Modes of Network Governance: Structure, Management, and Effectiveness

This article examines the governance of organizational networks and the impact of governance on network effectiveness. Three basic models, or forms, of network governance are developed focusing on their distinct structural properties. Propositions are formulated examining conditions for the effectiveness of each form. The tensions inherent in each form are then discussed, followed by the role that management may play in addressing these tensions. Finally, the evolution of governance is explored. Networks have been widely recognized by both scholars and practitioners as an important form of multi-organizational governance. The advantages of network coordination in both public and private sectors are considerable, including enhanced learning, more efficient use of resources, increased capacity to plan for and address complex problems, greater competitiveness, and better services for clients and customers (see Alter and Hage 1993; Brass et al. 2004; Huxham and Vangen 2005). However, despite much progress made by researchers studying networks of organizations over the past 15 years and more, there is still a considerable discrepancy between the acclamation and attention networks receive and the knowledge we have about the overall functioning of networks. By network functioning, we refer to the process by which certain network conditions lead to various network-level outcomes. Understanding the functioning of networks is important since only then can we better understand why networks produce certain outcomes, irrespective of whether networks result from bottom-up processes or are the product of strategic decisions made by network participants or government officials. This article addresses these issues by focusing on the critical role of network governance and its impact on network effectiveness. We recognize that the study of effectiveness has been problematic at both organizational (Goodman and Pennings 1977) and network (Provan and Milward 2001) levels, especially regarding the key question, ‘‘effectiveness for whom?’’. Yet despite problems regarding conceptualization and measurement, An earlier version of this article was presented at the Public Management Research Association meeting, September 29 to October 1, 2005, at the University of Southern California, Los Angeles, California. Address correspondence to the author at [email protected]. doi:10.1093/jopart/mum015 Advance Access publication on August 2, 2007 a The Author 2007. Published by Oxford University Press on behalf of the Journal of Public Administration Research and Theory, Inc. All rights reserved. For permissions, please e-mail: [email protected] JPART 18:229–252 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom effectiveness is a critical concept, for both researchers and practitioners, that cannot simply be ignored. Network effectiveness is defined here as the attainment of positive networklevel outcomes that could not normally be achieved by individual organizational participants acting independently. Although individual organizational participants may, and probably should benefit as well, effectiveness is viewed here at the network level. The specific type of network-level outcome considered is not, however, defined by us but depends on the particular constituency assessing the functioning of the network (see Provan and Milward 2001). This implies that we do not consider a certain outcome a priori as the correct one because each presents a potentially valid point of view. Such outcomes might include strengthened community capacity to solve public problems like crime, homelessness, or health care; improved integration of critical services to vulnerable populations; regional economic development; and responsiveness to natural or made-made disasters. Although networks have been studied from a variety of perspectives, surprisingly little attention has been paid to the governance of whole organizational networks. This broader focus is what Powell et al. (2005, 1133) recently referred to as ‘‘illuminating the structure of collective action.’’ In part, the reason for adopting a more narrow perspective may simply be that organizational scholars are used to studying organizations, not multiorganizational arrangements (Salancik 1995). In addition, developing a deep understanding of network governance requires collection of data on multiple networks, which can be time consuming and costly. Although there is a growing literature on networks as a unit of analysis, the majority of this work has been descriptive (cf. Agranoff and McGuire 2003; Goldsmith and Eggers 2004; Huxham and Vangen 2005; van Bueren, Klijn, and Koppenjan 2003). Finally, there seems to be some reluctance among many who study networks to discuss formal mechanisms of control. A common assumption is that since networks are collaborative arrangements, governance, which implies hierarchy and control, is inappropriate (Kenis and Provan 2006). ORGANIZATIONAL VERSUS NETWORK GOVERNANCE Governance is a topic that has long been studied by organizational scholars (cf. Mizruchi 1983; Westphal and Zajac 1995). Traditionally, governance in business firms has focused on the role of boards of directors in representing and protecting the interests of shareholders (Fama and Jensen 1983). Governance has also been studied in the nonprofit context, although the focus here has generally been on the role of boards of trustees, as representing and protecting the interests of community members or other politically important constituencies (Provan 1980). In public management, governance refers not to the activities of boards, but mainly, to the funding and oversight roles of government agencies, especially regarding the activities of private organizations that have been contracted to provide public services (Hill and Lynn 2005). A critical role for governance in all these sectors, and consistent with principal-agent theory, is to monitor and control the behavior of management, who are hired to preside over the day-to-day activities of running the organization (Eisenhardt 1989; Fama and Jensen 1983). Although there is much recent evidence that boards do not necessarily take their responsibilities seriously enough (i.e., Enron Corporation), board members do have a legal obligation to perform their duties and are