The architecture of cooperation: managing coordination costs and appropriation concerns in strategic alliances

Ranjay Gulati

Corporations have increasingly seen alliances as attractive vehicles through which they can grow and expand their scope, and the rate at which interfirm alliances have been formed in the last two decades has been unprecedented (Harrigan, 1986; Anderson, 1990). A notable characteristic of this growth has been the increasing diversity of interfirm alliances. The nationalities of partners, their motives and goals in entering alliances, and the formal structures used to organize the partnerships have all become increasingly varied. The variety of organizing structures implies that firms face numerous choices in structuring their alliances. This study examines why firms choose the specific governance structure they do in alliances. It explores some of the conditions at the inception of an alliance that influence the formal structure used to govern it.

An alliance is commonly defined as any voluntarily initiated cooperative agreement between firms that involves exchange, sharing, or co-development, and it can include contributions by partners of capital, technology, or firm-specific assets (e.g., Harrigan, 1986; Parkhe, 1993a; Gulati, 1998). The governance structure of the alliance is the formal contractual structure participants used to formalize it. Prior research has distinguished among such formal structures in terms of the degree of hierarchical elements they embody and the extent to which they replicate the control and coordination features associated with organizations, which are considered to be at the hierarchical end of the spectrum (e.g., Pisano, Russo, and Teece, 1988; Pisano, 1989; Gulati, 1995a). At one end are joint ventures, which involve partners creating a new entity in which they share equity and that most closely replicate the hierarchical control features of organizations. At the other end are alliances with no sharing of equity that have few hierarchical controls built into them.

Organizational scholars have long studied the basis for hierarchical controls within organizations and viewed them as a mechanism to manage uncertainty. Prior research on contract choices in alliances and the extent of hierarchical controls they embody has been influenced primarily by transaction cost economists, who have focused on the appropriation concerns in alliances, which originate from pervasive behavioral uncertainty and contracting problems (e.g., Pisano, Russo, and Teece, 1988; Pisano, 1989; Balakrishnan and Koza, 1993). Following this perspective, scholars have suggested that hierarchical controls are an effective response to such concerns at the time the alliance is formed. Thus, the greater the appropriation concerns, the more hierarchical the likely governance structures for organizing the alliance. The logic for hierarchical controls as a response to appropriation concerns is based on their ability to assert control by fiat, provide monitoring, and align incentives. The operation of such a logic was originally examined in make-or-buy decisions (e.g., Monteverde and Teece, 1982; Walker and Weber, 1984; Masten, Meehan, and Snyder, 1991). The same logic by which firms choose between the extremes of making or buying is also expected to operate, once firms have decided to form an alliance, in their choice of governance structure: when firms anticipate appropriation concerns, they are likely to organize alliances with more hierarchical contracts.

While researchers have made significant advances in classifying alliance governance structures and in identifying their determinants, our understanding of alliances is limited by two factors inherent in much of that research. First, the research on alliances focuses on the anticipated appropriation concerns as the primary basis of the choice of governance structure (Williamson, 1985, 1991). Building on the idea that an important feature of hierarchical controls is their ability to manage potential moral hazards, transaction cost economists suggest that hierarchical controls arise in alliances when participants anticipate such concerns. Even resource dependence theorists, who have looked primarily at the origin of ties rather than their structure, have suggested similar moral hazard concerns as a reason why firms may transform pure exchange relations into power relations (Pfeffer and Nowak, 1976).

While appropriation concerns originating from contracting obstacles, combined with pervasive behavioral uncertainty, can clearly be an important concern, once firms decide to enter an alliance, there is another set of concerns that arises from anticipated coordination costs. By coordination costs we mean the anticipated organizational complexity of decomposing tasks among partners along with ongoing coordination of activities to be completed jointly or individually across organizational boundaries and the related extent of communication and decisions that would be necessary.(1) Coordination considerations are extensive in alliances. Litwak and Hylton (1962: 399) noted that the specialized coordination in interorganizational relations is a challenge, "since there is both conflict and cooperation and formal authority structure is lacking." As a result, the anticipated interdependence resulting from the logistics of coordinating tasks can create considerable uncertainty at the outset of an alliance that is different from appropriation concerns. The uncertainty for participants concerns the way activities will be decomposed and integrated and the extent to which there is likely to be an ongoing need for mutual adaptation and adjustment.

The distinction between coordination costs and appropriation concerns can be understood with a hypothetical example.

Imagine that an alliance is formed between two firms that have complete confidence in each other and face no appropriation concerns whatsoever. Despite this frictionless situation, they must still coordinate the division of labor and the interface of activities and products between them. This creates considerable uncertainty that alliance partners consider at the time they form an alliance and attempt to answer in structuring the relationship. Hierarchical controls can be an effective solution in situations of high anticipated coordination costs. As noted by Barnard (1938), Chandler (1977), Thompson (1967), and others, an important basis for hierarchical controls is' their ability to provide superior task coordination, especially in situations involving high interdependence and coordination.

A second concern with prior research on the governance structure of alliances is that researchers in this area have broadly classified alliances as equity or nonequity and have considered the presence of shared equity in alliances as synonymous with hierarchy. They have justified this combination on the grounds that all equity alliances have similar incentive properties - shared ownership and controls. The use of equity to indicate hierarchical controls is logical, given the emphasis of prior researchers on the agency features of such controls, because equity provides an effective means to address agency concerns, but the presence of equity sharing also masks differences across each type of structure and provides only a partial assessment of the original basis for classifying the governance structure of alliances: degree of hierarchical control. Each governance structure not only presents distinct levels of hierarchical controls, but how such controls manifest themselves may also differ. Furthermore, implicit in this typology is the idea that the presence of hierarchical controls in alliances is an either- or proposition, which is questionable.

The present study addresses these problems by proposing several hypotheses in which we examine the influence of both anticipated coordination costs and appropriation concerns at the outset of alliances in explaining the choice of governance structure. We suggest that this choice is not only determined by concerns of appropriation, as previously suggested, but is also influenced by concerns about coordination costs resulting from the extent of interdependence expected by the partners. We use Thompson's (1967) distinction between pooled, sequential, and reciprocal interdependence to differentiate varying degrees of interdependence in three types of alliances. We test our predictions about alliance structure with comprehensive data on alliances formed between 1970 and 1989 in three industrial sectors worldwide.

THEORY

Coordination Costs and Interdependence as an Organizing Principle

Theories for the use of hierarchical controls have primarily focused on explaining structure within organizations and the choice between markets and organizations. The motor behind the theories for hierarchical controls is the management of uncertainty, but different perspectives have emphasized alternative facets of uncertainty. Transaction cost economics has focused primarily on the manifest presence of behavioral uncertainty which, when combined with certain contracting situations, enhances appropriation concerns and makes it difficult and costly to write contracts (Williamson, 1985). This can be further compounded by exogenous shocks, which may require subsequent adaptation by the partners (Williamson, 1991). While the importance of behavioral uncertainty and appropriation concerns as a rationale for hierarchical controls is well understood, the role of anticipated coordination costs and the uncertainty associated with them as a basis for hierarchical controls has been less developed and may be equally important.

The importance of coordination costs as a basis for hierarchical controls has been emphasized for the organization of activities within firms (Lawrence and Lorsch, 1967; Thompson, 1967), and organizational sociologists have referred to hierarchical controls as superior information-processing mechanisms that result from the increasing division of labor and the uncertainty originating from the need to coordinate interdependent subtasks (Galbraith, 1977: 93). They have built on the work of Barnard (1938), who noted the ability of organizational hierarchies to mitigate the uncertainty resulting from coordination and control of complex and interdependent tasks by creating cooperation and coordination among organizational members, and Chandler (1977), who emphasized the significance of coordination in hierarchical structures. Also, in recent years, transaction cost economists have begun to examine issues such as "temporal specificity," or the importance of timing in receipt of goods or services, that are related to coordination costs (Masten, Meehan, and Snyder, 1991).

Concerns about anticipated coordination costs are particularly salient in strategic alliances, which can entail significant coordination of activities between the partners and yet have to be managed without the benefit of the structure and systems available in traditional hierarchies (Litwak and Hylton, 1962). It arises from the complexity of ongoing coordination of activities to be completed jointly or individually across organizational boundaries and the difficulties associated with decomposing tasks and specifying a precise division of labor across partners in the alliance, all of which require ongoing communication and decisions. The extent of such concerns in an alliance is best encapsulated by the level of interdependence that is necessary for the alliance partners to complete tasks (Thompson, 1967). At one extreme, partners in an alliance may anticipate minimal engagement between the partners and, at another, extensive coordination may be necessary. Many researchers have testified to the complexities associated with the interdependence of activities across partners in strategic alliances (Doz, Hamel, and Prahalad, 1989; Ring and Van de Ven, 1992), yet the implications of varying levels of anticipated interdependence and coordination costs for the governance structure of interfirm alliances remain unexplored.(2)

The concept of interdependence is a fundamental principle defining the costs of coordination within organizations that dates back to the early work of systems theorists (Ashby, 1956; Katz and Kahn, 1966) and was later developed by Thompson (1967) and others.(3) Scholars have applied it primarily to studying the internal design features of organizations and have devoted considerable effort to its elaboration and measurement. Organization design scholars have in fact referred to the challenges posed by interdependence as "coordination costs" (McCann and Galbraith, 1981). As interdependence or coordination costs increase, information-processing costs can also rise (Galbraith, 1977), as can pressure for fast responses (Emery and Trist, 1965) and conflict; ultimately, it can lead to a decline in performance (Pondy, 1970).

The extent of the anticipated interdependence between partners at the time they form an alliance can vary substantially and depends on the tasks included and the likely division of labor in the partnership, all of which are a function of the strategic rationale for the alliance. At one extreme, an alliance may have a simple division of labor with minimal ongoing adjustments that require each partner to share information about the progress of its initiatives for the partnership to achieve strategic goals. At the other extreme, the likely interdependence can be extensive, resulting from the anticipation of a complex and overlapping division of labor that will entail continuing mutual adjustments between partners and require each partner to link specific activities with other partners closely and regularly. The higher the anticipated interdependence between alliance partners, the greater the magnitude of expected coordination costs. While the anticipated interdependence across partners in an alliance may also influence the extent of the appropriation concerns partners experience (e.g., Stinchcombe, 1985), the primary concerns from interdependence are the administrative challenges of coordinating tasks between partners.

For two firms considering an alliance, the greater the need for ongoing task coordination and joint decision making between the partners in an alliance, the higher the anticipated level of interdependence and coordination costs. Each form of alliance governance structure provides differing degrees of control over and coordination of the activities in a partnership. As a result, firms will seek the governance structures for alliances that will provide the necessary ongoing oversight and coordination. The higher the interdependence among partners, the greater the amount of information they must process while the alliance is in progress (Galbraith, 1977). Partners in such alliances must evolve mechanisms through the governance structure to process the requisite information. Alliances with more hierarchical controls are capable of providing greater coordination and information-processing capabilities than those with fewer controls.

