Adaptation In Vertical Relationships

Strategic Management Journal Strat. Mgmt. J., 26: 415–440 (2005) Published online 11 March 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/smj.458

ADAPTATION IN VERTICAL RELATIONSHIPS: BEYOND INCENTIVE CONFLICT RANJAY GULATI,1 * PAUL R. LAWRENCE2 and PHANISH PURANAM3 1 2 3

Kellogg School of Management, Northwestern University, Evanston, Illinois, U.S.A. Harvard Business School, Harvard University, Boston, Massachusetts, U.S.A. London Business School, University of London, London, U.K.

In this study, we extend the analysis of adaptation in theories of economic organization beyond traditional considerations of incentive conflict (hold-up). We conceptualize adaptation as coordinated and cooperative response to change, and define the adaptive capacity of a vertical relationship as the ability to generate coordinated and cooperative responses across procurer and supplier to changes in procurement conditions. We draw on the concepts of differentiation and integration to dimensionalize the adaptive capacity of different modes of procurement. Using data on all component classes procured internally and externally by Ford and Chrysler, we show that different procurement modes differ in terms of their adaptive capacity and performance. We also show that performance differences across modes of procurement arise as a function of the match between adaptive capacity and adaptation requirements associated with the exchange, and not only the match between governance form and transaction hazards. Copyright  2005 John Wiley & Sons, Ltd.

INTRODUCTION Vertical (procurement) relationships have always been the favorite empirical domain of theorists of economic organization (Coase, 1937; Grossman and Hart, 1986; Williamson, 1975). Such relationships involve exchange between adjacent stages of the value chain, and they occur both within firms (e.g., between different functional or divisional areas within a firm) and between firms (e.g., between specialist design firms and specialist manufacturers). In recent years, the study of vertical relationships has come to be dominated by transaction cost economics (see Shelanski and Klein, 1995; David and Han, 2004, for reviews). Keywords: coordination; organization design; differentiation and integration; auto industry

∗ Correspondence to: Ranjay Gulati, Kellogg School of Management, Northwestern University, Jacobs Center, 2001 Sheridan Road, Evanston, IL 60208-2001, U.S.A. E-mail: [email protected]

Copyright  2005 John Wiley & Sons, Ltd.

In Oliver Williamson’s development of the theory, the focus is on ‘how parties engaged in a longterm contract can adapt effectively to disturbances. The need to craft contractual structures in which they have mutual confidence . . .’ is the key issue (Williamson, 1991b). Williamson also notes that in addition to incentive conflict, failures of adaptation may arise ‘because autonomous parties read and react to signals differently, even though their purpose is to achieve a timely and compatible combined response’ (Williamson, 1991a). Yet, this theoretical recognition of adaptation problems that might persist even in the absence of incentive conflict finds scant recognition in most prior research motivated by transaction cost economics.1 In this 1 We focus on transaction cost rather than property rights when discussing existing literature on the economic organization of vertical relationships. This is because the empirical strategy and management literature on vertical relationships is much more influenced by transaction cost economics than property rights economics (see Novak and Eppinger, 2001, for an exception).

Received 1 March 2000 Final revision received 10 November 2004

416

R. Gulati, P. R. Lawrence and P. Puranam

study, our goal is to broaden the conceptualization of constraints to adaptation in vertical relationships beyond incentive conflict to include constraints arising from limited responsiveness to changing exchange conditions and coordination failures. The central and most investigated proposition in transaction cost economics concerns a class of adaptation problems resulting from the potential for hold-up in vertical relationships (David and Han, 2004; Masten, 2002; Shelanski and Klein, 1995). When procurement must be supported by dedicated (relationship-specific) investments, the anticipated costs of the transaction increase. This is because dedicated investments by one party create scope for the other to renegotiate the contract opportunistically when circumstances change. By organizing such transactions under common ownership, muted incentives, enhanced monitoring, and the threat of sanctions can limit opportunistic behavior and facilitate cooperative adaptation (Williamson, 1985). However, we will argue in this paper that adaptation failures in vertical relationships can also occur for reasons other than hold-up (or concerns about hold-up). Even when there is no incentive conflict, bounded rationality can cause the parties to an economic relationship to fail to recognize profound changes in the economic environment, or generate a coordinated response to such changes (Camerer and Knez, 1996, 1997; Foss, 2001; March and Simon, 1958). The theoretical challenge is to analyze these aspects of adaptation in addition to traditional hold-up concerns within a parsimonious and integrated framework. Along the lines of Williamson (1991a, 1991b) we define the adaptive capacity of a vertical relationship (within or across firm boundaries) as the ability to generate coordinated and cooperative responses across procuring and supplying units to changes in exchange conditions. However, our approach is distinctive from most prior empirical research in that in addition to the traditional focus on cooperation as the key aspect of adaptation, we also emphasize responsiveness to change and coordination of responses among the parties. We draw upon some of the seminal research on organization design to analyze our broader conception of adaptation in procurement activities. We assess the adaptive capacity of different modes of organizing procurement using the concepts of differentiation and integration. In prior literature on organizations, these concepts have been used to formulate principles for designing subunits Copyright  2005 John Wiley & Sons, Ltd.

within organizations that could adapt to change (Daft, 2001; Lawrence and Lorsch, 1967a, 1967b; Nohria and Ghoshal, 1994). In this tradition, organizational performance was argued to depend on the match between environmental contingencies and the extent of differentiation and integration across organizational subunits. We extend this analytical approach to vertical relationships both between and within firms. We argue that different modes of procurement—make, buy, and ally—differ in terms of the extent of differentiation and integration between procuring and supplying units and therefore in their adaptive capacity for responding to changes in the exchange environment in a coordinated and cooperative manner. We therefore expect that the performance of a given procurement activity will depend on the match between the adaptive capacity of the specific mode of procurement and the need for adaptation in the specific exchange relationship that is in turn impacted by contextual factors associated with the exchange. We test this prediction through a switching regression model of procurement performance, estimated with data on the procurement arrangements used by Ford Motor Company and Chrysler Corporation for all their major components. Our results show that performance differences across modes of procurement arise as a function of the match between adaptive capacity and adaptation requirements associated with the exchange, and not only the match between governance form and transaction hazards.

THEORETICAL BACKGROUND It is interesting to note that around the time Ronald Coase was formulating his ideas on transaction costs and their effects on coordination in markets and firms (Coase, 1937), Chester Barnard (1938) was emphasizing the importance of adjustment processes in organizations. As with Coase, Barnard’s analysis framed key contributions by later scholars (e.g., Simon, 1945; March and Simon, 1958; Thompson, 1967; Lawrence and Lorsch, 1967a, 1967b), who saw the essence of organizational adaptation as the generation of integrated responses to changed circumstances. The followers of Barnard, however, emphasized the importance of information rather than incentives (Grant, Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships 1996; March and Simon, 1958; Simon, 1945; Thompson, 1967). Furthermore, while Coase’s disciples focused on the boundary of the firm by assessing factors that impacted the make-orbuy problem, those who followed Barnard focused primarily on intra-organizational coordination. In the decades that followed, research in the tradition of Barnard rapidly accumulated on the design attributes of complex organizations comprising multiple, interdependent subunits that enabled them to achieve coordinated adjustments to changes in their environment (Daft, 2001; Galbraith, 1977; Nadler and Tushman, 1998). We develop and extend this tradition of research on organizational adaptation initiated by Barnard to the inter-organizational context. We analyze the adaptive capacity of vertical relationships in terms of differentiation and integration (Dougherty, 2001; Ghoshal and Nohria, 1989; Lawrence and Lorsch, 1967a, 1967b; Nohria and Ghoshal, 1994). Differentiation refers to the differences across organizational subunits that arise as a consequence of their local adaptation to unit-specific tasks and environments (Dougherty, 2001). Differentiation at the subunit level increases the responsiveness of the aggregate organization (and hence its adaptiveness), as it creates organizational diversity. Differentiated and diverse organizational subunits can recognize and engage in a wider search for new opportunities when environmental conditions change (Cohen and Levinthal, 1990; Ethiraj and Levinthal, 2004; Lawrence and Lorsch, 1967b; Rivkin and Siggelkow, 2003). Integration refers to the achievement of collaboration between organizational subunits. It encompasses not only cooperation (alignment of interest) but also coordination (alignment of actions) (Camerer and Knez, 1996, 1997; Foss, 2001; Heath and Staudenmayer, 2000). Achieving integration between interdependent organizational subunits is necessary in order to respond effectively to change (Nadler and Tushman, 1998; Thompson, 1967; Van de Ven and Walker, 1984). By analyzing organizational attributes that emphasize responsiveness and coordination (such as differentiation and integration) in vertical relationships, we propose to complement the rich body of research that has focused primarily on the governance attributes of such relationships. We also extend prior research that has used the differentiation and integration constructs primarily to study the adaptive capacity of the internal Copyright  2005 John Wiley & Sons, Ltd.

417

organization of firms, by applying these constructs to the context of vertical relationships. In developing our theory, we will argue that different modes of organizing vertical relationships—internal procurement, market procurement, and alliances—vary systematically in the extent of differentiation and integration between supplying and procuring units. As a result, they vary in terms of their adaptive capacity: their capacity to respond in a coordinated and cooperative manner to changes in exchange conditions. These differences are reflected in performance differences across modes of procurement that face differing levels of adaptation pressures in the transaction and task environments. Those instances in which the adaptive capacity of the procurement arrangement matches the adaptation pressures faced perform better than those without such a match.

THE ADAPTIVE CAPACITY OF VERTICAL RELATIONSHIPS In most prior research on the delineation of firm boundaries, scholars have studied two different modes of organizing vertical relationships: firms can either ‘make’ (procure from an internal supplying unit within the firm) or ‘buy’ each component (from an external supplier outside the firm) necessary to complete their chosen product mandates (Coase, 1937; Williamson, 1975). In recent years, scholars have expanded this dichotomous choice to focus on other ‘hybrid’ forms of organization that are intermediate between make and buy (Bradach and Eccles, 1989; Dyer and Singh, 1998; Gulati, 1998; Helper et al., 2000; Poppo and Zenger, 2002; Williamson, 1991a; Zaheer, 1995). In this study, in addition to make and buy, we focus on vertical alliances (or ‘ally’), a hybrid form that indicates a relationship characterized by continuity between two independent firms operating at successive stages in a vertical chain of production, with both firms expecting the interaction to continue into the future (Heide and John, 1990). As opposed to arm’s-length exchanges, which tend to be on the open market, discrete, and short term, vertical alliances tend to have a long time horizon or to be open ended (Ring and Van de Ven, 1992, 1994). Since the exchange partners are ultimately not under the same legal ownership structure, however, the supplying and purchasing units retain Strat. Mgmt. J., 26: 415–440 (2005)

418

R. Gulati, P. R. Lawrence and P. Puranam

autonomy in decision making and rights over residual returns (Grossman and Hart, 1986). Our unit of analysis for this study is the vertical relationship that a firm undertakes for the procurement of a component. We treat each relationship as akin to two interacting subunits of an organization: the procuring and supplying units. The two units could be part of the same firm (internal procurement or ‘make’), or belong to different firms but are tied together by ongoing exchange (vertical alliances or ‘ally’) or arm’s-length contracting (market contracting, or ‘buy’). We propose that vertical relationships are characterized by varying levels of adaptive capacity as indicated by the associated levels of differentiation and integration between the procuring and supplying unit. The adaptive capacity of the relationship impacts its performance, conditional on contextual attributes associated with the exchange. We first describe how different modes of organizing vertical relationships—make, buy, ally—differ from each other in the extent of integration and differentiation, and then discuss the consequence of these differences for the performance of the vertical relationship. Differentiation in vertical relationships Differentiation is the ‘state of segmentation of the organizational system into subsystems, each of which tends to develop particular attributes in relation to the requirements posed by its relevant external environment’ (Lawrence and Lorsch, 1967b: 4). It refers to the degree to which organizational units have developed distinctive structural characteristics and their members have made behavioral accommodations to their environment. It thus captures organizational differences across the procuring and supplying organizational subunits as a consequence of specialization and local adaptation, and it results in organizational diversity. The concept encompasses not only the differentiation of formal structure and accompanying specialization but also differentiation in the behavioral and attitudinal attributes of members of units (Lawrence and Lorsch, 1967a, 1967b). The key dimensions along which the differentiation of subunits can be assessed include differences in (1) the degree of formal structure, (2) the interpersonal orientation of members, (3) the time horizon of members, and Copyright  2005 John Wiley & Sons, Ltd.

