decision making that leads to alliancesuccess or failure, and, in particular, focus on therelative and interaction effects of economic incen-tive alignment and communication. In doing so,we integrate across conceptual papers examiningstrategic alliances that have used either an eco-nomic game theoretic lens and highlighted therole of economic incentives (Gulati et al., 1994;Khanna et al., 1998) or a social psychology lensand highlighted the role of communication as amotivational solution (Zeng and Chen, 2003). Bycontinuing in this under-researched route of exam-ining strategic alliances using the social dilemmaparadigm, we contribute to the research literatureby developing and testing hypotheses based onsome of the propositions of both, and developingnovel ones that integrate across the two comple-mentary strands of research. Methodologically, wecontribute to the strategic management literatureby deploying underutilized but powerful experi-mental methods to isolate the effects of alternativecausal mechanisms; we hope that more researchersuse this well-established technique to address addi-tional strategic management issues.

Through the current empirical research, we canappreciate more fully the importance of organiza-tion activities that create a shared confidence in oneanother’s cooperativeness (Barnard, 1938; Miller,1992). In most real-world settings, people do nothave a dominant strategy to defect or to coop-erate. Instead, cooperation is rational when eachperson has a high level of confidence that oth-ers will cooperate. If an organization fails to gainand maintain convergent expectations of coopera-tion, the cooperative system can quickly unravel.In a world of limited information processing andbounded rationality (Simon, 1947), communicationcan be essential for obtaining ‘common knowl-edge’ that each strategic alliance partner intendsto cooperate, that each alliance partner knows theother exchange partners intend to cooperate, and soon (Schofield, 1985). Thus, our empirical findingsunderscore Barnard (1938)’s emphasis that build-ing a cooperative system requires inculcating beliefin the real existence of common purpose.

In conclusion, by undertaking the integrationand testing of complementary theories and under-lying mechanisms, we contribute to strategicalliance theory. Consistent with Khanna et al.’s(1998) propositions, we find empirical evidence

that payoff structures are critical for determin-ing strategic alliance outcomes, and that increasedheterogeneity in strategic alliance scope can resultin lower rates of success. Thus, an important con-tribution to the organizational theory literature isthat alliance partners need to give close attentionto the underlying economics of a strategic alliance,and to ensure that win-win situations are createdfor all members in the strategic alliance. Further-more, by integrating key insights from social psy-chology and classical organizational theory, wemake an important contribution to property rightstheory; that successful strategic alliance outcomesare more likely when both economic incentivesand communication are present. These two mech-anisms are thus complements, rather than sub-stitutes, in determining the success of strategicalliances.

ACKNOWLEDGEMENTS

We thank the University of Illinois ResearchBoard and the Ewing Marion Kauffman Founda-tion for financial support. Shravan Gaonkar pro-vided able technical assistance for the creation ofthe experiment programming code. We appreci-ate the support provided by Steven Sonka and thestaff at the National Soybean Research Labora-tory at the University of Ilinois. Finally, we grate-fully acknowledge helpful suggestions from twoanonymous reviewers, Jay Anand, Ranjay Gulati,Matthew McCarter, Gregory Northcraft, JoanneOxley, and seminar participants at the Universityof Illinois and the Academy of Management Meet-ings in 2003.

REFERENCES

Adner R, Helfat CE. 2003. Corporate effects anddynamic managerial capabilities. Strategic Manage-ment Journal , October Special Issue 24: 1011–1025.

Ahuja G. 2000. The duality of collaboration: inducementsand opportunities in the formation of interfirmlinkages. Strategic Management Journal , MarchSpecial Issue 21: 317–343.

Anand BN, Khanna T. 2000. Do firms learn to createvalue? The case of alliances. Strategic ManagementJournal , March Special Issue 21: 295–315.

Arend RJ, Seale DA. 2005. Modelling alliance activity:an iterated prisoners’ dilemma with exit option.Strategic Management Journal 26(11): 1057–1074.

Argyres N, Liebeskind JP. 1998. Privatizing the intellec-tual commons: universities and the commercialization



of biotechnology. Journal of Economic Behavior &Organization 35(4): 427–454.