liable if the organization they represent engages in illegal or irresponsible behavior. With some exceptions (cf. Goldsmith and Eggers 2004; Imperial 2005; Jones, Hesterly, and Borgatti 1997; Moynihan 2005; Park 1996), most literature on organizational Journal of Public Administration Research and Theory 230 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom networks does not explicitly address governance. The most obvious reason is that networks are comprised of autonomous organizations and, thus, are essentially cooperative endeavors. Since networks are not legal entities (we do not consider joint ventures and equitybased alliances to be true networks), the legal imperative for governance is simply not present as it is for organizations. For goal-directed organizational networks with a distinct identity, however, some form of governance is necessary to ensure that participants engage in collective and mutually supportive action, that conflict is addressed, and that network resources are acquired and utilized efficiently and effectively. Although all networks comprise a range of interactions among participants, a focus on governance involves the use of institutions and structures of authority and collaboration to allocate resources and to coordinate and control joint action across the network as a whole. These interactions are distinct from operational links, which are often dyad based including referrals, sharing of information, and joint programs. Even when mechanisms for governance have been discussed in the literature, they are generally discussed in terms of specific activities performed for a particular network, rather than in a comparative way. As a result, there has been no theory on the various forms of governance that exist, the rationale for adopting one form versus another, and the impact of each form on network outcomes. This is our focus. We define the term ‘‘network’’ narrowly. Our focus is on groups of three or more legally autonomous organizations that work together to achieve not only their own goals but also a collective goal. Such networks may be self-initiated, by network members themselves, or may be mandated or contracted, as is often the case in the public sector. When defined in this way, as multilateral collectivities, networks can become extremely complex entities that require explanations that go well beyond the dyadic approaches that have been traditionally discussed in the organization theory and strategic management literatures. Our definition focuses on what Kilduff and Tsai (2003) refer to as ‘‘goaldirected,’’ as opposed to ‘‘serendipitous’’ networks. Although goal-directed networks occur less frequently, they have become extremely important as formal mechanisms for achieving multi-organizational outcomes, especially in the public and nonprofit sectors where collective action is often required for problem solving (cf. Agranoff and McGuire 2003; Imperial 2005; Lemieux-Charles et al. 2005; Provan, Isett, and Milward 2004; Provan and Milward 1995). Especially in the European literature, which is less based on an individualistic fiction (Coleman 1990, 300–05), a substantial number of cases of goal-directed networks have been empirically described (Acevedo and Common 2006; Daguerre 2000; Entwistle et al. 2007; Sydow 2004; Teisman and Klijn 2002). Serendipitous interactions, of course, occur within goal-directed networks, resulting in coevolutionary trajectories that may prove advantageous or detrimental to network outcomes. However, unlike serendipitous networks, which develop opportunistically, goal-directed networks are set up with a specific purpose, either by those who participate in the network or through mandate, and evolve largely through conscious efforts to build coordination. Addressing complex issues that demand multilateral coordination, as is often the case in the public and nonprofit sectors (to deal with major disasters, increase economic activity in the region, address critical and complex health or human service needs, etc.), requires more than just achieving the goals of individual organizations (O’Toole 1997). It requires collective action and the governance of these activities. Although network governance may not be a legal issue, as with organizational governance, we argue that it is critical for effectiveness. Unlike organizations, networks must be governed without benefit of hierarchy Provan and Kenis Modes of Network Governance 231 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom or ownership. In addition, network participants typically have limited formal accountability to network-level goals and conformity to rules and procedures is purely voluntary. Most research on organizational networks can be broadly characterized by two basic approaches: the ‘‘network analytical’’ approach and the ‘‘network as a form of governance’’ approach, both of which are limited when it comes to analyzing network-level functioning and governance. Network analytical approaches focus mainly on micro-level, egocentric aspects of networks, building largely on work done by sociologists studying networks of individuals. This perspective has had a long history (Moreno 1934). Scholars have contributed especially to the description and explanation of network structural characteristics using such concepts as density, centrality, and structural holes (Burt 1992; Wasserman and Faust 1994). The units of observation are a set of objects called nodes, positions, or actors, and a set of present or absent relations among these objects referred to as edges, ties, or links (Knoke 1990). In network analytical approaches, the main objective can be either to describe, explain, or compare relational configurations or to use these configurations to explain certain outcomes. The functioning of organizational networks can be partially addressed using this approach, since we defined functioning as the process by which certain network conditions lead to network outcomes. The problem, however, is that for the most part, what gets analyzed and explained is not the network itself, but the ‘‘nodes’’ and ‘‘relations’’ that comprise the network (cf. Graddy and Chen 2006; O’Toole and Meier 2006). Apart from some notable exceptions (e.g., Owen-Smith and Powell 2004; Powell et al. 2005; Provan and Milward 1995; van Raaij 2006), the unit of analysis (i.e., the phenomenon to be