(4) the goals of members (Lawrence and Lorsch, 1967b; Dougherty, 2001). In vertical relationships, supplying units are likely to be more differentiated from procuring units if they do not belong to the same firm, as there are limits on the extent to which organizational units within a firm can become differentiated from each other. An enduring pattern of behavior in organizations is the tendency to enforce conformity, treating all units alike so that, regardless of their task, they will tend to share a culture, business systems, and human resource policies. The pressures toward conformity emanate from several sources: (1) inter-unit rivalry, driving a desire to be treated the same as everyone else (Bradach and Eccles, 1989); (2) the physical and administrative proximity of units, facilitating social comparison processes (Zenger and Hesterly, 1997); (3) the need to facilitate career paths across units; and (4) the need to simplify the management tasks facing corporate executives responsible for multiple units. In addition to conformity pressures, the career advancement systems in hierarchies, which favor generalists over specialists, can limit the degree of differentiation possible. In contrast, supplying units outside the firm are unlikely to face pressures toward conformity with each other, though they may face pressures to conform to other units within their own firms. We therefore expect that in vertical relationships the extent to which procuring and supplying units are organizationally differentiated from each other varies with the fact of their joint memberships in a firm. When they belong to different firms, we should expect greater differentiation between them than when they are in the same firm. We predict: Hypothesis 1: Supplying units will be more differentiated from procuring units in market procurement and vertical alliances, than in internal procurement. Integration in vertical relationships Integration is the ‘quality of the state of collaboration that exists among departments that are required to achieve unity of effort by the demands of the environment’ (Lawrence and Lorsch, 1967a: 11). While traditionally applied to departments within a firm, the notion of integration can also be used to understand Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships relationships between organizational units from different firms. Integration not only requires the alignment of interests (cooperation), but also the alignment of actions (coordination). The distinction between cooperation and coordination is seldom maintained in organizational research. This is despite periodic recognition by scholars of important differences between the two (Grant, 1996; Heath and Staudenmayer, 2000; Jacobides, 2005; Kogut and Zander, 1996; Simon, 1991). We explain this difference in some detail below in order to support the point that integration encompasses both cooperation and coordination. Integration as cooperation and coordination Problems of cooperation arise from conflicts of interest. Under assumptions of self-interest, or its stronger forms (such as opportunism), collectively beneficial outcomes fail to arise due to actions motivated by the private benefits to individuals. The canonical problem is the famous prisoner’s dilemma. Problems of hold-up, agency, and the tragedy of the commons are all variants on the prisoner’s dilemma (Camerer and Knez, 1996, 1997; Foss, 2001; Heath and Staudenmayer, 2000). In essence, the problem of cooperation is a problem of motivation. It is resolved by aligning interests through formal mechanisms such as contracting (where possible) (Williamson, 1975), common ownership of assets (Grossman and Hart, 1986; Hart, 1995), monitoring, sanctions (Williamson, 1985), and the prospect of future interactions (Baker et al., 2002; Heide and Miner, 1992). Informal mechanisms such as identification and embeddedness may also serve to align interests (Granovetter, 1985; Gulati, 1995a, 1995b; Gulati and Sytch, 2005; Kogut and Zander, 1996). In contrast, coordination problems arise due to the lack of shared and accurate knowledge about the decision rules that others are likely to use and how one’s own actions are interdependent with those of others (Geanakoplos, 1992; Malmgren, 1961; Milgrom and Roberts, 1992; Thompson, 1967). The experimental economics literature on ‘weakest link’ games illustrates coordination failures due to lack of knowledge of how others will act. These games show that uncertainty about others’ rationality Copyright  2005 John Wiley & Sons, Ltd.

419

can be a key constraint on successful coordination on an efficient equilibrium (Camerer and Knez, 1997; Gulati et al., 1994; Knez and Camerer, 1994, 2000; Van Huyck et al., 1990, 1991). Behavioral economists have also modeled coordination problems where the agents are unaware of how their actions are interdependent, but discover them through an iterative, bounded rational learning process (Camerer, 2003). The organizational literature on adaptation on rugged landscapes (Kauffman, 1993) addresses similar problems. Myopic agents discover the true nature of their interdependence through search processes of varying degrees of intelligence (Ethiraj and Levinthal, 2004; Gavetti and Levinthal, 2000; Levinthal and Warglien, 1999; Rivkin and Siggelkow, 2003; Siggelkow and Levinthal, 2003). While formal analysis of coordination is relatively new in the economics and organizational literature, classical discussions of interdependence and coordination in organization theory were closely linked to the problem of creating knowledge about others’ actions and interdependence of actions (e.g., March and Simon, 1958; Thompson, 1967). In this literature, organizing to achieve coordination relied on three broad categories of mechanisms: programming, hierarchy, and feedback—each of which served to enhance the predictability of other’s actions, and to increase knowledge about how actions are interdependent (Galbraith, 1977; Nadler and Tushman, 1998; Thompson, 1967; Tushman and Nadler, 1978). In sum, coordination problems refer to the difficulties of aligning actions. They can persist even when interests are aligned, i.e., when cooperation is achieved. Put more strongly, incentives, sanctions, monitoring, rewards, and punishments can help to achieve cooperation but are not sufficient to achieve coordination (Gulati and Singh, 1998). This is because cooperation problems are rooted in motivation, whereas coordination problems arise due to the cognitive limitations of individuals that deny them comprehensive knowledge of how others will behave in situations of interdependence, and how they are interdependent with others. Therefore, the achievement of integration—‘unity of effort’ in Lawrence and Lorsch’s terms (1967a, 1967b)—requires the resolution of both cooperation and coordination problems. Strat. Mgmt. J., 26: 415–440 (2005)

420

R. Gulati, P. R. Lawrence and P. Puranam

Integration in make, ally and buy In this section, we will argue that the extent of integration between procuring and supplying units varies systematically across different modes of procurement, as these modes differ in the mechanisms available to generate cooperation and coordination. Cooperation between organizational units can arise when there are appropriate formal and informal incentives. For instance, the use of bonuses tied to firm-level profits can create incentives for cooperative behavior across departments (Gupta and Govindarajan, 1986; Kretschmer and Puranam, 2004). Cooperation can also arise through processes of identification, which infuse members of organizational subunits with a willingness to cooperate and exert effort for the goals of the aggregate organization (Barnard, 1938; Ghoshal and Moran, 1996; Lawrence and Lorsch, 1967b). In the context of vertical relationships, transaction cost economists have argued that cooperation between procuring and supplying units is likely to be higher when they both belong to the same firm. Within a firm, formal mechanisms such as authority and incentives influence cooperation between organizational units, aided by the continuity of association that common membership in a firm entails (Williamson, 1985). The quality of coordination between procuring and supplier units is also expected to be superior when they belong to the same firm. While governance is often construed as a set of mechanisms that facilitate cooperation (such as fiat, sanctions, and monitoring), it also represents a collection of coordination mechanisms (Gulati and Singh, 1998). Within firms, coordination between units is aided by the possibility of centralized decision making and the authority to design and use modes of coordination such as programming, hierarchy, and feedback (March and Simon, 1958; Thompson, 1967; Galbraith, 1977). Programming involves prior agreement on what actions must be taken and when (e.g., standards, schedules), and therefore enables coordination by enhancing the predictability of others’ actions. When programming proves insufficient, hierarchical elements such as a single source of authority and centralized decision making can enable coordination by dedicating individuals to the task of coordination, and allowing them to be informed about and even decide how different interdependent actors should behave. When it Copyright  2005 John Wiley & Sons, Ltd.

is important to be informed of others’ actions on an ongoing basis, or when the nature of interdependence must be discovered in an iterative fashion, then mechanisms that provide feedback (such as co-location and teams) enable mutual adjustment on an ongoing basis (Thompson, 1967; Van de Ven and Delbecq, 1974). In addition to the formal mechanisms of coordination, a number of elements of informal organization, such as shared experience, culture, leadership, norms, and precedent increase confidence about how others will behave and how one should behave in a given situation, and so enable coordination. The accumulation of shared experience and continuity of association is likely to be strong within firms, as its boundaries dictate who interacts with whom on a regular basis and often imply spatial collocation. Thus, co-membership within a firm can lead to the creation of shared understanding of the task environment and the interdependence it embodies (Weick, 1993, 1995) as well as shared values and norms that serve to make the actions of others more predictable (Kogut and Zander, 1996). Such formal and informal mechanisms that aid coordination are typically unavailable in arm’slength relationships characteristic of market procurement. Without common ownership, the design and implementation of coordination mechanisms to link the procuring and supplying units can involve costly and time-consuming negotiations even in the absence of opportunism (Conner and Prahalad, 1996; Hart, 1995). The low expected frequency of future exchanges in arm’s-length contracting may also make it harder to amortize the fixed costs of setting up such coordination structures (Dyer and Singh, 1998; Helper et al., 2000). The brevity of the relationship also implies a lack of time for partners to learn one another’s ways of doing business and develop multiple communication links or shared norms, values and beliefs. Further, differentiation itself may act as an impediment to integration (Lawrence and Lorsch, 1967a, 1967b). Differentiated organizational units that have cognitive and emotive differences (and therefore, unknown differences in decision rules and understanding of interdependence) may find it harder to achieve coordination than undifferentiated units (Dougherty, 2001). As we have argued earlier (Hypothesis 1), external procurement is characterized by higher levels of differentiation, and so may face greater challenges in achieving integration. These factors conspire to make it hard Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships to achieve the same degree of close coordination between procuring and supplying units in market procurement as in internal procurement. Vertical alliances can approximate some of the features of internal procurement in terms of coordination and cooperation. Long-term alliances can generate a state of cooperation between partners due to the ‘shadow of the future’ (Axelrod, 1984). With each partner anticipating doing business with the other well into the future, cooperation between them is more likely. Each sees the benefits from future interactions as outweighing the immediate pay-offs from noncooperative behavior and thus may choose to cooperate (Baker et al., 2002). The influence of such a time horizon on cooperative behavior has been observed in a variety of settings ranging from laboratory experiments (Murnighan and Roth, 1983) to firm–union relationships (Das and Teng, 1998). In addition to the assurance effects that arise in repeated interactions, trust between partners may also develop over time as a consequence of opportunities to share information and learn about each partner’s proclivities toward trustworthy behavior (Gulati, 1995a, 1995b, 1998). While these forms of relational contracting may provide collaborative effects weaker than those in ownership, which has the advantages of absolute authority and complete discretion to modify incentives, they may nonetheless be significant. Alliances may also benefit from mechanisms of coordination unavailable in market procurement (Bensaou and Venkatraman, 1995; Gulati and Singh, 1998). Repeated interactions between partner firms may justify investments in mechanisms of coordination, such as inter-firm organizational structures and information technology integration (Zaheer and Venkatraman, 1994), and may also give rise to a superior capacity for coordinating with each other through the formation of inter-firm routines (Dyer and Singh, 1998). Such routines embody knowledge about how coordinating agents will behave. Frequent and continued contact between the personnel of procuring and supplying units may also enable the formation of shared representations of the task environment, also enhancing coordination. Thus, internal procurement and vertical alliances have access to enablers of integration between units that are unavailable in arm’s-length transactions. We therefore expect: Copyright  2005 John Wiley & Sons, Ltd.

