Alliances With Competitors How to Combine and Protect Key Resources

Alliances With Competitors: How toCombine and Protect Key Resources?

Will Mitchell, Pierre Dussauge and Bernard Garrette

Our study addresses two main questions: First, what types of alliances do firms tend to createwhen combining different kinds of resources? Second, what governance mechanisms dofirms set up to coordinate and protect resources when they use them for different alliances?We examine 227 alliances between competitors in Asia, North America, and Europe. We firstidentify two types of alliances: scale alliances in which the partner firms contribute similarresources, and link alliances in which the partners contribute complementary resources. Wefind that firms contributing R&D and production resources tend to form scale alliances,while firms contributing marketing resources tend to enter into link alliances. We also findthat firms are more likely to choose stronger protection mechanisms for link alliances, whichcreate greater appropriation risks, while they tend to seek higher levels of coordination inscale alliances.

Creating and Protecting Resources:Formation and Governance of ScaleAnd Link Alliances BetweenCompetitors

T his study examines the formation ofstrategic alliances, with the goal of

moving us toward a more general under-standing of the relationship between twostreams of strategy research. During the pasttwo decades, two major themes concerningfirm-specific resources have emerged withinthe field of strategy. First, since Nelson andWinter’s (1982) and Wernerfelt’s (1984) argu-ments revived and developed earlier insightsfrom Edith Penrose (1959) and others, manystrategy researchers have focused on the roleof firm-specific resources in shaping a firm’scompetence. Second, building on William-son’s (1975, 1985) work concerning the appro-priation risks associated with idiosyncraticresources, much research attention has focusedon identifying mechanisms by which firmsprotect the value of their resources. Follow-ing Williamson (1999), we will refer to thesethemes as the governance and competenceperspectives on strategy. The competence andgovernance themes have developed in parallel,with some attempts to integrate the argumentsbut, more often, with competence and govern-ance researchers talking past each other. We

attempt to bring together these themes bystudying the types of resources and types ofgovernance mechanisms that firms use whenforming alliances with competitors in order toincrease efficiency in their existing activities(scale alliances) or to expand into new activitiesor markets (link alliances).

On two key dimensions, concerning con-ceptual systemization and units of analysis,the two strands of research have differentstrengths and weaknesses. First, the twostrands have proceeded with different em-phasis on conceptual systemization and gen-eralizability. Governance research has beenrelatively systematic. Williamson and thosewho have drawn on his insights have devel-oped a cohesive logic that underlies theidentification and protection of idiosyncraticresources (e.g., Williamson, 1991b). Compe-tence research tends to be somewhat morefragmented. At this point, there is at least apartial consensus in the strategy field thatresources exist and that they influence whatfirms do (e.g., Barney, 1986; Conner, 1991;Amit & Shoemaker, 1993). Many studies nowuse competence arguments as research moti-vations, sometimes by inferring the existenceof resources and other times by explicitly oper-ationalizing dimensions of resources (e.g.,Mitchell, 1991; Henderson & Cockburn, 1994;Karim & Mitchell, 2000). Despite the commonusage, though, there is little consensus about

ALLIANCES WITH COMPETITORS 203

# Blackwell Publishers Ltd 2002. 108 Cowley Road, Oxford OX4 1JFand 350 Main St, Malden, MA 02148, USA. Volume 11 Number 3 September 2002

how to generalize the concept of resources orhow to identify general influences of re-sources on strategic action. On the system-ization dimension, therefore, governanceresearch has proceeded somewhat fartherthan competence research.

Second, the two strands have emphasizeddifferent units of analysis. Governance researchhas emphasized individual transactions asunits of analysis, emphasizing contractualand organizational modes by which firmscan protect the value of transaction specificassets that relate to particular exchanges.While this approach has offered conceptualclarity, the approach has also tended tounder-emphasize the intertwined nature ofmany transactions, in the sense that firmsoften must make decisions that affect manyexchanges, involving many different assets.Competence research, by contrast with govern-ance research, tends to address more aggre-gated units of analysis, involving intertwinedsets of physical assets and organizationalsystems that together provide key influenceson what firms do and how they perform.Competence research often stresses the role ofthe firm in coordinating the development anduse of firm-specific resources. With respect tounits of analysis, competence research maycome closer than governance research tomany of the inter-twined strategic issues thatfirms face.

We attempt to bring together the compe-tence and governance arguments to investi-gate how firms undertake a particular type ofstrategic action, alliances among competitors.We focus on two questions. First, what typesof resources do firms tend to use for differenttypes of alliances? Second, what governancemechanisms do firms use to create andprotect resources when they use them fordifferent alliances? We have three conceptualgoals. First, we attempt to show that thenature of a firm’s resources influences itsstrategy choices; in doing so, we build oncompetence research that offers elements of ageneral framework for identifying resources.Second, we attempt to demonstrate howincentives to create and protect firm-specificresources tend to influence managerial action.Third, at a more specific level, the study helpsdescribe the multiple roles that alliances be-tween competitors play in modern economies,as firms attempt to reinforce their strategicpositions in their existing markets and expandthroughout the world.

Studying alliances between competitorsprovides a suitable context for exploringissues that arise from competence and gov-ernance arguments. The competence view ofthe firm suggests that firms’ competitive

advantages derive from their preferentialaccess to idiosyncratic resources, especiallyresources that stem from tacit knowledge-based routines (Penrose, 1959; Wernerfelt,1984; Conner, 1991; Amit & Shoemaker,1993). Although firms gain advantages frompossessing idiosyncratic resources, strategistsand organizational theorists dating to Com-mons (1934), Coase (1937), Barnard (1938),Simon (1957), Richardson (1972), and othershave long recognized that no one businesscan create all resources needed to prosper andgrow. Authors in several research traditionsargue that interfirm alliances provide a meansof pooling resources held by different firms inorder to exploit new business opportunitiesand to increase the efficiency of existingbusiness activities. Collaboration is often aneffective way of pooling resources that aresubject to a high degree of knowledge-basedmarket failure because interfirm collaborationhelps facilitate ongoing interpersonal contactbetween the allied firms (Itami & Roehl, 1987;Mitchell & Singh, 1993, 1996; Gulati, 1998).Moreover, collaboration provides a means forfirms to protect the value of their resourcesthrough financial and organizational safe-guards against opportunistic behavior (Teece,1986; Hennart, 1988; Bresser, 1988; Kogut,1988; Jorde & Teece, 1990; Williamson, 1991a;Chi, 1994). At the same time, firms usuallycannot fully protect their resources fromappropriation by partners, because the sameorganizational and individual processes thathelp pool the firms’ resources also tend toexpose the resources to the partners (Zajac& Olsen, 1993; Sobrero & Roberts, 1996).Although firms can attempt to create crediblebilateral commitments that align the resourcecoordination, creation, and protection incen-tives of the partners (Oxley, 2001), completealignment is often impossible owing to themultiplicity of organizational and personalinteractions and incentives that arise duringthe course of an alliance. Thus, collaborationprovides a variety of potential benefits thatstem from the combination of partner re-sources, as well as potential risks that firmswill attempt to minimize by choosing the bestavailable governance mechanisms.

Our empirical analysis examines 227 alli-ances between competitors in Asia, NorthAmerica, and Europe between 1952 and 1996.The alliances include firms operating in thetelecom-electronics, auto, aerospace, andother sectors. The industries in the sampletend to be oligopolistic, open to internationaltrade, R&D intensive, subject to significanteconomies of scale, and globally competitive.Alliances in such industries offer a highpotential for efficiency gains and expansion

CREATIVITY AND INNOVATION MANAGEMENT204

Volume 11 Number 3 September 2002 # Blackwell Publishers Ltd 2002

benefits, with few opportunities for limitingcompetition ( Jacquemin, Buigues & Ilzkovitz,1989; Millington & Bayliss, 1995).

Background and Predictions

Assumptions and implications for alliances

Table 1 reviews our basic assumptions.Williamson (1999) argues that a theory ofthe firm must specify five conceptual ele-ments, including behavioral assumptions,units of analysis, description of the firm,purpose of the firm, and efficiency criteria.Williamson also argues that conceptual argu-ments require empirical testing. We outlinethe assumptions and discuss their implica-tions for alliances. We then attempt to testpredictions that arise from the conceptualbase in the empirical context of alliancesbetween competitors.

Our behavioural assumptions include poten-tial self-interest plus bounded rationality withfirm-specific foresight. The implication for alli-ances of potential self-interest and boundedrationality is that alliance contracts are notself-enforcing, so that firms must organizealliances in ways that protect their resourcesas well as use other firms’ resources andcreate new resources. We assume that econ-omic actors have the capacity to look aheadand recognize opportunities and risks, butthat a firm’s experience shapes its foresight.This assumption of firm-specific foresightsuggests that firms commonly recognizepotential opportunities to gain efficiencies orexpansion via alliances with other organiz-ations, but that different firms will havedifferent expectations about the potentialoutcome of an alliance. Because of differentexpectations, the allies may create asymmetricgovernance mechanisms.

Table 1. Assumptions

Elements of theory Governance perspective Competence perspective Competence perspective:Alliance implications

1. Behavioral assumptions Bounded rationality, withforesight; potential self-interest.

Bounded rationality, with firm-specific foresight; potential self-interest.

Firms must organize alliances inways that protect their resourcesas well as use other firms’resources and create newresources. Partners may havedifferent perspectives on alliancepurposes and governance.

2. Units of analysis Transactions Routines (tacit, co-specialized,organizationally-embedded),which combine to formresources. Use of resourcesgenerates value. Productioncosts are outcomes of resources.

Routines and resources areimperfectly tradeable. Firmsoften need alliances to gainaccess to other organizations’resources.

3. Description of the firm Structure for governingtransactions. Governanceemphasizes protection.

Structure for governing routinesand resources. Governanceincludes coordination, creation,and protection.

Firms require mechanisms togovern the use of routines andresources that they use inalliances.

4. Purposes served Economizing on sum ofproduction costs & transactioncosts, where transaction costsstem from alignment betweentransaction attributes &governance structure of currentand future transactions. Takesproduction costs as exogenous tofirm.