Barnard CI. 1938. The Functions of the Executive.Harvard University Press: Cambridge, MA.

Barzel Y. 1989. An Economic Analysis of PropertyRights . Cambridge University Press: Cambridge, UK.

Camerer CF. 1991. Does strategy research need gametheory? Strategic Management Journal , WinterSpecial Issue 12: 137–152.

Chan S, Kensinger J, Keown A, Martin J. 1997. Dostrategic alliances create value? Journal of FinancialEconomics 46(2): 199–221.

Chi T. 1994. Trading in strategic resources: necessaryconditions, transaction cost problems, and choice ofexchange structure. Strategic Management Journal15(4): 271–290.

Coase RH. 1937. The nature of the firm. Economica 4:386–405.

Coase RH. 1960. The problem of social cost. Journal ofLaw and Economics 3: 1–44.

Crawford V. 1998. A survey of experiments oncommunication via cheap talk. Journal of EconomicTheory 78(2): 286–298.

Croson R, Donohue K. 2006. Behavioral causes of thebullwhip effect and the observed value of inventoryinformation. Management Science 52(3): 323–336.

Davis S. 2003. Handbook of Research Methods inExperimental Psychology . Blackwell: Oxford, UK.

Demsetz H. 1967. Towards a theory of property rights.American Economic Review 57: 347–359.

Doz YL. 1996. The evolution of cooperation in strategicalliances: initial conditions or learning processes.Strategic Management Journal , Summer Special Issue17: 55–83.

Dyer D, Kagel J, Levine D. 1989. A comparison of naı̈veand experienced bidders in common value auctions:a laboratory analysis. Economic Journal 99(March):108–115.

Dyer JH. 1997. Effective interfirm collaboration: howfirms minimize transaction costs and maximizetransaction value. Strategic Management Journal18(7): 535–556.

Dyer JH, Hatch NW. 2006. Relation-specific capabilitiesand barriers to knowledge transfers: creating advan-tage through network relationships. Strategic Manage-ment Journal 27(8): 701–719.

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

Eggertsson T. 1990. Economic Behavior and Institutions .Cambridge University Press: Cambridge, UK.

Farrell J, Rabin M. 1996. Cheap talk. Journal ofEconomic Perspectives 19(3): 103–118.

Foss K, Foss NJ. 2005. Resources and transaction costs:how property rights economics furthers the resource-based view. Strategic Management Journal 26(6):541–553.

Friedman D, Sunder S. 1994. Experimental Methods: APrimer for Economists . Cambridge University Press:Cambridge, UK.

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

Gottschlag O, Zollo M. 2007. Interest alignment andcompetitive advantage. Academy of ManagementReview 32(2): 418–437.

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

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

Gulati R, Khanna T, Nohria N. 1994. Unilateral com-mitments and the importance of process in alliances.Sloan Management Review 35(3): 61–69.

Gulati R, Singh H. 1998. The architecture of cooperation:managing coordination costs and appropriationconcerns for strategic alliances. Administrative ScienceQuarterly 43(4): 781–814.

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

Hamilton BH, Nickerson JA. 2003. Correcting forendogeneity in strategic management research.Strategic Organization 1(1): 51–78.

Hardin G. 1968. The tragedy of the commons. Science.162: 1243–1248.

Harsanyi JC, Selten R. 1988. A General Theory of Equi-librium Selection in Games . MIT Press: Cambridge,MA.

Hart O. 1995. Firms, Contracts, and Financial Structure.Clarendon Press: Oxford, UK.

Hazlett, D. 1997. A common property experiment with arenewable resource. Economic Inquiry 35: 858–861.

Hennart JF. 1988. A transaction costs theory of equityjoint ventures. Strategic Management Journal 9(4):361–374.

Holmstrom B. 1982. Moral hazard in teams. Bell Journalof Economics 13(2): 324–340.

Inkpen AC. 2000. A note on the dynamics of learningalliances: competition, cooperation, and relative scope.Strategic Management Journal 21(7): 775–779.