421

Hypothesis 2: Supplying units will be more integrated with procuring units in internal procurement and vertical alliances, than in market procurement. We can summarize the relationships between the three modes of organizing vertical relationships as follows: Internal procurement is characterized by weak differentiation and strong integration between procuring and supplying units. Market procurement is characterized by weak integration and strong differentiation between procuring and supplying units. Vertical alliances may provide more balanced levels of integration and differentiation. This is because vertical alliances provide more integration than market procurement and more differentiation than internal procurement (see Figure 1). The adaptive benefits of integration and differentiation in vertical relationships We now turn to the effects of differentiation and integration in coping with the adaptation pressures that arise in vertical relationships to procure different kinds of components. We analyze two basic categories of adaptation pressures: (a) changes in the transaction environment for the component—due to fluctuations in demand, supply, or technology; and (b) the need for mutual adjustment in the task environment associated with the component—due to strong interdependencies between the activities of the supplier and buyer. We argue below that, on the one hand, these two factors together influence the magnitude of adaptation requirements in a given vertical exchange. On the other hand, the extent of differentiation and integration between supplying and procuring units Differentiation Integration

Make

Figure 1.

Ally

Buy

Differentiation and integration in vertical relationships Strat. Mgmt. J., 26: 415–440 (2005)

422

R. Gulati, P. R. Lawrence and P. Puranam

impacts the adaptive capacity of the vertical relationship. The match between adaptive capacity and adaptation requirements influences the performance of the relationship. Effects of adaptation pressures in transaction environment The transaction environment in vertical relationships refers to the set of commercial contingencies (such as demand, supply, and technology conditions) that influence the terms and content of the transaction (Williamson, 1985, 1991a, 1991b). Adaptation pressures arise in the transaction environment when these attributes of the transaction change. We define ‘transaction instability’ as the extent and rate of change in demand and technology relevant to a particular transaction.2 For the vertical relationship to continue in a fruitful manner, such changes must first of all be recognized, and then adapted to. Ambiguity and lack of the information about changes in the transaction environment may make responsiveness difficult (Daft and Lengel, 1986). Differentiation between the supplying and procuring units contributes to the adaptive capacity of a vertical relationship under conditions of transaction instability by enhancing its responsiveness to such changes in two ways. First, differentiation can help create ‘requisite variety’ (Ashby, 1968), as a collection of subunits of an organization that are diverse in terms of organizational attributes is more likely to include at least one subunit that closely matches environmental contingencies (Lawrence and Lorsch, 1967a, 1967b). Cohen and Levinthal in their discussion of organizational absorptive capacity make a similar point: under conditions of rapid and uncertain change, they claim that it is best for the organization to ‘expose a fairly broad range of ‘receptors’ to the environment’ (Cohen and Levinthal, 1990: 132). Second, differentiation can enable parallel and therefore broad-ranging search for new opportunities that emerge due to changed conditions. 2 Instability is distinct from complexity, which refers to the heterogeneity or diversity in key exogenous characteristics relevant to an organizational unit (Thompson, 1967; Lawrence and Lorsch, 1967a, 1967b). Instability and complexity together form key dimensions of uncertainty as conceptualized by organization theorists (Daft, 2001). Given our focus on adaptation, we restrict our attention to instability, as complexity is a static aspect of the organizational environment.

Copyright  2005 John Wiley & Sons, Ltd.

Using simulations of organizational adaptation on rugged landscapes, several scholars have noted that internally differentiated organizational structures can benefit from parallel exploration (Rivkin and Siggelkow, 2003; Ethiraj and Levinthal, 2004; Siggelkow and Levinthal, 2003). This is because the search processes of differentiated organizational subunits are themselves likely to differ in terms of objectives, heuristics, and learning processes. The parallel operation of different search processes by organizational subunits effectively expands the search scope of the parent organization. Through requisite variety and parallelism, differentiation creates adaptive capacity in vertical relationships by making them sensitive to changes in the transaction environment. This capacity to sense and respond is valuable when the transaction environment is unstable (Dess and Beard, 1984; Duncan, 1972; Mintzberg, 1979). Transaction instability can adversely affect the performance of vertical relationships, though the magnitude of the effect will depend on the capacity of the relationship to adapt to transaction instability. Under the principle of ‘fit’ between the adaptive capacity of a vertical relationship and the pressures towards adaptation it faces, we expect that the performance of modes of procurement characterized by a low degree of differentiation between procuring and supplying units is likely to be most adversely affected by transaction instability. Since internal procurement offers the least scope for differentiation between procurer and supplier (Hypothesis 1), its performance should be significantly depressed relative to other modes of procurement under conditions of transaction instability. We therefore expect: Hypothesis 3: Transaction instability has more adverse effects on the performance of internal procurement than on the performance of market procurement and vertical alliances. Effects of adaptation pressures in task environment The task environment in vertical relationships refers to the division of labor across the value chain and the ongoing pattern of interactions between upstream and downstream activities (Bensaou and Venkatraman, 1995; Gulati and Singh, 1998). Interdependence across organizational subunits Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships is an important source of adaptation pressures within the task environment (Van de Ven and Delbecq, 1974). Two activities may be said to be interdependent when the value of performing one activity depends on how the second activity is performed (Milgrom and Roberts, 1990, 1995; Thompson, 1967). In Thompson’s classic typology, the nature of interdependence is relatively well known in situations he referred to as ‘pooled’ and ‘sequential’ interdependence. Interdependence is ‘pooled’ when tasks depend on each other in a simple additive manner; each task renders a discrete contribution to the whole, and each is supported by the whole (Thompson, 1967: 54, 64). Interdependence is ‘sequential’ when the outputs of one task form the inputs of the other. Achieving maximum combined value across activities in these two cases requires that each task be performed essentially independently (but in the proper sequence) as well as it can be. The situation is more complicated when the true nature of interdependence between tasks is unknown ex ante, or if ongoing communication is necessary to create awareness about others’ actions. Thompson referred to such situations as ‘reciprocal’ interdependence. In this case, the nature of interdependence implies a state of continuous adaptation between units as ongoing decision making about task allocation (i.e., planning) and continuous communication (i.e., mutual adjustment) is required (Thompson, 1967: 54–55, 64). In addition to coordination problems, reciprocal interdependence can also create significant cooperation problems because of the dangers of free-riding (Petersen, 1992; Wageman and Baker, 1997). When several individuals must contribute to a task, but it is hard to verify their marginal contributions, then collective under-investment of effort may occur (Alchian and Demsetz, 1972; Holmstrom, 1982). Thus, reciprocal task interdependence can create significant cooperation and coordination problems. In vertical relationships, reciprocal interdependence can arise when the design and production of the procured item require significant levels of joint activity by the buyer and supplier. This can involve extensive interactions between the two for the supplier to understand and meet the specific requirements of the buyer, as well as ongoing interactions as the supplier fulfills the procurement order. Unlike the case when designs can be ‘tossed over the wall’ from the buyer to the supplier, reciprocal Copyright  2005 John Wiley & Sons, Ltd.

423

interdependence requires continuous adaptation in the vertical relationships as both parties engage in joint problem solving (Iansiti, 1998; Monteverde, 1995). Integration between units enables them to manage reciprocal interdependence effectively (Thompson, 1967; Lawrence and Lorsch, 1967a, 1967b). Formal mechanisms such as incentives and information-processing structures, and informal elements such as identification and culture, ensure that adaptation by the procuring and supplying unit takes place in a cooperative and coordinated manner. We therefore expect the performance of procurement modes that are characterized by low degrees of integration to be most adversely affected by task interdependence. Since market procurement has the least capacity for achieving integration between supplier and procurer (Hypothesis 2), we expect: Hypothesis 4: Reciprocal task interdependence has more adverse effects on the performance of market procurement than on the performance of internal procurement or vertical alliances. Joint effects of adaptation pressures in transaction and task environment Our last hypothesis pertains to the joint effects of transaction instability and task interdependence on the performance of vertical relationships. We expect that task interdependence and transaction instability have superadditive effects on the need for adaptation in vertical relationships—put differently, we expect interdependence to magnify the effect of instability on the need for adaptation and vice versa. Transaction instability requires adaptation through search for solutions to the contractual difficulties posed by changes in technology or market conditions. However, interdependence imposes constraints on that search process, as successful adoption of the new solution will require coordinated action by procurer and supplier. Therefore, interdependence makes adaptation to instability more difficult. In the absence of interdependence, the search for and adoption of new solutions is a unilateral affair. Conversely, task interdependence requires ongoing coordination and mutual adjustment between procurer and supplier units as they engage in their respective activities. Instability in transaction conditions magnifies the extent of mutual adjustment required, as the procurer and Strat. Mgmt. J., 26: 415–440 (2005)

424

R. Gulati, P. R. Lawrence and P. Puranam

supplier need to adapt not only to the consequences of each other’s ordinary activities due to interdependence, but also extraordinary changes due to instability. In the absence of instability, mutual adjustment due to interdependence is a ‘routine’ affair (Nelson and Winter, 1982). There are important performance consequences of the superadditive joint effects of transaction instability and task interdependence. Modes of procurement that are able to adapt in some degree to both instability and interdependence suffer lower performance penalties from their joint effects, compared to modes that specialize in adaptation to any one. This suggests the existence of complementarities between differentiation and integration when there are significant adaptation pressures in both the task and transaction environment. For instance, under conditions of high task interdependence and transaction instability, neither differentiation nor integration alone provides adequate adaptive capacity. Instead, some combination of the two is required. Recent studies using agentbased simulation models (Rivkin and Siggelkow, 2003; Siggelkow and Levinthal, 2003) also highlight the complementarities between organizational mechanisms that enable parallel search (such as a differentiated organizational subunits engaged in parallel search processes) and the coordination of interactions between agents engaged in parallel search (for instance through an integrated evaluation of the alternatives they generate). The unique organizational position occupied by vertical alliances arises from their ability to generate greater differentiation than internal procurement and greater integration than market procurement (Hypotheses 1 and 2). Thus, unlike internal procurement or market procurement, the adaptive capacity of vertical alliances is evenly based on both differentiation and integration, whereas the adaptive capacity of internal procurement is based primarily on integration, and that of market procurement is based primarily on differentiation. This suggests that vertical alliances will have superior adaptive capacity compared to internal or market procurement, when both instability and interdependence are present. We therefore predict: Hypothesis 5: The joint effects of transaction instability and reciprocal task interdependence on the performance of vertical alliances will Copyright  2005 John Wiley & Sons, Ltd.

be less adverse than that on other modes of procurement.

METHODS Sample and data collection We utilize a survey of lead buyers of a variety of component areas at the Ford Motor Company and at the Chrysler Corporation conducted in 1995. Drawing on a previous study of the automobile sector (Monteverde and Teece, 1982) and our discussions with informants in the automobile industry, we developed a list of 120 components that go into most automobiles. The comprehensiveness of this list was verified with several executives in the industry and was also compared against component lists used by the firms to monitor their own parts quality. In-depth interviews with managers at Ford Motor Company and Chrysler Corporation preceded the design of the questionnaire and influenced many of the items included. We conducted a total of 37 interviews (16 at Chrysler; 21 at Ford). We ensured that we spoke with managers responsible for both external and internal sourcing. These included individuals in purchasing, quality control, platform management, and engineering operations. The initial interviews were exploratory and open ended but focused on the characteristics of each individual’s particular component, the type of sourcing arrangement and supplier used, its relative performance, and the pros and cons of alternative sourcing arrangements for that component. In later interviews, we sought clarifications on key constructs for this study, including transaction instability, task interdependence, differentiation, integration, and performance. The survey instrument was designed on the basis of these interviews and items from prior studies on vertical relationships. The survey was thoroughly pretested with several groups of senior managers at the participating companies to remove ambiguities and examine the face validity of our measures. Three groups of five executives each at the two companies went through the survey together to identify questions that were unclear or subject to multiple interpretations, revealed sensitive information, were difficult to answer, or were subject to social desirability bias. We incorporated the detailed feedback from these groups and Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships then sought additional input on the revised instrument from them to make additional changes. To reduce response bias, we also scattered questions measuring each construct across the survey and used multiple response formats. For each component, senior managers at Ford and Chrysler provided us with the names of individuals with oversight over the sourcing of that component. We mailed our survey to these individuals.3 Each survey respondent provided data on the component he or she was in charge of sourcing, as well as data on the two largest suppliers (or one, if only one existed) for that component. Unlike prior studies that aggregate all transactions relevant to a particular component (e.g., Monteverde and Teece, 1982), our analysis is truly at the transaction level. We ensured that each survey respondent was the expert for a given component by verifying his or her expert status with the controller’s office in each company. Our cover letter to respondents indicated that they had been identified as an expert on the acquisition of a given component and asked for the return of the uncompleted survey and a nomination of an expert if this were not the case. Survey implementation took several standard steps to ensure a good response rate. Sixty-four executives responded from Ford, and 67 executives responded from Chrysler, representing response rates of 53 percent and 56 percent, respectively, and a total response rate of 55 percent. We obtained data usable for this study on 222 procurement relationships, though data limitations reduced the effective number of observations for some multivariate analyses. We checked for non-response bias by comparing the characteristics of the components for which we received responses against those for which we did not receive a response. For each company, we assessed whether the components in our sample differed significantly from those which were not in our sample, on two key characteristics of components identified in prior research: type of sourcing and engineering complexity. We relied on Monteverde and Teece’s (1982) ratings of the primary types of sourcing of all components in automobiles and ratings of their engineering complexity as a basis for this comparison. We used the Kolmogorov–Smirnov two-sample test to assess the possibility of differences in the distribution of components in and outside the sample across these 3 Respondents were distinct from the senior managers who helped in pretesting the survey.