Economizing on the sum ofproduction costs and governancecosts. Multi-faceted costdimensions create substantialambiguity concerningeconomizing choices and scopefor self-interested choices.

A firm’s governancemechanisms for inter-organizational resources mustattempt to coordinate andprotect the value of currentresources, plus create andprotect the value of newresources.

5. Efficiency criterion Relative efficiency of overall setof current and future firmtransactions, based on feasiblealternatives.

Relative efficiency of current andfuture use of overall set of firmresources, based on feasiblealternatives.

A firm seeks the best availablemechanisms for jointlyprotecting and creatingresources that fall within theactivities of an alliance.


# Blackwell Publishers Ltd 2002 Volume 11 Number 3 September 2002

Our fundamental unit of analysis is theroutine, which closely relates to the concept ofresources. Routines are identifiable patterns ofactivity embodied in human or capital assets(Nelson & Winter, 1982; Winter, 1990; Dosi,Marengo & Fagiolo, 1996) and contain muchof the knowledge of what a firm can accom-plish (Hannan & Freeman, 1989). Routinesconsist of multiple related transactions thattake place over time either within a firm orvia interaction with external parties. Routinesare often tacit, either because they are intrin-sically uncodifiable or because they requirethe interactive participation of multiple people.Routines also tend to be co-specialized withother routines and to be embedded in broaderorganizational contexts.

Several routines combine together to createparticular resources. Resources, which weview as synonymous with capabilities, arestocks of knowledge, skills, financial assets,physical assets, human capital, and othertangible and intangible factors (Wernerfelt,1984; Grant, 1991; Amit & Schoemaker, 1993).Resources tend to be only semi-decomposableinto their underlying routines, so that re-sources also provide relevant units of analy-sis, in addition to routines. In our discussionof alliances, we will emphasize the joint use offirms’ resources as the purpose of alliances.We will refer to a resource typology that in-cludes R&D, production, and marketing re-sources (Capron, Dussauge & Mitchell, 1998).Firms create new resources by creating newroutines and recombining existing routines innovel ways. Resources and their underlyingroutines are often firm-specific and imper-fectly tradable, owing to their tacitness, co-specialization and organizational nature.

The need for alliances arises from theimperfect tradability of routine and resources.Firms often need to ally with other organiz-ations in order to extract value from under-utilized resources they possess, either throughmore efficient use of existing resources or bycreating new resources. Allying firms maypool similar resources in order to gain greaterefficiency, so long as increased economies ofscale more than outweigh the governance costof alliances. In addition, allying firms maywish to combine the routines that underliedifferent types of resources in order to createvaluable new resources, again accounting forgovernance costs.

Our description of the firm and our view ofthe purpose of the firm involve assumptionsconcerning the role of the firm in governingresources. We describe a firm as a governancestructure, where governance includes coordi-nating the use of existing resources, creatingnew resources, and protecting the value of

resources. This view closely follows Coriatand Dosi (1998), who argue that a firm is aparticular set of routines that result from theco-evolution between corporate patterns ofknowledge distribution and mechanisms ofgovernance. Governance mechanisms includeformal and informal incentive and controlsystems, legal regimes, organizational struc-tures, and corporate cultures (Argyres, 1996;Liebeskind, 1996). Governance mechanismsare often shaped by path dependency andlocal search, which arise from the tacitness,co-specialization, and organizational em-beddedness of routines. In turn, the purposeof the firm is to economize on the combina-tion of production and governance costs. Ourapproach implies that production costs arepartly an endogenous outcome of firm-specificresources and governance mechanisms. Thatis, production costs vary with the nature of afirm’s resources and the effectiveness withwhich a firm governs the use and creation ofresources.

The alliance implication of our assumptionsconcerning the description and purpose of thefirm is that a firm must create governancemechanisms for its inter-organizational activi-ties. A firm’s inter-organizational governancemechanisms must attempt to increase and pro-tect the value of the firm’s current resources,as well as create and protect the value of newresources (Child & Faulkner, 1998). Thus, theinter-organizational governance mechanismsneed to address resource coordination to en-sure efficient use of current resources andresource creation to support expansion, as wellas protection of the value of the resources thatfall within the coordination and creationactivities of the alliance.

Our efficiency criterion is of the bestavailable value of current and future use ofroutines, by which we mean that a firm seeksthe best available mechanisms to jointlyprotect and create resources. In this paper,our emphasis will be on factors that differ-entiate the types of resources that firms usefor alliances that create either efficiency orexpansion opportunities, along with the pro-tection mechanisms that the firms use toprotect the value of the resources.

Overall, our conceptual approach combinesthe protection emphasis of governance re-search with the coordination emphasis of thecompetence research. The key differencebetween our approach and transaction costeconomics, which is the core theory of thegovernance approach, is that we focus onroutines rather than individual transactions.In turn, this leads us to emphasize coordina-tion and creation roles for governance inaddition to a protection role. This dual



emphasis on protection and creation creditsthe firm with a critical coordination role inboth enhancing the value of existing re-sources and creating new resources. Ourfocus on routines as the fundamental unitbrings our approach close to that of evolu-tionary economics (Nelson & Winter, 1982),with the primary difference being that wefocus on relationships among firms in the useof resources rather than on independentsearch activities by single firms.

Our summary argument concerning alli-ances is that alliances provide opportunitiesfor firms to pool imperfectly tradable re-sources in order to gain greater efficiency inthe use of existing resources as well asopportunities to create new resources. Thefirms must create governance mechanismsthat both combine the resources effectivelyand protect the value of the resources. Thenature of the resources creates incentives toseek either efficiency or expansion benefitsfrom the alliance. Some types of resourcesprovide greater potential for efficiency gains,when a firm accounts for the governance costsof creating and protecting the resources aswell as potential scale economies. Other typesof resources provide greater opportunities forexpansion gains, again accounting for govern-ance costs. We will focus on two types ofalliances (link and scale), two sets of resources(geographic and functional), and two sets ofgovernance mechanisms (equity holdings andjoint ventures). We first develop predictionsconcerning the types of resources that firmswill tend to use for link and scale alliances.We then develop predictions concerninggovernance mechanisms that the firms willuse to protect and create resources in the twotypes of alliances.

Formation of link and scale alliances

Industry analysts and academic researchersreport a growing incidence and importance ofalliance activity among competitors for manypurposes, including technology and productdevelopment, joint manufacturing, and marketentry (Doz, 1996; Harbison and Pekar, 1998;Park & Russo, 1996; Sakakibara, 1997). Increas-ingly, researchers distinguish between twobasic types of alliances between competitors,which we refer to as link alliances and scalealliances (Porter & Fuller, 1986; Hennart,1988; Dussauge, Garrette & Mitchell, 2000).The distinction between link and scale alliancesarises from the symmetry or asymmetry of thepartners’ resource contributions to an alliance,which reflect different strategic purposes(Dussauge & Garrette, 1997). Scale alliancesare partnerships to which partners contribute

similar resources, while link alliances arepartnerships to which partners contributesubstantially different resources. Scale alli-ances allow firms to gain greater efficiency intheir existing business activities, while linkalliances allow firms to combine complemen-tary resources in order to expand theirbusiness activities. This distinction is similarto Sakakibara (1997), who defines alliances interms of cost-sharing and skill-sharing moti-vation. We define the concept of scale andlink alliances on the basis of the similarity ordifference in the resources that the partnerfirms contribute to the alliance. Empirically,we examine three types of resources, includ-ing R&D resources, production resources, andmarketing resources.

Although a substantial literature addressesalliance formation (e.g., Harrigan, 1985; Teece,1986; Contractor & Lorange, 1988; Hennart,1988; Kogut, 1988; Oliver, 1990; Williamson,1991a), this literature has not fully exploredthe strategic factors that differentiate incen-tives to ally with competitors for expansion orefficiency. In particular, it is not clear whichresources firms tend to use for efficiency andexpansion purposes when they ally withcompetitors or, in turn, how the firms governthe use of resources in link and scale alliances.

Although both scale and link alliances maycreate advantages for the allying firms, thenature of the advantages differ (Dussauge &Garrette, 1995). Because they are based onsimilar contributions from the partner firms,scale alliances produce significant economiesof scale for collaborative activities, therebyallowing the firms to reduce excess capacity(Hennart, 1988). Such scale alliances caninclude joint R&D efforts, the joint productionof a component or end product, or jointmarketing of the allies’ goods. Such scalealliances provide a way of avoiding, or atleast postponing, mergers in industries un-dergoing strong concentration processes.

Link alliances, in contrast with scale alli-ances, combine complementary resourcesfrom the partners (Porter & Fuller, 1988).Link alliances include partnerships in whichone partner provides market access to pro-ducts that another firm developed, such thatthe two allies create a form of customer-supplier relationship. Therefore, link alliancescreate opportunities for the partnering firmsto undertake immediate expansion withinthe current markets of one or other of thepartners. In addition, link alliances createopportunities for firms to learn from theirpartners and to use this learning as the basisfor future expansion beyond the scope ofthe alliance (Hamel, 1991; Khanna, Gulati &Nohria, 1998; Dussauge, Garrette & Mitchell,



2000). Link alliances sometimes involve jointmanufacturing, as long as the other compo-nents of the value-chain remain distributedbetween the partners.

A central proposition of this paper is thatpartners will tend to use different types ofresources in alliances, depending on whetherthe firms seek efficiency or expansion benefitsfrom the alliance. In order to predict therelationship between particular types of re-sources and alliance types, we distinguishseveral dimensions of geographic and func-tional resources. Tables 2a and 2b list theresources, along with several control vari-ables. We first discuss geographic resources,next turn to functional resources, and thenoutline several other strategic factors thatmight underlie the formation of scale andlink alliances.

By geographic resources, we mean the re-sources that arise as a result of a firm’s ac-tivities in different geographic home markets.A literature that focuses on distinctionsamong national innovation infrastructures(Nelson, 1991) suggests that firms based in

different geographic contexts tend to developdiffering resources. We use two aspects ofgeographic resources, including the parents’geographic origins and the alliance’s geo-graphic market.