Ireland RD, Hitt MA, Vaidyanath D. 2002. Alliancemanagement as a source of competitive advantage.Journal of Management 28(3): 413–446.

Kagel JH, Roth AE (eds). 1995. The Handbook ofExperimental Economics . Princeton University Press:Princeton, NJ.

Kahneman D, Knetsch J, Thaler R. 1990. Experimentaltests of the endowment effect and the Coase theorem.Journal of Political Economy 98(6): 1325–1348.

Kahneman D, Slovic P, Tversky A. 1982. Judgmentunder Uncertainty: Heuristics and Biases . CambridgeUniversity Press: Cambridge, UK.

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



Kale P, Singh H. 2007. Building firm capabilities throughlearning: the role of the alliance learning processin alliance capability and firm-level alliance success.Strategic Management Journal 28(10): 981–1000.

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

Khanna T, Gulati R, Nohria N. 2000. The economicmodeling of strategy process: ‘clean models and‘dirty hands’. Strategic Management Journal 21(7):781–790.

Kim J, Mahoney JT. 2005. Property rights theory,transaction costs theory, and agency theory: anorganizational economics approach to strategicmanagement. Managerial and Decision Economics26(4): 223–242.

Knez M, Camerer CF. 1994. Creating expectationalassets in the laboratory: coordination in weakest-linkgames. Strategic Management Journal , Winter SpecialIssue 15: 101–120.

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

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

Komorita SS, Parks CD. 1994. Social Dilemmas . Brown& Benhmark: Dubuque, IA.

Ledyard J. 1995. Public goods: a survey of experimen-tal research. In The Handbook of Experimental Eco-nomics , Kagel JH, Roth AE (eds). Princeton Univer-sity Press: Princeton, NJ; 111–181.

Leiblein MJ, Miller DJ. 2003. An empirical examinationof transaction and firm-level influences on the verticalboundaries of the firm. Strategic Management Journal24(9): 839–859.

Lerner J, Merges RP. 1998. The control of technologyalliances: an empirical analysis of the biotechnologyindustry. Journal of Industrial Economics 46(2):125–156.

Libecap GD. 1989. Contracting for Property Rights .Cambridge University Press: New York.

Luo Y. 2008. Procedural fairness and interfirm coopera-tion in strategic alliances. Strategic Management Jour-nal 29(1): 27–46.

Madhavan R, Koka BR, Prescott JE. 1998. Networks intransition: how industry events (re)shape interfirmrelationships. Strategic Management Journal 19(5):439–459.

Mahoney JT. 2005. Economic Foundations of Strategy .Sage: Thousand Oaks, CA.

Malmgren HR. 1961. Information, expectations and thetheory of the firm. Quarterly Journal of Economics75(3): 399–421.

McEvily B, Marcus A. 2005. Embedded ties and theacquisition of competitive capabilities. StrategicManagement Journal 26(11): 1033–1055.

McEvily B, Zaheer A. 1999. Bridging ties: a sourceof firm heterogeneity in competitive capabilities.Strategic Management Journal 20(12): 1133–1156.

Messick D. 1991. On the evolution of group-basedaltruism. In Game Equilibrium Models , Selten R (ed).Springer Verlag: Berlin: 304–328.

Miller GD. 1992. Managerial Dilemmas: The PoliticalEconomy of Hierarchy . Cambridge University Press:Cambridge, UK.

Mowery DC, Oxley JE, Silverman BS. 1996. Strategicalliances and interfirm knowledge transfer. StrategicManagement Journal , Winter Special Issue 17:77–91.

North DC. 1990. Institutions, Institutional Change, andEconomic Performance. Cambridge University Press:Cambridge, UK.

Olson M. 1965. The Logic of Collective Action . HarvardUniversity Press: Cambridge, MA.

Orbell JM, Dawes RM, van de Kragt AJC. 1990. Thelimits of multilateral promising. Ethics 100: 616–627.

Oxley JE. 1997. Appropriability hazards and governancein strategic alliances: a transaction cost approach.Journal of Law, Economics & Organization. 13(2):387–409.

Oxley JE. 1999. Institutional environment and themechanism of governance: the impact of intellectualproperty protection on the structure of inter-firm alliances. Journal of Economic Behavior &Organization 38(3): 283–309.