Copyright  2005 John Wiley & Sons, Ltd.

425

two variables (Siegel, 1956). The results of this test indicate that sample selection bias is not an issue with these data. Measures Each respondent was asked to answer some general questions about the exchange conditions relevant to his or her component. The respondent then identified the two principal suppliers of that component and answered a detailed set of identical questions about each of them. We used multi-item formative scales for most of our constructs. The items were based on a survey of the literature as well as fieldwork and the pretests. The key variables used in the analysis are reported in Table 1, along with details of the items used to construct them. Procurement mode The procurement mode is the specific organizational arrangement used to procure a component. Respondents identified each of two primary supply relationships for the component as falling into one of three categories defined in the questionnaire: (a) purchasing arrangement with external suppliers characterized by relatively shorter-term contracts and competitive bidding (market procurement); (b) purchasing arrangement with external supplier characterized by relatively longerterm or open-ended contracts (vertical alliance); and (c) purchasing arrangement with an internal division of their company (internal procurement). The response to this question was corroborated with two other survey items in which respondents indicated the percentage of the given component sourced through each of the three types of arrangements, and the expected duration of each relationship. Of the 222 exchanges in our sample, 21 were organized as internal procurement, 132 were organized as vertical alliances, and 69 were organized as market procurement. Differentiation Eight survey items were used to measure the degree of differentiation of the supplier of a component vis-`a-vis the procurer. The dimensions for this construct capture both structural and behavioral elements and measure the difference between supplier and procurer organizations in terms of speed of decision making, flexibility, Strat. Mgmt. J., 26: 415–440 (2005)

426

R. Gulati, P. R. Lawrence and P. Puranam

Table 1.

Key constructs

Construct Differentiation

Integration

Reciprocal task interdependence

Transaction instability

Performance of vertical relationship

Items Difference in the scores assigned to procurer and supplier on following organizational attributes: 1. Decision making (1 = quick, 7 = slow) 2. Culture (1 = rigid, 7 = flexible) 3. Information systems (1 = clear, 7 = complicated) 4. Time orientation (1 = long term, 7 = short term) 5. Style (1 = informal, 7 = formal) 6. Procedures (1 = bureaucratic, 7 = streamlined) 7. Employee benefits (1 = generous, 7 = limited) 8. Pay scales (1 = high, 7 = low) 1. This supplier has adapted its organization and management methods to work effectively with your organization (1 = strongly disagree, 7 = strongly agree) 2. During the past year, how often were there significant disagreements between your business unit and this supplier? (1 = very rarely, 7 = very frequently) [reverse coded] 3. How easy are the negotiations between your business unit and this supplier over sharing the burden of cost when your business unit requests engineering changes? (1 = very easy, 7 = very difficult) [reverse coded] 4. How easy are the negotiations between your business unit and this supplier over sharing the burden of cost when the supplier’s raw material costs increase (1 = very easy, 7 = very difficult) [reverse coded] 5. Problems that arise in the course of this relationship are treated as joint rather than individual responsibilities (1 = strongly disagree, 7 = strongly agree) Please indicate your opinion about the nature of the component (Scale: 1 = strongly disagree, 7 = strongly agree) 1. Its design requires contributions from both parties 2. Its production requires ongoing contributions from both parties 3. It takes significant amount of time and effort to understand your company’s specific requirements for this component 4. A significant amount of engineering effort is required in designing and developing this component 5. Producing this component generates specialized expertise in the producer Please indicate your opinion about the nature of the component (Scale: 1 = strongly disagree, 7 = strongly agree) 1. Significant future technological improvements are expected in its design 2. There is significant unpredictability or technological shifts in its production process 3. Significant fluctuations are expected in its monthly volume requirements 4. There is significant uncertainty in its annual volume estimates Your opinion about the attractiveness of this supplier compared to the best alternative supplier for this component 1. Price competitive (Scale: 1 = much less attractive than alternative, 7 = much more attractive than alternative) 2. Support and services 3. Flexibility in production 4. Product quality 5. Product innovations 6. Overall performance

information systems, time horizon, formalization, bureaucratization, employee benefits, and pay scales (Lawrence and Lorsch, 1967a, 1967b). The Cronbach alpha measure for reliability for this construct is satisfactory (0.78). Confirmatory factor analysis yielded an adjusted goodness of fit index of 0.94. Copyright  2005 John Wiley & Sons, Ltd.

Integration Five survey items captured the degree to which a state of cooperation and coordination existed between the supplying and procuring unit for each component and were used to compute a single scale for integration (Lawrence and Lorsch, 1967a, Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships 1967b). The Cronbach alpha measure of reliability for this construct was 0.72, and the adjusted goodness of fit index from confirmatory factor analysis was 0.98. Transaction instability The instability in exchange conditions associated with each component was broadly assessed in terms of demand and technological developments. These encompass measures of volume uncertainty and technological uncertainty (Heide and John, 1990; Walker and Weber, 1984, 1987). To construct the scale, we used four items. The reliability of the scale was satisfactory (alpha = 0.73) and the adjusted goodness of fit index from confirmatory factor analysis was 0.99. Reciprocal task interdependence We used a scale constructed from five items to measure the nature of task interdependence between procurer and possible supplier for each component. The items assessed the degree to which simultaneous contributions from supplier and procurer are required to design and produce the component, as well as the need for extensive interactions between the two for the supplier to understand and meet the specific requirements of the buyer (Thompson, 1967). The scale constructed from these items had satisfactory reliability (alpha = 0.71), and the adjusted goodness of fit index from confirmatory factor analysis was 0.97. Performance of vertical relationship Objective data on exchange performance is considered confidential and is hard to collect. As a result, we used items capturing multiple facets of the satisfaction of the procurers with the supplier in comparison with alternative suppliers. While we recognize that the satisfaction of the procurer with the supplier is not identical to exchange performance, we expect that it should be strongly correlated with it. We used six measures rating supplier performance along the dimensions of price, innovativeness, flexibility, and quality. The Cronbach alpha for this scale was 0.89, and the adjusted goodness of fit index from confirmatory factor analysis is 0.91. Copyright  2005 John Wiley & Sons, Ltd.

427

Control variables Several control variables that might impact the dependent and independent variables were used in the analysis. First, we controlled for assembler differences by including a dummy variable to differentiate Ford from Chrysler. Second, we limited informant bias by including a variable to control for the informant’s experience with the supplier (Kumar et al., 1993; Phillips, 1981). This was measured as the logarithm of the number of months that the respondent had personally dealt with the supplier. Third, we included a measure of the duration of the procuring relationship between the procurer and supplier (Kotabe et al., 2003). This was measured as the logarithm of the number of years the assembler has purchased the component from the supplier. Fourth, we controlled for supplier and buyer asset specificity, which are important alternative explanations from the transactions cost perspective. These variables were measured using single items, in which the respondent indicated agreement with the following assertions: ‘This supplier has made significant investments in terms of equipment, facilities, and engineering designed specifically to meet the buyer’s supply requirement for the component’ (supplier specificity) and ‘Your company has made significant investments in tooling and equipment that are specific to your relationship with the supplier’ (buyer specificity). Finally, we controlled for the volume of transactions for the component, as well as across all components between the supplier and assembler firm. These were categorical measures of the annual dollar value of purchases from a supplier for the focal exchange and from all exchanges between the buyer and supplier, respectively. The boundaries of the qualitative categories are less than $10 million, $50 million, $100 million, or more than $100 million. Analysis techniques Hypotheses 1 and 2 predict differences in differentiation and integration across the three modes of procurement. We therefore used OLS regression models to test them. To test Hypotheses 3, 4, and 5, we need to assess the relative impact of transaction instability, task interdependence, and their joint effect on the performance of different modes of procurement. Simply regressing the performance of each procurement mode on indicator variables Strat. Mgmt. J., 26: 415–440 (2005)

428

R. Gulati, P. R. Lawrence and P. Puranam

that classified that procurement relationship into make, buy, and ally, along with interaction terms with instability and interdependence, would not be appropriate. Unobserved features of the exchange (such as component, supplier, or procurer characteristics) could simultaneously influence the choice of procurement mode as well as performance outcomes (Hamilton and Nickerson, 2003; Shaver, 1998). It would then be difficult to draw valid conclusions about the effect of instability and interdependence on the performance of the mode of procurement. We therefore used a switching regression model that attempts to account for possibly endogenous choices of procurement modes. This model is estimated in two stages. First, a multinomial logit model is used to explain modes of procurement. The estimates from this model are used to calculate a non-selection hazard into each mode of procurement for each observation, which reflects the effect of unobservable variables that influence the procurement mode decision. Second, we estimate separate OLS regressions for subsamples of observations on each procurement mode, in which we include the non-selection hazard as a control variable. By its inclusion, the non-selection hazard controls for unobservable features of the relationships that might simultaneously affect the choice of mode and performance (Gulati and Nickerson, 2004; Poppo and Zenger, 1998; Shaver, 1998).

RESULTS Table 2 reports the summary statistics and pairwise correlations between variables used in our analysis. The largest correlation between two independent variables is 0.66 (between component procurement volume and total procurement volume). Collinearity therefore does not appear to be a significant concern. However, we calculated the variation inflation factors for all estimated models as a precaution, and these were well within acceptable limits. Several pairwise correlations are significant at the 1 percent level. For instance, integration is strongly correlated with relationship performance. It is also noteworthy that the correlations between reciprocal interdependence and supplier and buyer specificity are small (0.13 and 0.11), lending support to our arguments that these are distinct constructs. Supplier and buyer specificity themselves are Copyright  2005 John Wiley & Sons, Ltd.

only correlated 0.22. Our interview data suggest that in the automotive industry suppliers are more likely to make specific investments than assemblers. Table 3 reports results from OLS models in which the dependent variables are the extent of differentiation and integration between procuring and supplying units. For the specification in which differentiation is the dependent variable (column 1), only 193 observations were available. We used robust standard errors to calculate t-statistics in order to minimize the effect of heteroscedasticity. We tested Hypotheses 1 and 2 by introducing dummy variables for the mode by which the procuring relationship was organized. We controlled for assembler (Ford or Chrysler), buyer experience with supplier, length of procuring relationship, and component and total volume of procurement. Both models are significant, with R 2 of 12 percent and 8 percent. Examining the coefficients in column 1, we find that vertical alliances and market procurement relationships are characterized by greater differentiation than internal procurement relationships (reference category). We thus conclude that Hypothesis 1 is strongly supported. Vertical alliances and market procurement do not appear to differ significantly in differentiation. In column 2, the dependent variable is integration and the reference category is market procurement. We find evidence in favor of Hypothesis 2, as internal procurement and vertical alliances appear to be characterized by significantly higher levels of integration between procuring and supplying units than market procurement. Thus, Hypothesis 2 is also supported. The coefficient of internal procurement is larger and statistically different from that of vertical alliances. Thus, vertical alliances appear to have differentiation levels comparable to market procurement, and integration levels lower than internal procurement. Among the control variables, total volume of procurement and the length of respondents’ business association with the supplier (buyer history) have negative effects on differentiation. This may be due to convergence in organizational elements across buyer and supplier due to extensive and lengthy interactions. Procurement history has a negative effect on integration. This may indicate the constraining effects of ties. Since tie duration is not the focus of our analysis, we defer further investigation of these results to future research. Strat. Mgmt. J., 26: 415–440 (2005)

Copyright  2005 John Wiley & Sons, Ltd.