Hierarchical controls in strategic alliances. Contractual relationships such as alliances can include several hierarchical elements embedded in their structure (Stinchcombe, 1985), including (1) a command structure and authority systems to put it in place, as well as systems for certifying which communications are authoritative, (2) incentive systems that facilitate performance measurement and link rewards to performance, (3) standard operating procedures that allow quick decisions to be made by anticipating those decisions in advance, (4) dispute resolution procedures that bypass courts and markets by specifying a hierarchy of entities or individuals to which appeals can be made, and (5) non-market pricing systems, such as cost-plus systems, which enable greater precision in remuneration when changes in specification are made. These hierarchical elements are present to varying degrees in different governance structures for alliances. While incentive systems and non-market pricing highlight some of the agency features of hierarchical controls previously discussed by transaction cost economists, the other elements concern the coordination capabilities of hierarchical controls in alliances. For instance, the command structure, authority systems, and standard operating procedures all make it easier to coordinate tasks between partners by clarifying decision-making procedures and anticipating issues before they arise.

Hierarchical elements in alliances can effectively address the anticipated coordination costs resulting from interdependence for several reasons. The standard operating procedures, command structure, and authority systems in hierarchical governance structures in alliances typically include planning, rules, programs, or procedures, which March and Simon (1958) identified as key means for task coordination. Planning involves presetting schedules, outcomes, and targets; and rules, programs, and procedures emphasize formal controls in the form of decisions made a priori for various likely scenarios. All of these serve the common purpose of minimizing communication, simplifying decision making, reducing uncertainty about future tasks, and preventing disputes (Pondy, 1977). In alliances, hierarchical controls institutionalize, or formalize, interactions between partners (Van de Ven, 1976). In addition to regularizing meetings between partner representatives, such hierarchical controls may also formally designate roles for the partners. In joint ventures, which may involve the greatest extent of hierarchical controls, this type of control is implemented at the time of inception, when formal roles are allocated. Formalization makes the division of labor and the interactions between partners more predictable and allows joint decisions to be made more by rules than by exception. Because hierarchical controls clarify boundaries on decisions and activities, they simplify decision making (Galbraith, 1977).

While the mutual agreement about roles and responsibilities limits negotiation on each issue and preempts conflicts between interdependent units in the future, hierarchical governance structures for alliances may also include procedures for resolving disputes that arise as partners deal with task coordination and the division of labor. These procedures anticipate disputes and try to bypass adjudication by the courts by specifying a hierarchy of avenues for dealing with such concerns as they arise. Such alternative forums not only limit the scope of disputes, they also allow partners to discover joint solutions to more effective coordination. The presence of such systems in hierarchical structures gives them confidence that such issues will be adequately dealt with if they arise.

The incentive systems and non-market pricing elements of hierarchical controls most typically occur in alliances that create autonomous entities, such as joint ventures. Creating a separate entity makes it easier to monitor each party's contributions and the performance of joint activities while also aligning incentives for each of them. It also reduces ongoing market price haggling between partners by locating all resources and expenses within a single entity that is jointly owned. The creation of an autonomous entity can also simplify coordination in alliances. Such alliances provide a high level of discretion to the entity in which the joint activities are being conducted, which provides a dedicated management with a mandate to make decisions that optimize the activities contributed by each partner toward the accomplishment of their joint goals. An autonomous alliance such as a joint venture thus echoes Galbraith's (1977) notion of self-contained tasks, since the discrete entity is provided with its own set of resources to perform the assigned task. Such an alliance addresses anticipated coordination costs by limiting future discussions between partners on the precise division of labor and focuses attention on the outputs, since the inputs have already been agreed upon at the outset. It delineates clear roles for each partner, both through the separate management team and through its own board of directors. More important, it disperses decision making to local management, which leads to concerted decision making by the alliance partners. Much like Chandler's (1977) M-form organization, many hierarchical alliances such as joint ventures have their own strategic planning and resource allocation capability and their own monitoring and control apparatus.

Such autonomous structures can promote interactive planning, in which alliance partners frequently engage in collaborative objective-setting processes. Finally, hierarchical controls in alliances can address anticipated coordination costs by facilitating coordination through informal means, creating a sense of shared purpose that can motivate and guide individual participants and minimize conflict among them (Barnard, 1938; Blau, 1972; Ghoshal and Moran, 1996). Together, these attributes give hierarchical governance structures superior coordination capabilities and make them appropriate in situations of high interdependence and coordination costs.

Each governance structure for alliances is typically associated with distinct types and levels of hierarchical controls, which makes the choice across structures one of choosing the appropriate level of hierarchical controls depending on the extent of anticipated coordination costs and interdependence at the time of the inception of the alliance:

Hypothesis 1: The greater the anticipated interdependence in an alliance, the more hierarchical the governance structure used to organize it.

Appropriation Concerns and the Governance Structure of Strategic Alliances

Several researchers have used a transaction cost logic to study the governance structure of strategic alliances. Although transaction costs, broadly defined, encompass a wide range of elements, the basic thrust of the transaction cost argument as applied to strategic alliances builds on Alchian and Demsetz's (1972) discussion of team production and focuses on appropriation concerns that originate from the pervasive presence of behavioral uncertainty, combined with the difficulties of specifying intellectual property rights, and by the challenges of contractual monitoring and enforcement (Teece, 1986; Oxley, 1997).(4) Theorists talk about the appropriability of rents, usually referring to the ability of firms to capture rents generated by their innovative activities in an industry (Teece, 1986; Levin et al., 1987; Anand and Khanna, 1997). In contrast, when we discuss appropriation concerns in alliances, we are referring to the firm's concern about its ability to capture a fair share of the rents from the alliance in which it is engaged. Such concerns arise from the uncertainties associated with future specifications, cost uncertainties, and problems in observing partners' contributions, all of which aggravate the potential for moral hazards. Such appropriation concerns occur to varying degrees in most alliances.

Prior researchers have linked the anticipation of appropriation concerns at the time the alliance is formed with the specific governance structure used to formalize the alliance, suggesting that the greater the potential concerns, the more hierarchical the contract used (Pisano, Russo, and Teece, 1988; Pisano, 1989; Oxley, 1997). According to this logic, more hierarchical controls provide greater incentive alignment than fewer hierarchical controls (Alchian and Demsetz, 1972; Klein, Crawford, and Alchian, 1978; Demsetz, 1988). In discussing hierarchical governance structure, transaction cost economists typically focus on its agency features, which they view as addressing appropriation concerns through control mechanisms such as fiat, providing monitoring, and aligning incentives (Williamson, 1975, 1985). Hierarchical structures are thus thought to be more applicable when concerns of appropriation are potentially high. Empirical evidence for such explanations of alliance structure has been provided by Pisano (1989), in a study of the biotechnology sector, and by Pisano, Russo, and Teece (1988), in a study of the telecommunications sector.

Concerns about appropriation can vary with the specific circumstances of an alliance at its inception and arise because of the difficulty of writing complete contracts. This difficulty is exacerbated when technology exchange or sharing is involved and when the limits of the technology being transacted upon are difficult to specify (Merges and Nelson, 1990; Anand and Khanna, 1997; Oxley, 1997). Two conditions influence the extent to which such concerns arise: the presence of a technology component in the alliance and the appropriability regime in the industry. Each of these factors can independently and together influence the extent of appropriation concerns of firms entering an alliance by affecting the difficulties associated with specifying property rights and monitoring and enforcing the agreement.

Technology component in alliance. A primary basis from which prior researchers have examined concerns of moral hazards and appropriation in alliances is the presence of a technology component (Pisano, Russo, and Teece, 1988; Pisano, 1989), which can affect the extent of possible monitoring problems and the possibility of unobserved violation of contracts. Monitoring problems in technology alliances result from the ambiguity surrounding two key issues: What is the technology being transferred, and what are the limits to its use (Anand and Khanna, 1997)? In alliances encompassing technology, it can be difficult to circumscribe, bound, monitor, and codify the knowledge to be included within the alliance, which may lead to concerns about free-riding and possible appropriation of key technology by the partner. Such concerns are further compounded by the peculiar character of knowledge as a commodity, which makes it difficult for parties to assess accurately the value of the commodity being exchanged without complete information from the partner, who may not want to reveal such information because it is proprietary (Winter, 1964; Arrow, 1974; Teece, 1980: 28). This dilemma, which is called the knowledge paradox, can further aggravate concerns about appropriation of rents resulting from poor monitoring possibilities in such exchanges (Barzel, 1982; Hennart, 1988; Balakrishnan and Koza, 1993). The difficulty of transferring tacit research and development (R&D) know-how across organizations adds to these problems (Teece, 1980; Silver, 1984; Mowery and Rosenberg, 1989). Coordination costs can also be a concern in technology alliances, but they are likely to be salient only in a subset that may involve bilateral exchange or joint development. The primary concern of participants entering alliances with any technology component is with the anticipated appropriation concerns. As a result, according to such a logic, firms entering an alliance with a technology component are likely to prefer hierarchical alliances:

Hypothesis 2a: Alliances with an expected technology component are more likely than those without a technology component to be organized with more hierarchical governance structures.

Appropriability regime. Another factor likely to influence the level of appropriation concerns is the strength of the appropriability regime of the industry, which is the degree to which firms are able to capture the rents generated by their innovations (Anand and Khanna, 1997). tn a tight appropriability regime, firms can retain the profits they earn from their proprietary resources, while in a loose regime, these profits are subject to involuntary leakage or spillovers to other firms. The strength of the appropriability regime of an industry is related to patent strength, the value of first-mover advantage, and the ability to maintain the secrecy of an innovation (Teece, 1986; Levin et al., 1987). For alliances, a firm's concerns about appropriation will vary depending on the industry in which the alliance occurs and the degree to which the appropriability regime in the industry is tight or loose (Teece, 1986). If participants in an alliance believe that the appropriability regime is strong, because patent protection is significant, that they can keep trade secrets, or that their first-mover advantage is sufficiently large, they are likely to be less concerned about appropriation in an alliance, and this will be reflected in the formal governance structure used for the alliance. As a result, there should be a relationship between the appropriability regime of the industry and the governance structure of alliances, with more hierarchical structures expected in industries with weak appropriability regimes:

Hypothesis 2b: Alliances in an industry in which appropriability regimes are weak are more likely to be organized with more hierarchical governance structures than are alliances in an industry in which appropriability regimes are strong.

The effects of the appropriability regime are likely to be particularly acute if the alliance has a technology component, because the effects of the appropriability regime principally operate when there is a knowledge component in the alliance. As a result, partners in technology alliances are likely to experience concerns from appropriability regimes more acutely than those in alliances without a technology component. More specifically, the appropriation concerns anticipated in an alliance are likely to be amplified when potential alliance partners consider a technology alliance in an industry with a weak appropriability regime. Thus, the effect of the appropriability regime on the governance structure of alliances is likely to be moderated by the presence of a technology component in the proposed alliance:

Hypothesis 2c: The negative relationship between the strength of the appropriability regime in the industry and the extent of hierarchical governance structures of alliances will be stronger for alliances with a technology component than for those without one.

