Abstract
While strategic alliances have emerged in recent years as common and important structural vehicles for business development, surprisingly little is known about how collaborative activities are organized and administered within these governance structures. We see classic organizational scholarship as useful insofar as it both provides clear classifications that distinguish alternative intraorganizational designs and explicates how they affect the inner workings of organizations. Existing alliance classification schemes based on type of collaborative activity, partner characteristics, or legal structure, on the other hand, rarely delineate important differences of how collaborative work is organized among partners. We seek to redress this shortcoming by developing a framework of alliance structural parameters based on classic organizational design considerations. Specifically we identify and discuss five key design parameters for alliances: the structural interface between partners, the structural “intraface” within partners, and the specialization, formalization, and centralization of the alliance organization. We show how consideration of these five parameters provides a deeper understanding of alliance governance and suggest how partner organizations can achieve differential levels of connectivity and steering for their collaborative ventures.
Strategic alliances, purposive relationships between firms that share compatible goals and strive for mutual benefits (Ireland, Hitt, & Vaidyanath, 2002; Mohr & Spekman, 1994), have received extensive attention by practitioners and scholars alike during the past two decades (Gulati, 1998, 2007; Kale & Singh, 2009; Parmigiani & Rivera-Santos, 2011). These research efforts have significantly advanced our awareness of the reasons why firms enter into alliances, when alliances create value, and which organizations benefit most from alliances.
Compared to the wealth of scholarship on the antecedents and consequences of organizations’ decision to ally—the before and after—we know much less about what goes on in alliances. Only relatively few studies have opened the black box of alliances’ internal operations (see, e.g., Ariño & de la Torre, 1998; Inkpen & Currall, 2004; Kumar & Nti, 1998; Schreiner, Kale, & Corsten, 2009). As a result, alliances often appear as vacuous, abstract strategic vehicles—deals without organizational or structural substance.
This inattention to the administrative body of interorganizational collaborative relationships is reflected in existing typologies for alliances. Beside activity- and partner-characteristics-based classification schemes, the predominant method of distinguishing alliances is on the basis of their legal foundation—contracts and equity investments—often denoted as the alliance’s “governance structure” (Contractor & Lorange, 1988; Das & Teng, 2001; Gulati, 1998; Yoshino & Rangan, 1995).
Its limited usefulness as a shorthand for actual governance arrangements in an alliance notwithstanding, 1 the legal-structure-based distinction is particularly problematic when it is used as a proxy for alliances’ internal administrative arrangements, as a predictor of alliance dynamics and alliance outcomes. These accounts usually rely on transaction cost economics to suggest that different legal structures afford more or less hierarchical means of coordination and control to alliance partners, and therefore decisively alter the internal operation of the alliance (Gulati & Singh, 1998; Mowery, Oxley, & Silverman, 1996). Such an exclusive reliance on legal form as an explanation for organizational behavior and performance of traditional organizations would appear unusual—after all, we would not expect privately held companies, for example, to be categorically superior or inferior at learning or coordination or to be more efficient or adaptive compared to public corporations. Yet it is common for strategic alliance research.
Gulati and Singh (1998: 783) recognize this problem, noting that the common focus on levels of equity ownership in alliances “masks differences across each type of structure and provides only a partial assessment of the . . . degree of hierarchical control.” Similarly, Reuer and Ariño (2007: 315) remark that “formal governance mechanisms vary a great deal within alliance classes, and that it is plausible that some non-equity alliances may in fact offer partners fairly substantial control just as some equity alliances may be subject to fairly low levels of control, which runs counter to typical depictions of these two classes of collaborative agreements in the literature.” And just as the legal-structure-based classification obscures the underlying levers of coordination and control alliance partners employ to address collaborative risks (Das & Teng, 1996), it also leaves unanswered questions regarding how partners ensure value creation and operational efficiency in the relationship (Zajac & Olsen, 1993).
While legal structure thus seems to offer limited explanatory potential and insight into key strategic and operational concerns of interorganizational partnerships, the role of the alliance’s organizational structure remains largely unexplored. As Grandori (1997: 899) noted, “[W]hile for the internal organization of firms a wide and theoretically based repertoire of organizational forms and procedures for organizational design have been developed, such tools are underdeveloped in the relatively new field of inter-firm organization.” In classic organizational scholarship (Blau & Scott, 1962; Gulick & Urwick, 1937; Lawrence & Lorsch, 1967b; Simon, 1957; Taylor, 1911; Weber, 1974) structure has been central to explaining organizations’ conduct and adaptiveness in particular contexts, and how they succeed or fail in leveraging their resource endowments. A closer examination of alliances’ organization structure, therefore, appears to be a promising direction to address unanswered questions about alliance dynamics and performance.
Hence, we suggest a new typology of alliance structure rooted in the classic organizational design literature. 2 In the tradition of Mintzberg (1979) and Galbraith (1973) we develop a multidimensional framework for alliance organizational structures that goes beyond the predominant legal-structure-based distinctions. With our framework we seek to capture key differences in the internal administrative arrangements made to manage the division of labor, integrate disparate activities, enable collaborative value creation, and foster productive relationships among employees from different partner organizations. Specifically, we suggest five dimensions of alliance organizational structure. The interface and intraface dimensions capture which organizational members are involved in the alliance and how they are connected to each other within and across partnering organizations. Specialization captures the degree to which organizational members involved in the alliance focus exclusively on alliance management tasks. Formalization describes the codification and standardization of alliance activities, and centralization the allocation of decision-making authority in the partnership.
This framework for alliances’ organizational structure makes a number of contributions to the alliance literature. First, it highlights key differences between organizational economic and organization theory approaches to alliance design. While the former provides a robust analytical framework for issues of opportunism and forbearance, the latter foregrounds efficiency, effectiveness, and adaptiveness as central considerations for alliance structures (Li, Zhou, & Zajac, 2009). Second, it integrates and further develops recent alliance scholarship on individual elements of alliance designs, such as contractual control and coordination provisions (Vanneste & Puranam, 2010), decision-making committees (Reuer & Devarakonda, 2012), and internal alliance functions or departments (Kale, Dyer, & Singh, 2002). Jointly the five dimensions of alliance organizational structure suggested in our framework provide a nuanced description of the connectivity among partners, and their provisions for the steering of joint efforts. By unbundling alliance governance into five key design components, our framework provides insights into design trade-offs that go beyond binary choices, such as equity versus nonequity alliance structures. Third, the framework allows a more penetrating assessment of an alliance’s distinctive challenges, and its likely success of implementing plans and attaining set goals (Das & Teng, 2001). Specifically, it sheds light on the structural determinants of key alliance dynamics—such as interorganizational coordination, learning, and the development of interpartner trust.
We begin our argument with a brief review of existing alliance classification schemes and their key advantages and disadvantages. Here we focus particularly on existing classifications’ shortcomings for examining interorganizational coordination, learning, and trust. We then develop in detail the five parameters of alliance organizational structure—interface, intraface, specialization, formalization, and centralization—and elaborate how they affect coordination, learning, and trust. We then use these five parameters to discuss the design challenges facing alliance partners when crafting alliance designs and address the implications of our approach for alliance research and practice.