Differentiation Integration Relationship performance Transaction instability ∼ Reciprocal interdependence ∼ Interdependence × instability Supplier asset specificity Buyer asset specificity Procurement history [ln(year)] Buyer history [ln(months)] Component procurement volume Total procurement volume Ford Procurement mode

222 222

222

222

222

222

222

222

222

222

222

0.55 2.49

3.21

2.60

3.05

2.45

4.74

5.70

0.27

0.00

0.00

193 1.06 222 −0.18 219 4.81

∼ centered ∗ p < 0.01 in a two-tailed test

13 14

12

11

10

9

8

7

6

5

4

1 2 3

0.50 0.66

1.00

1.05

0.83

0.77

1.68

1.10 0.13

0.12 0.03

0.01

0.12

0.22∗

1

7

0.09

0.00 0.06

0.15 −0.10

0.15

0.11

0.11

0.06

1

8

0

1

10

12

13

0.01 1 0.01 0.18∗

0.66∗ 1

1

11

0.33∗ −0.10 0.10 −0.12

0.27∗ −0.01

0.21∗

0.11

1

9

0.09 0.04 −0.14 0.06 −0.21∗ −0.15

0.05

0.03

0.11

0.04 −0.03

0.13 −0.03

0.06

0.15 −0.14 0.22∗ ∗ 0.22 −0.06 −0.09

0.01 −0.06

1.00 4.00 −0.16 0.00 1.00 −0.04 1.00 3.00 0.17

0.04 −0.11

1.00 4.00 −0.08

0.19∗

0.06

1

6

0.11 −0.05

0.13

0.00

0.13

1.00 5.48 −0.14 −0.08

0.01

0.11

0.04 −0.11

0.03

0.28∗

1

5

0.28∗ −0.16

0.31∗

1

4

0.00 4.32 −0.10 −0.17 −0.13

1.00 7.00

0.11

2.00 7.00 −0.15

0.14

0.01

0.85 −2.13 4.71 0.02

0.07

0.80 −3.10 1.50 −0.23∗ −0.00

0.02 −0.06

1

3

0.99 −2.36 3.14 −0.18∗

2 1 0.25∗

1

0.68 0.00 4.38 1 1.39 −5.00 7.00 −0.16 1.00 1.67 7.00 −0.06

Obs. Mean S.D. Min. Max.

Descriptive statistics

Variable

Table 2.

Adaptation in Vertical Relationships 429

Strat. Mgmt. J., 26: 415–440 (2005)

430

R. Gulati, P. R. Lawrence and P. Puranam

Table 3. Differences in organizational attributes across procurement modes (OLS) Differentiation Make Buy Ally Procurement history Buyer history Component volume Total volume Ford Constant R2 F n d.f.

0.70∗∗∗ 0.15 0.66∗∗∗ 0.14 0.05 0.06 −0.13∗ 0.07 0.05 0.07 −0.12∗ 0.07 −0.06 0.13 1.03∗∗∗ 0.39 0.12 5.84∗∗∗ 193 7

Integration 0.82∗∗ 0.38 0.35∗ 0.19 −0.37∗∗ 0.15 −0.18 0.14 0.13 0.13 −0.06 0.105 0.45∗∗ 0.21 0.61 0.41 0.08 2.13∗∗ 222 7

∗ p < 0.10; ∗∗ p < 0.05; ∗∗∗ p < 0.01 in a two-tailed test. Numbers below coefficients are robust standard errors.

To test Hypotheses 3, 4, and 5, we used a switching regression model (Shaver, 1998; Hamilton and Nickerson, 2003). We estimate this model in two steps. First, we estimated a multinomial logit model to predict the choice of procurement mode (make, buy, or ally).4 In addition to transaction instability, reciprocal interdependence, and their interaction, which are the variables of theoretical interest, we controlled for assembler (Ford or Chrysler), supplier, and buyer asset specificity, buyer experience with supplier, length of procuring relationship, and component and total volume of procurement. Component and total procurement volume function as instrumental variables in our model, and do not appear in the second-stage models predicting performance. While they are expected to affect 4

A multinomial logit model imposes no conditions on the rank ordering of buy, ally, and make. Unlike transaction cost economics, which views these modes of procurement in terms of increasing degrees of hierarchy, our hypotheses do not suggest a clear rank-ordering in terms of adaptive capacity. In additional analyses to assess the robustness of our results, we used an ordered probit model instead of a multinomial logit model in the first stage of our switching regression; the pattern of support for our hypotheses is unaltered. Copyright  2005 John Wiley & Sons, Ltd.

procurement mode, we assume that component and total procurement volume do not directly affect the performance of the procurement mode once we control for instability, interdependence, asset specificity, procurement, and buyer history and assembler.5 Data limitations reduced the number of observations at the first and second stage of the analysis to 219 and 216 respectively.6 The results for the first-stage multinomial model are presented in Table 4. The model is significant, with a pseudo-R 2 of 24 percent. The baseline category in Table 4 is vertical alliances (‘Ally’), so that the coefficients must be interpreted as affecting the odds of choosing internal procurement (column 1) or market procurement (column 3), relative to the odds of choosing alliances. We note that our instrumental variables (component and procurement volume) do affect procurement mode, as assumed. Across all three categories (make, buy, and ally) component and procurement volume are jointly significant at the 5 percent level. While we use the results in Table 4 as an intermediate step in estimating a switching regression model, these results are worth noting for their own merit. Reciprocal interdependence between supplier and procuring unit and buyer asset specificity make internal procurement more likely relative to alliances. Internal procurement is also likely to have a longer history, and be characterized by greater buyer asset specificity. The coefficients in column 4 suggest that supplier asset specificity makes market procurement less likely than alliances, but buyer asset specificity has the opposite effect. These results suggest that in the automotive industry, alliances are created to protect the idiosyncratic investments of suppliers rather than buyers, whose position as monopsonists may allow them to use market contracting with impunity. While not the focus of our research, we believe that further analysis of the distinct effects of buyer and supplier asset specificity may prove fruitful. Reciprocal interdependence appears to increase the odds of 5 Such an assumption is necessary in order for the estimated equations to be econometrically identified, and appears to be theoretically plausible as well as supported by prior empirical work (e.g., see Poppo and Zenger, 1998). Further, by using two instrumental variables for one endogenous variable (procurement mode) we are able to directly test and verify the assumption that they are exogenous in the performance model (Wooldridge, 2003). 6 Three observations lacked data on performance, and three others turned out to be significant influence points.

Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships Table 4.

431

Procurement mode choice (multinomial logit) Make 1

Transaction instability Reciprocal interdependence Instability × interdependence Supplier asset specificity Buyer asset specificity Procurement history Buyer history Component procurement volume Total procurement volume Ford Constant LR χ 2 N d.f. Pseudo R 2

−0.33 0.34 1.17∗∗ 0.47 0.03 0.33 0.43∗∗ 0.21 2.52∗∗∗ 0.83 −0.30 0.55 0.55+ 0.37 0.10 0.46 −2.50∗ 1.11 −11.18 5.21 48.74∗∗∗ 219 19 0.24

Ally 2 −0.30 0.34 1.11∗∗ 0.54 −0.27 0.41 0.01 0.32 0.44∗∗ 0.21 2.50∗∗∗ 0.82 −0.26 0.57 0.58+ 0.39 −0.16 0.50 −2.54∗ 1.16 −10.99 5.33 49.39∗∗∗ 219 20 0.25

3

Buy 4 −0.12 0.17 0.60∗∗ 0.27

B A S E C A T E G O R Y

−0.45∗∗ 0.18 0.27∗∗ 0.13 −0.38+ 0.24 −0.20 0.23 0.47∗∗ 0.22 −0.75∗∗∗ 0.24 −0.24 0.36 3.28 1.27 48.74∗∗∗ 219 19 0.24

5 0.03 0.20 0.53∗ 0.29 −0.44∗ 0.24 −0.44∗∗ 0.19 0.25∗ 0.13 −0.37+ 0.24 −0.18 0.23 0.51∗ 0.23 −0.82∗∗∗ 0.25 −0.22 0.37 3.51∗∗∗ 1.33 49.39∗∗∗ 219 20 0.25

∗ p < 0.10; ∗∗ p < 0.05; ∗∗∗ p < 0.01 in a two tailed test; + p < 0.10 in a one tailed test. Numbers below coefficients are robust standard errors.

selecting ‘buy’ as opposed to ‘ally,’ but this puzzling result disappears in the full specification (column 5). The significant negative interaction between instability and interdependence suggests that for non-trivial levels of instability the net effect of interdependence on the odds of selecting buy relative to ally is negative. We also note that reciprocal interdependence and asset specificity have distinct and independent effects on the choice of procurement modes, so that our belief in the distinctiveness of these constructs is reinforced. We now turn to the second step in estimating the switching regression model. Using the predicted probabilities from the multinomial logit reported in Table 4, we constructed the hazard for non-selection into each procurement mode (Hamilton and Nickerson, 2003; Lee, 1982). Table 5 presents the results from the second step of the switching regression procedure, in which we regress the performance of each Copyright  2005 John Wiley & Sons, Ltd.

procurement relationship on the non-selection hazard, the extent of transaction instability, reciprocal interdependence, their interaction and other control variables except component and total procurement volumes. The switching regression model was estimated separately in each subsample of transactions, i.e., for all transactions classified as ‘make,’ ‘buy,’ and ‘ally.’ All models are significant, with R 2 ranging from 16 percent to 58 percent. Hypotheses 3, 4, and 5 require comparison of coefficients across these models, as we are interested in testing the differences between the marginal effects of transaction instability, reciprocal interdependence, and their interaction across the three modes of procurement. In addition to robust standard errors (reported below each coefficient), in order to facilitate intermodel comparison of coefficients, we calculated robust standard errors from a combined variance–covariance matrix using the seemingly unrelated estimation algorithm in STATA 8.2. These Strat. Mgmt. J., 26: 415–440 (2005)

432

R. Gulati, P. R. Lawrence and P. Puranam

Table 5.

Performance of procurement modes (switching regression) Make

Non-selection hazard Transaction instability Reciprocal interdependence Instability × Interdependence Supplier asset specificity Buyer asset specificity Procurement history Buyer history Ford Constant R2 F N d.f.