Phene A, Fladmoe-Lindquest K, Marsh L. 2006. Break-through innovations in the U.S. biotechnology indus-try: the effects of technological space and geo-graphic origin. Strategic Management Journal 27(4):369–388.

Plott CR. 1982. Industrial organization theory and exper-imental economics. Journal of Economic Literature20(4): 1485–1527.

Poppe M, Utens L. 1986. Effects of greed and fear ofbeing gypped in a social dilemma situation withchanging pool size. Journal of Economic Psychology7(1): 61–71.

Reuer JJ, Ragozzino R. 2006. Agency hazards andalliance portfolios. Strategic Management Journal27(1): 27–43.

Reuer JJ, Zollo M, Singh H. 2002. Post-formationdynamics in strategic alliances. Strategic ManagementJournal 23(2): 135–151.

Rodan S, Galunic C. 2004. More than network structure:how knowledge heterogeneity influences managerialperformance and innovativeness. Strategic Manage-ment Journal 25(6): 541–562.

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

Rousseau JJ. 1754/1984. A Discourse on Inequality .Penguin Books: London, UK.

Saloner G. 1991. Modeling, game theory, and strategicmanagement. Strategic Management Journal WinterSpecial Issue 12: 119–136.

Samuelson L. 2005. Economic theory and experimentaleconomics. Journal of Economic Literature 43:63–107.

Santoro MD, McGill JP. 2005. The effect of uncer-tainty and asset co-specialization on governance in



biotechnology alliances. Strategic Management Jour-nal 26(13): 1261–1269.

Sarkar MB, Echambadi R, Harrison JS. 2001. Allianceentrepreneurship and firm market performance.Strategic Management Journal , June–July SpecialIssue 22: 701–711.

Schofield N. 1985. Anarchy, altruism, and cooperation: areview. Social Choice and Welfare 2(3): 207–219.

Scott WR. 1987. Organizations: Rational, Natural, andOpen Systems . Prentice-Hall: Englewood Cliffs, NJ.

Shan W, Walker G, Kogut B. 1994. Interfirm cooperationand startup innovation in the biotechnology industry.Strategic Management Journal 15(5): 387–394.

Simon HA. 1947. Administrative Behavior . Macmillan:New York.

Simon HA. 1957. Models of Man: Social and Rational .John Wiley & Sons: New York.

Skyrms B. 2003. The Stag Hunt and Evolution of SocialStructure. Cambridge University Press: Cambridge,UK.

Smith A. 1776/1937. The Wealth of Nations . Meuthen:New York.

Smith V. 1976. Experimental economics: induced valuetheory. American Economic Review 66: 274–279.

Smith V. 2000. Bargaining and Market Behavior: Essaysin Experimental Economics . Cambridge UniversityPress: New York.

Song M, Calantone RJ, Di Benedetto CA. 2002. Compet-itive forces and strategic choice decisions: an exper-

imental investigation in the United States and Japan.Strategic Management Journal 23(10): 969–978.

Tiwana A. 2008. Do bridging ties complement strongties? An empirical examination of alliance ambidex-terity. Strategic Management Journal 29(3): 251–272.

Treece JB, Miller KL, Melcher RA. 1992. The partners.BusinessWeek . 10 February, 102–107.

Villalonga B, McGahan AM. 2005. The choice amongacquisitions, alliances, and divestitures. StrategicManagement Journal 26(13): 1183–1208.

Williamson OE. 1975. Markets and Hierarchies . FreePress: New York.

Williamson OE. 1996. The Mechanisms of Governance.Oxford University Press: New York.

Zaheer A, Bell GG. 2005. Benefiting from networkposition: firm capabilities, structural holes, andperformance. Strategic Management Journal 26(9):809–825.

Zaheer A, Venkatraman N. 1995. Relational governanceas an interorganizational strategy: an empirical testof the role of trust in economic exchange. StrategicManagement Journal 16(5): 373–392.

Zajac EJ, Bazerman MH. 1991. Blind spots in industryand competitor analysis. Academy of ManagementReview 16(1): 37–56.