Review of Existing Alliance Classifications
Alliances are a particularly complex organizational phenomenon. Utilized across a broad range of contexts, alliances can involve a wide variety of configurations of partners, involve the pursuit of a multitude of specific goals, and exhibit various levels of commitment and investment from partners. This phenomenological complexity renders most attempts at developing generalized theories of alliancing behavior and alliance outcomes problematic. Some scholars have acknowledged this challenge by carefully limiting their object of study to a narrower subset within the broad phenomenological space—they focus, for example, exclusively on R&D alliances (Li, Eden, Hitt, & Ireland, 2008), joint ventures (Inkpen & Currall, 2004; Kogut, 1988), international alliances (Oxley & Sampson, 2004), or buyer–supplier alliances (Mesquita & Brush, 2008).
Some scholars have attempted to deal with the phenomenological complexity of alliances more directly by suggesting classification criteria for them. Our review of the alliance literature revealed three common sets of criteria for distinguishing types of alliances. 3
Distinctions of alliances along activity-domain-, partner-characteristics-, and legal-structure-focused criteria are helpful when explaining alliance formation and partnering risks. For example, we can reasonably expect a horizontal explorative R&D alliance between firms operating in the same industry and markets (e.g., the alliance between Apple and Google for the early iPhone models’ key apps) to be wrought with more appropriation concerns and rivalry than an alliance among noncompetitors for sharing logistics infrastructure (e.g., the Nabisco–Dole alliance for pooling of trucking resources). We would similarly expect a multiparty alliance composed of firms from different industries and geographic regions to be more difficult to coordinate than a dyadic alliance between two biotech firms colocated in North Carolina’s Research Triangle Park.
Due to their focus on initial conditions of the alliance (Who is involved? What are the declared goals? What is the negotiated deal structure?), activity-domain-, partner-characteristics-, and legal-structure-focused criteria are less suitable for explaining systematic differences in alliance dynamics (How do partners deal with the various challenges and risks of their collaborative efforts over time?) and can provide only a partial account of alliance outcomes (What do partners ultimately get out of their collaboration?; Doz, 1996; Khanna, Gulati, & Nohria, 1998; Reuer, Zollo, & Singh, 2002). More specifically, existing criteria sets for classifying alliances hamper advancement of our understanding of alliance dynamics and outcomes due to three key shortcomings: (a) they obscure substantial differences in coordination arrangements made within the same type of alliance, (b) they help explain downside alliance risk, and risk mitigation choices, but not value creation potential or alliance learning, and (c) they ignore interpersonal social and political dynamics in alliances that can impede or improve trust. The following paragraphs elaborate on these three issues.
Obscuring Key Differences
An understanding of how work is organized within an organization is an important prerequisite to explaining how organizations evolve and adapt to their environments (Burns & Stalker, 1961; Cyert & March, 1963; Simon, 1957) and to explain performance differences (Dalton, Todor, Spendolini, Fielding, & Porter, 1980). Central to the organization of work are mechanisms for coordination, that is, the alignment and adjustment of activities across organizational units (Gulati, Wohlgezogen, & Zhelyazkov, 2012; Puranam, Raveendran, & Knudsen, 2012), and mechanisms for control, that is, the evaluation and interventions that ensure individual efforts contribute to organizational goals (Ouchi, 1979). To better understand how alliances evolve and adapt, and how they perform, then, we need a better understanding of coordination and control in alliances. Unfortunately, existing alliance typologies do not directly consider such internal administrative arrangements.
Some studies have suggested legal structures as a useful proxy for internal administrative arrangements in the alliance (Das & Teng, 2001; Gulati & Singh, 1998). However, alliance practitioners’ accounts (Bamford, Gomes-Casseres, & Robinson, 2003) call into question whether legal structure really provides an unequivocal differentiation of distinctive internal organization of an alliance. They reveal that contractual alliances can involve elaborate, hierarchical coordination provisions (e.g., the Walmart and Procter & Gamble alliance) or only rudimentary coordination (e.g., the alliance between fashion retailer the Gap and cause marketing organization (RED)). Likewise some joint ventures (JVs) exist primarily as legal vehicles—for example, to more easily attract funding or commercialize partners’ pooled or jointly created IP—with only threadbare administrative structures, while others take the form of complex, fully fledged multinational corporations (e.g., Dow-Corning JV). In short, alliance organizational structures can be designed in a variety of ways, to a large degree independent of the chosen legal structure of the relationship. An exclusive focus on legal structure conceals this variety of structural profiles and discourages closer empirical attention to those specific internal coordination arrangements.
Focus on Alliance Risk, Not Value
A large number of alliance studies point to high failure rates (Kale et al., 2002), to high transaction costs involved in negotiating and monitoring alliance deals (Argyres & Mayer, 2007), and to problematic uncertainties related to the appropriation of alliance benefits (Park & Ungson, 2001). In response they often provide suggestions for the selection of partners and legal structures to reduce failure risks, transaction costs, and misappropriation. With its strong focus on the guarding against downsides of collaboration, alliance research gives relatively little insight into sources of variation in alliance value creation (Zajac & Olsen, 1993). Complementarity of partners’ resources and capabilities have been suggested as an important basis for value creation (Eisenhardt & Schoonhoven, 1996; Harrison, Hitt, Hoskisson, & Ireland, 2001) but is often defined post hoc as differences that happened to create value—differences that failed to create value are labeled as incompatibilities. Evidence regarding which measures help the synergistic combination of partners’ resources abounds in the intraorganizational literature (Cohen & Levinthal, 1990; Grant, 1996), but in the alliance literature evidence remains sporadic. While research on alliance learning overcomes some of these shortcomings (Inkpen & Tsang, 2007; Lam, 1997; Lane & Lubatkin, 1998; Lane, Salk, & Lyles, 2001), existing alliance typologies provide little help with distinguishing successful and unsuccessful cases of alliance learning: Are partnerships between similar or dissimilar firms more likely to yield learning benefits? Are JVs better than contracts for alliance learning? These questions remain largely unanswered, and the underlying enablers of productive knowledge exchange and integration remain largely unexamined.
Ignoring Interpersonal Social/Political Dynamics in Alliances
The antecedents and consequences of trust in strategic alliances have long been a focal interest of alliance scholars, and there is broad consensus that trust plays an important role in explaining alliancing behavior and ultimately alliance success (Gulati & Nickerson, 2008; McEvily & Zaheer, 2006; Rousseau, Sitkin, Burt, & Camerer, 1998). Most studies on trust in alliances, however, regard the partnering organizations as monolithic entities, treat trust as an organization-level variable, and assume away the messy interpersonal dynamics inherent in boundary work (Zaheer, McEvily, & Perrone, 1998). They imply individuals within an organization all have equal levels of trust in a partner organization (“Can I trust the partner organization given what I know/believe about the organization as a whole?”). Recently, however, scholars have increasingly suggested that interorganizational trust rests at least partially on interpersonal trust that exists among individual members of partnering organizations (“Can I trust the partner organization given what I know/believe about the individual(s) from that organization.”). 5
Existing alliance classifications do not allow us to predict which alliances succeed in fostering interpersonal trust and which fail. If we recognize interpersonal trust as an important foundation for interorganizational trust and seek to explain how such trust develops, we need to understand what kind of opportunities exist for individual members of the partnering organizations to interact, and what context is provided for these interactions (Castelfranchi & Falcone, 2010).