Buy

a

b

1.29 1.13 (0.84) −0.52∗∗ (∗∗∗ ) 0.20 (0.15) 0.54 (+ ) 0.50 (0.36)

2.21 (∗ ) 1.84 (1.31) −0.58∗∗ (∗∗∗ ) 0.23 (0.16) 1.10 (+ ) 0.90 (0.64) 0.20 0.33 (0.23) 0.27 (∗ ) 0.22 (0.16) 0.63 (∗∗ ) 0.45 (0.32) 1.91 (∗∗ ) 1.25 (0.88) −0.32 0.63 (0.44) −1.51 (∗ ) 1.25 (0.89) −8.53 (∗ ) 6.90 (4.89) 0.58 7.60∗∗∗ 20 10

0.26 (∗ ) 0.20 (0.15) 0.44 (∗ ) 0.30 (0.23) 1.26 (∗∗ ) 0.85 (0.63) 0.04 0.38 (0.28) −1.04 0.87 (0.65) −5.30 4.54 (3.37) 0.57 11.28∗∗∗ 20 11

a 0.34 0.48 (0.44) −0.26 0.17 (0.16) 0.23 0.21 (0.20) 0.14 0.15 (0.14) 0.01 0.07 (0.07) −0.37 0.27 (0.25) 0.66∗∗∗ (∗∗∗ ) 0.15 (0.14) −0.60∗∗∗ (∗∗∗ ) 0.24 (0.22) 2.91∗∗∗ (∗∗∗ ) 1.04 (0.97) 0.28 4.00∗∗∗ 65 56

Ally b

a

b

0.32 0.48 (0.44) −0.25 0.17 (0.16) 0.23 0.23 (0.22) −0.20 0.26 (0.24) 0.14 0.15 (0.14) 0.00 0.07 (0.06) −0.35 0.27 (0.25) 0.65∗∗∗ (∗∗∗ ) 0.14 (0.13) −0.59∗∗ (∗∗∗ ) 0.24 (0.22) 2.93∗∗∗ (∗∗∗ ) 1.03 (0.95) 0.29 4.00∗∗∗ 65 55

0.35 0.24 (0.23) 0.01 0.08 (0.08) 0.03 0.12 (0.12)

0.32 0.24 (0.23) −0.07 0.09 (0.08) 0.08 0.12 (0.01) 0.21∗∗ (∗∗ ) 0.09 (0.08) 0.36∗∗∗ (∗∗∗ ) 0.09 (0.09) −0.03 0.05 (0.05) 0.01 0.11 (0.11) 0.00 0.11 (0.10) 0.22 0.21 (0.20) 2.54∗∗∗ (∗∗∗ ) 0.59 (0.57) 0.18 2.71∗∗∗ 131 121

0.36∗∗∗ (∗∗∗ ) 0.10 (0.09) −0.05 0.05 (0.05) 0.02 0.11 (0.11) −0.00 0.10 (0.10) 0.26 0.20 (0.20) 2.63 0.58 (0.56) 0.16 2.56∗∗ 131 122

∗ p < 0.10; ∗∗ p < 0.05; ∗∗∗ p < 0.01 in a two-tailed test; + p < 0.10 in a one-tailed test. Numbers below coefficients are robust standard errors. Significance marks and standard errors in parentheses are based on a common robust covariance matrix across models for all three modes of procurement.

results and the resulting significance levels are reported in brackets.7 To test Hypothesis 3, we compare the coefficients of transaction instability across the column ‘a’ models for Make, Ally and Buy, which only includes the main effects for instability and interdependence. Hypothesis 3 predicts that the adaptation pressure arising from transaction instability has more adverse effects on the performance of internal procurement than on market procurement or vertical alliances. We find that the coefficient of instability is not different from zero for Ally and Buy, but is significant and negative for Make. Therefore, we conclude that Hypothesis 3 is supported in our data. Hypothesis 4 predicts that the adaptation pressure due to reciprocal 7 Using a common variance–covariance matrix to construct robust standard errors is useful because it accounts for the effect of (a) unequal subsample sizes and (b) correlations in the error terms across the three equations on the standard errors used to statistically compare coefficients across models.

Copyright  2005 John Wiley & Sons, Ltd.

interdependence between procuring and supplying units affects market procurement more adversely than internal procurement or vertical alliances. Comparing the coefficient of reciprocal interdependence across Make, Buy, and Ally (column ‘a’), we find that it has a marginally significant positive effect on the performance of Make, but has an effect indistinguishable from zero in Buy and Ally. We therefore conclude that Hypothesis 4 is not supported, as external procurement does not do significantly worse than alliances under conditions of task interdependence (though it does marginally worse than internal procurement). To test Hypothesis 5, we constructed an interaction term between transaction instability and reciprocal interdependence. Centered versions of these variables were used in order to reduce collinearity. The inclusion of the interaction term does not alter the basic pattern of results for the main effects. In addition, we find that the coefficient on the interaction term is positive and significant for Ally, but Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships is indistinguishable from zero for Make and Buy. This is consistent with Hypothesis 5, which predicts that the joint effects of transaction instability and task interdependence on the performance of vertical alliances will be less adverse than that on other modes of procurement. We therefore conclude that Hypothesis 5 is supported. Among the control variables used to test Hypotheses 3, 4, and 5, we find that supplier asset specificity has a positive effect on the performance of internal procurement and vertical alliances, but not on the performance of market procurement. This is consistent with standard transaction cost predictions concerning the hold-up problem. Buyer asset specificity has a positive effect on the performance of internal procurement only. Analogous to our results in Table 4, these results also point to the asymmetry between buyer and supplier asset specificity. We also find that procurement history has a positive effect on the performance of internal procurement, while buyer history has a positive effect on the performance of market procurement. These results indicate a difference between the performance effects of organizational and individual histories of prior interaction in internal and market procurement. Finally, we note that the nonselection hazard is not significant in our models. However, a Hausman specification test indicates that the coefficients in the models change significantly on the inclusion of the non-selection hazard. This suggests that effects due to self-selection of procurement modes are operating in our data (and we have controlled for them), though we may have been unable to quantify them precisely (see also Shaver, 1998, for similar findings in the context of international expansion).

DISCUSSION In this study, our goal has been to broaden our understanding of the constraints to adaptation in exchange relationships beyond incentive conflict to include constraints arising from limited responsiveness to changing exchange conditions and coordination failures. We have proposed that the organizational attributes of differentiation and integration provide a parsimonious and integrated approach toward analyzing adaptive capacity in terms of cooperation, coordination and responsiveness. Our empirical analysis shows that the observed levels of differentiation and integration Copyright  2005 John Wiley & Sons, Ltd.

433

vary systematically with the type of procurement mode. Thus, different modes of organizing procurement relationships differ not only in the ability to control hazards of opportunistic renegotiation but also in their capacity to generate coordinated responses to adaptive pressures in the task and transaction environment. We have also argued that these differences in adaptive capacity across procurement modes result in performance differences across modes of exchange. Our results generally support this argument as well; the marginal effect of transaction instability is most adverse on internal procurement, and the marginal joint effect of instability and interdependence is least adverse on vertical alliances. However, we are unable to detect any differences between alliances and market procurement in terms of the effects of task interdependence. Perhaps the differences in integration between these two modes of procurement may be small (as also indicated by the results in Table 3). However, these slight differences in integration levels may be leveraged dramatically by the complementarity between differentiation and integration, as evidenced by the robust support for Hypothesis 5. Thus, we conclude that the adaptation perspective can be useful in understanding performance differences across modes of procurement in terms of a match between adaptive capacity and adaptation needs. The adaptation perspective we develop revives an extensive tradition of research on adaptation within organizations (Galbraith, 1977; Nadler and Tushman, 1997; Nohria and Ghoshal, 1994; Daft, 2001; Donaldson, 2001), which has focused on the issues of coordination and responsiveness that have been central to our discussion of adaptive capacity. In this tradition, scholars emphasized the importance of limited rationality rather than incentive conflict as constraints to adaptation, and saw the essence of organizational adaptation as the generation of integrated responses to changed circumstances (Grant, 1996; March and Simon, 1958; Simon, 1945). Our study extends this research by showing that its central ideas are applicable to adaptation within and between organizations, and by linking them to recent research on coordination and adaptation (also see Jacobides and Winter, 2005). Our study also extends transaction cost analyses of vertical relationships beyond the usual considerations of incentive conflict. Unlike the Strat. Mgmt. J., 26: 415–440 (2005)

434

R. Gulati, P. R. Lawrence and P. Puranam

hold-up problem, which emphasizes opportunistic renegotiation as the primary constraint on adaptation in exchange relationships, we also focus on the limitations posed by bounded rationality in responding in a coordinated manner to changed circumstances (Conner and Prahalad, 1996; Hodgson, 2004; Simon, 1991). The novelty of our approach lies in suggesting that different governance modes used to organize procurement not only differ in their consequences for aligning interests, but also in their ability to align actions through processes that affect the recognition of changes in exchange conditions, and the generation of coordinated responses to such changes. We have intentionally focused on sources of adaptation pressures other than volume uncertainty and asset specificity. For instance, transaction instability combines the concepts of volume and technological uncertainty (Walker and Weber, 1984, 1987). Since we focus on the cognitive limitations of individuals that prevent organizations from responding to changes in the transaction environment, we expect changes arising from demand and supply conditions (volume uncertainty) and technology (technological uncertainty) to both create constraints on adaptation. Thus, these sources of uncertainty are seen as equivalent in terms of their impact on the need for adaptation in vertical relationships. Whereas an analysis from the perspective of hold-up would suggest that the inability to predict volume requirements in exchange relationships is likely to bestow an adaptive advantage on internal procurement, the perspective on adaptation we develop suggests that internal procurement is disadvantaged in such cases (Hypothesis 3).8 At the same time, our results are broadly consistent with the idea in transactions costs economics that an important organizational cost of internal procurement arises from bureaucratic inefficiencies (Williamson, 1985). Constraints on differentiation limit adaptation in internal procurement, and may be seen as an instance of bureaucratic costs arising from reasons other than the dulling of incentives within hierarchies. Similarly, the construct of asset specificity in discussions of hold-up is related to, but distinct from, the concept of task interdependence as used 8 It is also worth noting that David and Han (2004: 52) concluded after an extensive review of the empirical literature on transaction cost economics that ‘there does not seem to be a clear relationship between uncertainty and either the choice of governance form or the level of transaction costs.’

Copyright  2005 John Wiley & Sons, Ltd.

in the adaptation perspective presented here. Asset specificity refers to the need to make investments specific to a relationship, which once made have a significantly lower value in other uses. This creates quasi-rents and the possibility of opportunistic behavior in a bid to appropriate them (Klein et al., 1978). Interdependence, in contrast, refers to the existence of complementarities between tasks; the value of performing one task is a function of how the other is performed (Milgrom and Roberts, 1995; Sobrero and Roberts, 2001). Coordinating interdependent activities to maximize their combined value involves informationprocessing activities such as decision making and communication to allocate the tasks among individuals and enable ongoing adaptation and mutual adjustment between them as the tasks are executed (Thompson, 1967; Galbraith, 1977; Tushman and Nadler, 1978; Gulati and Singh, 1998; Sobrero and Roberts, 2001). Investments in coordination activities may or may not lead to quasi-rents. For instance, a number of studies on the effect of alliance experience on alliance performance find that learning about managing alliances with one partner transfers reasonably well to other alliance partners (Anand and Khanna, 2000; Kale et al., 2000). Conversely, investments in dedicated assets may or may not imply ongoing mutual adjustment, once the investment is made. For instance, firms in technology partnerships may agree on investing in a specific technology standard, and such investments may be highly specific, but, once made, standards reduce the need for ongoing coordination (Baldwin and Clark, 2000; Schilling, 2000).9 Like any study, ours also has certain limitations. We depend on a single respondent for data on each procurement relationship, thus suggesting the possibility of common method bias. Our approach of contacting the most knowledgeable informant is consistent with most prior research on vertical relationships (e.g., Artz and Brush, 2000; Heide and John, 1990; Kotabe et al., 2003; Poppo and Zenger, 1998, 2002). As other researchers have noted (Artz and Brush, 2000), typically only one individual interacts closely with each supplier on an ongoing basis. To mitigate concerns about respondent bias, we controlled for the history of the respondent’s interaction with the supplier 9 David and Han’s review (2004: Table 4) also shows that existing studies rarely if ever distinguish between the two constructs of task interdependence and asset specificity.

Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships and used exploratory factor analysis to verify that different measures indeed tap into different constructs. Our data about exchange performance are open to criticism for possibly being one sided, as our measures represent the opinion of the buyers alone. However, we believe that the buyer’s opinion of performance must correlate reasonably well with the suppliers, failing which suppliers would have exited the relationship. The average duration of the procurement relationship in our sample was 15 years. Another important caveat is that our measure of exchange performance is based on perceptual items measured in reference to the best alternative supplier for the component. Heterogeneity across components makes it necessary to measure performance relative to suppliers in that component class. While a measure of economic performance is undoubtedly preferable, such data are very difficult to collect for reasons of confidentiality. Relying on a perceptual measure comparing the supplier to all other suppliers for the same component is also problematic, for we would then be asking respondents to construct an implicit average across all other suppliers known to them. Our measurement strategy creates a clear anchor point (the best alternative). A problem with such a strategy arises if unobserved attributes of the component influence both the covariates of interest as well as the performance of the best alternative (and therefore the relative performance of the focal supplier). However, our switching regression approach helps alleviate this problem, as the procedure controls for unobserved features of components that influence a key covariate (the choice of procurement mode) and supplier performance. Analogous to assumptions about asset specificity in prior work, the adaptation perspective we developed makes the simplifying assumption that reciprocal task interdependence is an exogenous variable. Some scholars have suggested that interdependence is not completely exogenous, but firms actively design technical and organizational interfaces that can lower coordination needs across organizational and technical subsystems (Galunic and Eisenhardt, 1996, 2001; Puranam and Singh, 2002; Sanchez and Mahoney, 1996; Schilling, 2000). Ignoring the joint determination of transaction properties and organizational form is a common failing. Further research on the possible endogeneity of task interdependence along the lines of Copyright  2005 John Wiley & Sons, Ltd.