We expect expansion incentives to be par-ticularly strong for alliances involving firmswith disparate national origins. Firms based indifferent countries can often combine differentcomplementary resources in link alliances thatdraw on their different national environmentsto create expansion opportunities. In order toexplore the impact of increasingly divergentgeographic origins, we will distinguish em-pirically between parents from the samecountry, parents from different countrieswithin a single continent, and parents fromdifferent continents.

Hypothesis 1a. Alliances between compe-titors with different geographic origins aremore likely to be link alliances than scalealliances.

By contrast, we expect that alliances cover-ing broader geographic markets will offer

Table 2a. Alliance Formation Factors

H Alliance formation factors Type Resource combination opportunities

H1a Different geographic origins ofparents

Link Combine resources that draw on different national environments to create newresources needed to expand activities.

H1b Broader geographic marketcoverage

Scale Combine similar resources to achieve scale needed for greater sales in more extensivemarkets.

H2a R&D resources Scale Link alliances involving R&D resources create expansion opportunities, but also createappropriation risks. Firms are most likely to combine R&D with a competitor forgreater scale, which has a longer term commercialization horizon, than to contributetheir R&D resources to a competitor’s production or marketing resources, which wouldhave immediate competitive risks.

H2b Production resources Scale Combine production resources to achieve scale efficiency.

H2c Marketing resources Link Use one firm’s marketing resources to sell goods that use another firm’s developmentor production resources.

Multiple functional resources ? The incentives of one type of resource may tend to dominate the other, or there may besubstantial heterogeneity.

H3a Competitive asymmetry Link In link alliances, relative size is not be an issue so long as a partner possesses neededresource. By contrast, it often is not worth while for a firm to engage in a scale alliance asmaller partner, because the cost of collaborating will exceed the expected scalebenefits.

H3b Number of partners Scale Scale economies opportunities increase with multiple partners, while governance costsfor disparate resources in link alliances increase with multiple partners.

Year ? Expansion opportunities might have arisen more frequently in recent years, but firmshave long had incentives to ally for expansion. Recent competitive pressure may haveincreased the incentives to undertake scale alliances for greater efficiency.

Prior alliances among partners ? Allies with prior experience might be more willing to undertake expansionopportunities together, but experience might also lead to independent competition forexpansion opportunities rather than continued cooperation.



greater opportunities for economies of scaleand thus will tend to be scale alliances.Broader market coverage offers a greaterpotential for efficiency gains which, in turn,requires greater business scale; this createsincentives for partners to pool similar re-sources within scale alliances in order toachieve efficient size. For example, globalmarkets offer particularly strong oppor-tunities for firms to increase efficiency bypooling similar resources and therefore favorthe formation of scale alliances. In contrast,fragmented markets require access to locally-specific resources in order for firms to expandinternationally, which, in turn, tends to leadto the formation of link alliances. We willdistinguish empirically between alliances thatsell goods within a single continent (fewalliances in our sample limit their sales to asingle country) and alliances that sell goodsacross multiple continents.

Hypothesis 1b. Alliances between compe-titors that provide sales coverage for broadergeographic markets are more likely to bescale alliances than link alliances.

We now turn to functional resources. Byfunctional resources, we mean resources thatfirms use for R&D, production, and market-ing. Several typologies in the literature thathas become known as the resource-basedview of the firm suggest classes of functionalresources. Amit and Schoemaker (1993) dis-tinguish between R&D resources (technologicalresources, R&D resources, product develop-ment speed), manufacturing resources, andmarketing resources (brand management, dis-tribution channels, buyer-seller relationships,user base, customer service, business reputa-tion). Similarly, Teece and Pisano (1994) andTeece, Pisano, and Shuen (1997) distinguishbetween technological resources, productionresources, and customer-related resources.Chatterjee and Wernerfelt (1991) distinguishbetween R&D resources, production resources,and marketing resources. Three categories offunctional resources stand out in these dis-cussions, together encompassing the commer-cialization sequence of product development,production, and marketing. Capron, Dussauge,and Mitchell (1998) show that R&D, produc-tion, and marketing resources are particularlycommon targets for inter-business usage fol-lowing business acquisitions, while Dussauge,Garrette, and Mitchell (2000) show that thethree types of resources are important elementsof alliance activity. In addition to functionalR&D, production, and marketing resources,one might also investigate the complementaryfunctions of general management and financialsupport (Barney, 1986; Teece, 1986; Chatterjee

& Wernerfelt, 1991; Teece, Pisano & Shuen,1997; Capron, Dussauge & Mitchell, 1998). Forthis paper, we will focus on the three-partfunctional sequence of R&D, production, andmarketing resources.

When competing firms ally, we expect scaleefficiency incentives to be particularly strongfor alliances that involve R&D or productionresources. R&D resources may offer substan-tial opportunities for link alliances in order toexpand, but such link alliances involving R&Dresources would create appropriation riskswhen combined with a competitor’s produc-tion or marketing resources (Hamel, 1991;Hennart, Roehl & Zietlow, 1999). Therefore,we expect firms to be reluctant to combinetheir R&D resources in link alliances withcompetitors. Instead, when forming linkalliances, firms will often exclude R&D re-sources from the activities of the alliance andwill contribute only outputs of their proprie-tary R&D resources, such as existing designsand previously developed products. In con-trast, firms are most likely to pool R&D re-sources with a competitor in order to achievegreater scale, which has a longer-term com-mercialization horizon and lesser immediaterisks. In addition, because all partners con-tribute R&D resources in scale alliances, themutual hostage situation thus created limitsopportunism (Oxley, 2001).

Hypothesis 2a. Competing firm alliancesthat involve R&D resources are more likelyto be scale alliances than link alliances.

Production resources often offer opportu-nities for two or more firms to pool resourcesin order to achieve efficient size. Because ofthe extent of scale economies and experienceeffects in production, firms possessing under-utilized production resources will often formalliances with partners that contribute similarresources in order to achieve efficient size orreduce excess capacity. Therefore, we expectthe efficiency potential created by alliancesinvolving production activities to lead to theformation of scale alliances rather than linkalliances.

Hypothesis 2b. Competing firm alliancesthat involve production resources are morelikely to be scale alliances than linkalliances.

We expect expansion incentives to be par-ticularly strong for competing-firm alliancesthat involve marketing resources. Alliancesinvolving marketing activities provide oppor-tunities to expand by using one firm’smarketing resources in order to sell goodsbased on another firm’s development or



production resources. Thus, marketing linkalliances will often use one firm’s marketingsystems to sell the other partner’s products,rather than combine one partner’s productionor R&D resources with the other competitor’smarketing resources in order to produce andmarket a new product specifically developedwithin the scope of the alliance.

Hypothesis 2c. Competing firm alliancesthat involve marketing resources are morelikely to be link alliances than scalealliances.

To test hypotheses 2a to 2c, we will comparealliances that involve only a single functionalresource to alliances that involve combinationsof different types of resources. We expect theprediction to hold for alliances that involveonly one type of resource. That is, we expectalliances involving only R&D or productionresources to tend to be scale alliances, whilealliances involving only marketing resourcesto tend to be link alliances.1 We do not haveuniform expectations concerning alliancesthat combine marketing resources with R&Dand/or production resources. In such cases,it is possible that the market expansion in-centives for link alliances may dominate or,conversely, that the R&D and productionprotection and efficiency incentives for scalealliances will dominate.

We will consider four other firm-levelfactors that also might underlie competitors’alliance formation. These four variables includeasymmetric competitive strength, number ofpartners, year of formation, and prior allianceexperience.

Competitive asymmetry is likely to favorlink alliances, in which the partners attemptto create expansion opportunities. In linkalliances, the firms seek partners with com-plementary skills and capabilities. Provided apartner possesses the needed skills, its rela-tive size should not arise as an issue in linkalliances. By contrast, if partners seek econ-omies of scale, each partner will only agree toengage in the collaboration if the increasedvolume that the other partner contributes islarge enough to produce savings that willoutweigh the costs of governing the collab-oration. In other words, it often is not worthwhile for a firm to engage in a scale alliancewith a smaller partner, because the govern-ance cost of collaborating will exceed theexpected scale benefits. This argument isconsistent with Hennart’s (1988) transactioncost theory of joint ventures. The argumentalso corresponds to the view of Porter andFuller (1986), who argue that partners thathave dissimilar strengths and weaknessestend to form X coalitions (link alliances).

Hypothesis 3a: Alliances involving com-petitors with asymmetric competitive posi-tions are more likely to be link alliancesthan scale alliances.

Alliances with more than two partners aremore likely to favor efficiency gains than tocreate expansion opportunities. Indeed, scaleeconomies tend to increase with the numberof partner firms involved in an alliance, thuscreating incentives to form scale alliances. Thevery similarity in the partners’ contributionstends to keep governance costs relatively low.In contrast, combining complementary re-sources of multiple partner firms becomesincreasingly complex and increases govern-ance costs, while, at the same time, theincremental contribution of each additionalpartner to the expansion potential of thealliance tends to decrease. Moreover, appro-priation risks are likely to increase with thenumber of partners involved in the alliance.Therefore, we expect most multiple partneralliances to be scale alliances.

Hypothesis 3b: Alliances involving morethan two partner firms are more likely tobe scale alliances than link alliances.

The influences of the other two firm-levelfactors that might tend to favor either linkor scale alliances, formation year and prioralliance experience, are ambiguous. Alliancesformed in more recent years might tend toinvolve expansion, if popular perceptionsabout changes in alliance objectives arecorrect, but, alternatively, alliances have longbeen important to firm expansion so thatthere is no clear prediction. Allies with priorexperience in alliances together might bemore willing to undertake expansion oppor-tunities together. Alternatively, experiencemight also lead to independent competitionfor expansion opportunities, so that priorexperience might be predict scale alliancesrather than link relationships.