We can conclude that existing alliance typologies provide little support to identify systematic differences in alliances regarding the key issues of coordination, learning, and trust. Existing classifications, in particular those relying on legal-structure-based criteria, obscure and mask more than they clarify important differences in alliances dynamics and outcomes because they fail to penetrate the alliance “black box” and to distinguish different internal organizations of alliances.
Promising Developments for a Reconceptualization of Alliance Structure
Following calls for a more nuanced analysis of administrative structures and processes in interorganizational relations (Grandori, 1997; Provan & Milward, 2002), a number of recent studies have moved beyond partner-, activity-, or legal-structure-based categorizations of alliances to theorize differences in alliance dynamics and outcomes. They have assessed directly—through surveys, field-based qualitative research, or in-depth contract analysis—the concrete administrative provisions established by partnering organizations, and their impact on collaborative processes and outcomes.
Some of these recent studies provide a closer look at how exactly partner firms are connected through interpersonal networks (Berends, Garud, Debackere, & Weggeman, 2011; Davis, 2011; Davis & Eisenhardt, 2011) or committee structures (Reuer & Devarakonda, 2012). Others have focused on how decision making is structured in alliances (Bensaou & Venkatraman, 1995), often through detailed analysis of alliance contracts (Mellewigt, Decker, & Eckhard, 2012) to identify clauses that define technical or administrative arrangements, and to explain under which circumstances partners emphasize these arrangements (Vanneste & Puranam, 2010). Last, a number of studies have examined the establishment and role of an alliance management function in organizations (Kale et al., 2002; Schreiner et al., 2009). This stream of work highlights the importance of structural and processual anchors within partnering organizations for shaping alliancing behavior and ensuring positive alliance outcomes.
Taken together, these recent studies make important advances in opening the black box of alliance organization and, with their emphasis on organizational structures and processes, hark back to classic organization design concerns. 6 We integrate these disparate streams of research to develop a comprehensive framework of key organizational—not legal—design parameters of alliances. This framework will help us to better explain differences in coordination, learning, and trust in interorganizational relations.
Organizational Design Dimensions of Alliance Structure
Conceptual Basis
Formal organizational structures group and connect individuals within an organization. Departments, committees, boards, teams, and task forces create temporary or permanent channels for individuals to come together and interact (Allison, 1971; Blau & Scott, 1962; Lawrence & Lorsch, 1967a; Simon, 1957). Role descriptions and task assignments typically specify reporting or supervisory duties, and thus create connections among individuals. 7 Essentially then, organizational structures establish networks of ties among organizational members. These ties in turn are fundamental to effective administration, as Thompson has recognized: “Administration is not something done at one level in the organization, but is a process spanning and linking levels” (Thompson, 1967: 149).
Alliance structures, likewise, group and connect individuals of partnering organizations across organizational boundaries (Bouty, 2000; Davis & Eisenhardt, 2011). They connect decision makers and contributors from various positions within partnering organizations (e.g., from different levels of hierarchy) and assign responsibilities for the formulation of alliance strategy and for the operation of the alliance—in short, they allow partners to administer their collaborative efforts. Thus, what we often conceptualize as a single tie between allying organizations is on closer inspection a complex set of ties among the individual employees and managers from the partnering organizations (Adobor, 2006; Alter, 1990; Lee, 2011; Matthiesen, 1971; Seabright, Levinthal, & Fichman, 1992). And whereas the existence of an interorganizational tie provides little information about the administrative makeup of the relationship, the interpersonal network among partner organizations’ employees provides us with a highly useful point of departure to analyze how the partnership is structured and managed.
Who exactly is involved in an alliance’s interpersonal networks? Arguably the most important individuals involved are the boundary spanners, those individuals who are traditionally recognized for connecting the organization with its environment (Aldrich & Herker, 1977), and which in our context interact directly with members of the other partner organizations. The set of relations among boundary spanners we designate as the interface among alliance partners. In addition to these boundary spanners, any other individual in the respective member organizations who is formally involved in and (at least partially) responsible for governing, managing, operating, or contributing to the alliance, even if he or she does not directly interact with members of the external partners, is also part of the network (see Figure 1). These individuals receive information, decisions, or requests from boundary spanners, or feed information and requests to boundary spanners. We call the network of ties between non–boundary spanners and boundary spanners within the partner organizations the intraface of the respective partner.
Visualizing Alliance Organizational Structure
We further typify the alliance’s interpersonal network with regard to task assignments and specification of decision-making roles. Specifically, we suggest as key dimensions of alliance design specialization, the degree to which organizational members involved in the alliance focus exclusively on alliance management tasks, formalization, the definition of rules, procedures, and plans, and centralization, the degree to which alliance decisions are concentrated within the organizational hierarchies of the partnering organizations. 8 All three dimensions have been consistently regarded as fundamental structural properties in classic organization design scholarship (Pugh, Hickson, Hinings, & Turner, 1968). More recent empirical work on the professionalization and differentiation of alliance management in organizations (Kale et al., 2002; Vlaar, van den Bosch, & Volberda, 2007) and on the topography and structuring of decision-making channels (Grandori & Soda, 1995; Reuer & Devarakonda, 2012) also suggests that these three dimensions are also crucial for alliance dynamics and outcomes.
This structural configuration of the alliance, we argue, has significant implications for communication and decision making, and for flexibility and adaptiveness in the alliance—independent of the chosen legal structure of the alliance. In the subsequent paragraphs we elaborate and illustrate each of these five dimensions of an alliance’s organizational structure and explain their impact on the key alliance concerns of coordination, trust, and learning (see Table 1). 9
Alliance Design Parameters and Their Impact on Key Alliance Issues
Interface
A key structural parameter of alliances is the network of interpersonal ties among the partner organizations’ boundary-spanning employees—the interface among alliance partners. Boundary spanners represent their respective organization to the other partners; they are the voice and the eyes and ears for their organization in the alliance (Aldrich & Herker, 1977). They are key conduits for information exchange among partners and act as agents for their organization in the partnership. 10 As such, they are crucial to alliance success (Hoegl, Weinkauf, & Gemuenden, 2004; Provan & Milward, 1995).