Trust and Governance Structure

Researchers have argued that although expectations of trust ultimately reside with individuals, it is possible to think of interfirm trust in economic transactions (Macaulay, 1963; Zucker, 1986; Gulati, 1995a, 1998). The benefits of trust in economic transactions such as alliances have been strongly emphasized. As Arrow (1974) has suggested, trust is perhaps the most efficient mechanism for governing economic transactions. Trust between alliance partners at the time of an alliance formation should address both coordination-cost and appropriation concerns and thus reduce the need for hierarchical controls in the alliance. This is consistent with the notion that three primary control mechanisms govern economic transactions between firms: price, authority, and trust (Bradach and Eccles, 1989). When there is trust, firms no longer consider hierarchical controls to be necessary (Powell, 1990; Ring and Van de Ven, 1992; Gulati, 1995a).

A key consequence of the embeddedness of economic transactions such as alliances in a social structure of trusting relationships is that the partners are likely to have greater confidence in the predictability of each other's actions and thus anticipate lower appropriation concerns when they form an alliance (Granovetter, 1985; Gulati and Gargiulo, 1999). Thus, the presence of interfirm trust at the time of alliance announcement, which obliges partners to behave loyally, can address appropriation concerns by making it easier to assess each other's likely behavior and to enforce property rights. In Levine and White's (1961) terms, such firms are likely to have greater domain consensus and fewer conflicts between them as a result. Entering alliances with partners whom one trusts can also significantly reduce adverse selection problems (Balakrishnan and Koza, 1993).

In addition to mitigating appropriation concerns, trust can alleviate concerns about coordination costs. Interfirm trust can be an extraordinary lubricant for alliances that involve considerable interdependence and task coordination between partners. Firms that trust each other are likely to have a greater awareness, or a willingness to become aware, of the rules, routines, and procedures each follows. Furthermore, since interorganizational trust typically results from the social structure of prior interactions, firms may have developed routines together to enable ease in joint interaction with each other from their prior experience (Gulati, 1995b; Gulati and Gargiulo, 1999). Thus, trusting firms may have greater competence in transacting with each other, which makes the interface between them easier to manage, and the information processing requirements associated with anticipated coordination costs are more easily addressed. While the presence of trust may not allow us to discriminate between coordination-cost and appropriation concerns, trust is distinctive in its ability to address both types of concerns. As a result, the presence of trust between partners is likely to promote fewer hierarchical controls in the alliances between them, not only because concerns of appropriation and behavioral uncertainty are effectively addressed but also because coordination costs are easily managed:

Hypothesis 3: Alliances in which there is less trust between partners are more likely to be organized with more hierarchical governance structures than are those in which there is greater trust.

Classifying Governance Structures

Prior research has generally classified the governance structure of interfirm alliances in terms of their hierarchical components and has typically differentiated alliances by the presence or absence of equity, with alliances involving equity considered to be more hierarchical than nonequity exchanges (Hennart, 1988; Pisano, Russo, and Teece, 1988; Pisano, 1989; Teece, 1992). Equity alliances include any exchange agreement in which the partners share or exchange equity. These include agreements in which partners create a new entity in which they share equity as well as those in which one partner takes an equity interest in the other. Equity has been considered an indicator of hierarchy because it is considered to be an effective mechanism for managing the rent appropriation concerns associated with partnering (Pisano, Russo, and Teece, 1988; Parkhe, 1993a; Moon and Khanna, 1995). Thus, Teece (1992: 20) suggested: "Equity stakes provide a mechanism for distributing residuals when ex ante contractual agreements cannot be written to specify or enforce a division of returns." In joint ventures, this occurs by creating a mutual hostage in the form of shared equity that helps align the interests of all the partners, inasmuch as each partner is concerned about the value of its equity in the alliance. In minority equity investments, the investing partner has an interest in the value of its equity holdings, while the recipient of investments can be legally required to furnish certain verified information to its investors. In equity alliances, the effective shared equity stakes of the firms vary by case, but beyond a certain threshold, the shared ownership structure is expected to provide an effective hierarchical control over the exchange.

While such a scheme is parsimonious, it masks differences in hierarchical controls across different types of structures and ignores the original basis for classifying the governance structure of alliances: degree of hierarchical controls. Also, because this typology considers the presence of equity sharing as a heuristic to indicate hierarchical controls, there has been little serious consideration of the specific governance structures of alliances or the precise levels of hierarchical controls in each and how they manifest themselves. For instance, joint ventures and minority investments provide varying levels of hierarchical control in a partnership, with joint ventures incorporating more hierarchical elements than minority investments. To varying degrees, each includes several "mechanisms for collecting information, deciding, and disseminating information to resolve conflicts and guide interdependent actions" (Galbraith, 1977: 40). In contrast with joint ventures, minority investments typically do not have a separate organizational and administrative structure and are thus relatively limited in their capacity to coordinate activities across partners. In addition to the exchanged equity, joint ventures entail separate administrative entities with their own management structures. Thus, in combining types, researchers have failed to consider the varying degree of hierarchical controls provided by each form of alliance. Moreover, because the presence of equity is an either- or proposition, the hierarchical controls in alliances have been treated similarly, without adequate justification.

We depart from prior efforts here by presenting a typology of alliance structure that does not treat the presence of equity as synonymous with hierarchical controls but, rather, defines three distinct types of alliance governance structures - joint ventures, minority investment, and contractual alliances - and the magnitude and specific types of hierarchical controls typically present in each of them. In our previous discussion we described the dimensions of hierarchical controls in alliances to include the following: command structure and authority systems, incentive systems, standard operating procedures, dispute resolution procedures, and non-market pricing systems. Together, these dimensions encompass both the agency and coordination features of hierarchical controls that are likely to be part of various types of alliances. While there may be some variation in the extent and type of hierarchical controls in specific instances, each of the three types of governance structures are typically associated with specific levels and forms of hierarchical controls. Our typology of alliance structures is the basis for testing the hypotheses on the factors likely to influence the choice among alternative governance structures with differing extents of hierarchical controls present.

At the hierarchical end of the spectrum are joint ventures, which occur when partners create a separate entity in which each owns a portion of the equity. In such alliances, a separate administrative hierarchy of managers oversees day-today functioning and addresses contingencies as they arise. This provides an independent command structure and authority system with clearly defined rules and responsibilities for each partner. The autonomous unit enables creation of an incentive system, because each partner is concerned about the value of its equity in the joint venture. Furthermore, pricing discussions are internalized by the joint venture, which controls much of the input and output resource flows to and from the partner organizations. As part of creating a joint venture entity, partner firms also typically put in place standard operating systems and dispute resolution procedures. Together, these features allow joint ventures to include considerable hierarchical controls within them.

Minority alliances, in contrast, include partnerships in which the firms work together without creating a new entity. Instead, one partner or a set of partners takes a minority equity position in the other (or others). Such alliances introduce a weaker form of control, with a degree of hierarchical controls intermediate between that in joint ventures and that in contractual alliances (Herriott, 1996). Hierarchical supervision is typically created by the investing partner joining the board of directors of the partner that received the investment. The presence of one or more individuals on the board of the partner in a minority alliance introduces a fiduciary role into the relationship and is also a vehicle for hierarchical controls. Since boards ratify most major decisions, the presence of an individual from the investing organization on the investee board ensures that the investor has some form of command and authority system. A concern for the value of its equity provides appropriate incentives for the investor. Furthermore, disputes are easier to resolve through board member intervention if necessary. Finally, while there may or may not be many standard operating procedures with such alliances, board representation does create a forum in which both partners exchange information and can initiate and ratify decisions on a regular basis. Beyond board-level interactions, day-to-day activities are jointly coordinated by the partners and negotiated on an ongoing basis.

Alliances in the third category, contractual alliances, do not involve the sharing or exchange of equity, nor do they entail the creation of new organizational entities. Lacking any shared ownership or administrative structure, contractual alliances are considered more akin to arm's-length market exchanges. Members of the partner firms work together directly from their own organizational confines. Few if any command structures, authority systems, incentive systems, standard operating systems, dispute resolution procedures, or non-market pricing systems are necessarily part of such arrangements. Ongoing activities are jointly coordinated, and new decisions are negotiated by the partners. Contractual alliances include unidirectional agreements such as licensing, second-sourcing, and distribution agreements and bidirectional agreements such as joint contracts and technology exchange agreements. While some of the hierarchical elements discussed may occur in some contractual alliances, they are not necessarily widespread and do not occur on a systematic basis.

METHODS

Sample

The data used in this study are drawn from all alliances announced between 1970 and 1989 in the biopharmaceuticals, new materials, and automobile sectors. Biopharmaceuticals includes both biotechnology firms and pharmaceutical firms.

The new materials sector includes ceramics, polymers, composites, and metals. The automotive sector includes both assemblers and suppliers within the automotive industry. A trade-off made in the design of this research was to limit the sample to three sectors. We ensured that the industries chosen represented a broad spectrum. While biopharmaceuticals and new materials are considered emerging industries, automotives is classified as more mature. Nonetheless, we exercised caution in interpreting the results from the pooled sample, and the models presented in this study were reestimated separately for each sector to ensure that the results were not sensitive to particular sectors.

We combined data from various sources. More than half the data on alliance announcements came from the Cooperative Agreements and Technology Indicators (CATI) database, collected by researchers at the University of Limburg. The CATI data were collected by examining public announcements in specialized technical journals, books, and business periodicals for various sectors and in the popular business press. We collected additional data on alliance announcements and details on the specific attributes of all alliances from numerous sources, including industry reports, industry-specific articles reporting alliances, and additional materials made available by industry consultants. Multiple sources, when they were available, helped us to verify the alliance's strategic rationale, technology components, and governance structure. For the automotive industry, the sources consulted include Automotive News, Ward's Automotive Reports, U.S. Auto Industry Report, Motor Industry of Japan, and the Japanese Auto Manufacturers Forum; for the biopharmaceutical sector, Bioscan, Ernst & Young reports, and the Biotechnology Directory; for the new materials sector, reports from the Office of Technology Assessment and the Organization for Economic Cooperation and Development; for all sectors, Predicast's Funk and Scott Index of Corporate Change.

The objective of the data collection effort was to cover comprehensively the alliances formed within the selected industries. Only alliances that had actually been formed were included - reports of probable future alliances were excluded. The complete alliance dataset includes information on 1,570 alliances formed by American, European, and Japanese firms between 1970 and 1989. We took pains to ensure that the dataset fully covered all alliances in each industry for each year of the study period. It is impossible to gauge the extent or consequences of the possible sample selection bias resulting from including only publicly announced ventures. Since our panel included only large and prominent firms in each industry, which usually get extensive press coverage, we expect this concern to be much less than with a sample of small firms, which receive less public scrutiny.

In our data collection efforts, we used precise criteria to facilitate the coding, which is described further below, and took several steps to ensure coding reliability. First, prior to starting the coding, we carefully controlled the dichotomous choice process by developing a list of synonyms for each choice. We clarified and refined explicit coding rules using 50 alliance announcements not in the sample. The general coding rule applied was to code only explicit references to each choice. Multiple public announcements were consulted from a wide variety of sources listed above. Because the dichotomous choices were clearly specified, the rules for coding were kept simple and straightforward, and multiple sources were consulted, the actual coding of alliances was not a complex task. Also, the clear specification of categories and the simplified coding rules boosted the reliability of our coding. We further attempted to ensure test-retest reliability by recoding a small number of alliances periodically after some time had elapsed since the original coding. Throughout the coding process, the results of the recoding were almost identical to the previous results, and the agreement rate ranged from .96 to 1.00. Overall, we believe that this process resulted in highly reliable coding of the alliance data.