It would be desirable also to control fordifferential industry growth and concentra-tion as influences on alliance formation, butthe multi-period and multi-national scope ofthis study makes such measurement im-possible. The focus of our argument is onfirm-level issues, however, rather than onindustry-level trends. Moreover, the industryvariables help address differences acrosseconomic sectors, while the alliance foundingyear and geographic variables help addressinter-period and inter-region differences.

In summary, the hypotheses address thetendency of competing firms to use geo-graphic resources and functional resourcesfor scale and link alliances. We expect that



scale alliances among competitors will pri-marily involve R&D resources, productionresources, broad geographic market coverage,competitors with similar competitive posi-tions and multiple partners. We expect thatlink alliances will emphasize marketing re-sources, competitors with different geographicorigins, and competitors with asymmetriccompetitive positions.

Governance mechanisms forprotecting, coordinating, andcreating resources

Scale and link alliances tend to requiredifferent governance mechanisms. We expectthe mechanisms to reflect the need to co-ordinate the use of existing resources in scalealliances and create new resources in linkalliances, while also protecting the value ofresources in either case. As we noted earlier,we will focus on resources as the analytic unit,subsuming the concept of routines withinthat of resources. We will examine two setsof governance mechanisms, including parentequity holdings and joint venture structures.The two governance choices have different im-plications for resource coordination, creation,and protection. Table 2b summarizes thevariables and predictions.

Parent equity holdings arise as a means ofprotecting firms from opportunistic behaviorby partners, because the firms either unilater-ally or jointly can exercise a degree of owner-ship control over their partners (Geringer &Hebert, 1989; Gomes-Casseres, 1990; Harrigan,1986; Hennart, 1991; Killing, 1983; Pisano,1989). Firms have greatest need of such equity

holdings in the case of link alliances, whenthey face greater uncertainty about the re-sources that alliance activities might createand about unwanted resource transfers. Inscale alliances, by contrast, the ownershipcosts that equity holdings entail will oftenout-weigh protection value.

Hypothesis 4. Parent equity holdings willbe more common in link alliances than inscale alliances.

By contrast with equity holdings, the for-mation of a joint venture, rather than simplyundertaking a contractual relationship, mayprovide coordination and creation roles, aswell as a protection mechanism. A jointventure provides a protection mechanism bygiving the partners’ direct influence overalliance activities via their financial stake inthe joint venture (Williamson, 1983). A jointventure also provides coordination and cre-ation roles by forming a focal entity to whichthe allies can direct their human resourcesand other assets for the alliance’s efforts tocombine their existing resources and possiblycreate new resources (Killing, 1983; Geringer& Hebert, 1989; Oxley, 2001). The coordinationand creation roles can include both manage-ment of the use of the allies’ resources and,in some cases, active functional use of thepartners’ resources in a joint venture facility.

It is not clear whether joint ventures will bemore common for different types of alliances.Scale alliances require that the firms pooltheir similar resources and actively undertakecoordinated use of the resources in order toachieve greater scale efficiencies. A jointventure organization provides a mechanismfor coordinating this joint use. Some link

Table 2b. Alliance Governance Mechanisms

H Alliance governance mechanisms Type Resource governance roles: Protection, coordination, creation

H4 Parent equity holdings Link Protection: Protect the value of resources that alliance activities might create (lessincentive to incur ownership costs in scale alliances).

H5a Unbalanced joint ventures Link Protection & creation: Create and protect new resources that arise fromcomplementary resources (may also coordinate use of similar resources but unbalancedstructure is less necessary in such cases).

H5b Balanced joint ventures Scale Coordination: Coordination while using similar resources in scale allliances, whichincur fewer appropriation risks than link alliances.

H6 Sequential joint ventures Scale Coordination: Coordinate use of similar resources in scale alliances. Uncommon forlink alliances, because sequential joint ventures raise appropriation risks withoutproviding functional organization to create new resources.

Integrated joint ventures Both Coordination & creation: Coordination while using similar resources in scale alliances;creation while using complementary resources in link alliances.



alliances, by contrast, may require less jointuse of the allies’ resources than in the case ofscale alliances. Instead, the allies may often beable to use contractual relationships to governthe complementary use of their resources thatthe firms draw upon at different stages in theproduction process. The resource creationactivities of such link alliances may then tendto take place within the parent firms, ratherthan within a joint venture entity. None-theless, many link alliances will requireinteraction among the partners in order tocreate resources, so that joint ventures mightwell be equally common for link alliances asfor scale alliances.

Once one makes the first distinction betweenjoint ventures versus contractual alliances,different forms of joint ventures provideinsights concerning firms attempt to balancethe protection, coordination, and creationincentives. Two variants of joint ventures areparticularly interesting in this context.

The first variant concerns balanced andunbalanced joint ventures. Unbalanced jointventures, which are joint ventures in whichone partner has greater share of equity thanother partners, provide the leading share-holder with greater resource protection thanbalanced joint ventures because the greaterownership share allows some degree ofunilateral control. We expect unbalanced jointventures to be common in the case of linkalliances, in which concerns about losing thevalue of key resources to a partner are par-ticularly strong. By contrast, firms will tend touse balanced joint ventures for scale alliances,to which firms contribute similar resourcesand need have less concern about appropri-ation by a partner.

Hypothesis 5a. Unbalanced joint ventureswill be more common in link alliances thanin scale alliances.

Hypothesis 5b. Balanced joint ventureswill be more common in scale alliancesthan in link alliances.

The second variant compares integrativeand sequential joint ventures. Integrative jointventures are joint ventures that create a stand-alone alliance organization, while sequentialjoint ventures are ventures in which eachpartner carries out part of the alliance activitieswithin its own facility and then passes that parton to its partner for further activity (Park andRusso, 1996). In an integrative joint venture,the joint venture organization has the poten-tial to provide hands-on coordination andcreation roles in the joint use of the firms’resources, while the joint venture organizationfor a sequential joint venture coordinates only

the establishment of legal and functionalcooperation frameworks. Owing to the dis-tinct organizational presence of integrativejoint ventures, which blend personnel andother resources of the parents, such venturesprovide greater opportunities than sequentialjoint ventures for the partners to combinetheir resources.

We expect sequential joint ventures to beparticularly uncommon in link alliances.Sequential joint ventures provide only a co-ordinating organization, without providing afunctional organization for resource creation.Link alliances are likely to rely on either con-tractual relationships or integrative joint ven-tures, rather than sequential joint ventures.Integrative joint ventures will be appropriatefor link alliances in which the partners requireactive ongoing coordination of the combina-tion of their complementary resources. Con-tractual alliances will be appropriate for linkalliances in which the partners are more in-terested in learning from each others’ re-sources than in undertaking joint combination,and can rely on individual-level coordinationto achieve the desired learning. Moreover,sequential alliances involve the risk of creatinga coordinating organization that may allowcompetitors to gain substantial access to re-sources that the competitors lack, and therebyrisk appropriation of the value of the resources,without the benefit of providing a functionalorganization that can create new resources inwhich all partners will have an ownershipshare. Therefore, few link alliances will besequential joint ventures.

Hypothesis 6. Sequential joint ventures willbe more common in scale alliances than inlink alliances.

By contrast, integrative joint ventures mayarise for both scale alliances and link alliances.Scale alliances may use integrative jointventures in order to achieve functional co-ordination of their similar resources andthereby obtain scale economies. As we notedabove, link alliances may use integrative jointventures in order to combine dissimilar andcomplementary resources in order to createnew resources.

In summary, this section argues that differ-ent governance mechanisms address differentelements of coordination, creation, and protec-tion in the use of the resources that competingfirms contribute to alliances. We argue thatfirms can at least partly achieve the jointneeds of resource coordination, creation, andprotection, counter to arguments that the goalstend to be opposed (e.g., Ghoshal & Moran,1996; Sobrero & Roberts, 1996). Parent equityholdings emphasize protection mechanisms.



Joint ventures provide coordination andcreation mechanism, but may also createappropriation risks, which will be particu-larly prevalent for link alliances. Balancedjoint ventures and sequential joint ventures,which raise appropriation risks, will tend tobe suited to coordination and pooling ofsimilar resources in scale alliances. Unbalancedjoint ventures, which provide protectionmechanisms for one of the parents, will tendto be suited to the creation of new resourcesfrom recombination of complementary re-sources in link alliances. Integrative jointventures provide opportunities to either co-ordinate the use of similar resources or createnew resources from complementary re-sources. Empirically, we will also examinehow combinations of the governance mechan-isms align with scale and link alliances. It ispossible that combinations of mechanismsthat emphasize protection will tend to alignwith link alliances (e.g., parent equity hold-ings and unbalanced joint ventures), whilecombinations that emphasize coordinationbenefits for similar resources will tend to alignwith scale alliances (e.g., balance sequentialjoint ventures).

Overall, this paper focuses on how thefunctional and geographic resource scope ofthe alliances and parent firms distinguishcompetitors’ strategic objectives to coordinatethe use of existing resources and to createnew resources. The predictions address theopportunities and risks that firms face whenthey form alliances with competitors. Thestudy helps describe the multiple roles thatalliances between competitors play in moderneconomies, as firms attempt to reinforce theirstrategic positions in their existing marketsand expand throughout the world.

Data, Variables, And StatisticalMethods

We tested our hypotheses on a set of 227alliances among competing firms in a range ofmanufacturing industries. We define strategicalliances as arrangements between two ormore independent companies that choose tocarry out a project or operate in a specificbusiness area by coordinating the necessaryskills and resources jointly rather than eitheroperating on their own or merging their oper-ations. The alliances in our sample includeequity joint ventures as well as contractualpartnerships that did not entail the formationof a separate legal entity.