The interface between partners can vary in strength, that is, the number and type of boundary spanners involved, the number of connections among them, and the intensity of their interaction. 11 We can term these parameters of interface strength the scope, density, and activity. The scope of the interface reflects the location of boundary spanners within their respective organizations. Boundary spanners involved in the alliance may come from different domains such as functional, geographic, or business units (horizontal scope) and from different levels of hierarchy (vertical scope). The density of the interface describes the number of ties that exist among boundary spanners from partnering organizations. While a low-density interface features only few connections among boundary spanners, a high-density interface provides multiple redundant pathways for communication and a higher degree of network closure among boundary spanners. The activity of the interface describes the frequency and intensity of interaction among boundary spanners (Bensaou & Venkatraman, 1995; Bouty, 2000). 12
The Renault–Nissan alliance, founded in 1999, features an especially broad, dense, and active interface. 13 The two carmakers are connected through numerous cross-company teams and steering committees that identify and implement cost savings, joint marketing, and joint R&D projects, and which constitute a horizontally and vertically wide scope and high-density interface. In addition, the partners have established a 50–50 JV, the “Renault–Nissan BV.” It serves as the seat of the alliance board, which is responsible for alliance strategy, centralizes logistics and IT services for much of the alliance, and houses the Renault–Nissan Purchasing Organization, which negotiates supplier contracts on behalf of both parent firms and supports the implementation and monitoring of these contracts. 14 Last, both firms have appointed a team of dedicated alliance directors who oversee and advise the cross-company teams and committees to accelerate best-practice sharing and increase synergies, that is, to stimulate more activity across the interface, but are also responsible for identifying and counteracting activities within the two partner organizations that endanger alliance synergies. 15
The alliance interface has significant impact on coordination, learning, and trust among partners. First and foremost, the scope, density, and strength of linkages between the partners affect coordination among partners. The interface provides the critical linkages that integrate partners’ decentralized activities and contributions to the alliance. It crucially determines the information-processing capacity between partners (Aldrich & Herker, 1977; Galbraith, 1974) but also the quality of interactions. Given individuals’ difficulties with identifying relevant interdependencies (Heath & Staudenmayer, 2000; Sherman & Keller, 2011) and uncertainties (Milliken, 1987), a broader and denser network of boundary spanners can triangulate perceptions of interdependencies and uncertainties, and cover individuals’ cognitive blind spots (Zajac & Bazerman, 1991). Such a quantitatively and qualitatively improved cumulative attentional capacity of the alliance organization improves coordination, both in a static sense, that is, the formulation of agreements and plans for alliance activities and monitoring to ensure congruence of actual partnering activities with those agreements, and in a dynamic sense, that is, the identification of and adjustment to changing conditions such as new emerging interdependencies and uncertainties. However, a broad interface can also cause partners to struggle more to impose control over what information is shared and how decisions are made: the more extensively partners interface with each other, the more likely unintended, incorrect, or inconsistent communication and decisions occur before alliance managers can intervene to prevent them. In addition, greater effort is required to redirect this larger group of boundary spanners in the face of changing external or internal circumstances for the alliance. Beyond a certain threshold, the complexity and control disadvantages of a strong interface can be expected to eclipse its information-processing advantages.
Beyond their role in coordination, boundary spanners’ ties also play a central role in developing trust among partnering organizations (Perrone, Zaheer, & McEvily, 2003). Since an important driver of interorganizational trust is the interpersonal trust among members of both organizations, establishing and cultivating a broad set of strong ties that give more opportunities for members of partnering organizations to interact, to get to know and understand each other, and independently witness benevolence and integrity of managers from a partner organization puts the alliance on a broader foundation of affective, benevolence-based trust (Greenhalgh, 2001). It makes the alliance less vulnerable to the particular interpersonal “chemistry” or potential antipathy between any two boundary spanners. In addition, a higher density of ties in the boundary spanner network means a higher degree of network closure, which supports the development of affective trust (Coleman, 1988). The higher degree of exposure to individuals from partner organizations and their activities in a broad, dense, and active interface can potentially also provide a basis for developing instrumental, competence-based trust, that is, confidence in partners’ abilities to deliver results and perform tasks reliably and as agreed (Das & Teng, 1998; Levin & Cross, 2004; McAllister, 1995). However, outcomes, rather than process, may be more important for the development of instrumental trust. In that regard, transparency of the processes used to achieve outcomes may pose a risk for assessments of instrumental trust: an organization may be content with outcomes achieved by a partner but object to the way in which they are achieved (as a consequence of knowing “how the sausage is made”) and thus may, despite reliable outcomes, question that partner’s underlying abilities.
Last, the interface affects alliance learning. A single boundary spanner representing his or her organization in the alliance can quickly become overburdened with trying to understand a partner organization’s internal structures, processes, resources, capabilities, and so on and may struggle to adequately transmit and translate key insights about the partner back to his or her own organization. Hence a narrow interface may limit partners’ ability to develop alliance-specific routines (Zollo, Reuer, & Singh, 2002) that would help render the present collaborative venture more efficient and effective, and it may also limit partners’ ability to innovate jointly and search more broadly for collaborative opportunities. A broader interface can remedy these limitations. The more boundary spanners from different levels and functions are involved in the interface, and the denser and more active their linkages, the more rapidly organizations are likely to accumulate knowledge about each other and the task at hand and about how to leverage and integrate each other’s resources and capabilities (Ngai, Jin, & Liang, 2008; Nielsen, 2005; Uzzi, 1997). Hence a broad, dense, and active interface can promote exploitation-focused learning, that is, the refinement and improvement of existing processes and capabilities (March, 1991). It may also enable more explorative forms of learning, that is, development of innovative, new processes and capabilities, especially when mutual exposure to multiple representatives from different units and functions may allow partners to uncover collaborative opportunities that otherwise would go unnoticed.
In many cases, alliance partners cannot fully define an optimal interface at the outset of the relationship: They may struggle to specify in advance which individuals should manage the alliance, and with which individuals from the partnering organizations they should interact. Furthermore, strategic changes in the alliance may necessitate change to the interface. Hence, the interface among partnering organizations usually evolves over time. Initial boundary spanners play a crucial role in steering this evolution: Given their direct interaction with the partner organizations, they can initiate new ties among the partners, and nominate and involve new boundary spanners from their own organizations (Davis & Eisenhardt, 2011; Sieg, Wallin, & Von Krogh, 2010)—thus contributing to an increasing scope or density of the interface.
Intraface
Significantly less examined than ties among alliance boundary spanners are the ties between the boundary spanners and their non-boundary-spanning colleagues within their own organizations. We call this parameter of the alliance structure the “intraface.” Whereas the interface describes an interorganizational network of ties, ties that reach out from an organization to its alliance partners, the intraface describes an alliance partner’s intraorganizational network of ties among employees involved in the alliance, ties that reach back into a participating organization’s internal structure.
The intraface part of the alliance structures is often less visible than the interface. Frequently, alliance contracts specify boundary spanners and boundary-spanning decision-making committees and task forces (Reuer & Devarakonda, 2012) and partners openly communicate in their press releases about which executives from the respective partner organizations are going to work together in the context of the alliance. But they usually do not detail the intraorganizational groupings and linkages that are created to implement the partnering agreement. However, the intraface established by partners is often crucial to both the value creation and the value capturing in an alliance as it connects the alliance activities back to parent organizations’ “regular” in-house activities and resources—its technical core. Some alliance studies (Corsten & Kumar, 2005) have identified adequate internal structures, that is, intrafaces, as a prerequisite to firms’ decisions to establish complex, embedded interorganizational relationships.
Procter & Gamble, for example, has established a complex internal organization to operate and develop the strategic alliance with its largest customer, Walmart (Galbraith, 2010). P&G has implemented the alliance intraface as a formal multidimensional matrix: members of global functional units, business units that sell products to Walmart, and regional units for countries in which Walmart operates, all report to P&G’s Walmart Team leader (in addition to their respective functional, regional, and business unit heads). P&G’s dedicated Walmart Team—which in 1992 contained 70 employees and by 2010 had grown to 250—serves as the direct interface to Walmart, coordinates and facilitates information exchange and collaborative initiatives with Walmart, and helps connect Walmart employees to P&G’s relevant internal experts and vice versa when needed. P&G’s internal organizational structure for the alliance has been instrumental to effectively leverage its internal capabilities for its downstream partner, and identify and implement diverse new partnering opportunities—from joint operations and logistics projects like RFID and, more recently, QR code implementations, to joint marketing initiatives, such as the 2012 London Olympics in-store promotions, and the joint development of TV feature films for FOX and NBC with the “Family Film Night” initiative.