Variables

Dependent variable. We do not believe that treating the presence of equity as synonymous with hierarchical alliances is entirely appropriate. To assess more accurately the factors explaining the degree of hierarchy in alliances, we therefore conducted our analyses with three categories of alliances, arrayed in increasing order of hierarchical controls (hierarchy): contractual alliances (coded 0), minority equity investments (1), and joint ventures (2).

Interdependence. We were interested in assessing the levels of interdependence the partners in an alliance anticipated at the outset, when the alliance was announced. Using Thompson's (1967) distinction between pooled, sequential, and reciprocal interdependence is a parsimonious way of arraying the degree of interdependence in alliances that underlies coordination costs. These three types of interdependence, although previously discussed in the intraorganizational context, can also be seen in the context of interfirm coordination of activities (Borys and Jemison, 1989). Pooled, or generalized interdependence, denotes situations in which "each part renders a discrete contribution to the whole, and each is supported by the whole," is "coordinated by standardization, and is least costly in terms of communication and decision effort" because it does not require any serial ordering of activities (Thompson, 1967: 54, 64). It exists in alliances when organizations pool their resources to achieve a shared strategic goal, the common benefits arise from combining resources into a shared pool, and each partner uses resources from the shared pool. These relatively small interdependencies entail low coordination requirements but provide partners with benefits from the pooled resources.

In situations of sequential interdependence, the activities of each partner are distinct and are serially arrayed so that the activities of one partner precede those of another, resulting in a higher degree of coordination than in pooled interdependence. Coordination in a sequentially interdependent alliance thus goes beyond the pooling of resources to include the order in which the product or service moves from one organization to the other. The partner producing the original product or service has to perform the task as laid out in plans for the alliance, and the subsequent activities in the alliance then have to be performed in a coordinated fashion for the overall strategy to be successful.

Reciprocal interdependence occurs when units come together to exchange outputs with each other simultaneously. Such exchange entails a pooling of resources by different units, but in addition, each unit is simultaneously dependent on the other because its outputs are the other's inputs. In contrast to pooled interdependence, reciprocal interdependence is more interactive and requires ongoing mutual adjustment by both units and continuous adaptation to each other's circumstances. Each unit must continually anticipate the other's output stream and communicate its own production schedule to the other. These efforts require both units to work closely to ensure that there is requisite mutual adaptation and adjustment.

These three types of interdependence can be arrayed on a scale, with reciprocal encompassing the highest coordination costs due to the need for extensive coordination across the partners (Thompson, 1967). Somewhat less uncertain is sequential interdependence. Least uncertain is pooled interdependence, in which close ongoing coordination is not essential, and coordination demands are limited to broadly aligning the activities of the partners toward joint success.

Following prior research, we identified the anticipated interdependence in an alliance at the time of its inception from its underlying logic of value creation (Borys and Jemison, 1989; Zajac and Olsen, 1993). Alliances are usually formed to create value in a way that each partner alone could not. Different logics for value creation require distinctly different levels of coordination between the partners and hence are indicative of different types of interdependence (Borys and Jemison, 1989: 241). For instance, an alliance in which two partners seek to create value by one of them distributing the other's products is likely to have lower coordination and interdependence than another in which the logic for creating value involves both partners coming together to develop a new product. In Thompson's (1967) classic formulation as well, interdependence among units in an organization was embedded in the logic by which they created value through interacting with each other. That is, the logic for value creation led to distinct levels and types of interaction between adjacent units in a value chain and indicated the level of interdependence between those units.

We gauged the anticipated presence of reciprocal, sequential, or pooled interdependence in an alliance from the strategic rationales given by each partner for its participation in the alliance. The rationales provided by each partner for an alliance at the time of its announcement are an excellent indicator for the interdependence they anticipate, because each rationale suggests a distinct logic for value creation that is associated with a specific level of interdependence necessary for accomplishing the rationale. From an extensive review of the alliance literature, we identified eight rationales that provided a comprehensive picture of all the value creation logics of the partners entering an alliance: (1) sharing costs/risks, (2) access to financial resources, (3) sharing complementary technology, (4) reducing the time span of innovation, (5) joint development of new technology, (6) access to new markets, (7) access to new products, and (8) sharing production facilities (Contractor and Lorange, 1988; Hagedoorn, 1993). We assessed these rationales of an alliance from the public announcement and coded them separately for each partner in the alliance. In most instances, we examined multiple announcements in a variety of public sources, including the industry-specific trade journals mentioned above, to gauge this measure accurately. The eight categories are not mutually exclusive, and an alliance could include multiple strategic objectives for any partner. Each alliance was coded as involving reciprocal, sequential, or pooled interdependence, using the classification scheme discussed below. Two dummy variables, reciprocal and sequential, capture this distinction and were used to test hypothesis 1. The default category included instances of pooled interdependence. A comparison of the coefficients of the two dummy variables allowed us to look at the differences in effects across reciprocally and sequentially interdependent alliances.

We classified alliances as reciprocally interdependent if the strategic rationales of the partners included sharing complementary technology, jointly reducing the time needed for innovation, or joint development of new technology. Such alliances include those in which the partners are actively seeking to learn from the alliances to broaden or deepen their skills or to develop new skills jointly, all of which require crucial ongoing inputs from all partners and involve high levels of interdependence. Reciprocally interdependent alliances overlap with but are not synonymous with alliances encompassing a technology component. For instance, not all technology alliances are bilateral learning ties, and some can thus include a unilateral transfer of technological know-how that does not create reciprocal interdependence. Also, reciprocally interdependent alliances may involve the joint development of marketing or distribution skills and not include any technology component.

Sequentially interdependent alliances include partnerships in which the output of one partner is handed off to the other, for whom it is an input. We classified an alliance as involving sequential interdependence both when one partner sought to expand its market access or tap into new markets and did so through an alliance with a partner that had marketing and distribution prowess in those markets and when an alliance involved one partner gaining access to new products provided by the other.

Pooled interdependent alliances exist when alliance partners do not depend on each other for inputs or outputs but, rather, pool resources toward shared activities that need not be coordinated on a regular basis. We classified alliances as involving pooled interdependence when partners came together to share high costs and risks, to share financial resources for expensive endeavors, or to build joint production facilities.

Because the unit of analysis here is the individual alliance and not firms, and all partners usually have a voice in determining the alliance's formal governance structure, we wanted to capture the highest level of interdependence anticipated by the partners entering the alliance. We therefore conservatively coded each alliance with the highest level of interdependence anticipated by either partner within it. Alliances in which a partner had multiple strategic rationales or in which the partners had differing rationales were thus placed in one of the three categories according to the highest level of interdependence among them. As a result, we classified an alliance with elements of both reciprocal and sequential interdependence as reciprocal, one with sequential and pooled interdependence as sequential, and so forth. This coding is consistent with Thompson's notion that the three types of interdependence can be arrayed on a scale in which reciprocal interdependence may include elements of sequential and pooled interdependence, and sequentially interdependent situations may also have some pooled elements.

To ensure that our findings were robust, we also ran our estimations by coding this variable with alternative specifications in which we arrayed all eight original dimensions on a single ordinal scale of interdependence. We did this by first constructing a single variable that took values from 1 to 8 and, second, by introducing seven dummy variables for the eight categories. Our results were consistent with those obtained using Thompson's three-way typology.

Appropriation concerns. We included separate measures for each facet of appropriation concerns partners are likely to anticipate at the time they are forming an alliance: presence of a technology component in the alliance and appropriability regime of the industry. Following prior empirical research, we first included a measure, R&D, to capture the presence of a technology component within the alliance (1 = technology component, 0 = no technology component). R&D alliances included those that encompassed a technology component in the agreement. They could involve exchange, unilateral transfer, sharing, or co-development of technology or elements of all the above. Such alliances could encompass basic R&D, product development, or any other technology-related efforts. Non-R&D alliances typically included those that primarily involved production, distribution, or marketing. We used this variable to test hypothesis 2a on the role of appropriation concerns resulting from the presence of a technology component in determining the governance structure of alliances.

We also assessed the magnitude of appropriation concerns by including measures to capture systematic differences in appropriability regimes across industries. We controlled for sector differences with two dummy variables, new materials and automotive; biopharmaceuticals was the default sector. These variables allowed us to test hypothesis 2b on the role of appropriation concerns resulting from the appropriability regime of the industry in determining the governance structure of alliances. Prior research suggests that biopharmaceuticals has the strongest appropriability regime and automotives the weakest, with new materials lying in between (Levin et al., 1987; Arora and Gambardella, 1994). While it is possible that the strength of appropriability regimes may have changed over our 20-year observation period, our discussion with experts suggests that there have been no dramatic changes in any of these industries to alter the relative levels of appropriability regimes across them. Thus, while there may have been shifts in the absolute levels of the strength of appropriability regimes, relative differences across the three seem to have remained stable. Since our concern is with the relative differentiation across the three, any changes over time should not affect our findings.

We used the interaction between the presence of technology and industry participation to test hypothesis 2c, which predicted that the presence of a technology component in an alliance would moderate the effect of appropriability regimes on the structure of the alliance.

Trust. We included several measures to capture interorganizational trust and test hypothesis 3, suggesting that trust can reduce the likelihood of hierarchical controls in alliances. One mechanism through which such trust is built is through prior alliances (Ring and Van de Ven, 1992; Gulati, 1995a; Gulati and Gargiulo, 1999). The idea of trust emerging from prior contact is based on the premise that through ongoing interaction, firms learn about each other and develop trust around norms of equity, or knowledge-based trust (Shapiro, Sheppard, and Cheraskin, 1992). Prior ties can also promote deterrence-based trust, resulting from viewing prior ties as possible hostages, which deters partners from untrustworthy behavior because they are concerned about potential sanctions, including the dissolution of prior alliances and loss of reputation. Firms having prior alliances with each other will trust each other more than partners who have no prior history with each other. Although it is possible that a prior experience may be a negative one, those firms with bad prior experiences are unlikely to form subsequent alliances with each other. In fact, entering a repeated tie can be a way to mitigate adverse selection problems, because the firms can have reliable firsthand information on each other from prior interactions (Balakrishnan and Koza, 1993). Thus, repeated interaction between two firms can be considered one reasonable indicator of trust between them.

We used an indicator, repeated ties, to record the number of prior alliances the two firms had entered into since 1970 to test hypothesis 3 (0 = first-time alliance). We also examined whether the nature of prior ties (i.e., if they were joint ventures, minority equity investments, or contractual alliances) influenced the governance structure chosen in subsequent alliances. Since this measure could also be capturing experience-related effects resulting from the partners developing routines for working with each other, we included several other measures for trust as well (Nelson and Winter, 1982).