investigated) in this literature is not the complete network but a node (ego) or a dyad. In these studies, findings are related to questions of whether or not the way an actor is embedded in a network has an effect on the outcomes of the actor (such as level of innovation, performance, and learning) (Ahuja 2000) or on describing and explaining the birth, death, effectiveness, etc. of dyadic relationships (cf. Larson 1992; Ring and Van de Ven 1994; Uzzi 1997). Consequently, this literature tells us little about the functioning of networks, because networks are seldom treated as the unit of analysis. The network as a form of governance approach, in contrast, does treat networks as the unit of analysis. Network is viewed as a mechanism of coordination, or what has often been referred to as network governance. Starting with Williamson’s (1975) Markets and Hierarchies, a rich literature has developed on different forms of governance over the last two decades. As seen from an economic perspective, this literature challenged the conventional wisdom that the market is the only efficient system of nonhierarchical coordination. From an organization and administrative science perspective, the most innovative aspect of this literature is that it made clear that organizations cannot be taken as something for granted (see Perrow 1986) and that other forms of coordination, such as networks, can equally achieve goals. Consequently, a discussion unfolded as to whether networks are simply a combination of elements of market and hierarchy, and could, therefore, be placed on a continuum between market and hierarchy, or whether they would be better understood as unique forms of governance in their own right (see Powell 1990). This literature moved toward treating networks as discrete forms of governance, characterizing them as having unique structural characteristics, modes of conflict resolution, bases of legitimacy, etc. (cf. Jones et al. 1998; Raab 2004). Although the governance approach considers networks as the unit of analysis, the tradition has been for networks to be treated as undifferentiated forms, as if they all could Journal of Public Administration Research and Theory 232 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom be characterized in the same general way (e.g., Jones, Hesterly, and Borgatti 1997; Powell 1990). This may be due to the fact that for the most part, networks were seen as a ‘‘new’’ and ‘‘positive’’ mode of coordination that needed to be distinguished from markets and hierarchies. A functionalist argument dominated, claiming that networks are a response to failures of markets, failures of hierarchical coordination, and to societal and technological developments. The implication was, and continues to be, that despite problems, networks in general can produce positive outcomes that would not be possible in a market or a hierarchy. What we propose here is to combine the network analytical and ‘‘governance’’ perspectives. The governance perspective is valuable in that the network itself is considered to be the unit of analysis. Networks are forms of social organization, which are more than the sum of the actors and their links and which deserve to be studied in their own right (O’Toole 1997). The network analytical perspective contributes another central idea to our work—that networks are a set of actors or nodes, with relationships between these nodes as being either present or absent. Thus, networks are considered to vary with regard to their structural patterns of relations. Consistent with this logic, we view network as a variable, examining different network governance configurations and the conditions for the effectiveness of each form. Only by demonstrating that networks with different configurations have different networklevel effects can a rationale for developing network-level theories be established. To build our case, we formulate a number of propositions about the relationship between the configuration of network governance and network effects. We also argue that the role of management is critical for effective network governance, especially regarding the handling of tensions inherent in each governance form. Essentially, our focus is not on networks as a means of governance, but on the governance and management of networks themselves. The article proceeds as follows. First, three basic forms of network governance are identified. Second, we focus on development of four contingency conditions that are likely to affect the successful adoption of each governance form. Specific propositions are developed. Although many factors can contribute to network effectiveness, our arguments are built around the assumption that there is a rationale for utilizing one form over another and that there are consequences for selection of each form of governance. Network managers must recognize these consequences and address them appropriately if positive network outcomes are to be realized. Next, we focus on three tensions that result from choice of network governance form. We outline these tensions and discuss which specific tensions are likely to be most prevalent for each of the three forms. Finally, we discuss the evolution of network governance. If the effectiveness of a form is tied strongly to a set of critical contingencies, then what happens when these contingencies change? We explore the pattern of change from one form to another. FORMS OF NETWORK GOVERNANCE Based on a review of the literature on whole networks (rather than dyad-based network relationships; cf. Provan, Fish, and Sydow 2007), coupled with our own extensive observations, network governance forms can be categorized along two different dimensions. First, network governance may or may not be brokered. At one extreme, networks may be governed completely by the organizations that comprise the network. Every organization Provan and Kenis Modes of Network Governance 233 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom would interact with every other organization to govern the network, resulting in a dense and highly decentralized form. This is what we call shared governance. At the other extreme, the network may be highly brokered, with few direct organization-to-organization interactions, except regarding operational issues such as the transfer of business, clients, information on services, and the