As this example illustrates, even purely contractual alliances can involve formal structures within partner organizations, composed of boundary spanning and non-boundary-spanning employees involved in the partnership. The internal network among these individuals, its composition and structure, constitutes the intraface. It can vary in strength for different partners and can be characterized—analogously to the interface—by scope, density, and the level of activity across its ties. 16
The intraface affects coordination, learning, and trust within the alliance. It is important for coordination among partners, since boundary spanners are often merely information conduits and relays who need other organizational members to plan, evaluate, and perform crucial activities for the alliance. Boundary spanners’ ties to internal specialists can be instrumental for assessments of feasibility (Do we really have the capabilities or capacity for the alliance?) and usefulness (What do we stand to gain from the alliance?) of planned joint projects, and for estimates of resource requirements and time schedules.
The configuration and quality of relations between non–boundary spanners and boundary spanners is vital to the effective implementation of alliance plans. These ties allow partners to identify and leverage those relevant tangible and intangible internal resources that boundary spanners themselves are not aware of or have access to or control over. As a result, the intraface strongly affects how partners address emerging issues in the alliance, that is, which resources they (can) bring into play to tackle these issues, and ultimately how successful they are in resolving problems over the course of the alliance life cycle.
The alliance partners’ intraface also helps ensure that jointly pursued activities are aligned with partners’ internal non-alliance-related activities and dynamically adjusted to environmental changes (Aggarwal, Siggelkow, & Singh, 2011; Das & Teng, 2000; Lavie & Singh, 2012). As the Renault–Nissan example illustrated, alliance teams need to ensure that alliance activities do not interfere with partners’ other internal activities, and conversely identify internal activities that may conflict with alliance activities. 17 As a consequence of the intraface’s role in facilitating problem solving and adaptiveness in the partnership, it can foster the development of instrumental trust.
The intraface is also crucial for alliance learning. It is often the non-boundary-spanning specialists in the organization, not the boundary spanners themselves, who benefit most from the information and knowledge gleaned from alliance partners. The intraface is instrumental for the internal diffusion of task and partner-related knowledge, and its absorption (Lane et al., 2001; Tsai, 2001). Boundary spanners’ linkages back into their own organizations ensure that their insights and information gleaned from partners reach those who most need them and can best act on them.
Specialization
The interface and intraface describe the scope of and relations among individuals involved in the alliance, but they say little about their level of involvement. This level of involvement is described by the third parameter in our framework—the specialization within the alliance. Based on the traditional understanding of specialization as the number of different tasks a single individual in an organization has to carry out (Mintzberg, 1980), specialization within the alliance refers to the degree to which alliance activities are differentiated from other organizational activities within the partners’ internal organization. 18
Differentiation can be the cause or consequence of structural separation (Lawrence & Lorsch, 1967a) of individuals involved in the alliance from the “normal running business” of their organization. Such separation is often achieved by forming JV companies to house alliance-related activities. It can also be achieved by establishing dedicated alliance units within the partner organizations, which group together individuals from different organizational units (e.g., marketing, legal, procurement, logistics, IT, etc.) to jointly work on an individual alliance—representing a high degree of specialization—or on multiple alliances—representing a medium degree of specialization. 19 Often, however, employees involved in an alliance have numerous other, non-alliance-related responsibilities within their respective organizations. For them, the alliance is just one of many projects they have to “juggle,” and their level of alliance specialization is low. The higher the number of alliance-specialized positions and the higher their degree of specialization, the higher the degree of specialization in the alliance structure overall. Figure 2 illustrates variations of alliance structures along the specialization and intraface parameters.
Illustrative Variations of Alliance Structures With Regard to Specialization and Intraface Parameters
Pfizer serves an example of increasing specialization for its alliance activities. In 2002 the company changed its R&D organizational structure and introduced a specialized vice president position for “worldwide research and technology alliances,” a role that reported directly to the executive board member responsible for global R&D. 20 At the time, the VP oversaw Pfizer’s more than 500 alliances (the alliance portfolio rapidly grew to more than 1,000 alliances in the subsequent years). His department supported the formation of new alliances by convening internal experts from across the corporation for evaluations of potential partners, and supported their due diligence efforts with IT infrastructure and tools for data collection and evaluation of commercial and academic drug R&D worldwide. To support the management of established alliances, the department deployed an IT system titled “Strategic Technologies, Alliances and Relationships” (STAR), which helped alliance teams to manage information throughout the alliance life cycle, from an initial contact with a partner through to the completion of a partnership. 21 These activities of the headquartered global alliance management organization are today supplemented by dedicated alliance management positions in many of Pfizer’s local subsidiaries.
Specialization in the alliance has implications for coordination, learning, and—to a lesser degree—trust. Alliance specialists are able to improve static and dynamic coordination in the alliance (Schreiner et al., 2009). Since specialized managers can focus more of their limited attentional resources on their area of specialization rather than on miscellaneous other responsibilities (cf. Perrow, 1977), alliance management specialists are more likely to notice incompatibilities of action, misallocation of resources, and omissions of action in the alliance, and more able to dynamically adjust to environmental or internal changes. The coordination benefits and the symbolic value of specialized alliance roles and units—a signal of the importance of the alliance for the organization, and the status of the “alliance professional”—can foster instrumental trust among partners.
In addition, increasing levels of specialization support exploitative learning as alliance managers’ attentional focus encourage the continuous improvement and optimization of partnering processes and systems. Explorative learning, however, is more likely with medium levels of specialization, that is, alliance managers dealing with multiple alliances, which exposes individuals to a broader scope of partnering opportunities and challenges (Sidhu, Volberda, & Commandeur, 2004) and allows individuals to assess a focal alliance with more cognitive distance in the context of the entire alliance portfolio and alliance strategy (Nooteboom, Van Haverbeke, Duysters, Gilsing, & Van Den Oord, 2007).
Dedicated alliance units and roles may initially increase alliance administrative costs, but in the long run may contribute to lower overall costs and higher value from partnering, if their specialization contributes to better preventing coordination failures, resolving partner conflict, and streamlining identification, formation, and management of alliances.
Formalization
Specialization of individuals or organizational units on alliance tasks is often related to formalization, the specification and standardization of rules, procedures, plans, and documentation to guide alliance activities (Pugh et al., 1968). 22 The greater the variety of tasks and contingencies that are covered by rules, procedures, and documentation requirements, and the more detailed the prescribed standardized responses are, the more highly formalized is the alliance structure (Albers, 2010). Often these formal arrangements can be found as technical or administrative clauses in alliance contracts (Vanneste & Puranam, 2010), but also in nonlegal documents such as employee handbooks, operation manuals, guidelines, and so on. As Grandori and Soda (1995: 197-198) note, “The substance of an inter-firm cooperative agreement . . . can vary substantially in its degree of formalization. . . . [T]he whole body of literature on networks shows that the extent to which inter-firm relationships are formalized . . . is an important dimension of inter-firm organizing—as it is of any sort of organizing.” Recent studies have examined the performance implications of formalization in interorganizational relationships (Vlaar et al., 2007), noting both the potential benefits that accrue from a greater level of clarity about responsibilities and processes, and the threat of rigidity and bureaucratic drag.