Zeng M, Chen XP. 2003. Achieving cooperation inmultiparty alliances: a social dilemma approach topartnership management. Academy of ManagementReview 28(4): 587–605.



TECHNICAL APPENDIX: PARAMETERSAND RESULTING EQUILIBRIA

Key characteristics of all treatments

Players Five playersEndowment 20 units per periodValue $1,000 per unit when used

for private productionInitial level 100 units in allianceThreshold for success 150 unitsDepreciation 33 percentContinuation probability 80 percent

‘Common benefits’ or bonus earned by treatment

Player Low benefit High benefit Mixed benefit

Player 1 4000 35 000 35 000Player 2 4000 35 000 35 000Player 3 5000 40 000 40 000Player 4 5000 40 000 5000Player 5 5000 40 000 5000

Stage game equilibria

High common benefit treatment:

There are multiple equilibria in this treatment. Thepayoff dominant equilibrium outcomes involve theset of five managers allocating enough resourcesfor the strategic alliance activities to succeed. As istypical of these situations, there are multiple waysthat this goal can be achieved.

For example, consider the unique symmetricpayoff dominant equilibrium. Each firm contributes10 units to the strategic alliance. The strategicalliance collects 50 units over its starting value of100 units, and therefore meets its threshold for suc-cess. Each firm thus earns its own bonus ($35,000or $40,000), plus $10,000 in private consumptionof its remaining units. So payoffs to each firm are$45,000 or $50,000 in this equilibrium.

To see that this is indeed an equilibrium ofthe stage game, consider a unilateral deviation byone player. First note that no firm has an incen-tive to deviate by contributing more than 10 units.Excess contributions cost the firm ($1,000 per unit)and bring no additional benefit. Second, considerdownward deviations. For example, imagine thata firm decided to contribute zero instead of theirequilibrium strategy of 10. Their earnings fromprivate consumption would increase to $20,000.

However, the alliance would not meet its thresh-old, thus they would not earn their bonuses. Thisdeviating firm would earn only $20,000 instead of$45,000 (or $50,000). No firm has an incentive todeviate by contributing less than their ‘share’ of10. Thus this allocation of resources represents anequilibrium of the stage game (and, it turns out, ofthe infinitely repeated game as well).

Of course, there are many equilibria of thisgame; any outcome in which the strategic allianceis successful and each firm earns at least $20,000is a payoff dominant equilbrium. But, these equi-libria are differentially attractive to the differentplayers (that is, each firm prefers the equilibriumin which they allocate fewer resources toward thealliance activities and the others allocate more).For example, consider the allocation (20, 0, 20,10, 0). Exactly 50 are allocated to the strate-gic alliance, so it succeeds and each firm earnsits bonus. Each firm also earns $1,000 for eachunit of information it keeps, thus the net earn-ings for Firms 1–5 are ($35,000, $55,000, $40,000,$50,000, $60,000). Again, no firm wants to devi-ate by contributing more, as those contributionswould simply be wasted. No firm wants to deviateby contributing less, as this would involve the lossof their bonus and result in earning only $20,000.Thus, each firm is playing a best response to thestrategies of the other firms, and this contributionprofile is also a payoff maximizing equilibrium(although an asymmetric one). Of course, Firm5 prefers this equilibrium to the symmetric one,while Firm 1 prefers the symmetric equilibrium tothis one. There are many such asymmetric equilib-ria (in this treatment, 116,601 of them); the task ofthe firms in the strategic alliance is to coordinateon one. Earnings for each firm range from $35,000to $55,000 ($60,000), depending on which equilib-rium they are in (as can be seen in the asymmetricexample above).

There is also a unique (risk dominant) equilib-rium in which no resources are allocated towardthe strategic alliance activity. Each firm receives$20,000, yet no firm has an economic incentiveto deviate by contributing more. Imagine one firmconsidering a deviation of contributing all 20 units.They would still not reach the threshold to collecttheir bonus, and their earnings would reduce to$0. Given that other firms are playing their part ofthis strategy, the best response is to also contributezero.