The discussion of trust in alliances has been extended to comparisons of international and domestic and multilateral and bilateral alliances and is used here to further test hypothesis 3. Prior research indicates systematic differences in the behavior of participants in alliances involving partners of different nationalities (Parkhe, 1993b) and also in choices between modes of entry into new geographic markets (Kogut and Singh, 1988; Singh and Kogut, 1989). Recent evidence also suggests systematic differences in the level of patent protection afforded by different countries (Mansfield, 1993). Researchers have argued that cross-border alliances have greater obstacles for building trust and a concomitant higher potential for appropriation concerns than domestic alliances because the difficulties of specifying intellectual property rights, legally enforcing intellectual property, and monitoring partner activities are greater among cross-border firms (Pisano, 1990; Oxley, 1997). As a result, greater trust is expected in domestic alliances than in others. To assess whether alliances between cross-regional partners are likely to have more hierarchical controls and to examine differences across local partner alliances in different global regions, we included three dummy variables, USA, Europe, and Japan, indicating whether an alliance was between partners in those regions. The default was a cross-region alliance (1 = partners of that region, 0 = partners of different regions).

Increasing the number of partners in an alliance can also limit the level of trust between alliance partners (Parkhe, 1993b). Including more partners in an alliance can make identifying and realizing common interests more difficult, which complicates the task of ensuring trust between alliance partners. Furthermore, simply having more partners makes it less likely that all the partners will trust all others in the alliance. Monitoring each partner's contributions and introducing appropriate sanctions in the face of free-riding is harder to implement when there is a large group of participants involved. This makes it difficult to introduce incentive structures that may foster trust. To capture any effects that arose from the number of partners in the alliance, we computed that number and recoded it as a dummy variable, multilateral, with a value of 1 if an alliance was multilateral and a value of 0 if an alliance was bilateral.

Controls. We controlled for temporal trends in alliances and included dummy variables for each year. We included 19 dummy variables for the 20-year period covered in the study, with the default year being 1970. For simplicity of presentation, the results for these dummy variables are not reported in the tables. We also included two control variables, percent joint venture and percent minority investment, assessing the influence of the frequency with which specific types of alliances had been announced in each industry on the choice of governance structure. We counted the number of alliances announced in an industry in the prior year and computed the percentage of those that were of each type. Percent joint venture and percent minority investment capture the percentage of alliances that were joint ventures and minority equity investments, respectively, in the previous year. In a limited way, this calculation tests for the institutionalist claim that firms may mimic the contracts other firms in the industry use to organize their alliances (Fligstein, 1985; Davis, 1991). An alternative interpretation of the variables is that they capture the net effect of the various macro-economic factors within an industry that may influence the formation of particular types of alliances (Amburgey and Miner, 1992).

Table 1 describes the variables included in the analysis and summarizes the predicted signs for the effects of each independent variable.

[TABULAR DATA FOR TABLE 1 OMITTED]

Statistical Model

We assessed the choice between joint ventures, minority equity investments, and contractual alliances with a multinomial logistic regression model. Since the choice among these three alternatives was deemed to be a single step and not nested, we rejected the alternative of using a conditional logit. We later reestimated all models with conditional logit and ordered logit models and found that the results were robust. The general specification of the multinomial logistic regression model applied here was as follows:

ln ([P.sub.i]/[P.sub.o]) = a + [b.sub.j][X.sub.j],

where [P.sub.i] is the probability of an event occurring for the jth case. The two possible events are defined here as a minority equity alliance (i = 1) or a joint venture (i = 2). [P.sub.o] is the probability of a contractual alliance.(5) [X.sub.j] is the vector of independent variables.

RESULTS

Table 2 presents descriptive statistics and correlations for all variables. The descriptive statistics indicate that of the alliances in our sample, 52 percent involved reciprocal interdependence, 24 percent involved sequential interdependence, and the remainder involved pooled interdependence. The total sample of 1,570 alliances included 769 alliances in the new materials industry, 345 alliances in automotives, and 456 alliances in biopharmaceuticals. There was considerable geographic diversity in our sample as well: 27 percent were among U.S. firms, 13 percent were among Japanese firms, 22 percent among European firms, and the remainder were cross-region alliances. Also, 32 percent of the sample involved more than two partners. Of the total alliances, 31 percent were joint ventures, 23 percent were minority equity investments, and the remainder were contractual alliances. Overall, the results point to the diversity of alliances included within the pooled sample of all three industries. The correlations show no significant problems of multicollinearity. The dependent variable is moderately correlated with both dummy variables capturing interdependence (reciprocal and sequential).

[TABULAR DATA FOR TABLE 2 OMITTED]

Table 3 presents the results of the multinomial logistic regression analysis as estimated with LIMDEP 7.0. In this set of analyses we examined the choice between joint ventures, minority equity investments, and contractual alliances. These results provide a detailed assessment of the choices firms make when entering alliances and the factors that may be guiding this choice. Each model estimates coefficients for the choice of minority equity investments and for joint ventures against the default category of contractual alliances. We later compared the sets of coefficients to examine the choice between minority equity investments and joint ventures. Overall, the directionality and significance of the coefficients are consistent with the hypotheses presented here. Furthermore, all models included have significant explanatory power, as demonstrated by the chi-square test on the observed log likelihoods. The negative and significant coefficient for the intercept term suggests that, on average, minority equity investments and joint ventures were used less often than contractual alliances.

Model 1 in table 3 includes the control variables only, and model 2 shows results with the addition of the two measures of interdependence, reciprocal and sequential. Alliances can be arrayed by low to high interdependence as pooled, sequential, and reciprocal. They can have contracts that vary from less to more hierarchical and that range from contractual alliances to minority investments to joint ventures. To test hypothesis 1, we assessed whether alliances with higher levels of interdependence use more hierarchical contracts than those with lower levels by comparing the presence of the three alternative types of interdependence with the use of the three alternative types of governance structure. The results in model 2 support hypothesis 1, which predicted that alliances with higher interdependence [TABULAR DATA FOR TABLE 3 OMITTED] are likely to be organized with more hierarchical contracts than are those with less interdependence. This result holds true in models 3 and 4 as well. The positive coefficients for reciprocal in model 2 indicate that both joint ventures and minority equity investments are more likely than contractual alliances when interdependence is reciprocal than when it is pooled. A t-test of the difference between the coefficients for reciprocal under joint ventures and minority equity investments further supports hypothesis 1 and shows that joint ventures, which encompass the most hierarchical controls, are more likely than minority equity investments when interdependence is reciprocal rather than pooled. The results in model 2 suggest that hierarchical contracts such as minority equity investments and joint ventures are more likely for sequentially interdependent alliances than for alliances with pooled interdependence. The positive coefficients for sequential in model 2 indicate that both joint ventures and minority equity investments are more likely than contractual alliances when interdependence is sequential than when it is pooled. A t-test of the difference between the coefficients for sequential under joint ventures and minority equity investments further supports hypothesis 1 and shows that joint ventures are more likely than minority equity investments when interdependence is sequential rather than pooled. A comparison of the coefficients of reciprocal and sequential in model 2 under joint ventures and minority equity investments further suggests that reciprocally interdependent alliances (vs. those with sequential interdependence) will be more likely to prefer joint ventures or minority equity investments over contractual alliances. Finally, a t-test to compare the coefficients of reciprocal and sequential for joint ventures and minority equity investments shows that joint ventures are more likely than minority equity investments when interdependence is reciprocal rather than sequential.

Overall, not only are the two interdependence indicators significant in predictable ways in our models, but the significant improvement in log likelihood and chi-square statistics in model 2 indicates a much better model fit. This finding clearly indicates the added value of incorporating coordination costs and interdependence into our analysis.

Results of models 3 and 4 support hypotheses 2a-2c. The positive coefficient for R&D supports hypothesis 2a and shows that alliances with a technology component are more likely than those without such a component to be organized with hierarchical governance structures, and under such circumstances, firms will prefer joint ventures and minority equity investments over contractual alliances. A t-test of the difference in coefficients of R&D for joint ventures and minority equity investments shows further support for hypothesis 2a and suggests that firms prefer joint ventures over minority equity investments when an alliance includes a technology component than if it does not.

The industry dummy variables, which indicate the appropriability regime, provide mixed support for hypothesis 2b. We expected more hierarchical controls for alliances in sectors with weaker appropriability regimes. The dummy variable for new materials, where the appropriability regime is intermediate between the three sectors, suggests that compared with biopharmaceuticals, where the appropriability regime is stronger, joint ventures are more likely than contractual alliances in new materials, which is consistent with hypothesis 2b. Contrary to our expectations, however, there is no significant difference in the use of minority equity investments and contractual alliances between new materials and biopharmaceuticals. In the automotive sector, where the appropriability regime is the weakest, consistent with our expectations, both joint ventures and minority equity investments are more likely than contractual alliances when this sector is compared with biopharmaceuticals, where the appropriability regime is strongest. The inclusion of dummy variables for each sector does not reveal whether the remaining main effects differ across the industries. To assess if the other effects observed differ systematically across industries, we estimated unrestricted models for each industry separately (results not reported here). The signs of the coefficients indicated that the postulated directionality and significance of the other main effects observed in the pooled sample did indeed hold true for each sector.

In model 4, we introduced the interaction term between R&D and the dummy variables for industry to test hypothesis 2c and examine whether the effect of industrial sector was more salient for alliances with a technology component than those without one. That is, we tested whether the presence of a technology component in an alliance moderates the influence of the appropriability regime of the industrial sector on the extent of hierarchical controls in alliances. We expected hierarchical controls in alliances to be greatest when alliances have a technology component and are in an industry with a weak appropriability regime. As expected, the positive and significant coefficient for the interaction between automotive and R&D shows that technology-based alliances are more likely to be joint ventures and minority equity investments than to be contractual alliances in the automotive sector, where appropriability regimes are relatively weak, than they are in biopharmaceuticals, where appropriability regimes are stronger. Contrary to our expectations, the statistically insignificant coefficient for the interaction between new materials and R&D suggests that technology alliances in the new materials sector with an intermediate appropriability regime are no different in their governance structure than in the biopharmaceutical sector, where the appropriability regime is strongest.

Model 3 also includes several measures of the extent of trust developed among alliance partners. The negative coefficient for repeated ties, a measure of interorganizational trust, supports hypothesis 3 and indicates that repeated ties are less likely than first-time alliances to be organized as joint ventures or minority equity investments than as contractual alliances. A comparison of the coefficients further supports hypothesis 3 and suggests that repeated ties are less likely to be organized as joint ventures than as minority equity investments. We also assessed the role of repetition of different types of alliances and found that regardless of type of prior ties, repeated ties were less likely than first-time alliances to be joint ventures or minority equity investments (results not reported here).

We also introduced dummy variables for nationality of partner and whether alliances were multilateral or bilateral to further assess the role of trust on governance structure proposed in hypothesis 3. The negative and significant coefficient for Japan and Europe suggests that alliances involving firms from only those regions are less likely than cross-regional alliances to use joint ventures and minority equity investments than contractual alliances. This is consistent with our intuition that there is likely to be greater trust in alliances involving regionally similar partners than cross-regional partners, which in turn is reflected in the governance structure of the alliances. The insignificant coefficient for USA suggests that alliances between American partners are no different from cross-regional alliances in their governance structure and is contrary to our expectations in hypothesis 3. In addition, contrary to hypothesis 3, the positive but insignificant coefficients for multilateral across all models indicates that multilateral alliances are statistically no more likely than bilateral alliances to be joint ventures or minority equity investments than to be contractual alliances.