Formalization affects coordination, learning, and trust in the alliance. Specifying standardized responses, plans, metrics, and documentation of past actions and decisions enhances transparency of “who is doing what, when” in the alliance and aides monitoring (Callahan & MacKenzie, 1999). It allows actors to anticipate, with some certainty, what others are doing in specific circumstances, thus facilitating planning, joint performance of tasks, and feedback-based adjustments (Galbraith, 1973; Lawrence & Lorsch, 1967b). This predictability can engender instrumental trust (Walter, Kellermanns, & Lechner, 2012).
Formalization can, however, limit dynamic coordination in the alliance, its ability to adapt to unforeseen changes, challenges, or opportunities: Per definition it restricts the range of individuals’ responses to issues and occurrences, and the prescribed standardized responses are bound to imperfectly fit the specific issues encountered by alliance partners. Once partners have committed to detailed formal arrangements they may be hesitant to question and renegotiate these arrangements (Reuer & Ariño, 2002).
The standardization of procedures and structures for interactions allows organizations to scale alliancing activities to very large partner portfolios. Disney’s consumer products licensing department and Apple’s developer relations team, for example, utilize highly standardized contracts and processes as a baseline to streamline interactions with literally thousands of partners. Oracle has utilized these standardized procedures to effectively automate partnering (Dyer, Kale, & Singh, 2001): Potential partners use a web interface to select their desired tier of partnership, provide detailed information about their own capabilities, review terms and conditions, and sign up as a partner without ever having to speak with Oracle’s alliance management specialists.
With regard to learning, formalization offers both benefits and drawbacks. To the degree that formalized procedures distill and encapsulate experiential or vicarious learning, formalization fosters the diffusion and adoption of best practices in the alliance—thus improving exploitation-focused learning. They make insights about how best to respond to specific alliance issues available even to those individuals who have little alliance experience. In some cases, existing rules and procedures can become a focal point for improvement efforts (Feldman & Pentland, 2003). However, standards and rules can lead individuals to process new situations and information less deliberately and thoroughly, thus limiting the identification of issue-response misfit and inhibiting even exploitation-focused learning (Schulz, 1998). Exploration-focused learning is generally inhibited by formalization.
Formalization can also affect the development of trust in the partnership. Perrone and colleagues (2003) suggest that role autonomy allows boundary spanners to engage in discretionary action, which provides clues to their “motives and intentions” and thus engender interpersonal affective trust. This finding corresponds with the more general argument that narrowly defined, restrictive partnering agreements can undermine the development of goodwill and affective trust among partners (Greenhalgh, 2001). Formalization can contribute to instrumental trust among partners, if rules and standards help to instill confidence in the predictability and reliability of partnering activities. However, formal rigidities can also limit partners’ ability to leverage internal resources and capabilities for the partnership, and may even suppress partners’ existing change management skills that would allow the alliance to adapt to changing conditions. As a result, partners may misattribute alliance problems and failures—that are really rooted in the constraining structure chosen for the partnership—to partners’ inabilities and incompetence (Gulati et al., 2012).
Toyota, for example, seeks to avoid the learning and trust problems that can result from high formalization. Its relationships—and contracts—with its supplier community are often deliberately low in formalization, in some cases represent little more than a general statement of intent (Dyer & Nobeoka, 2000). They rely on implicit social pressure for commitment and cooperation, and partners’ willingness and initiative to adapt and improve the relationship.
Centralization
Centralization in the alliance structure describes the locus of decision-making rights (Albers, 2010), specifically who holds the authority to make planning and implementation decisions for the alliance. Since authority in many alliances is shared between alliance partners—involving joint decision making by one or more representatives from both partner organizations (from whatever unit or level of hierarchy within those organizations)—it makes sense to think about the locus of decision as residing not within a particular role but within a decision-making channel. Allison (1969: 710) describes such channels as “regularized ways of producing action concerning types of issues, [which] structure the game by pre-selecting the major players.”
An alliance may be vertically centralized, concentrating decision-making authority in a single channel, for example in an alliance management board exclusively involving senior executives. Alternatively, partners may establish multiple decision-making channels, involving different sets or committees of representatives to develop consensus solutions for different domains of alliance issues (Reuer & Devarakonda, 2012). Such delegation of decision authority leads to a shift in individuals’ attentional focus and more effective problem solving (Grant, 1996; Perrow, 1977). Decentralized decision-making channels can be functionally oriented (such as channels for strategy, marketing, logistics, or intellectual property issues) or focused on technologies, industries, or geographic regions within which the alliance is active. Depending on the differential authority given to individual channels, they can constitute hierarchical layers in the alliance structure. For example, an alliance board composed of top managers may set the direction and budget for the alliance, and specialized committees of middle managers may determine the implementation of plans and detailed allocation of resources. When multiple differentiated and hierarchically ordered decision-making channels exist, the alliance structure is vertically decentralized.
The Star Alliance—a contractual alliance among 27 airlines from Africa, America, Asia, Australasia, and Europe—provides a good example of a vertically decentralized alliance structure: Partners have established multiple decision-making channels for different sets of alliance issues and also set up an equity JV organization to support the collaboration (German-based Star Alliance Service GmbH). The CEOs of the partnering airlines compose the alliance’s Chief Executive Board and meet twice a year on issues such as alliance membership (e.g., addition of new members), governance changes (e.g., expansion on working groups/boards), and finance (e.g., budget and bonus payments of Star Alliance GmbH staff). On the next lower level, the Alliance Management Board, consisting of all alliance managers of the member firms, prepares and implements the decisions taken on the executive level. Members of the Alliance Management Board have a senior hierarchical position in their own organization to be familiar with their CEO’s priorities, and to be able to quickly distribute and access relevant information in their organizations. In most cases they work full-time on alliance issues (Star Alliance and other alliances) and are supported by a significant number of staff. The Star Alliance Service GmbH comprises 70 employees, and its functional organizational structure mirrors that of an airline (network management, sales, revenue management, etc.). The JV’s CEO and VPs have no formal authority over the member airlines, but they exert informal influence and foster cooperation by developing proposals for collaborative initiatives. The JV’s CEO participates in and facilitates the Chief Executive Board meetings. And the JV’s VPs work closely with member airlines’ alliance managers to identify and formulate new projects, and are involved in all of the more than 50 existing work groups within the alliance that topically fall within their area of responsibility (i.e., network, revenue, etc.).
The staffing of decision-making channels also allows us to distinguish the degree of horizontal centralization in the alliance, that is, the number of actors involved in a given decision-making channel. A low level of horizontal centralization would involve multiple representatives from each partner in established decision-making channels. A medium level of horizontal centralization could, for example, comprise one manager from each partner. High horizontal centralization would grant the authority to make strategic choices on behalf of all partners to a single manager. This is usually the case in JV arrangements, where specific decision-making domains are assigned to individual JV managers (e.g., the JV’s CEO, CFO, CTO, etc.) rather than committees of parent firms’ representatives. But such concentration of decision-making authority also occurs in purely contractual alliances that follow a “trading” logic, that is, where partners bring together complementary competences: In those cases the partner organization that has most competences or experience in a particular domain may have decision authority over that domain. In the previously described Renault–Nissan alliance each company leads the alliance’s new engine design efforts and orchestrates alliance resources for their respective area of expertise—Renault for diesel and Nissan for gasoline engines. 23 Horizontal decentralization essentially allows alliance partners to make alliance-related planning and resource allocation decisions independently. Aldrich and Sasaki (1995) identify such decentralized arrangements as typical for Japanese research consortia, where partners shun the establishment of joint R&D facilities and instead fully decentralize research projects to member organizations and simply share results and benefits at the end.