The alliances in this study involved partner-ships between competitors, that is, firms thatoperated in the same industries. We based

industry categorization on descriptions of thealliances’ business areas. The industry defini-tions approximate a three-digit to four-digitlevel in the U.S. Standard Industrial Categor-ization classification, although we did not useformal SIC-type classifications because of theinternational nature of the data and ofinconsistencies and unavailability of differentnational classification systems. The mostfrequent industry in our analysis is telecom-electronics (35 percent), followed by auto-mobiles (29 percent), aerospace (19 percent)and a mixture of other industries, such aspharmaceuticals, medical devices, chemicals,and foods (17 percent). Business areas in thetelecom-electronics cases included publicswitching equipment, PBX, radiotelephoneequipment, mainframe computers, personalcomputers, consumer electronics, and semi-conductors. Business areas in the auto in-dustry cases included cars, trucks, engines,and transmissions. Business areas in theaerospace cases included commercial air-planes, military airplanes, airplane engines,helicopters, helicopter engines, missiles, andspacecraft. For each alliance, we checkedsecondary sources, industry analysts, andexecutives to determine that each partnerhad prior activities in the alliance businessarea.

The alliances in the sample involve partnerfirms from North America (U.S. and Canada),Western Europe (Sweden, Italy, Britain,Germany, France, the Netherlands, Spain,Switzerland, and Finland), or Asia ( Japanand Korea) and entail operations in one ofthese three continental zones. We includedonly agreements that operated within at leastone of the partners’ home markets. Thus, weexcluded agreements such as the GeneralMotors-Toyota joint venture in Australia andthe Autolatina alliance that Ford and Volks-wagen formed in Brazil and Argentina. Wealso did not collect information on agreementsconcerning the supply of components and sub-assemblies from one manufacturer to anotherbecause such exchanges are closer to markettransactions than to strategic alliances. Inaddition, we excluded government-sponsoredresearch consortia, such as those sponsoredby the European Commission and by MITI inJapan. The resulting data focus on strategicalliances that involve the partners’ corebusinesses and markets.

Each data point in our sample correspondsto an agreement between two or morepartners, covering a specific business area.For example, in aerospace we consideredagreements involving commercial airplanes,military airplanes, airplane engines, heli-copters, helicopter engines, missiles and



spacecraft. Each alliance operates in one of thethree above-mentioned geographic zones. Toidentify alliances in which reorganizationsoccurred, we categorized collaborative activ-ities into four main functions, including R&D,manufacturing, assembly, and marketing.Each alliance corresponds to a specific alloca-tion of R&D, manufacturing, assembly, andmarketing activities among the partners.Thus, an alliance between an American anda European telecommunications equipmentmanufacturer by which they each agree tomarket one of the other’s products in theirrespective home markets would include twocases: one in which the European partnermarkets the American system in Europe, anda second in which the American partnermarkets the European system in NorthAmerica.

We gathered the data for the study fromsecondary sources such as industry reports,publications of manufacturers’ associations,and journals specializing in specific indus-tries. Examples of the publications includeAutomotive News or Aviation Week and SpaceTechnology. Reports of all the alliances in thesample occurred in published sources. Tradeassociations and private research institutespublish annual updates on alliances andcollaborative ventures for the industries thataccount for a significant portion of our sample(automobile, aerospace, telecom-electronics),which makes it possible to trace allianceformation, reorganization, and terminationon a yearly basis. In addition, when informa-tion necessary for our study was not availablefrom these sources, we complemented thedata by interviewing industry analysts andcompany executives. Park and Russo (1996)report using a similar supplementary inter-view approach to complement archival sourcedata. To avoid perception biases, we relied onvariables describing a factual event or situa-tion, rather than using variables that reflectedmanagers’ opinions.

Alliance formation variables and methods

The dependent variable for the empiricalanalysis of alliance formation is the alliancetype. We set a dummy variable equal to 1 forlink alliances and 0 for scale alliances. To dothis, we classified possible contributions to analliance into three categories that distinguishedbetween technical, production, and marketingactivities: (i) research, technology develop-ment, and product design, (ii) manufacturingfacilities and resources, and (iii) marketingand sales networks and resources. We thenexamined the respective contributions of eachpartner. When, based on the three categories,

all the contributions of the partners over-lapped, we considered a partnership to be ascale alliance. When, in at least one of thethree categories, all contributions came fromone partner, we considered a partnership tobe of a link alliance. Two authors of this studyeach coded the variable independently. Wethen asked an industry expert to indepen-dently classify alliances in the automobile,aerospace, data processing, electronics, andtelecom industries. After undertaking thisprocess, we dropped eleven ambiguous casesbecause of conflicting coding. Table 3a reportssummary statistics for the alliance formationvariables.

We recognize that the routines that compriseresources within a common category will differacross firms. For instance, the routines thatmake up the marketing resources of one firmwill differ from the routines that make up themarketing resources of another firm. Thedifferences among the routines thus providean aspect of link alliance to all alliances amongfirms. However, an assumption concerningthe alliance categorization is that the routinesthat comprise the resources within a commoncategory are more alike than the routines thatcomprise resources in different categories. Thisassumption allows us to classify alliances thatinvolve resources within common categoriesas scale alliances.

We defined several variables to address thepredictions concerning alliance formation.We defined three dummy variables to denotealliance activities, based on whether thealliance involved R&D, production, and/ormarketing activities. We considered a functionto be within the scope of the alliance if thefirms performed the tasks pertaining to thisfunction as a direct element of the alliance’sactivities. For example, if firms formed analliance to market an existing product in anew geographic zone, we considered thealliance to involve marketing but not R&D.If the alliance led to the development of a newproduct, then we considered its scope toencompass R&D. We defined several vari-ables for geographic coverage of the allianceand the allies. We determined whether themarket of the alliance covered Europe, Asia,North America, or a combination of thesezones. Three variables denoted alliances, suchas the Rover Honda and NUMMI alliances,that sold goods only within one continent. Acomparison variable denoted alliances, suchas Airbus, that also sold their output outsidethe continent in which the firms based thealliance. We defined four 0–1 dummy vari-ables to denote partnerships involving parentfirms from the same country, same continent,or different continents.



Table 3a. Summary Statistics for Link Formation Variables (N = 227)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

1 Link alliances 1 71.00 70.61 70.16 0.41 70.27 0.05 0.21 0.06 70.51 0.04 0.30 0.32 70.13 70.44 0.42 0.14 0.01 70.012 Scale alliances 71.00 1 0.61 0.16 70.41 0.27 70.05 70.21 70.06 0.51 70.04 70.30 70.32 0.13 0.44 70.42 70.14 70.01 0.013 R&D activities 70.61 0.61 1 0.24 70.12 0.18 70.05 70.18 70.19 0.40 0.04 70.21 70.39 0.02 0.33 70.41 0.05 0.16 70.164 Production activities 70.16 0.16 0.24 1 0.30 0.02 70.31 0.11 70.27 0.19 0.10 70.18 70.21 70.12 0.18 70.18 0.08 70.07 0.075 Marketing activities 0.41 70.41 70.12 0.30 1 70.17 70.08 0.21 0.05 70.13 70.08 0.12 0.15 70.29 70.01 0.14 0.09 70.23 0.236 Multi-firm alliances 70.27 0.27 0.18 0.02 70.17 1 70.15 70.09 0.04 0.25 70.10 70.14 70.06 0.00 0.17 70.13 70.06 70.04 0.047 Year of formation 0.05 70.05 70.05 70.31 70.08 70.15 1 70.05 0.10 70.28 0.18 0.03 0.11 0.07 70.13 0.01 0.07 0.11 70.118 Competitive asymmetry 0.21 70.21 70.18 0.11 0.21 70.09 70.05 1 0.08 0.00 70.02 0.02 0.02 70.04 70.16 0.07 0.14 0.02 70.029 Prior alliances among partners 0.06 70.06 70.19 70.27 0.05 0.04 0.10 0.08 1 70.10 70.03 0.04 0.13 70.07 0.07 0.08 70.11 70.10 0.10

10 Zone, global 70.51 0.51 0.40 0.19 70.13 0.25 70.28 0.00 70.10 1 70.55 70.28 70.28 0.08 0.22 70.25 70.03 70.23 0.2311 Single zone, Europe 0.04 70.04 0.04 0.10 70.08 70.10 0.18 70.02 70.03 70.55 1 70.33 70.33 70.11 0.13 70.15 0.11 0.38 70.3812 Single zone, North America 0.30 70.30 70.21 70.18 0.12 70.14 0.03 0.02 0.04 70.28 70.33 1 70.17 0.10 70.25 0.07 0.12 70.07 0.0713 Single zone, Asia 0.32 70.32 70.39 70.21 0.15 70.06 0.11 0.02 0.13 70.28 70.33 70.17 1 70.05 70.22 0.48 70.23 70.15 0.1514 Parent same continent 70.13 0.13 0.02 70.12 70.29 0.00 0.07 70.04 70.07 0.08 70.11 0.10 70.05 1 70.28 70.25 70.21 70.06 0.0615 Parent inter-continent 70.44 0.44 0.33 0.18 70.01 0.17 70.13 70.16 0.07 0.22 0.13 70.25 70.22 70.28 1 70.46 70.40 70.03 0.0316 Parent inter-continent, Asia 0.42 70.42 70.41 70.18 0.14 70.13 0.01 0.07 0.08 70.25 70.15 0.07 0.48 70.25 70.46 1 70.36 70.05 0.0517 Parent inter-continent, Eur-NAm 0.14 70.14 0.05 0.08 0.09 70.06 0.07 0.14 70.11 70.03 0.11 0.12 70.23 70.21 70.40 70.36 1 0.13 70.1318 Industry telecom-electronics 0.01 70.01 0.16 70.07 70.23 70.04 0.11 0.02 70.10 70.23 0.38 70.07 70.15 70.06 70.03 70.05 0.13 1 71.0019 Other industries 70.01 0.01 70.16 0.07 0.23 0.04 70.11 70.02 0.10 0.23 70.38 0.07 0.15 0.06 0.03 0.05 70.13 71.00 1

Mean 0.52 0.48 0.53 0.63 0.79 0.11 82.8 0.74 0.21 0.32 0.39 0.15 0.15 0.13 0.34 0.29 0.24 0.35 0.65s.d. 0.50 0.50 0.50 0.48 0.41 0.31 8.3 0.44 0.41 0.47 0.49 0.35 0.35 0.33 0.48 0.46 0.43 0.48 0.48Minimum 0 0 0 0 0 0 52 0 0 0 0 0 0 0 0 0 0 0 0Maximum 1 1 1 1 1 1 96 1 1 1 1 1 1 1 1 1 1 1 1