The first column of table 3 also reports the base model including all the control variables. Our estimations included a dummy variable for each year but one in all models. The results for these dummy variables (not reported in tables for ease of presentation) broadly confirm the growing frequency of minority investments and contractual alliances. In addition, the positive and significant coefficient for percent joint ventures and percent minority equity investments suggests that the use of these alliances positively affects firms' use of those types of alliances in that industry.

To assess the possible influence of firm-level attributes on the choice of governance structure, we conducted a separate analysis with data on all alliances formed by a subgroup of firms. Given the diversity of firms in our sample from the U.S., Europe, and Japan, it was not possible to collect financial information on all firms. For the subgroup analysis, we selected the 50 largest firms in each sector and collected information on their size, performance, liquidity, and solvency. We reestimated all our models with this sample and included variables to assess the role of partner differences by computing a ratio of the smaller to the larger financial item to assess the effect of differences across partners on their choice of governance structure. The results (not reported here) for coordination costs and appropriation concerns after controlling for these firm attributes were consistent with those we obtained with our original sample. The fact that we still observed our hypothesized findings gives us greater confidence in our results. Among the ratios themselves, the only one that was significant was size, indicating that the greater the difference in size between the partners, the more likely they were to use joint ventures or minority equity investments than contractual alliances.

DISCUSSION

The findings in this study shed light on some of the factors that underlie how firms choose between the diversity of contracts available to formalize their alliances. Using a typology of three types of alliance structure and the magnitude and type of hierarchical controls present in each, we found that both the extent of coordination costs and appropriation concerns in an alliance can predict the use of particular governance structures. By directly modeling the influence of anticipated coordination costs on the governance structure of alliances after accounting for concerns about appropriation, the study provides a window into the multiple logics used by alliance participants in determining the governance structure used to formalize alliances.

The results provide strong support for the importance of coordination costs. An important finding was that the greater the anticipated coordination costs arising from interdependence associated with a strategic alliance at the time of its formation, the more hierarchical was the governance structure used to formalize it. Our findings confirm that reciprocally interdependent alliances are likely to have structures with greater hierarchical control than those with sequential interdependence, which in turn are likely to have more hierarchically organized alliances than those with pooled interdependence. This result suggests that the deliberations underlying the choice of alliance structure are not dominated by concerns of appropriation alone, as previously suggested, and that considerations associated with managing coordination costs resulting from the interdependence of tasks across partners are also salient. Such considerations have yet to be examined for interorganizational relationships and, as our results suggest, they merit serious consideration. Given that coordination costs are a key element in the choice of alliances, they may influence the fundamental choice of firm boundaries as well and could be an exciting topic for future research.

We also examined the role of appropriation concerns and behavioral uncertainty, highlighted in previous research, in guiding the choice of alliance structure. The results provide mixed support for the appropriation hypothesis. Consistent with previous work (Pisano, 1989), alliances involving a technology component were likely to use more hierarchical structures than those that did not involve one. Contrary to our expectations, however, the differences across industrial sectors, which also reflect varying appropriability regimes, do not entirely explain the choice of governance structure. This is problematic in the comparison across new materials and biopharmaceuticals, and the results show that while joint ventures are more likely than contractual alliances in the former than in the latter, there is no significant difference in the use of minority equity investments and contractual alliances. The interpretation of this null result is not unambiguous, as this result could be influenced by additional unmeasured factors, such as localized institutional norms, historical imprinting of behavior by industry participants, and the intensity, diversity, and niche-based dynamics of competition in those industries.

The results for appropriation concerns were also confirmed by looking at the simultaneous influence of the sector in which the alliance occurred and the presence of a technology component in the alliance. The results suggest that the presence of a technology component in alliances enhances the influence of the appropriability regime of the industry on the governance structure used. In particular, the combination of a technology component and an alliance in a sector with a weak appropriability regime increases the likelihood of firms choosing hierarchical governance structures.

We used several measures to assess the influence of interfirm trust as another factor that can affect the choice of governance structure of alliances because it can address not only appropriation concerns but also potential concerns about coordination costs. Results for the first indicator of trust, the presence of repeated ties, are consistent with our expectations: repeated ties diminish the use of hierarchical controls in alliances. This result holds true even after we separated out the history of alliances by specific types of alliance. Thus, a prior history between the firms matters, regardless of the type of prior alliance the firms entered.

The results for the regional origin of partners, which we also proposed would capture trust, reveal some interesting trends. The comparison between local versus cross-region alliances was broadly consistent with our expectations of greater trust in local than in cross-regional alliances, but breaking down local alliances by region suggests some provocative issues not fully explored here. While the results for European alliances are consistent with our predictions, contrary to our expectations, Japanese domestic alliances are no different from cross-regional alliances in their use of minority equity investments or contractual alliances, though they do differ in their use of joint ventures or contractual alliances. Even more conspicuous is the absence of significant differences in the governance structure of alliances between American partners and cross-regional alliances. While these results suggest some systematic differences in the level of trust between local and cross-regional alliances that is reflected in the governance structure of alliances, several alternative interpretations are possible for the results for each region. These differences may be the result of appropriation concerns resulting from greater difficulties in specifying and enforcing property rights and monitoring problems in cross-regional alliances than in local alliances, or they may be due to the greater coordination challenges and coordination costs in cross-regional alliances than in local alliances. Local alliances in each region may also be influenced by localized institutional contexts deeply embedded in normative practices and authority structures (Hamilton and Biggart, 1988). Or perhaps there are historical and legal circumstances that mandate or encourage the use of particular governance structures for alliances. These results, along with those for sectoral differences, reveal some interesting patterns that remain to be explored in future research.

Overall, the results for the effect of prior trust confirm that it is important to consider alliances between firms as occurring within a rich social context in which firms are embedded (Gulati, 1998). This context channels valuable information between firms and can thus influence not only the formation of new alliances and the choice of partners (Gulati, 1995b) but also the specific structures used to formalize alliances. While this study has considered the implications of relational embeddedness resulting from the direct ties in which firms are placed, it would also be instructive to explore the role of structural embeddedness resulting from the positions firms occupy in the overall network on the governance structure of alliances (Granovetter, 1992). This could encompass not only an examination of firm location in the interorganizational network but also consideration of the extent to which the network itself has become differentiated (Gulati and Gargiulo, 1999).

Another facet not considered is the influence of the social structure of dependence between partners on the governance structure of alliances. A prominent line of research has outlined the social structure of resource interdependence as an important determinant of tie formation between firms (Pfeffer and Nowak, 1976; Burt, 1982; Oliver, 1990; Gulati, 1995b; Chung, Singh, and Lee, 1998; Dyer and Singh, 1998; Gulati and Gargiulo, 1999). This suggests that resource considerations and a quest for complementary skills are powerful propellants for the formation of new ties between firms, including joint ventures. While this research has pointed to the catalysts for new alliances, it has neglected to examine the factors that may influence the choice among an array of contracts possible for an alliance. Partners in alliances may indeed seek to mitigate concerns of dependence and power arising from resource asymmetries by using particular governance structures (Baker, 1990; Aghion and Tirole, 1994).

The results of the control variables suggest that both the time trend and the frequency with which other industry participants used particular structures also influence the choice of governance structure in alliances. The positive influence of frequency of prior alliances by industry participants on the choice of structure in an alliance reflects the likely occurrence of imitation or of industry imperatives not captured by other variables included in our study (Westphal, Gulati, and Shortell, 1997). This effect disappeared, however, once we introduced the hypothesized variables in our framework. The findings in this study help us understand some of the reasons why the composition of structures used may have shifted over time from more to less hierarchical. In particular, the insignificance of most of the dummy variables capturing time, once we introduced the measures for interdependence, suggest that the time trend may have been capturing the changing patterns in underlying interdependence in alliances. Thus, a likely explanation for the observed time trend may be that firms are entering alliances with lower levels of interdependence.

This study focused on the decision firms make at the outset of an alliance about its governance structure. Our discussion with legal experts who have been involved in the creation of many such alliances confirmed that firms entering alliances do indeed face such a decision once they have agreed on the logic of value creation and the scope of activities to be included within the partnership. There are numerous other questions that arise, however, once we consider the dynamic evolution of the alliance. It would be useful to examine how the governance structure of alliances changes as the expectations and goals of partners in the alliance evolve (Gulati, Khanna, and Nohria, 1994; Doz, 1996; Khanna, Gulati, and Nohria, 1998), as well as the extent of asset-specific investments undertaken by the partners and how these may change over time.

Our results also have implications for fundamental questions about the alternative bases for hierarchical controls in alliances and for the design of organizations. We have suggested that hierarchical controls arise in alliances when participants anticipate either coordination costs or appropriation concerns. The implications of hierarchical controls on the ultimate success or failure of alliances remains an interesting but separate question. Addressing this would entail not only a precise specification of conditions under which hierarchical controls may be beneficial, it would also involve a clear definition of performance for alliances, which itself can be problematic (Gulati, Kumar, and Zajac, 1998). Furthermore, it could be useful to separate the enabling and coercive elements within hierarchical controls that could promote or inhibit the ultimate success of alliances (Adler and Borys, 1996).

Powell (1990) and others have suggested that hybrid forms such as alliances may not necessarily be a halfway house between the twin pillars of market and hierarchy but, instead, could be considered distinctive forms of governance in their own right. This study focuses exclusively on alliances and does not examine the entire spectrum of exchanges from market to hierarchy. While it utilizes hierarchical controls as a key dimension to distinguish among alliances, this does not imply that alliances as a whole are situated between markets and hierarchies on a single scale. The place of alliances in the realm of organizational forms is an important question and must remain the object of future theoretical and empirical inquiry.

While this study focused on the origin of hierarchical controls in alliances, its findings can provoke an examination of the importance of coordination costs for the question of why firms exist. Recent efforts to question the singular importance of opportunism-based transaction costs economics have introduced the role of knowledge and its application to business activities as a basis for why firms exist (Conner and Prahalad, 1996). This work could easily be expanded to examine the role of coordination costs as an important basis for why firms exist and their influence on the scale and scope of firms.

Together, our findings lend credence to our typology of alliance governance structure, in which we arrayed alliances from less to more hierarchical. The systematic evidence that the choice between the three alternative alliance structures is influenced significantly and in theoretically predictable ways by both interdependence and appropriation concerns enhances our claim that these categories indicate distinctive types of alliances, each of which incorporates specific levels of hierarchical controls. The distinction we offer is not only theoretically relevant, it is also parsimonious and easy to operationalize. In prior research, numerous schemes to classify alliances have been offered, but when it comes time to translate them into operational definitions with which to classify data, they fall prey to arbitrary and overlapping categorizations. In this paper, we present a distinction that is both robust and theoretically relevant and that we hope will help to guide other scholars. Nonetheless, numerous directions for extending this typology are also possible. It would be useful, for instance, to examine the extent to which there may be differences in the contracts within each of these categories. The five facets of hierarchical controls in alliances that we introduced can provide an informative set of dimensions along which this variation could be assessed. This would entail looking at the actual covenants of the contracts and observing the extent to which each of the five elements may be present in them (Stinchcombe, 1985).