Centralization has important ramifications for coordination, learning, and trust in the alliance. With regard to coordination, centralization has both advantages and disadvantages. Horizontal and vertical centralization make coherence and consistency in decision making easier, and facilitate and accelerate conflict resolution—by limiting the number of actors involved in decision making, and by limiting the number of channels within which conflicts can occur. Hence, decision makers in a centralized alliance structure will be better positioned to ensure alignment and give direction to joint efforts—and to do so more quickly—compared to those acting from within decentralized structures. But when faced with a large volume or range of coordination issues, they may quickly push beyond their limits of attention and coordination skills. Vertically and horizontal decentralized decision-making structures, in contrast, facilitate coordination because they allow decisions to be made closer to where problems occur (Grant, 1996), enhance information-processing capacity of the alliance organization by establishing multiple channels (Galbraith, 1974), and involve multiple managers in these decision-making channels, which allow them to collectively overcome individual attentional constraints and cognitive blind spots.
Decentralization may also foster alliance learning. A higher degree of individuals’ exposure to alliance issues and participation in alliance decisions in decentralized structures encourages more alliance exploitation-focused learning. And the number and diversity of actors participating in decision making, and the more open and dialectic decision practices they may need to employ to find agreements, may encourage exploration-focused learning as well (cf. Coopey & Burgoyne, 2000).
The higher degree of involvement and participation in decision making can also contribute to interpersonal attachment and affective trust. However, high levels of centralization are not necessarily anathema to development of interorganizational trust. Insofar as centralization of decision authority renders alliance activity more predictable and allows partners to better control the quality of decision making (even at the price of flexibility), it may contribute to instrumental trust.
Discussion—Toward a Holistic Perspective on Alliance Design
So far we have discussed the alliance design parameters in isolation. However, the “configuration hypothesis” proposed by organizational design scholars (Hage & Aiken, 1967; Meyer, Tsui, & Hinings, 1993; Miller & Friesen, 1984; Mintzberg, 1979; Wiechmann, 1974) suggests organizational effectiveness and efficiency depend on a “consistent” combination of individual design parameters. Others have noted the equifinality of organizational design alternatives (Doty, Glick, & Huber, 1993; Gresov & Drazin, 1997). This implies that advantages and disadvantages of structural parameter choices can be balanced out in different combinations and configurations.
Consistent with the configuration logic, the five parameters of alliance structure should be considered in combination, not isolation (see Figure 3). 24 The interface, intraface, and specialization jointly determine the social capital of the alliance. They describe how thoroughly and strongly partners are connected, how deeply embedded their relationship is. Formalization, centralization, and specialization jointly determine how effectively the alliance can be directed, and how such direction is achieved.
Interconnections Between Design Parameters
Given the various advantageous and disadvantageous implications of individual design parameters for coordination, learning, and trust described in the last section, combinations of these parameters to enable and balance connection and direction can stand in some tension. For example, the deeper the connection between partners, the more costly and ineffectual high levels of centralization and formalization tend to be. At the same time, deeper connections due to stronger interfaces and intrafaces also represent more organizational complexity in the alliance, which requires some degree of formalization and centralization for the alliance not to devolve into chaos.
Individual parameters can partially complement or compensate each other. An alliance structure that lacks what partners would consider a sufficiently strong connection can be compensated to some degree by giving more direction to the communication and decision-making processes to enhance efficiency and effectiveness of interactions and joint decisions. Likewise, an alliance for which partners find it difficult to formally establish a clear direction can be compensated by deepening the connection among partners and thus providing a less imposed bureaucratic, more emergent relational basis for managing the partnership. Toyota’s approach to managing its supplier community, which we described earlier, can serve as an example.
In the following paragraphs we discuss common design challenges partnering organizations face when making choices regarding the five design parameters and their combination.
Bottlenecks
Structural bottlenecks in an alliance can occur when partners establish a weak interface and/or a high degree of vertical and horizontal centralization, while operating comparatively strong intrafaces within their respective organizations. Bottlenecks can be a deliberate means of exerting control over information flow and decision making to reduce the relational complexity of the partnership, better control information exchange, and avoid contradictions, misunderstandings, and ambiguities. However, bottlenecks can be a significant impediment to overall alliance operational effectiveness and individual partners’ benefit from the partnership. The complexity of operational details, and volume of information-processing and communication required to effectively connect partnering organizations—especially when partners seek to manage complex interdependencies—can overwhelm the brokering boundary spanners and cause decision backlogs, and lower levels of coordination and learning.
Bottleneck problems can be compensated by increasing specialization of existing boundary spanners, increasing decentralization, or adjusting formalization—thus improving boundary spanners’ attentional capacity and focus on alliance issues, placing the burden of identification, deliberation, and resolution of partnering problems on more shoulders, and providing some rules to prioritize alliance management tasks and to reduce information glut.
Decision-Making Style
In addition to ensuring fit between the interface and intraface dimensions of alliance structure, alliance designers also need to consider how to align centralization with the interface and intraface. This is primarily a decision about the desired decision-making style for the alliance (Adler & Borys, 1996)—is it intended to be top-down or bottom-up, swift or deliberative?
An alliance with a broad and dense interface, for example, may benefit from a low degree of vertical centralization, that is, the diffusion of decision-making authority among multiple decision-making channels. This would allow boundary spanners at multiple levels of the alliance to immediately act on relevant new information as they become aware of it. Such colocation of information and decision making (Grant, 1996) can help improve speed and quality of individual decisions, but can come at the cost of diminished coordination among boundary spanners, and alignment and consistency across decisions.
An alliance with broad and dense interface and horizontal decentralization, that is, the involvement of multiple representatives from each partner organization in key decision-making channels, fosters a consensual decision-making culture in the alliance and encourages a broader search for opportunities for improving alliance effectiveness and efficiency and for new collaborative initiatives (cf. Siggelkow & Rivkin, 2005). It can also help ensure broad support of selected responses (Friedkin, 2004; Janis, 1972). However, it is likely to reduce decision speed, especially when channel participants come from different functional domains, and when the intraface within alliance partner organizations is broad—thus giving decision makers a broader set of internal constituents and issues to consider.