Low common benefit treatment:

There is one equilibrium in this treatment wherenobody contributes anything toward the alliance.Given that others are not contributing, no individ-ual exchange partner has an economic incentive tocontribute unilaterally to the alliance.

Given low common benefits for all players, thereare no payoff dominant equilibria in which thestrategic alliance is successful. In particular, themost each firm is willing to contribute to thealliance is the amount of its bonus, (4, 4, 5, 5, 5)units = 23 units, not enough to reach the thresholdof resources necessary for success.

Mixed common benefit treatment:

As in the high common benefit treatment, there isagain a risk dominant equilibrium for the mixedbenefit where no firm contributes anything (zero).As before, no firm has an economic incentiveto deviate from this outcome by contributing, astheir contributions would simply be lost and noresources earned.

There are also a series of payoff dominant equi-libria, but fewer than in the high common benefittreatment. In particular, the symmetric contributionprofile of before (where each firm contributes 10)is no longer an equilibrium. While the high-valuedfirms would be happy with this arrangement, thelow-valued firms would each earn only $15,000each ($5,000 from their bonus, and $10,000 fromtheir retained resources), less than the $20,000 theywould earn if they simply kept their resources.

Still, there are many payoff dominant equilibriaof this game; any outcome in which the alliance issuccessful and each firm earns at least $20,000 isa payoff dominant equilbrium. For example, con-sider the contribution profile (16, 10, 20, 4, 0). Thealliance is successful, and the firms earn ($39,000,$45,000, $40,000, $21,000, $25,000). Each firmearns at least as much as it would by keeping itsresources ($20,000), eliminating the temptation todeviate by contributing less. As before, no firm istempted to deviate by contributing more, as theseresources are simply wasted.

Earnings from the risk dominant equilibrium are$20,000 for each firm. Earnings from the payoffdominant equilibrium vary between $20,000 and$60,000, depending on the firm’s bonus level andamount contributed.

Repeated game equilibria

In the preceding analysis we solved for the stagegame equilibria of the game. However, we didnot consider the possibility of individuals takingresources from the strategic alliance, only devi-ations which involved them not contributing. Inorder to constrain taking behavior, we rely on theinfinitely repeated nature of the interaction.

High common benefit treatment:

Imagine an equilibrium in which four firms arecontributing together 50 units to the strategicalliance. The fifth firm is considering playing theirequilibrium strategy of contributing zero, or tak-ing the entire 150 units available in the strategicalliance. We assume that all other players will playa trigger strategy, that is, once the pool has beenharvested, they will not contribute in the future.(Note, this is a reasonable assumption since, evenif everyone contributes all 20 units, it will take atleast two periods of full contribution to reach thethreshold of 150).

If our fifth firm takes the entire common pool,it earns $150,000 in this period. However, itthen loses the bonuses it would have receivedover the rest of the game. Since the continua-tion probability is 80 percent, it thus loses $35,000+ 0.8∗$35,000 + 0.8∗35, 000 + . . . = $175,000 >

$150,000. Thus this fifth firm would prefer to playits part of the equilibrium (contributing 0) to devi-ating and taking the strategic alliance resources inthis treatment.

Note that this example is the worst-case sce-nario for the equilibrium. We have maximized theamount the deviator can earn from taking (all 150)and assumed the lowest bonus firm ($35,000). Theresult is strengthened when we consider the otherfirms, or lower earnings from taking.

Low common benefit treatment:

In the low common benefit treatment, the onlyequilibrium is an inefficient one. This means thatin the first round, each of the five firms has aneconomic incentive to raid the strategic alliance,and never to contribute again.

Mixed common benefit treatment:

In the mixed common benefit treatment, the anal-ysis from high common benefit applies to the



high-valued firms. For the low-valued firms, theyinstead have an economic incentive to raid thecommon pool; their losses of the low bonus fromraiding are outweighed by the current benefit of

the raiding. Thus the challenge for the high-valuedfirms is to persuade the low-valued firms to leavethe strategic alliance pool untouched (at the veryleast).


the role of incentives and communication in strategic ...terpconnect.umd.edu/~rajshree/research/36...

Documents