AL

LIA

NC

ES

WIT

HC

OM

PE

TIT

OR

S215

#B

lackw

ellP

ub

lishers

Ltd

2002V

olum

e11

Nu

mber

3S

eptember

2002

We defined several variables to addressother formation factors, including multi-firmalliances, alliance founding year, competitiveasymmetry, alliance experience, and industryidentity. A 0–1 dummy variable denotedalliances that had more than two partners.We defined a variable to denote the calendaryear in which the firms founded the alliance.We defined a variable to denote the com-petitive asymmetry of the partners. We oper-ationalized the concept of asymmetry bycomparing the sales of the partner firms inthe industry in which the firms created thealliance. The asymmetry measure is appro-priate because the alliance partners competein the same industries and product lines,which we checked from secondary sources,industry analysts, and company executives.We considered a partnership to be asym-metric when, at the time the firms created thealliance, the sales in the focal industry of oneof the partner firms were at least twice aslarge as the sales of the other partner. Franko(1971) and Ravenscraft and Scherer (1987)used similar factors. The alliance experiencevariable noted whether two or more of thepartners in an alliance had formed an alliancewith each other within the ten years beforethe formation of the focal alliance. Finally, toaddress industry differences and to investi-gate alliances in the telecom sector, which isthe largest sector in our data and has beenparticularly dynamic (Garrette & Quelin,1994), we defined two 0–1 dummy variablesto distinguish alliances set up in the telecom-electronics industry and in other sectors.

We used maximum likelihood binomiallogistic regression to test the formation propo-sitions. The logistic regression models tookthe form Ln Pi/(1-Pi) = bXi. In this equation,Pi is the probability that alliance i will be alink or scale alliances. A vector of covariatesXi with coefficient vector b, including anintercept, linearly affects the log odds of theprobability. The effect of a one-unit change ofcovariate j on the probability that an alliancewill be a particular type is bjPi(1-Pi). We usedthe logistic regression procedure of the SASstatistical package to obtain the estimates.Logistic regression provides a well-acceptedtechnique for estimating the likelihood thatdiscrete outcomes will occur.

Governance mechanism variables andmethods

To test the governance mechanism predic-tions, we defined several dummy variablesfor parent equity holdings and the varioustypes of joint ventures. An equity variabletook a value of 1 if an alliance partner held

equity in the other parent. A joint venturevariable denoted if the alliance took the formof an equity joint venture organization. Wedistinguished between balanced and unbal-anced joint venture ownership where, in thecase of two partner alliances, balanced owner-ship corresponds to 50—50 joint ventures. Wedistinguished between integrative and se-quential joint ventures (Park & Russo, 1996).Integrative joint ventures assign alliance manu-facturing activities to a jointly owned jointventure facility. In sequential joint ventures,the firms allocate all activities to individualpartners in a sequential path, with no jointoperations within a separate joint venturefacility. We also created two-way interactionterms among the governance mechanismsthat had consistent direction for their singlevariable predictions to test for augmentingjoint relationships among governance modes.We then used the alliance type (link or scale)variable that we described earlier as the inde-pendent variable for the governance analysis.Table 3b reports summary statistics for thegovernance variables.

We chose a correlation approach to test thegovernance mechanisms hypotheses. In thisapproach, we estimated correlation relation-ships between each governance mechanismand the alliance type variable. Each of themechanisms represents a different outcomechoice, so that a more complicated approachto the statistical analysis would require alarge number of models. Moreover, we viewlink formation and governance as a two-stageprocess in which firms first chose an alliancetype, depending on their strategic objectives,and then chose a governance mode that isappropriate for the type of alliance. That is,we view resource characteristics as the deter-minant of alliance type, with alliance typethen determining governance mode. With thistwo-stage approach, the appropriate test ofthe governance hypotheses is to examine thesimple relationships between alliance typeand governance mode, rather than undertakea more complicated analysis in which alliancetype and resource characteristics jointly de-termine governance mode. Indeed, such amultiple regression approach would entailendogeneity among the independent vari-ables on the right hand side of the regressionequations.

In summary, we have gathered data that ishighly relevant to testing the hypotheses thatwe develop in this paper. The data includemore than 200 alliances among competingfirms from Europe, North America, and Asia,in several industrial sectors. The data provideoperational measures of alliance types, re-source characteristics, and alliance governance



Table 3b. Summary Statistics for Governance Variables (N = 227)

20 21 22 23 24 25 26 27 28 29 30

20 Parent equity holding 1 70.19 70.17 70.03 70.05 70.15 0.39 70.33 70.17 70.09 70.1521 Joint venture (v. Contract alliance) 70.19 1 0.65 0.49 0.34 0.79 0.19 0.60 0.31 0.30 0.2822 Balanced joint venture 70.17 0.65 1 70.35 0.35 0.43 70.14 0.92 0.35 0.46 0.4323 Unbalanced joint venture 70.03 0.49 70.35 1 0.02 0.48 0.40 70.32 70.02 70.16 70.1524 Sequential JV (Coordination organization 70.05 0.34 0.35 0.02 1 70.30 0.07 0.33 0.91 0.86 0.8225 Integrative JV (Coordination & functional org.) 70.15 0.79 0.43 0.48 70.30 1 0.15 0.39 70.27 70.26 70.2426 Equity holding * Unbalanced joint venture 0.39 0.19 70.14 0.40 0.07 0.15 1 70.13 70.07 70.06 70.0627 Nonequity * Balanced joint venture 70.33 0.60 0.92 70.32 0.33 0.39 70.13 1 0.39 0.43 0.4728 Nonequity * Sequential joint venture 70.17 0.31 0.35 70.02 0.91 70.27 70.07 0.39 1 0.84 0.9029 Balanced sequential joint venture 70.09 0.30 0.46 70.16 0.86 70.26 70.06 0.43 0.84 1 0.9430 Nonequity * Balanced JV * Sequential JV 70.15 0.28 0.43 70.15 0.82 70.24 70.06 0.47 0.90 0.94 1

Mean 0.22 0.52 0.32 0.21 0.11 0.41 0.04 0.28 0.10 0.09 0.08s.d. 0.41 0.50 0.47 0.41 0.32 0.49 0.20 0.45 0.30 0.28 0.27Minimum 0 0 0 0 0 0 0 0 0 0 0Maximum 1 1 1 1 1 1 1 1 1 1 1

AL

LIA

NC

ES

WIT

HC

OM

PE

TIT

OR

S217

#B

lackw

ellP

ub

lishers

Ltd

2002V

olum

e11

Nu

mber

3S

eptember

2002

mechanisms. The sample supports the use ofstraightforward statistical techniques to testthe hypotheses.

Results

Alliance formation

Table 4 reports the results of the allianceformation analysis. The results provide strongsupport for most formation hypotheses. Theoverall statistical fit of the models is reason-able, with pseudo R-square statistics (theproportion of the loglikelihood of a modelwith no covariates that the reported modelexplains) of 0.64.

The results in Table 4 are consistent withboth geographic resource predictions. Ashypothesis 1a predicted, allies with differentgeographic origins are most likely to formlink alliances. The results apply to parentsthat have home bases on different continents,with similar results for intercontinental alli-ances involving Asian, European, and NorthAmerican firms. The parent geographic dis-persion result does not hold for parents fromdifferent countries within the same continent,likely because there has been substantialdiffusion of resources among countries withinEurope and within North America.

As hypothesis 1b predicted, Table 4 showsthat alliances with narrower geographic cover-

Table 4. Logistic Regression Estimates of Associations with Link and Scale AlliancesPositive Coefficient means Association with Link Alliances

Linkprediction

Coef s.e.

Parent geographic originParent same continent (a) H1a + 70.55 0.87Parent inter-continent, Asia (a) H1a ++ 1.71 0.89 *Parent inter-continent, Eur-NAm (a) H1a ++ 1.67 0.87 *Alliance geographic marketSingle zone, Europe (b) H1b ++ 2.18 0.69 ***Single zone, North America (b) H1b ++ 3.39 1.06 ***Single zone, Asia (b) H1b ++ 2.02 1.07 *Alliance functional resources (c)R&D resources H2a – 73.12 0.85 ***Production resources H2b – 71.93 0.72 ***Marketing resources H2c + 3.79 1.06 ***Production & marketing resources 2.50 0.79 ***R&D, production, marketing 71.24 0.48 ***

Competitive asymmetry H3a + 0.87 0.64 #Multi-firm alliances H3b – 71.70 1.16 #

Other factorsYear of formation 70.10 0.04 ***Prior alliances among partners 71.25 0.91Industry telecom-electronics (d) 1.02 0.60 *

Intercept 5.96 3.20 *

Loglikelihood ratio 201.8 ***Cases (Link alliances) 227 (118)Pseudo R-square 0.64

(a) Compared to parents from the same country(b) Compared to market coverage that extends beyond a single continent(c) Resource variables are mutually-exclusive mean effects dummy variables(d) Compared to industries other than telecom-electronics (auto, aerospace, other)* p5.10, ** p5.05, *** p5.01 (two-tailed tests of coefficients; one tailed test of loglikelihood ratio chi-square); # p5.10 (one-tailed test)



age tend to be link alliances. Alliances thatlimit coverage to only a single continent,whether the continent be Asia, Europe, orNorth America, are more likely to be linkalliances, while alliances that provide marketcoverage beyond a single continent tend to bescale alliances.

The results in Table 4 are also consistentwith the functional resource predictions. Ashypotheses 2a and 2b predicted, firms areunlikely to employ R&D or productionresources in link alliances. As hypothesis 2cpredicted, firms tend to employ marketingresources in link alliances. The table alsoexamines cases in which alliances involvemore than one type of resource. The resultsshow that the marketing influence towardslink alliances tends to drive alliances thatcombine production and marketing re-sources. By contrast, alliances that combineall three types of resources tend to be scalealliances, likely because of the concern aboutR&D appropriation that arises in link alli-ances among competitors.