Silos
A number of studies have extolled the benefits of specialization and structural separation of boundary spanners, highlighting in particular the positive impact of these design choices on the development of alliance capabilities and routines, and the establishment of administrative provisions that are tailored to the specific relationship rather than based on an amalgamation of partners’ existing administrative structures and practices (Gulati & Singh, 1998; Kale et al., 2002; Schreiner et al., 2009). Disadvantages of high levels of specialization and structural separation, beyond the obvious costs involved in such a dedicated resource commitment for alliance administration, have received less attention. Yet grouping together those who are (principally) involved in managing an alliance or set of alliances, and isolating them from the “regular” internal business operations can impede information flow, hamper coordination and integration of alliance activities with relevant other in-house activities, and limit the degree to which alternative interpretations of internal and external events relevant to the alliance are reconciled and brought to bear on important alliance decisions (Thomas & Trevino, 1993). The siloing of alliance managers and alliance activities can also hinder the transfer of knowledge acquired in the alliance to nonparticipating members within a partner organization. Last, structurally isolated boundary spanners may find it more difficult to identify and engage additional organizational members whose contributions are required by the alliance, or who would benefit from the alliance’s output. 25
The disadvantages of high degrees of boundary spanner specialization and separation can be partially compensated by a broad, strong intraface. It can help connect alliance activities with internal operations, facilitate coordination and knowledge transfer, and support the identification and engagement of additional organizational members in the alliance effort.
Asymmetries
Alliance scholars have noted that partnerships among dissimilar organizations are more risky than those among similar organizations. Harrigan (1988) and Hall (1984), for example, have suggested that dissimilar organizational cultures and management procedures can cause alliances to struggle and fail. While Harrigan’s and Hall’s remarks apply directly to asymmetric levels of formalization among partners, we can extend the logic of their argument to the other four elements of alliance structure. 26 Unless they match an asymmetric division of labor among partners, asymmetries in the alliance’s interface, intraface, and specialization can create bottleneck problems. In addition, a relatively narrower, sparser, and weaker intraface likely lowers a partner’s absorptive capacity and disadvantages him or her in alliance learning races (Khanna et al., 1998). Last, different approaches to centralization can cause materially and symbolically problematic decision processes: A low level purchasing manager interfacing with a supplier’s CEO or head of marketing may not be able to identify or realize possible value creation opportunities, and may—in some cultural contexts—anger the partner by not being a status equal.
Concluding Remarks
In this article we have developed an organization-design-focused framework for classifying alliances. The framework’s five parameters—interface, intraface, specialization, formalization, centralization—provide a nuanced description of alliance structures and their effect on coordination, learning, and trust in alliances and thus also enabled us to outline design challenges that arise out of tensions and trade-offs between individual design parameters.
Our alliance design framework can complement established legal structure classifications. It can help, for example, distinguish different structural types of contractual alliance, or different types of JVs. This provides opportunities for more nuanced assessments of design antecedents (When do organizations tend to establish more centralization in JVs than in contractual alliances?) and consequences (Under which circumstances do broad interface JVs or broad interface contractual alliances perform better?).
Adoption of our administrative and behavioral perspective on design can yield some consistent and some divergent predictions of structural choices and their impact on alliance dynamics and outcomes relative to a transaction cost economics (TCE) perspective (Parkhe, 1993). Both perspectives agree that organizational form chosen for a collaborative relationship is critical to its evolution and performance and that structural arrangements can address transaction concerns regarding opportunism and value appropriation. However, the two perspectives differ with regard to how to respond to uncertainty and complexity in collaborative relationships. Williamson (1991), for example, attributes a greater adaptability to disturbances caused by environmental uncertainty to hierarchical governance modes. A number of TCE-informed alliance studies also have suggested more hierarchical modes of governance for relationships characterized by high uncertainty and complexity (Gulati & Singh, 1998; Mowery et al., 1996). Consistent with the organizational design literature, our perspective would suggest the inverse—that dynamic and turbulent environments require organic, more flexible alliance organizational structural arrangements.
Our framework also helps deepen our understanding of the underlying organizational structural determinants of advantages and disadvantages frequently attributed to JV-based partnerships relative to contractual partnerships. The control and coordination advantages attributed to JVs (Gulati & Singh, 1998), for example, can be traced to the high level of specialization of individuals involved in the JV, and the strong intraface among those individuals within the JV organization. How much JV partners trust each other (Park & Russo, 1996) and how much they learn from each other (Inkpen & Currall, 2004) depends chiefly on the strength of the interface between the parent organizations and the JV, and on the centralization and formalization of decision making.
We recognize that the alliance structure dimensions suggested in this article are empirically more difficult to obtain and to measure than an alliance’s legal structure. However, such empirical difficulty can be overcome—as it has been in studies of intraorganizational structures (Donaldson, 1995, 1996). Close examination of alliance contracts, surveys that capture aspects of alliance organizational structure not explicated in contracts, policy-capturing studies (Thomas, Mellewigt, & Zajac, 2012), as well as field-based qualitative research are all effective empirical strategies to advance our understanding of the inner workings of alliances. Scholarly attention to the micro-mechanisms of interorganizational collaboration is likely to highlight behavioral issues that emerge from the social, psychological, and political aspects of alliance management, presenting opportunities for both richer theory and greater relevance to practitioners. We hope that our design framework can help highlight how elements of alliance structures can affect and be affected by such behavioral processes, providing new direction for future research in this area.
By unbundling alliance governance arrangements we particularly encourage research designs that help identify common alliance design configurations, the antecedents of their deployment, and their performance consequences (cf. Albers, 2005, 2010). For example, partners’ alliance design choices may be influenced by designs they have utilized for prior partnerships—thus suggesting a possible path dependency of structural characteristics of different alliances over time. Contingency factors such as uncertainty or complexity faced by alliance partners may also determine structural choices. Research examining these influences promises to contribute to a better understanding of agency in alliance design, the relative importance of and interrelationships between different design parameters, and the possible equifinality of different combinations under specific circumstances. In addition, longitudinal alliance studies may reveal transitions from different structural templates over time (e.g., shift from more formalization and centralization during the “setup” phase of the partnership to less formalization and centralization once the alliance reaches a “steady state” for its regular operation), complementing extant work on alliance dynamics (Ariño & de la Torre, 1998; Doz, 1996; Reuer & Ariño, 2002; Ring & Van De Ven, 1994; Zajac & Olsen, 1993).
Practitioners can also benefit from our alliance design framework. It encourages them to look beyond binary legal options for alliances and provides systematic guidance for establishing and tweaking administrative arrangements for alliances. Our discussion of individual design parameters’ benefits and drawbacks for coordination, learning, and trust allows alliance designers to more closely relate the alliance’s organizational setup to partners’ strategic goals and specific administrative preferences. The discussion of the individual parameters has also highlighted the equifinality of structural options to support and enhance coordination, learning, and trust in the alliance. For example, we discussed how coordination can be enhanced by increasing specialization or formalization, how learning can be stimulated by a stronger intraface or more decentralization, and how trust can be built through a more extensive interface or through less formalization. In other words, the alliance organizational structure can be designed in a multitude of ways to accommodate partners’ goals and preferences. We believe our alliance design framework provides a useful language for both alliance researchers and practitioners and, more important, offers a deeper understanding of strategic alliance governance and management.
Footnotes
Acknowledgements
This article was accepted under the editorship of Deborah E. Rupp. We gratefully acknowledge the helpful comments of Asli Arikan, Werner Delfmann, Guido Möllering, Kathryn Pavlovich, and Markus Reihlen. Previous versions of this article were presented at the Academy of Management Annual Meeting in Philadelphia, at the Strategic Management Society Conference in Cologne, and at seminars at Northwestern University, the University of Cologne (Germany), and the University of Waikato (New Zealand).
The first author acknowledges financial support from the German Academic Exchange Service (DAAD) during his visit to Northwestern University.