In this study we have considered two perspectives that suggest differing circumstances that influence the extent of hierarchical controls in alliances: an economic approach, which highlights appropriation concerns, and an organizational approach, which emphasizes coordination costs. Overall, the results have several ramifications. First, they suggest that the formation of exchange relations between firms is not entirely dominated by appropriation concerns. Coordination costs, arising from decomposing tasks between partners and the requisite ongoing coordination and management of tasks across partners, are also important and play at least an equally significant role. Underlying this study is the fundamental question about the relative importance of alternative bases for hierarchical controls in alliances. Both perspectives discussed point to the salience of bounded rationality and concomitant uncertainty as important considerations for the emergence of hierarchical controls. But each specifies the role of different facets of uncertainty experienced by alliance participants at the time of forming an alliance as important in their decision to introduce hierarchical controls in the formal structure. While one perspective highlights the salience of anticipated appropriation concerns resulting from contracting hazards and manifest behavioral uncertainty, the other points to uncertainty arising from the anticipation of the extent of ongoing task coordination and the complexity of decomposing the division of tasks. We do not claim that our perspective, highlighting the role of coordination costs, constitutes the exclusive explanation for hierarchical controls in alliances; hence, we investigate the simultaneous influence of both sets of factors. There can be overlaps in the occurrence of coordination costs and appropriation concerns. As a result, it can be difficult to distinguish between the influence of these two empirically. While it may be difficult to isolate the occurrence of each it its pure form in the absence of the other, as was true with our hypothetical example discussed earlier, we empirically demonstrate the distinct role of coordination costs in guiding the choice of governance structure in alliances.

Second, the results in this study highlight the fundamental issue of the origin of hierarchical controls in alliances: whether they arise from a concern with coordination costs and interdependence across partners or result from anticipated appropriation concerns. These results show that both sets of factors are important considerations for alliances. Firms choose governance structures both to manage anticipated coordination costs and to address appropriation concerns. This finding is consistent with our belief that hierarchical controls are more than mechanisms to control opportunism and provide incentive alignment across partners; they also provide an organizational context that determines the rules of the game and creates an administrative architecture within which the partnership proceeds. This architecture provides alliance partners with the ability to coordinate tasks and responsibilities between themselves in a way that meets their own needs for value creation and allays their particular concerns about the alliance. Our findings thus begin to explain the variety of alliances that firms form with their partners and the different structures they build together.

We thank seminar participants at the Academy of Management (1997) and Northwestern University, the anonymous reviewers for ASQ, Associate Editor Mark Mizruchi, and the following individuals for helpful comments in revising this paper: Christina Ahmedjian, Gautam Ahuja, Jeff Dyer, Martin Gargiulo, Heather Haveman, Tarun Khanna, Michael Leiblein, Nitin Nohria, Willy Ocasio, Joanne Oxley, Woody Powell, Huggy Rao, Mark Shanley, Brian Silverman, Art Stinchcombe, James Westphal, and Ed Zajac. We would also like to thank Linda Johanson for her able editorial guidance. Financial support from the J. L. Kellogg Graduate School of Management and the Wharton School is gratefully acknowledged.

1 This concept is distinct from transaction cost economists' coordination costs, which refer to the agency costs resulting from the growth of organizations, which provide "decreasing returns to the entrepreneur function" (Coase, 1937: 340). Such costs, also referred to as organization (Masten, Meehan, and Snyder, 1991) or management costs (Demsetz, 1988), are the costs of organizing resources within firm boundaries (Alston and Gillespie, 1989) and are not easily applicable to exchange relationships such as alliances. Our concept is akin to Williamson's (1991) in his discussions of different forms of adaptation, although he focuses primarily on the requisite coordination for adapting activities to exogenous disturbances and on appropriation concerns and the incentive alignment requirements rather than the administrative challenges of coordination itself.

2 Davis, Kahn, and Zald (1990) connected governance structure and interdependence in the context of interactions between nation states and across organizations using research on interorganizational ties as an analogue to illuminate how nation states behave. Gerlach and Palmer (1981) looked at changing levels of sociopolitical interdependence in societal evolution and explored the antecedents of interdependence as well as its consequences for the emergence of new governance institutions.

3 Teece (1992: 8) acknowledged that an innovator's quest for complementary assets can lead to varying degrees of interdependence and co-specialization, but he focused on the relative degree of mutual dependence resulting from specialization of assets for the alliance; the greater the mutual dependence, the larger the co-specialization. In contrast, our usage of interdependence focuses on the partners' anticipation of the extent of complexity in decomposing tasks and the degree to which it will entail ongoing mutual adjustment and adaptation to accomplish joint tasks, which is akin to Teece's discussion of "coupling."

4 Strictly speaking, behavioral uncertainty is considered to be exogenous and ubiquitous, and it is the concomitant occurrence of contracting hazards such as asset-specific investments and appropriability concerns that aggravate the challenges of writing contracts to cope with sequential adaptation and lead to appropriation or moral hazard concerns (Williamson, 1991). While behavioral uncertainty is treated as a given, it is an important assumption that creates a necessary but perhaps not sufficient condition for moral hazards.

5 In applying the multinomial logistic regression model, we tested for the independence of irrelevant alternatives by applying the Hausman and McFadden (1984) test. Results suggest that the null hypothesis cannot be rejected and that all three choices are indeed independent of one another.

REFERENCES

Adler, Paul S., and Bryan Borys

1996 "Two types of bureaucracy: Enabling and coercive." Administrative Science Quarterly, 41: 61-89.

Aghion, Phillipe, and Jean Tirole

1994 "On the management of innovation." Quarterly Journal of Economics, 109:1185-1207.

Alchian, Armen, and Harold Demsetz

1972 "Production, information costs, and economic organization." American Economic Review, 62: 777-795.

Alston, Lee J., and William Gillespie

1989 "Resource coordination and transaction costs." Journal of Economic Behavior and Organization, 11: 191-212.

Amburgey, T. L., and A. S. Miner

1992 "Strategic momentum: The effects of repetitive, positional, and contextual momentum on merger activity." Strategic Management Journal, 13: 335-348.

Anand, Bharat, and Tarun Khanna

1997 "Intellectual property rights and contract structure." Harvard Business School Working Paper, 97-016.

Anderson, Erin

1990 "Two firms, one frontier: On assessing joint venture performance." Sloan Management Review, 31: 19-31.

Arora, Ashish, and Alfonso Gambardella

1994 "The changing technology of technological change: General and abstract knowledge and the division of innovative labor." Research Policy, 23: 523-532.

Arrow, K.

1974 The Limits of Organization. New York: W. W. Norton.

Ashby, W. Ross

1956 Introduction to Cybernetics. London: Chapman and Hall.

Baker, Wayne E.

1990 "Market networks and corporate behavior." American Journal of Sociology, 96: 589-625.

Balakrishnan, Srinivisan, and Mitchell P. Koza

1993 "Information asymmetry, adverse selection and joint ventures: Theory and evidence." Journal of Economic Behavior and Organization, 20: 99-117.

Barnard, Chester I.

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

Barzel, Y.

1982 "Measurement costs and the organization of markets." Journal of Law and Economics, 25: 27-48.

Blau, Peter M.

1972 "Interdependence and hierarchy in organizations." Social Science Research, 1: 1-24.

Borys, Bryan, and David B. Jemison

1989 "Hybrid arrangements as strategic alliances: Theoretical issues in organizational combinations." Academy of Management Review, 14: 234-249.

Bradach, J. L., and R. G. Eccles

1989 "Markets versus hierarchies: From ideal types to plural forms." In W. R. Scott (ed.), Annual Review of Sociology: 97-118. Palo Alto, CA: Annual Reviews.

Burt, Ronald

1982 Toward a Structural Theory of Action. New York: Academic Press.

Chandler, Alfred D., Jr.

1977 The Visible Hand: The Managerial Revolution in American Business. Cambridge, MA: Harvard University Press.

Chung, Suenghwa, Harbir Singh, and Kyungmook Lee

1998 "Compatibility of capabilities and reputational capital as drivers of alliance formation." Strategic Management Journal (in press).

Coase, R. H.

1937 "The nature of the firm." Reprinted in G. J. Stigler and K. E. Boulding (eds.), A.E.A. Readings in Price Theory: 331-351. Homewood, IL: Irwin.

Conner, Kathleen, and C. K. Prahalad

1996 "A resource-based theory of the firm: Knowledge versus opportunism." Organization Science, 7: 477-501.

Contractor, Farok, and Peter Lorange

1988 Cooperative Strategies in International Business. Lexington, MA: Lexington Books.

Davis, Gerald F.

1991 "Agents without principles? The spread of the poison pill through the intercorporate network." Administrative Science Quarterly, 36: 583-613.

Davis, Gerald F., Robert Kahn, and Mayer Zald

1990 "Contracts, treaties, and joint ventures." In Robert L. Kahn, and Mayer N. Zald (eds.), Organizations and Nation States: New Perspectives on Conflict and Cooperation. San Francisco: Jossey-Bass.

Demsetz, Harold

1988 "The theory of the firm revisited." Journal of Law, Economics, and Organization, 4: 141-161.

Doz, Yves

1996 "The evolution of cooperation in strategic alliances: Initial conditions or learning processes?" Strategic Management Journal, Special Issue, 17: 55-83.

Doz, Yves, Gary Hamel, and C. K. Prahalad

1989 "Collaborate with your competitors and win." Harvard Business Review, 67: 133-139.

Dyer, Jeffrey H., and Harbir Singh

1998 "The relational view: Cooperative strategy and sources of interorganizational competitive advantage." Academy of Management Review, 23: 660-679.

Emery, Fred E., and Eric L. Trist

1965 "The causal texture of organizational environments." Human Relations, 18: 21-32.

Fligstein, Neil

1985 "The spread of the multidivisional form among large firms, 1919-1979." American Sociological Review, 50: 377-391.

Galbraith, Jay R.

1977 Organization Design. Reading, MA: Addison-Wesley.

Gerlach, Luther P., and Gary B. Palmer

1981 "Adaptation through evolving interdependence." In Paul C. Nystrom, and William H. Starbuck (eds.), Handbook of Organizational Design, 1: 323-384. Oxford: Oxford University Press.

Ghoshal, Sumantra, and Peter Moran

1996 "Bad for practice: A critique of the transaction cost theory." Academy of Management Review, 21: 13-48.

Granovetter, Mark

1985 "Economic action and social structure: A theory of embeddedness." American Journal of Sociology, 91: 481-510.

1992 "Problems of explanation in economic sociology." In N. Nohria and R. Eccles (eds.), Networks and Organization: Structure, Form and Action: 25-56. Boston: Harvard Business School Press.

Gulati, Ranjay

1995a "Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances." Academy of Management Journal, 38: 85-112.