435

Oxley and Sampson’s analysis of the endogeneity of transaction scope, or Novak and Eppinger’s simultaneous analysis of governance form and product complexity, would be useful extensions (Novak and Eppinger, 2001; Oxley and Sampson, 2004). Despite these limitations, we believe that our study makes three significant contributions to the literature on the organization of vertical relationships. First, this study is at once distinct from and complementary to the voluminous research on firm boundaries that has primarily built on arguments about hazards of hold-up and opportunistic renegotiation to demonstrate the antecedents and consequences of governance choice. Through this study, we are able to extend, articulate, and test a perspective on adaptation in vertical relationships, which is broader than that in traditional discussions of hold-up. Adaptation is not only cooperative adjustment; it is also coordinated adjustment. Because of limited rationality, even agents acting cooperatively may fail to successfully adapt in a coordinated manner (Weick and Roberts, 1993). While others have made these theoretical arguments as well (e.g., Heath and Staudenmayer, 2000; Kogut and Zander, 1992, 1996; Zajac and Olsen, 1993), we have been able to empirically establish that arguments emphasizing coordination and cooperation are both capable of explaining the organization and performance of vertical relationships. Our results show that asset specificity has predictable effects on mode choice and performance even after controlling for factors derived from an adaptation perspective, and vice versa. Thus, it may be time to move beyond debates about hold-up vs. other explanations to begin considering when certain explanations for economic organization are more useful than others. Some recent studies that are primarily motivated by the hold-up problem have nonetheless begun to recognize the importance of coordination issues in exchange relationships. For instance, Oxley and Sampson (2004) studied the factors that influence the choice of alliance scope and structure. They conclude: ‘Our study suggests that these cognitive limitations (as manifested in the importance of adequate absorptive capacity to support broad scope alliances involving joint manufacturing) are an important consideration, in addition to, though not to the exclusion of—concerns with mitigating the hazards of potential opportunism.’ Ryall and Sampson in their analysis of R&D collaboration Strat. Mgmt. J., 26: 415–440 (2005)

436

R. Gulati, P. R. Lawrence and P. Puranam

contracts note that contracts often embody nonenforceable clauses that seem to serve only coordinative functions. These clauses serve as blueprints for exchange as a means to ‘plan the collaboration, set partner expectations, and consequently reduce misunderstandings and costly mis-steps’ (Ryall and Sampson, 2004). Mayer and Argyres (2004) reach similar conclusions, and draw attention to the functions of contracts as knowledge repositories. It is interesting to note that the conditions under which they claim contracts will function as repositories of knowledge match precisely the ones in which we emphasize the adaptive advantages of alliances—the conjunction of instability and interdependence (Mayer and Argyres, 2004). Our study offers a systematic framework based on an extensive tradition of research to study aspects of adaptation in vertical relationships that are not reducible only to issues of cooperation, but also require consideration of coordination. Second, while not the focus of our analysis, several results on control variables in our study point to avenues for further research. Our analysis suggests that the distinction between buyer and supplier asset specificity may be theoretically fruitful. Our results show that buyer asset specificity makes internal and market procurement preferable to alliances, while supplier asset specificity makes market procurement less preferable than internal procurement or alliances. Whether these differences in the effects of supplier and buyer asset specificity merely reflect the institutional context of the automotive industry, or are meaningful in other contests, is a useful line of inquiry. We also find differences between the performance effects of organizational and individual histories of interaction in internal and market procurement. While procurement history enhances the performance of internal procurement, buyer history enhanced the performance of market procurement. Perhaps the benefits of past interactions are transferable from individuals to organizations to different degrees in firm and markets. Third, this study points to the unique organizational role occupied by alliances among various modes of vertical relationships. We predicted and found support for the argument that alliances represent balanced levels of differentiation as well as integration, whereas internal and market procurement emphasize either one or the other. We therefore expected (and found) that alliances adapt more Copyright  2005 John Wiley & Sons, Ltd.

Table 6. Adaptation requirements and optimal procurement mode Low transaction instability Low task Market interdependence procurement High task Internal interdependence procurement

High transaction Instability Market procurement Vertical alliances

successfully to the joint effects of transaction instability and reciprocal interdependence. Our results indicate that when vertical relationships are characterized by low interdependence and high transaction instability, market procurement or alliance may be the preferred mode of organization. In contrast, high levels of task interdependence between procuring and supplying units combined with fairly stable transaction conditions make internal procurement optimal. However, when high levels of instability combine with high levels of interdependence, then vertical relationships may be optimally organized through alliances. These results also have a puzzling aspect: we are unable to detect significant differences between alliances and market procurement in terms of differentiation, and only marginally in terms of integration. Therefore, an obvious question that arises is ‘why are not all external relationships organized as alliances rather than arms length arrangements?’ We speculate that the answer may depend on the costs of establishing each procurement mode (not considered in our analysis). If we assume that setting up market contracting is the least expensive alternative (Williamson, 1991a, 1991b), the normative implications for choosing procurement modes may be summarized as shown in Table 6.

CONCLUSION The literature on adaptation in vertical relationships has been dominated by considerations of hold-up, and increasingly by the debate with its critics who espouse relational norms as an alternative to formal ownership and contractual mechanisms to resolve hold-up (e.g., Granovetter, 1985; Gulati, 1995; Dyer and Singh, 1998; Baker et al., 2002; Poppo and Zenger, 2002). However, the view of adaptation espoused by proponents and Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships critics is essentially similar—cooperative adjustment, given the hazards of opportunistic behavior. In this study, we refocus attention on the impediments to adaptation that arise primarily from bounded rationality, and the organizational attributes that alleviate them. In doing so, we hope to re-emphasize a point that sometimes appears forgotten: that different governance modes are also different modes of organization, with consequences for adaptation beyond the resolution of incentive conflict.

ACKNOWLEDGEMENTS We thank Gautam Ahuja, David Hsu, Dovev Lavie, Kyle Mayer, Jackson Nickerson, Scott Stern, and Maxim Sytch for helpful comments.

REFERENCES Alchian AA, Demsetz H. 1972. Production, information costs, and economic organization. American Economic Review 62(5): 777–795. Anand BN, Khanna T. 2000. Do firms learn to create value? The case of alliances. Strategic Management Journal 21(3): 295–315. Artz KW, Brush TH. 2000. Asset specificity, uncertainty and relational norms: an examination of coordination costs in collaborative strategic alliances. Journal of Economic Behavior and Organization 41(4): 337–362. Ashby WR. 1968. An Introduction to Cybernetics. University Paperbacks: London. Axelrod RM. 1984. The Evolution of Cooperation. Basic Books: New York. Baker G, Gibbons R, Murphy KJ. 2002. Relational contracts and the theory of the firm. Quarterly Journal of Economics 117(1): 39–84. Baldwin CY, Clark KB. 2000. Design Rules: The Power of Modularity. Harvard Business School Press: Cambridge, MA. Barnard C. 1938. The Functions of the Executive. Harvard University Press: Cambridge, MA. Bensaou M, Venkatraman N. 1995. Configurations of interorganizational relationships: a comparison. Management Science 41(9): 1471–1493. Bradach JL, Eccles RG. 1989. Price, authority and trust: from ideal types to plural forms. Annual Review of Sociology 15: 97–118. Camerer C. 2003. Behavioural Game Theory. Russel Sage Foundation: Princeton, NJ. Camerer C, Knez M. 1996. Coordination, organizational boundaries and fads in business practices. Industrial and Corporate Change 5: 89–112. Camerer C, Knez M. 1997. Coordination in organizations: a game theoretic perspective. In Organizational Copyright  2005 John Wiley & Sons, Ltd.

437

Decision Making, Shapira (ed). Cambridge University Press: Cambridge, UK; 158–188. Coase RH. 1937. The nature of the firm. Economica 4: 385–405. Cohen WM, Levinthal DA. 1990. Absorptive capacity: a new perspective on learning and innovation. Administrative Science Quarterly 35(1): 128–153. Conner K, Prahalad CK. 1996. A resource based theory of the firm: knowledge vs. opportunism. Organization Science 7(5): 477–502. Daft RL. 2001. Organization Theory and Design. SouthWestern: Cincinnati, OH. Daft RL, Lengel RH. 1986. Organizational information requirements, media richness and structural design. Management Science 32(5): 554–572. Das TK, Teng B-S. 1998. Between trust and control: developing confidence in partner cooperation in alliances. Academy of Management Review 23(3): 491–512. David RJ, Han S-K. 2004. A systematic assessment of the empirical support for transaction cost economics. Strategic Management Journal 25(1): 39–58. Dess GG, Beard DW. 1984. Dimensions of organizational task environments. Administrative Science Quarterly 29(1): 52–74. Donaldson, L. 2001. The Contingency Theory of Organizations. Sage: Thousand Oaks, CA. Dougherty D. 2001. Reimagining the differentiation and integration of work for sustained product innovation. Organization Science 12(5): 612–631. Duncan RB. 1972. Characteristics of organizational environments and perceived environmental uncertainty. Administrative Science Quarterly 17(3): 313–328. Dyer JH, Singh H. 1998. The relational view: cooperative strategy and sources of interorganizational competitive advantage. Academy of Management Review 23(4): 660–680. Ethiraj SK, Levinthal D. 2004. Modularity and innovation in complex systems. Management Science 50(2): 159–173. Foss NJ. 2001. Leadership, beliefs and coordination: an explorative discussion. Industrial and Corporate Change 10(2): 357–388. Galbraith JR. 1977. Organization Design. AddisonWesley: Reading, MA. Galunic DC, Eisenhardt KM. 1996. The evolution of intracorporate domains: divisional charter losses in high-technology, multidivisional corporations. Organization Science 7(3): 255–283. Galunic DC, Eisenhardt KM. 2001. Architectural innovation and modular corporate forms. Academy of Management Journal 44(6): 1229–1250. Gavetti G, Levinthal D. 2000. Looking forward and looking backward: cognitive and experiential search. Administrative Science Quarterly 45(1): 113–138. Geanakoplos J. 1992. Common knowledge. Journal of Economic Perspectives 6(4): 53–83. Ghoshal S, Moran P. 1996. Bad for practice: a critique of the transaction cost theory. Academy of Management Review 21(1): 13–48. Strat. Mgmt. J., 26: 415–440 (2005)

438

R. Gulati, P. R. Lawrence and P. Puranam

Ghoshal S, Nohria N. 1989. Internal differentiation within multinational corporations. Strategic Management Journal 10(4): 323–337. Granovetter M. 1985. Economic action and social structure: the problem of embeddedness. American Journal of Sociology 91: 481–510. Grant RM. 1996. Toward a knowledge based theory of the firm. Strategic Management Journal , Winter Special Issue 17: 109–122. Grossman SJ, Hart OD. 1986. The costs and benefits of ownership: a theory of vertical and lateral integration. Journal of Political Economy 94(4): 691–720. Gulati R. 1995a. Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Academy of Management Journal 38(1): 85–113. Gulati R. 1995b. Social structure and alliance formation patterns: a longitudinal analysis. Administrative Science Quarterly 40(4): 619–654. Gulati R. 1998. Alliances and networks. Strategic Management Journal 19(4): 293–318. Gulati R, Khanna T, Nohria N. 1994. Unilateral commitments and the importance of process in alliances. Sloan Management Review 35: 61–69. Gulati R, Nickerson J. 2004. Interorganizational-trust, governance choice, and exchange performance. Working paper, Northwestern University. Gulati R, Singh H. 1998. The architecture of cooperation: managing coordination costs and appropriation concerns in strategic alliances. Administrative Science Quarterly 43: 781–794. Gulati R, Sytch M. 2005. Exploring the effects of organizational interdependence on performance: the role of embeddedness and power. Working paper, Kellogg School of Management, Northwestern University. Gupta AK, Govindarajan V. 1986. Resource sharing among SBUs: strategic antecedents and administrative implications. Academy of Management Journal 29(4): 695–715. Hamilton BA, Nickerson JA. 2003. Correcting for endogeneity in strategic management research. Strategic Organization 1(1): 51–78. Hart OD. 1995. Firms, Contracts and Financial Structure. Oxford University Press: Oxford. Heath C, Staudenmayer N. 2000. Coordination neglect: how lay theories of organizing complicate coordination in organizations. In Research in Organizational Behavior, Vol. 22, Staw BM, Staudenmayer N (eds). JAI Press: Greenwich, CT; 153–191. Heide JB, John G. 1990. Alliances in industrial purchasing: the determinants of joint action in buyer–supplier relationships. Journal of Marketing Research 27(1): 24–37. Heide JB, Miner AS. 1992. The shadow of the future: effects of anticipated interaction and frequency of contact on buyer–seller cooperation. Academy of Management Review 35(2): 265–291. Helper S, MacDuffie JP, Sabel C. 2000. Pragmatic collaborations: advancing knowledge while controlling opportunism. Industrial and Corporate Change 9(3): 443–488. Copyright  2005 John Wiley & Sons, Ltd.