The results in Table 4 moderately supporthypotheses 3a and 3b, concerning competitiveasymmetry and number of partners. As ex-pected, we find that alliances among partnerswith asymmetric competitive positions tendto be link alliances, although with onlymoderate significance. Similarly, alliancesinvolving multiple partners tend to be scalealliances, with moderate significance.

The other factors in column 1a also revealinfluences on alliance formation. Link alli-ances are less common in more recent years.Link alliances are also moderately less com-mon when the partners have previous experi-ence with each other. The telecom-electronicsindustry variable shows that link alliances aremore common in that sector than in the otherindustries. The recent technical, market, regu-latory, and competitive dynamics of thetelecom-electronics sector appears to havedriven a greater need for expansion-orientedlink alliances, which provide opportunities torecombine resources and create new resourcesthat attempt to respond to the rapidly changingenvironment.

Overall, the alliance formation results inTable 4 show that how firms apply geographicand functional resources has strong influenceson the types of alliances that they form. Linkalliances are most common with parents fromdifferent continents, with alliances that in-volve marketing resources, and with alliancesamong competitors with asymmetric compe-titive strength. Scale alliances are most com-mon for alliances with multi-continentalmarket coverage and for alliances involvingR&D or production resources.

Governance mechanisms

Table 5 reports the governance mechanismresults, showing the correlations between linkalliances and the different governance mech-anisms. The correlations provide support formost predictions, with some intriguing differ-ences emerging across the industry sub-samples. As hypothesis 4 predicted, link alli-ances tend to associate with parent equityholding. We also find that joint ventures aremore common among scale alliances, whichwe posed as an empirical question. As hypoth-esis 5a predicted, unbalanced joint venturesare somewhat more common among linkalliances, although this result is not statisticallysignificant. As hypothesis 5b predicted, bal-anced joint ventures are more common amongscale alliances. Consistent with hypothesis 6,sequential joint ventures are most commonfor scale alliances. Finally, consistent with ourarguments, integrative joint ventures associ-ate with both types of alliances.

The analysis of interactions among govern-ance mechanisms offers additional insights.Firms appear to use the basic protectionmechanisms we discussed earlier predomi-nantly as substitutes for one another ratherthan in a cumulative way. This implicationemerges from the observation that the singlevariable correlations with link alliances tendto be about as strong as the interactions of twoor more variables that have the same pre-dicted relationship with link alliances. In-deed, we find that the combination of equityholdings and unbalanced joint ventures hasno significant relationship with link alliances.Link alliances tend to include one form ofprotection, i.e., equity holdings or unbalancedjoint ventures, but do not systematicallycombine the two forms of protection. Scalealliances, in which unwanted resource trans-fers are a less salient issue, are more likely tocombine governance mechanisms that offer alow level of protection and also offer oppor-tunities for coordination via joint ventures.Nonetheless, the strength of the statisticalrelationship is similar to those that we foundwith the individual coordination mechanismsof balanced joint ventures and sequential jointventures.

Overall, the correlations between alliancetypes and governance mechanisms are onlymoderately strong. A likely explanation isthat firms address many of their concernsabout resource protection when they under-take alliances with competitors through thechoice of the most appropriate alliance typeand through the nature of the resources theycontribute to the alliance, as much as throughspecific safeguard mechanisms. For instance,



to limit the risk of appropriation, firms avoidcontributing R&D resources in the context oflink alliances, as the results of hypothesis 2ain Table 4 show. If firms tend not to contri-bute sensitive resources to alliances that wouldencompass strong appropriation hazards, thenthe need for highly protective governancemechanisms is limited. This suggests that firmsdo not rely primarily on governance mechan-isms to protect the value of the resources theyown or of resources they create through col-laboration with competitors.

Our study of alliances between competingfirms suggests that a fundamental choicefirms make in alliances involves the resourcesthey are willing to share with a partner ratherthan solely the organization they set up toprotect the value of these resources. Thisresult at first might appear to contradict theview that ‘the best strategy is to organize andoperate efficiently’ (Williamson, 1991b). Ourresults, though, are consistent with William-son’s (1975, 1985) arguments concerning in-centives for internal organization versusexternal forms of organization. That is, firmsappear to be more likely to retain internalcontrol of resources that would be difficult toprotect via contractual safeguards. Overall, thechoice of alliance type and resource contribu-tions suggests that firms make such decisionsin anticipation of maneuvering and strategiz-ing during interactions with competitors.

The results suggest limits to the ability ofcontractual relationships such as equity hold-

ings in partners and majority holdings in jointventures to provide safeguards in inter-organizational alliances. If firms anticipatedthat it would be possible to create strongprotection mechanisms for most allianceswith competitors, then they could choose toeconomize by contributing and combiningany under-utilized resources with any rele-vant partner in any type of alliance. In suchcases, the links between parent firm featuresand alliance type on the one hand, andbetween resources contributed and alliancetype on the other hand would then be weak.In contrast, the link between resource appro-priation hazards and governance mechanismswould be strong. Instead, we found results insupport of both the formation and govern-ance predictions, suggesting that firms bothattempt to take appropriation risks intoaccount when forming alliances and thenattempt to choose governance mechanismsthat address protection concerns, along withresource coordination and creation oppor-tunities.2

Conclusion

This study uses a competence perspective onfirm strategy to develop hypotheses concern-ing alliance formation and governance. Westart with the assumption of bounded ration-ality by potentially self-interested actors whoface firm-specific limits on their foresight.

Table 5. Correlations Between Alliance Type and Governance Mechanisms

Variables (n = 227) Hypothesis Correlationwith link alliance

Governance mechanismsParent equity holding H4 + 0.18Joint venture (v. Contract alliance) 70.12Unbalanced joint venture H5a + 0.08Balanced joint venture H5b – 70.20Sequential JV (Coordination organization) H6 – 70.13Integrative JV (Coordination & functional org.) 70.04

Governance mechanism interactionsEquity holding * Unbalanced jV H4 & 5a +,+ 0.01Nonequity * Balanced JV H4 & 5b –,– 70.16Nonequity * Sequential JV H4 & 6 –,– 70.13Balanced JV * Sequential JV H5b & 6 –,– 70.20Nonequity * Balanced JV * Sequential JV H4, 5b & 6 –,–,– 70.17

p50.10 in bold typeface



We then focus on routines and resources asthe fundamental units of analysis for firmstrategy, while viewing firms as structures forgoverning routines and resources. We arguethat resource governance has three aspects,including resource coordination, creation,and protection. We argue that the alliancesfirms form with their competitors will at-tempt to govern resource use in a way thateconomizes on resource coordination, creation,and protection. We find that firms tend tocontribute different types of resources to linkand scale alliances with competitors: firmsprimarily contribute R&D and productionresources to scale alliances, and marketingresources to link alliances. This is consistentwith the argument that different resourcesboth offer different opportunities for resourcecoordination and creation and also createdifferent appropriation risks. We also findthat firms tend to use different governancemechanisms for link and scale alliances: firmsare more likely to choose stronger protectionmechanisms for link alliances which creategreater appropriation risks, and tend to seekhigher levels of coordination in scale alliances.This is consistent with the argument that thegovernance mechanisms provide differentialopportunities to coordinate, create, and protectresources.

We view this research as part of an emergingstream of work that is attempting to developa more detailed conceptual basis for under-standing the business organization. We believethat such work needs to grapple with the jointdemands of conceptual clarity on the onehand and managerial ambiguity on the other.That is, a central challenge that faces re-searchers who are developing routine-basedtheories of the firm lies in defining a cumu-lative set of concepts and measures, whileretaining key inter-connections among thefactors that managers must address whenthey make decisions about what their firmswill do.

Notes

1. Link alliances that involve only marketingresources are alliances in which one partnerprovides a marketing system and the otherpartner provides the alliance with an existingproduct that the firm produces itself.

2. We also tested for the cases in which linksbetween alliance type and governance mechan-isms on those alliances did not fit the dominantmodel (link alliances including R&D, multi-partner link alliances or scale alliances includ-ing marketing), finding no relationship withgovernance mechanisms.

References

Amit, R. and Schoemaker, P. J. H. (1993) StrategicAssets And Organizational Rents. Strategic Man-agement Journal, 14, 33–46.

Argyres, N. (1996) Capabilities, TechnologicalDiversification And Divisionalization. StrategicManagement Journal, 17, 395–410.

Barnard, C. I. (1938) The Functions of The Executive.Harvard University Press, Cambridge, MA.

Barney, J. B. (1986) Strategic Factors Markets:Expectations, Luck, And Business Strategy.Management Science, 42, 1231–1241.

Bresser, R. (1988) Matching Collective and Compe-titive Strategies. Strategic Management Journal, 9,375–385.

Capron, L., Dussauge, P. and Mitchell, W. (1998)Resource Redeployment Following HorizontalAcquisitions in Europe and North America,1988–1992. Strategic Management Journal, 19,631–661.

Chatterjee, S. and Wernerfelt, B. (1991) The LinkBetween Resources And Type Of Diversification:Theory And Evidence. Strategic ManagementJournal, 12, 33–48.

Chi, T. (1994) Trading In Strategic Resources:Necessary Conditions, Transaction Cost Pro-blems, And Choice Of Exchange Structure.Strategic Management Journal, 15, 271–290.

Child, J. and Faulkner, D. (1998) Strategies of Co-operation. Oxford University Press, Oxford.

Coase, R. H. (1937) The Nature Of The Firm.Economica, 4, 386–405.

Commons, J. R. (1934) Institutional Economics: ItsPlace In Political Economy. Macmillan Company,New York.

Conner, K.R. (1991) A Historical Comparison OfResource-Based Theory And Five Schools OfThought Within Industrial Organization Econ-omics: Do We Have A New Theory Of The Firm?Journal of Management, 17, 121–154.

Contractor, F.J. and Lorange, P. (1988) Why ShouldFirms Cooperate? The Strategy and EconomicsBasis for Cooperative Ventures. In Contractor,F.J. and Lorange, P. Cooperative Strategies inInternational Business. Lexington Books, Lexing-ton, MA, 3–30.