like. Instead, network governance would occur by and through a single organization, acting as a highly centralized network broker, or lead organization, regarding issues that are critical for overall network maintenance and survival. At the mid-range, a single organization might take on some key governance activities while leaving others to network members. Alternatively, network members may divide governance responsibilities among various subsets, or cliques of network members, with no single organization taking on significant governance tasks. A second distinction regarding governance can be made in brokered networks by focusing on whether the network is participant governed or externally governed. As discussed below, participant-governed networks are, at one extreme, governed either collectively by the members themselves (i.e., shared), or at the other extreme, by a single network participant that takes on the role of a lead organization. Externally governed networks are governed by a unique network administrative organization (NAO), as discussed below, which may be either voluntarily established by network members or mandated as part of the network formation process. Each of these forms has certain key structural characteristics, which we identify below. Each form is utilized in practice for a variety of reasons, and no one model is universally superior or effective. Rather, each form has its own particular strengths and weaknesses, leading to outcomes that are likely to depend on the form chosen. Participant-Governed Networks The simplest and most common form is participant governance. This form is governed by the network members themselves with no separate and unique governance entity. Governance in this form can be accomplished either formally; for instance, through regular meetings of designated organizational representatives, or more informally, through the ongoing but typically uncoordinated efforts of those who have a stake in network success. At one extreme, participant-governed networks can be highly decentralized, involving most or all network members interacting on a relatively equal basis in the process of governance. This is what we refer to as shared participant governance. At the other extreme, the network may be highly centralized, governed by and through a lead organization that is a network member. Shared participant-governed networks depend exclusively on the involvement and commitment of all, or a significant subset of the organizations that comprise the network. Network participants are themselves responsible for managing internal network relationships and operations as well as external relations with such groups as funders, government, and customers. In health and human services, shared-governance networks are common, in part because networks are often considered to be an important way of building ‘‘community capacity’’ (Chaskin et al. 2001). Only by having all network members participate, on an equal basis, will participants be committed to the goals of the network. In business, shared governance may be used in smaller, multi-firm strategic alliances and partnerships (where multi-firm ownership is not involved) designed to develop new products or to attract new business in ways that could not be otherwise accomplished through the Journal of Public Administration Research and Theory 234 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom independent efforts of network members (Venkatraman and Lee 2004). The form has also been used in multilateral relations among investment banking firms and venture capitalists working to assemble a financial package for a project or business (Eccles and Crane 1988). When network governance is shared, it is the collectivity of partners themselves that make all the decisions and manage network activities. Power in the network, at least regarding network-level decisions, is more or less symmetrical, even though there may be differences in organizational size, resource capabilities, and performance. There is no distinct, formal administrative entity, although some administrative and coordination activities may be performed by a subset of the full network. In theory, the network acts collectively and no single entity represents the network as a whole. Later in the article, we discuss in depth the specific characteristics of this and the other forms of governance, which, we argue, explain the likely adoption of one form versus another. Lead Organization–Governed Networks While shared, participant governance may involve many or all network members, there are many situations that may not be conducive to such decentralized, collective selfgovernance. In particular, the inefficiencies of shared governance may mean that a far more centralized approach is preferred. At the extreme, network governance can occur through what we refer to as a ‘‘lead organization.’’ In business, lead organization governance often occurs in vertical, buyer-supplier relationships, especially when there is a single powerful, often large, buyer/supplier/funder and several weaker and smaller supplier/ buyer/resource recipient firms. The most obvious examples of this can be found in the Keiretsu models of Japanese manufacturing (Gerlach 1992) and similar models of cooperative buyer-supplier models in the United States (Uzzi 1999) and Europe (Inzerilli 1990; Lazerson 1995). In movie production, the lead organization may be a major film studio (Jones and DeFillippi 1996). It can also occur in horizontal multilateral networks, most often when one organization has sufficient resources and legitimacy to play a lead role. This is a model that often occurs in health and human services where there may be a core provider agency that assumes the role of network leader because of its central position in the flow of clients and key resources. In community health, this may be a hospital or a health clinic (Weiner and Alexander 1998); in mental health, it may be a community mental health center (Provan andMilward 1995); in community policing, it is likely to be the police department; in local health policy, it is likely to be the local health department (Brandes, Kenis, and Wagner 2003; Hoeijmakers et al. 2007). Teisman and Klijn (2002) also describe a government agency acting as a lead organization in development of the Rotterdam harbor, whereas Graddy and Chen (2006) focus on