Hodgson GM. 2004. Opportunism is not the only reason why firms exist: why an explanatory emphasis on opportunism may mislead management strategy. Industrial and Corporate Change 13(2): 401–418. Holmstrom B. 1982. Moral hazard in teams. Bell Journal of Economics 13(2): 324–341. Iansiti M. 1998. Technology Integration: Making Critical Choices in a Dynamic World . Harvard Business School Press: Boston, MA. Jacobides MG. 2005. Industry change through vertical dis-integration: how and why markets emerged in mortgage banking. Academy of Management Journal (forthcoming). Jacobides MG, Winter SG. 2005. The co-evolution of capabilities and transaction costs: explaining the institutional structure of production. Strategic Management Journal (forthcoming). Kale P, Singh H, Perlmutter H. 2000. Learning and protection of proprietary assets in strategic alliances: building relational capital. Strategic Management Journal 21(3): 217–237. Kauffman S. 1993. The Origins of Order: SelfOrganization and Selection in Evolution. Oxford University Press: Oxford, UK. Klein B, Crawford RG, Alchian AA. 1978. Vertical integration, appropriable rents, and the competitive contracting process. Journal of Law and Economics 21(2): 297–326. Knez M, Camerer C. 1994. Creating expectational assets in the laboratory: coordination in ‘weakestlink’ games. Strategic Management Journal 15(8): 101–119. Knez M, Camerer C. 2000. Increasing cooperation in prisoner’s dilemmas by establishing a precedent of efficiency in coordination games. Organizational Behavior and Human Decision Processes 82(2): 194–216. Kogut B, Zander U. 1992. Knowledge of the firm, combinative capabilities, and the replication of technology. Organization Science 3(3): 383–398. Kogut B, Zander U. 1996. What firms do? Coordination, identity, and learning. Organization Science 7(5): 502–518. Kotabe M, Martin X, Domoto H. 2003. Gaining from vertical partnerships: knowledge transfer, relationship duration and supplier performance improvement in the U.S. and Japanese automotive industries. Strategic Management Journal 24(4): 293–316. Kretschmer T, Puranam P. 2004. Realizing synergies in complex organizations: when are collaborative incentives useful? Working paper, London School of Economics. Kumar N, Stern LW, Anderson JC. 1993. Conducting interorganizational research using key informants. Academy of Management Journal 36(6): 1633–1652. Lawrence PR, Lorsch JW. 1967a. Differentiation and integration in complex organizations. Administrative Science Quarterly 12(1): 1–48. Lawrence PR, Lorsch JW. 1967b. Organizations and Environment: Managing Differentiation and Integration. Harvard University Press: Cambridge, MA. Strat. Mgmt. J., 26: 415–440 (2005)

Adaptation in Vertical Relationships Lee LF. 1982. Some approaches to the correction of selectivity bias. Review of Economic Studies 49: 355–372. Levinthal DA, Warglien M. 1999. Landscape design: designing for local action in complex worlds. Organization Science 10(3): 342–358. Malmgren HB. 1961. Information, expectations and the theory of the firm. Quarterly Journal of Economics 75(3): 399–422. March JG, Simon HA. 1958. Organizations. Wiley: New York. Masten SE. 2002. Modern evidence on the firm. American Economic Review 92(2): 428–432. Mayer KJ, Argyres NS. 2004. Learning to contract: evidence from the personal computer industry. Organization Science 15(4): 394–410. Milgrom P, Roberts J. 1990. The economics of modern manufacturing: technology, strategy, and organization. American Economic Review 80(3): 511–528. Milgrom PR, Roberts J. 1992. Economics, Organization and Management (1st edn). Prentice-Hall: Englewood Cliffs, NJ. Milgrom P, Roberts J. 1995. Complementaries and fit strategy, structure, and organizational change in manufacturing. Journal of Accounting and Economics 19(2/3): 179–208. Mintzberg H. 1979. The Structuring of Organizations. Prentice-Hall: Englewood Cliffs, NJ. Monteverde K. 1995. Technical dialog as an incentive for vertical integration in the semiconductor industry. Management Science 41(10): 1624–1638. Monteverde K, Teece DJ. 1982. Supplier switching costs and vertical integration in the automobile industry. Bell Journal of Economics 13(1): 206–214. Murnighan JK, Roth AE. 1983. Expecting continued play in prisoner’s dilemma games. Journal of Conflict Resolution 27(2): 279–300. Nadler DA, Tushman ML. 1997. Competing by Design: The Power of Organizational Architecture. Oxford University Press: Oxford, UK. Nadler DA, Tushman ML. 1998. Competing by design. Executive Excellence 15(8): 12–13. Nelson R, Winter S. 1982. An Evolutionary Theory of Economic Change. Harvard University Press: Cambridge, MA. Nohria N, Ghoshal S. 1994. Differentiated fit and shared values: alternatives for managing headquarters–subsidiary relations. Strategic Management Journal 15(6): 491–502. Novak S, Eppinger SD. 2001. Sourcing by design: product complexity and the supply chain. Management Science 47(1): 189–204. Oxley JE, Sampson RC. 2004. The scope and governance of international R&D alliances. Strategic Management Journal 25(8/9): 723–749. Petersen T. 1992. Individual, collective, and systems rationality in work groups: dilemmas and markettype solutions. American Journal of Sociology 98(3): 469–511. Phillips LW. 1981. Assessing measurement error in key informant reports: a methodological note on Copyright  2005 John Wiley & Sons, Ltd.

439

organizational analysis in marketing. Journal of Marketing Research 18: 395–415. Poppo L, Zenger T. 1998. Testing alternative theories of the firm: transaction cost, knowledge based and measurement explanations for make or buy decisions in information services. Strategic Management Journal 19(9): 853–877. Poppo L, Zenger T. 2002. Do formal contracts and relational governance function as substitutes or complements? Strategic Management Journal 23(8): 707–725. Puranam P, Singh H. 2002. Structural form in technology acquisitions: knowledge and coordination as antecedents. Working paper, London Business School. Ring PS, Van de Ven AH. 1992. Structuring cooperative relationships between organizations. Strategic Management Journal 13(7): 483–498. Ring PS, Van de Ven AH. 1994. Developmental processes of cooperative interorganizational relationships. Academy of Management Review 19(1): 90–119. Rivkin JW, Siggelkow N. 2003. Balancing search and stability: interdependencies among elements of organizational design. Management Science 49(3): 290–311. Ryall MD, Sampson RC. 2004. Do prior alliances influence contract structure? Evidence from technology alliance contracts. Working paper, University of Maryland. Sanchez R, Mahoney JT. 1996. Modularity, flexibility, and knowledge management in product and organization design. Strategic Management Journal 17(10): 63–76. Schilling MA. 2000. Toward a general modular systems theory and its application to interfirm product modularity. Academy of Management Review 25(2): 312–335. Shaver JM. 1998. Accounting for endogeneity when assessing strategy performance: does entry mode choice affect FDI survival? Management Science 44(4): 571–586. Shelanski HA, Klein PG. 1995. Empirical research in transaction cost economics: a review and assessment. Journal of Law, Economics and Organization 11(2): 335–362. Siegel S. 1956. Nonparametric Statistics for the Behavioral Sciences. McGraw-Hill: Tokyo. Siggelkow N, Levinthal DA. 2003. Temporarily divide to conquer: centralized, decentralized, and reintegrated organizational approaches to exploration and adaptation. Organization Science 14(6): 650–669. Simon H. 1945. Administrative Behavior (3rd edn). Free Press: New York. Simon HA. 1991. Organizations and markets. Journal of Economic Perspectives 5(2): 25–44. Sobrero M, Roberts EB. 2001. The trade-off between efficiency and learning in interorganizational relationships for product development. Management Science 47(4): 493–511. Thompson JD. 1967. Organizations in Action. McGrawHill: New York. Strat. Mgmt. J., 26: 415–440 (2005)

440

R. Gulati, P. R. Lawrence and P. Puranam

Tushman ML, Nadler DA. 1978. Information processing as an integrating concept in organizational design. Academy of Management Review 3(3): 613–625. Van de Ven AH, Delbecq AL. 1974. A task contingent model of work-unit structure. Administrative Science Quarterly 19(2): 183–198. Van de Ven AH, Walker G. 1984. The dynamics of interorganizational coordination. Administrative Science Quarterly 29: 598–621. Van Huyck JB, Battalio RC, Beil RO. 1990. Tacit coordination games, strategic uncertainty, and coordination failure. American Economic Review 80(1): 234–249. Van Huyck JB, Battalio RC, Beil RO. 1991. Strategic uncertainty, equilibrium selection, and coordination failure in average opinion games. Quarterly Journal of Economics 106(3): 885–911. Wageman R, Baker G. 1997. Incentives and cooperation: the joint effects of task and reward interdependence on group. Journal of Organizational Behavior 18(2): 139–159. Walker G, Weber D. 1984. A transaction cost approach to make-or-buy decisions. Administrative Science Quarterly 29(3): 373–392. Walker G, Weber D. 1987. Supplier competition, uncertainty, and make-or-buy decisions. Academy of Management Journal 30(3): 589–597. Weick KE. 1993. The collapse of sensemaking in organizations: the Mann Gulch disaster. Administrative Science Quarterly 38: 628–652. Weick KE. 1995. Sensemaking in Organizations. Sage: Thousand Oaks, CA.

Copyright  2005 John Wiley & Sons, Ltd.

Weick KE, Roberts KH. 1993. Collective mind in organizations: heedful interrelating on flight decks. Administrative Science Quarterly 38(3): 357–382. Williamson OE. 1975. Markets and Hierarchies: Analysis and Antitrust Implications. Free Press: New York. Williamson OE. 1985. The Economic Institutions of Capitalism. Free Press: New York. Williamson OE. 1991a. Comparative economic organization: the analysis of discrete structural alternatives. Administrative Science Quarterly 36: 269–296. Williamson OE. 1991b. The logic of economic organization. In The Nature of The Firm, Williamson OE, Winter S (eds). Oxford University Press: Oxford; 90–117. Wooldridge JM. 2003. Introductory Econometrics: A Modern Approach. Southwestern College Publications: Cincinnati, OH. Zaheer A. 1995. Relational governance as an interorganizational strategy: an empirical test of the role of trust. Strategic Management Journal 16(5): 373–392. Zaheer A, Venkatraman N. 1994. Determinants of electronic integration in the insurance industry: an empirical test. Management Science 40(5): 549–567. Zajac EJ, Olsen CP. 1993. From transaction cost to transactional value analysis: implications for the study of interorganizational strategies. Journal of Management Studies 30(1): 131–146. Zenger TR, Hesterly WS. 1997. The dissaggregation of corporations: selective intervention, high-powered incentives, and molecular units. Organization Science 8(3): 209–235.

Strat. Mgmt. J., 26: 415–440 (2005)