Coriat, B. and Dosi, G. (1998) Learning How ToGovern And Learning How To Solve Problems:On The Co-Evolution Of Competences, Conflicts,And Organizational Routines. In Chandler, A.,Hagstrom, P. and Solwell, O. (eds.), The DynamicFirm. Oxford University Press, Oxford, pp. 103–133.

Dosi, G., Marengo, L. and Fagiolo, G. (1996)Learning In Evolutionary Environments. Univ. ofRome working paper.

Doz, Y.L. (1996) The Evolution of Cooperation inStrategic Alliances: Initial Conditions, or LearningProcesses? Strategic Management Journal, 17, 55–83.

Dussauge, P. and Garrette, B. (1995) Determinantsof Success in International Strategic Alliances :Evidence from the Global Aerospace Industry,Journal of International Business Studies, 26, 505–530.

Dussauge, P. and Garrette, B. (1997) Anticipating theEvolutions and Outcomes of Strategic Alliances



Between Rival Firms International Studies ofManagement and Organization, 27, 104–126.

Dussauge, P., Garrette, B. and Mitchell, W. (2000)Learning from Competing Partners: Outcomesand Durations of Scale and Link Alliances inEurope, North America and Asia. StrategicManagement Journal, 21, 99-126

Franko, L.G. (1971) Joint Venture Survival in Multi-national Corporations. Praeger, New York.

Garrette, B. and Quelin, B. (1994) An EmpiricalStudy of Hybrid Forms of Governance Structure:The Case of the Telecommunication EquipmentIndustry. Research Policy, 23, 395–412.

Geringer, J.M. and Hebert, L. (1989) Control andPerformance of International Joint Ventures.Journal of International Business Studies, 20, 235–254.

Ghoshal, S. and Moran, P. (1996) Bad For Practice:A Critique of Transaction Cost Theory. Academyof Management Review, 21, 13–47.

Gomes-Casseres, B. (1990) Firm Ownership Pre-ferences and Host Government Restrictions: AnIntegrated Approach. Journal of InternationalBusiness Studies, 21, 1–22.

Grant, R. M. (1991) The Resource-Based TheoryOf Competitive Advantage: Implications ForStrategy Formulation. California Management Re-view, 33, 114–135.

Gulati, R. (1998) Alliances and networks. StrategicManagement Journal, 19, 293–317.

Hamel, G. (1991) Competition for Competenceand Inter-Partner Learning within InternationalStrategic Alliances. Strategic Management Journal,12, 83–103.

Hannan, M. T. and Freeman, J. H. (1989) Organiz-ational Ecology. Harvard University Press, Cam-bridge, MA.

Harbison, J. R. and Pekar, P., Jr. (1998) SmartAlliances: A Practical Guide To Repeatable Success.Jossy-Bass, San Francisco.

Harrigan, K. (1986) Managing for Joint VentureSuccess. Lexington Books, Lexington, MA.

Harrigan, K.R. (1985) Strategies for Joint Ventures.Lexington Books, Lexington, MA.

Henderson, R. and Cockburn, I. (1994) MeasuringCompetence? Exploring Firm Effects In Pharma-ceutical Research. Strategic Management Journal,15, 63–84.

Hennart, J.-F. (1991) The transaction costs theory ofthe multinational enterprise. In Pitelis, C. andSugden, R. (eds.), The Nature of the TransnationalFirm. Routledge, London.

Hennart, J.F. (1988) A Transaction Cost Theoryof Equity Joint Ventures. Strategic ManagementJournal, 9, 361–374.

Hennart, J.F., Roehl, T. and Zietlow, D.S. (1999)Trojan Horse or Workhorse? The Evolution ofU.S.-Japanese Joint Ventures in the UnitedStates. Strategic Management Journal, 20, 15–29.

Itami, H. and Roehl, T. (1987) Mobilizing invisibleassets. Harvard University Press, Cambridge, MA.

Jacquemin, A., Buigues, P. and Ilzkovitz, F. (1989)Horizontal Mergers and Competition Policy inthe European Community. European Economy, 40,125–143.

Jorde T.M. and Teece, D.J. (1990) Innovation andCooperation: Implications for Competition andAntitrust. Journal of Economic Perspectives, 4, 75–96.

Karim, S. and Mitchell, W. (2000) Reconfiguringbusiness resources following acquisitions in theU.S. medical sector, 1978–1995. Strategic Manage-ment Journal, Special Issue on the Evolution ofBusiness Capabilities, 21, 1061–1081.

Khanna, T., Gulati, R. and Nohria, N. (1998) TheDynamics of Learning Alliances: Competition,Cooperation and Relative Scope. Strategic Man-agement Journal, 19, 193–210.

Killing, J.P. (1983) Strategies for Joint Venture Success.New York : Praeger.

Kogut, B. (1988) Joint Ventures: Theoretical andEmpirical Perspectives. Strategic ManagementJournal, 9, 319–332.

Liebeskind, J. (1996) Knowledge, strategy, and thetheory of the firm. Strategic Management Journal,17, 93–107.

Millington, A.I. and Bayliss, B.T. (1995) Trans-national Joint Ventures between UK and EUManufacturing Companies and the Structure ofCompetition. Journal of International BusinessStudies, 26, 239–254.

Mitchell, W. (1991) Dual Clocks: Entry Order Influ-ences on Industry Incumbent and NewcomerMarket Share and Survival When SpecializedAssets Retain Their Value. Strategic ManagementJournal, 12, 85–100.

Mitchell, W. and Singh, K. (1993) Death of theLethargic: Effects of Expansion into New Tech-nical Subfields on Performance in a Firm’s BaseBusiness. Organization Science, 4, 152–180.

Mitchell, W. and Singh, K. (1996) Survival ofBusinesses using Collaborative Relationships toCommercialize Complex Goods. Strategic Man-agement Journal, 17, 169–195.

Nelson, R.R. (1991) Why Do Firms Differ and HowDoes it Matter? Strategic Management Journal, 12,61–74.

Nelson, R.R. and Winter, S.G. (1982) An Evolu-tionary Theory of Economic Change. HarvardUniversity Press, Cambridge, MA.

Oliver, C. (1990) Determinants of Interorganiza-tional Relationships: Integration and FutureDirections. Academy of Management Review, 15,241–265.

Oxley, J.E. (1999) Institutional Environment AndThe Mechanisms Of Governance: The Impact OfIntellectual Property Protection On The Struc-ture Of Inter-Firm Alliances. Journal of EconomicBehavior, 38, 283–310.

Park, S.H. and Russo, M.V. (1996) When Competi-tion Eclipses Cooperation: An Event HistoryAnalysis of Joint Venture Failure. ManagementScience, 42, 875–890.

Penrose, E.T. (1959) The Theory of Growth of the Firm.Basil Blackwell, London.

Pisano, G. (1989) Using Equity Participation ToSupport Exchange: Evidence From The Biotech-nology Industry. Journal of Law, Economics andOrganization, 5, 109–26.

Porter, M.E. and Fuller, M.B. (1986) Coalitionsand Global Strategy. In Porter, M.E., Competitionin Global Industries. Harvard University Press,Cambridge, MA, 315–344,

Ravenscraft, D.J. and Scherer, F.M. (1987) Life afterTakeover. Journal of Industrial Economics, 36, 147–156.



Richardson, G.B. (1972) The Organization OfIndustry. Economic Journal, 82, 883–896.

Sakakibara, M. (1997) Heterogeneity Of FirmCapabilities And Cooperative Research AndDevelopment: An Empirical Examination OfMotives. Strategic Management Journal, 18, 143–164.

Simon, H. A. (1957) Administrative Behavior. Mac-millan, New York.

Sobrero, M. and Roberts, E. B. (1996) The trade-off between efficiency and learning in inter-organizational relationships. Working Paper#3896, MIT Sloan School of Management.

Teece, D. J. (1986) Profiting From TechnologicalInnovation: Implications For Integration, Collab-oration, Licensing, And Public Policy. ResearchPolicy, 15, 285–305.

Teece, D. J., Pisano, G. and Shuen, A. (1997)Dynamic capabilities and strategic management.Strategic Management Journal, 18, 509–533.

Teece, D.J. and Pisano, G. (1994) The DynamicCapabilities Of Firms: An Introduction. Industrialand Corporate Change, 3, 537–556.

Wernerfelt, B. (1984) A Resource Based View OfThe Firm. Strategic Management Journal, 5, 171–180.

Williamson, O. E. (1975) Markets and Hierarchies.Free Press New York.

Williamson, O. E. (1983) Credible Commitments:Using Hostages To Support Exchange. AmericanEconomic Review, 73, 519–540.

Williamson, O. E. (1985) The Economic Institutions ofCapitalism. The Free Press, New York, NY.

Williamson, O.E. (1991a) Comparative EconomicOrganization: The Analysis of Discrete Structural

Alternatives. Administrative Science Quarterly, 36,269–296.

Williamson, O.E. (1991b) Strategizing, Economiz-ing, and Economic Organization. Strategic Man-agement Journal, 12, 75–94.

Williamson, O.E. (1999) Strategy research: Govern-ance and competence perspectives. University ofCalifornia at Berkeley working paper.

Winter, S.G. (1990) Survival, selection, and inheri-tance in evolutionary theories of organization. InSingh, J. V. (ed.), Organizational Evolution: NewDirections. Sage, Newbury Park, CA, 269–296.

Zajac, E. J. and Olsen, C. P. (1993) From Trans-action Cost To Transaction Value Analysis:Implications For The Study Of Interorganiza-tional Strategies. Journal of Management Studies,30, 131–145.

Will Mitchell is the J. Rex Fuqua Professorof International Management and Professorin Strategy at the Fuqua School of Businessof Duke University, in Durham, NorthCarolina.

Pierre Dussauge is Professor of StrategicManagement at HEC School of Manage-ment and Visiting Professor at the Uni-versity of Michigan Business School.

Bernard Garrette is Associate Professorof Strategic Management at HEC School ofManagement, in Paris, France.



Documents

Alliances With Competitors How to Combine and Protect Key Resources