the key role of lead organizations in governing child welfare networks in Los Angeles. In lead organization governance, all major network-level activities and key decisions are coordinated through and by a single participating member, acting as a lead organization. Thus, network governance becomes highly centralized and brokered, with asymmetrical power. A lead organization provides administration for the network and/or facilitates the activities of member organizations in their efforts to achieve network goals, which may be closely aligned with the goals of the lead organization. The lead organization may underwrite the cost of network administration on its own, receive resource contributions from network members, or seek and control access to external funding through grants or government funding. The role of lead organization may emerge from the members Provan and Kenis Modes of Network Governance 235 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom themselves, based on what seems to be most efficient and effective, or it may be mandated, often by an external funding source. Network Administrative Organization A third form of network governance is the NAO model. The basic idea is that a separate administrative entity is set up specifically to govern the network and its activities. Although network members still interact with one another, as with the lead organization model, the NAO model is centralized. The network broker (in this case, the NAO) plays a key role in coordinating and sustaining the network. Unlike the lead organization model, however, the NAO is not another member organization providing its own services. Instead, the network is externally governed, with the NAO established, either through mandate or by the members themselves, for the exclusive purpose of network governance. The NAO may be a government entity, or a nonprofit, which is often the case even when the network members are for-profit firms. For instance, Human and Provan (2000) describe two networks in the wood processing industry that were both guided by an NAO. All firms were for-profit but the NAOs were nonprofit (a 501(c)3 corporation under the U.S. federal revenue code). The NAO may even be a unique for-profit corporation, as with Nexia International, the global accounting network discussed by Koza and Lewin (1999). An NAOmay be modest in scale, consisting only of a single individual, often referred to as the network facilitator or broker, or it may be a formal organization, consisting of an executive director, staff, and board operating out of a physically distinct office (McEvily and Zaheer 2004; Provan, Isett, and Milward 2004). This latter form may be used as a mechanism for enhancing network legitimacy, dealing with unique and complex network-level problems and issues, and reducing the complexity of shared governance. These more formalized NAOs typically have board structures that include all or a subset of network members (Evan and Olk 1990; Provan, Isett, and Milward. 2004). The board addresses strategic-level network concerns, leaving operational decisions to the NAO leader. Government run NAOs are generally set up when the network first forms, to stimulate its growth through targeted funding and/or network facilitation and to ensure that network goals are met (Goldsmith and Eggers 2004). Such NAOs are established locally for purposes of accomplishing broad goals, such as those related to regional economic development (Gebauer, Nam, and Parsche 2005; Piore and Sabel 1984; Saxenian 1994). NETWORK GOVERNANCE AND EFFECTIVENESS So far, discussion has focused solely on describing the various forms of governance. However, a key contribution we hope to make is to develop a theoretical rationale for the adoption of one form over another in a way that can predict the successful attainment of network-level outcomes, or what some have referred to as network effectiveness (Provan and Milward 1995, 2001). Specifically, we argue that based on what is known about networks and network interactions, there are certain critical contingencies (Drazin and Van de Ven 1985) that can be identified to explain whether or not a particular form of network governance is likely to be effective. It is not unreasonable to argue that choice of one governance form or another will be based, at least in part, on the discretion of key network decision makers. Network managers can and do choose forms for reasons like Journal of Public Administration Research and Theory 236 at U nirsity of A lska A ncorage on A uust 0, 2011 jpart.oxjournals.org D ow nladed fom mimicry, past experience, and personal preference. However, we argue that choices based on these sorts of factors alone are not consistent with the structural characteristics of the governance form itself and, thus, if adopted, will have only a limited chance of being successful. The point can also be made that many networks are mandated and, thus, no choice of form is even possible. However, many networks are not mandated, and even for those that are, decisions about network governance do not simply emerge out of thin air. Rather, they are determined by decision makers, like government policy officials, who may want to base their decisions about how the network will be governed on evidence regarding what form seems most likely to be effective under a particular set of conditions. We propose that the successful adoption of a particular form of governance will be based on four key structural and relational contingencies: trust, size (number of participants), goal consensus, and the nature of the task (specifically, the need for network-level competencies). We make no claim that these are the only contingencies that are relevant. Rather, we argue that based on the network literature and on our understanding of networks and network governance, these particular factors are important and can explain considerable variance in the choice of one form over another. In general, we argue that as trust becomes less densely distributed throughout the network, as the number of participants gets larger, as network goal consensus declines, and as the need for network-level competencies increases, brokered forms of network governance, like lead organization and NAO, are likely to become more effective than shared-governance networks. Table 1 summarizes the specific relationships we propose.