1995b "Social structure and alliance formation patterns: A longitudinal analysis." Administrative Science Quarterly, 40: 619-652.

1998 "Alliances and networks." Strategic Management Journal, 19: 293-317.

Gulati, Ranjay, and Martin Gargiulo

1999 "Where do interorganizational networks come from?" American Journal of Sociology (in press).

Gulati, Ranjay, Tarun Khanna, and Nitin Nohria

1994 "Unilateral commitments and the importance of process in alliances." Sloan Management Review, 35 (3): 61-69.

Gulati, Ranjay, N. Kumar, and Edward Zajac

1998 "Interorganizational dynamics and their impact on alliance performance and stability." Working paper, Department of Organization Behavior, J. L. Kellogg Graduate School of Management, Northwestern University.

Hagedoorn, John

1993 "Understanding the rationale of strategic technology partnering: Interorganizational modes of cooperation and sectoral differences." Strategic Management Journal, 14: 371-385.

Hamilton, Gary, and Nicole Biggart

1988 "Market, culture, and authority." American Journal of Sociology (Supplement), 94: S52-S94.

Harrigan, Kathryn R.

1986 Managing for Joint Ventures Success. Lexington, MA: Lexington Books.

Hausman, J. A., and D. L. McFadden

1984 "Specification tests for the multinomial logit model." Econometrica, 52: 1219-1240.

Hennart, Jean-Francois

1988 "A transaction costs theory of equity joint ventures." Strategic Management Journal, 9: 361-374.

Herriott, Scott R.

1996 "Control of strategic alliances through partial acquisitions." Working paper, Department of Organization Behavior, Maharishi International University.

Katz, Daniel, and Robert L. Kahn

1966 The Social Psychology of Organizations. New York: Wiley.

Khanna, Tarun, Ranjay Gulati, and Nitin Nohria

1998 "The dynamics of learning alliances: Competition, cooperation, and relative scope." Strategic Management Journal, 19: 193-210.

Klein, Benjamin, Robert Crawford, and Armen Alchian

1978 "Vertical integration, appropriable rents and the competitive contracting process." Journal of Law and Economics, 21: 297-326.

Kogut, Bruce, and Harbir Singh

1988 "The effect of national culture on the choice of entry mode." Journal of International Business Studies, 19: 319-332.

Lawrence, P. R., and J. W. Lorsch

1967 Organization and Environment. Boston: Harvard Business School Press.

Levin, Richard, Alvin Klevorick, Richard Nelson, and Sidney Winter

1987 "Appropriating the returns from industrial research and development." Brookings Papers on Economic Activity, 3: 783-820.

Levine, Sol, and Paul E. White

1961 "Exchange as a conceptual framework for the study of interorganizational relationships." Administrative Science Quarterly, 5: 583-601.

Litwak, Eugene, and Lydia F. Hylton

1962 "Interorganizational analysis: A hypothesis on coordinating agencies." Administrative Science Quarterly, 6: 395-420.

Macauley, S.

1963 "Non-contractual relations in business: A preliminary study." American Sociological Review, 28: 55-67.

Mansfield, Edwin

1993 "Unauthorized use of intellectual property: Effects on investment, technology transfer, and innovation." In Michael Wallerstein, Mary Ellen Mogee, and Roberta Schoen (eds.), Global Dimensions of Intellectual Property Rights in Science and Technology: 107148. Washington, DC: National Academy Press.

March, James G., and Herbert A. Simon

1958 Organizations. New York: Wiley.

Masten, Scott E., James W. Meehan, and Edward A. Snyder

1991 "The costs of organization." Journal of Law, Economics, and Organization, 7: 1-25.

McCann, Joseph, and Jay R. Galbraith

1981 "Interdepartmental relations." In Paul C. Nystrom, and William H. Starbuck (eds.), Handbook of Organizational Design, 2: 60-84. Oxford: Oxford University Press.

Merges, R., and R. Nelson

1990 "On the complex economics of patent scope." Columbia Law Review, 90: 839-870.

Monteverdee, Kirk, and David Teece

1982 "Supplier switching costs and vertical integration in the auto industry." Bell Journal of Economics, 13: 206-213.

Moon, John J., and Tarun Khanna

1995 "Product market considerations in private equity sales." Working paper, Harvard Business School.

Mowery, David, and D. Rosenberg

1989 Technology and the Pursuit of Economic Growth. New York: Cambridge University Press.

Nelson, Richard, and Sidney Winter

1982 An Evolutionary Theory of Economic Change. Cambridge, MA: Belknap Press.

Oliver, Christine

1990 "Determinants of interorganizational relationships: Integration and future directions." Academy of Management Review, 15: 241-265.

Oxley, Joanne E.

1997 "Appropriability hazards and governance in strategic alliances: A transaction cost approach." Journal of Law, Economics, and Organization, 13: 387-409.

Parkhe, Arvind

1993a "Strategic alliance structuring: A game theoretic and transaction cost examination of interfirm cooperation." Academy of Management Journal, 36: 794-829.

1993b "Partner nationality and the structure-performance relationship in strategic alliances." Organization Science, 4: 301-324.

Pfeffer, Jeffrey, and Philip Nowak

1976 "Joint ventures and interorganizational interdependence." Administrative Science Quarterly, 21: 398-418.

Pisano, G. P.

1989 "Using equity participation to support exchange: Evidence from the biotechnology industry." Journal of Law, Economics, and Organization, 5: 109126.

1990 "The R&D boundaries of the firm: An empirical analysis." Administrative Science Quarterly, 35: 153-176.

Pisano, G. P., Michael V. Russo, and David Teece

1988 "Joint ventures and collaborative agreements in the telecommunications equipment industry." In David Mowery (ed.), International Collaborative Ventures in U.S. Manufacturing: 23-70. Cambridge, MA: Ballinger.

Pondy, Louis R.

1970 "Toward a theory of internal resource allocation." In Mayer N. Zald (ed.), Power in Organizations: 270-311. Nashville, TN: Vanderbilt University Press.

1977 "The other hand clapping: An information-processing approach to organizational power." In Tove H. Hammer, and Samuel B. Bacharach (eds.), Reward Systems and Power Distribution in Organizations: 56-91. Ithaca, NY: Cornell University Press.

Powell, Walter W.

1990 "Neither market nor hierarchy: Network forms of organization." In B. M. Staw and L. L. Cummings (eds.), Research in Organizational Behavior, 12: 295-336. Greenwich, CT: JAI Press.

Ring, Peter Smith, and Andrew H. Van de Ven

1992 "Structuring cooperative relationships between organizations." Strategic Management Journal, 13: 483-498.

Shapiro, D. L., B. H. Sheppard, and L. Cheraskin

1992 "In theory: Business on a handshake." Negotiation Journal, 8: 365-377.

Silver, M.

1984 Enterprise and the Scope of the Firm. Oxford: Martin Robertson Press.

Singh, Harbir, and Bruce Kogut

1989 "Industry effects on the choice of entry mode." Best Papers and Proceedings of the Academy of Management: 116-121.

Stinchcombe, Arthur L.

1985 "Contracts as hierarchical documents." In A. Stinchcombe and C. Heimer (eds.), Organization Theory and Project Management: 121-171. Bergen, Norway: Norwegian University Press.

Teece, David J.

1980 "Economies of scope and the scope of the enterprise." Journal of Economic Behavior and Organization, 1: 223-247.

1986 "Profiting from technological innovation: Implications for integration, collaboration, licensing, and public policy." Research Policy, 15: 285-305.

1992 "Competition, cooperation, and innovation." Journal of Economic Behavior and Organization, 18: 1-25.

Thompson, James D.

1967 Organizations in Action: Social Science Bases of Administration. New York: McGraw-Hill.

Van de Yen, Andrew H.

1976 "On the nature, formation and maintenance of relations among organizations." Academy of Management Review, 1: 24-36.

Walker, Gordon, and David Weber

1984 "A transaction cost approach to make-or-buy decisions." Administrative Science Quarterly, 29: 373-391.

Westphal, James, Ranjay Gulati, and Stephen Shortell

1997 "Customization or conformity? An institutional and network perspective on the content and consequences of TQM adoption." Administrative Science Quarterly, 42: 366-394.

Williamson, O. E.

1975 Markets and Hierarchies: Analysis and Antitrust Implications. New York: Free Press.

1985 The Economic Institutions of Capitalism. New York: Free Press.

1991 "Comparative economic organization: The analysis of discrete structural alternatives." Administrative Science Quarterly, 36: 269-296.

Winter, S.

1964 "Economic 'natural selection' and the theory of the firm." Yale Economic Essays, 4: 225-272.

Zajac, Edward J., and C. P. Olsen

1993 "From transaction cost to transactional value analysis: Implications for the study of interorganizational strategies." Journal of Management Studies, 30: 131-145.

Zucker, Lynne G.

1986 "Production of trust: Institutional sources of economic structure, 1840-1920." In B. M. Staw and L. L. Cummings (eds.), Research in Organizational Behavior, 8: 53-111. Greenwich, CT: JAI Press.

Ranjay Gulati [coauthor, "The Architecture of Cooperation: Managing Coordination Costs and Appropriation Concerns in Strategic Alliances"] is an associate professor of Organization Behavior at the J. L. Kellogg Graduate School of Management and adjunct associate professor of sociology at Northwestern University, Evanston, IL 60208-2001 (e-mail: r-gulati@nwu.edu). His research interests include the dynamics of social networks, with a focus on interorganizational embeddedness and strategic alliances. His forthcoming articles include "The Dark Side of Embeddedness: An Examination of the Influence of Direct and Indirect Board Irterlocks and CEO/Board Relationships on Interfirm Alliances," with James Westphal (Administrative Science Quarterly, forthcoming), and "Network Location and Learning: The Influence of Network Resources and Firm Capabilities on Alliance Formation" (Strategic Management Journal, forthcoming, 1999). He is currently researching the relational attributes of embedded ties and their influence on performance in the American automotive industry. He received his Ph.D. in organizational behavior from Harvard University.

Harbir Singh [coauthor, "The Architecture of Cooperation: Managing Coordination Costs and Appropriation Concerns in Strategic Alliances"] is a professor of management and chair of the Management Department at The Wharton Business School, University of Pennsylvania, Philadelphia, PA 19104 (e-mail: Singhh@wharton.upenn.edu). His research interests include firms' capabilities to manage acquisitions and alliances and the consequences of corporate restructuring. Recent publications include "When Does Corporate Restructuring Improve Economic Performance?," with Edward H. Bowman, Michael Useem, and Raja Bhadury (California Management Review, January 1998), "National Cultural Distance and Cross-Border Acquisition Performance," with Piero Morosini and Scott Shane (Journal of International Business Studies, March 1998), "Asset Redeployment, Acquisitions and Corporate Strategy in Declining Industries," with Jaideep Anand (Strategic Management Journal, 18:99-118), and "Corporate Restructuring: A Symptom of Poor Governance or a Solution to Past Managerial Mistakes?," with Constantinos Markides (European Management Journal, 15: 213-219). He earned his doctorate in strategic management from the University of Michigan.

COPYRIGHT 1998 Cornell University, Johnson Graduate School
COPYRIGHT 2000 Gale Group