IMPLICATIONS FOR RESEARCH AND PRACTICE
Contributions and Challenges
This paper provides several contributions to our understanding of network governance. We furnish a simple, integrated framework for understanding not only why firms disaggregate but the conditions under which they form into durable networks. We extend TCE by integrating task complexity and structural embeddedness into the TCE framework and by moving TCE from a dyadic to systems perspective. We extend the work on structural embeddedness by identifying exchange conditions which promote its development and also elaborate its role in social mechanisms. Finally, we delineate some key social mechanisms needed for networks to function effectively. Since these social mechanisms have been "only vaguely articulated and still incipient" (Uzzi, 1996b), we enhance understanding of network governance available in the extant literature. Although we identify a few key social mechanisms, these are not an exhaustive set. We expect future research to identify other social mechanismsin network governance.
A challenge in researching network governance is defining network membership. Because network governance exists to complete a project, product or service, this goal is an organizing principle around which the network is "draped" (Kadushin 1976). Network membership may be defined in terms of the firms relationship to the attainment of this goal, rather than by firm characteristics, such as size, SIC codes, or geographical location. From a research point of view, then, network membership is operationally defined by the relations an organization has with other firms in the network, rather than by an attribute of the organization itself. In social network analysis, this corresponds to taking a realist approach to boundary specification rather than a nominalist approach (Laumann, Marsden & Prensky, 1983:20-25). Since network governance is a select, persistent and structured set of autonomous firms, it is not enough to call an industry or region a network without examining relations among firms and how these relations complete a product or service.
Future Directions for Research
We identify several areas ripe for future research in network governance. A first area involves macrocultures and their content and development. The content of macrocultures in network governance is poorly understood. An important contribution involves identifying key values, norms, and assumptions that guide network participants. Uzzis (1996b) research is an important first step in this direction. However, more comparative work across network domains needs to take place to ascertain similarities and differences in values, goals, and assumptions of those within networks. A related issue involves identifying the processes of socialization and institutionalization and if these vary across networks in differing domains. For example, are third parties and institutions such as professional schools and industry events critical for all networks?
A second research stream examining the interaction of social mechanisms could provide important insights. Since the network form of governance involves implicit and open-ended contracts, social mechanisms are critical to networks functioning effectively. Thus, we must have a better understanding of how social mechanisms reinforce, substitute or undermine one another and how their combination influences performance. For example, are some combinations of social mechanisms more effective and under what conditions? In addition, we do not understand whether some social mechanisms are more important for predicting when networks emerge and others for when networks thrive.
A third research agenda concerns if there is an optimal size for network governance. Just as prior TCE work on governance has explored the tradeoffs involved in determining an optimal firm size, we need to consider how similar tradeoffs influence optimal network size. Presumably, as networks get larger, they can draw from more numerous and diverse resources, which would give them greater adaptability. However, greater size brings greater coordination and safeguarding problems. Too many partners places overwhelming demands on resources (time, energy, financial, etc.) as well as on the ability to define a common goal while minimizing competing claims (Gomes-Casseres, 1994.) Thus, there may be a negative correlation between network size and structural embeddedness, so that to maintain a certain level of embeddedness, networks must not get too large. We see network size as an empirical question answered through comparing many contexts in which network governance is found. To date, few comparative studies exist. In addition, the influence of size on governance and on performance of individual firms and the network needs to be assessed.
Fourth, power and its exercise within the network form of governance is an important topic. Any discussion of social structure raises questions of how such structures facilitate or constrain the exploitation of power. Power may be constrained in networks due to complex tasks high in human asset specificity. These tasks demand a high degree of creative problem solving, knowledge, and best efforts which are enhanced by a cooperative rather than an adversarial orientation. Those who are typically seen as powerful the prime contractors, distributors, or financiers become dependent on subcontractors to execute their tasks with best effort and financial integrity. In addition, a network structure of decoupled units performing complex tasks enhances the ability of parties to use two-step leverage, where a dependent actor gains leverage over a more powerful actor by developing a relationship with a third actor, to alter power relations (Gargiulo, 1993). This leverage may constrain those in more powerful positions from fully exploiting their power. Finally, power is influenced by output demand uncertainty. This means that power may be transitory since it is unclear that todays success will be tomorrows sucess. Thus, abuses of power could generate retaliation when power shifts due to rapidly changing markets. This prospect may constrain power abuse within network governance.
A fifth research agenda is determining whether networks result from searches for efficiency or from managerial fads and institutional processes. Although our arguments are based on efficiency considerations, we do not rule out the possibility of adoption for other reasons. The somewhat simultaneous, but isolated emergence of some networks (e.g. film, Italian textiles, Silicon Valley, dealmaking in investment banking) argues against fashion as the reason for their emergence. It seems unlikely (and we know of no accounts which suggest) that those forming these networks were looking to other industries as models. Though, more recently, fashion and institutionalization appear to play an increasing role. Recent popular press books advocate that firms form networks without carefully discriminating what conditions are necessary for networks to thrive.. We suggest that such processes may result in many experiments in network governance in a variety of industries. However, according to our theory, these experiments will fail unless the necessary exchange conditions and social mechanisms are in place to solve problems of adaptation, coordination and safeguarding.
We suggest testing institutional and efficiency explanations in three ways. First, by comparing networks in divergent industries (e.g., U.S. film industry, Maine lobster market, construction), efficiency explanations are more solidly supported when firms across a variety of domains institute networks to resolve similar exchange conditions. A second study is along the lines of Tolbert and Zuckers (1983) seminal work on the diffusion of civil reform. By assessing the conditions of earlier versus later network governance, we may tease out the relationship between efficiency and institutional explanations. A third study could compare failure or productivity rates of networks. If our efficiency explanation is correct, then networks with more of the exchange conditions and social mechanisms we have identified should have better adaptive fit to the environment which should be indicated by lower failure rates and greater productivity. If failure rate or productivity is independent of exchange conditions and social mechanisms for networks, then isomorphism processes may be a more viable explanation.
Network governance is increasingly important but poorly understood. Although the exchange conditions and social mechanisms we have identified make networks might seem a rare or difficult governance form to employ, we suggest that network governance will likely become more prevalent because these exchange conditions output uncertainty, human asset specificity, complex tasks are increasing. Several scholars have noted increased uncertainty in firms environments (Daft & Lewin, 1993; Volberda, 1996) which has been labeled the "hypercompetitive shift" (Thomas, 1996). Hypercompetition necessitates more rapid and flexible responses on the part of firms which suggests that the conditions needed for network emergence and viability will be increasingly common. In addition, work has increasingly shifted to knowledge-based modes where human asset specificity and the transfer of tacit knowledge across boundaries are important.
Research on network governance is not only of theoretical but of practical importance. The practical implications of our theoretical framework highlight the dangers for those who might seek to use network governance without the appropriate supporting social mechanisms. Without these mechanisms, both coordination and safeguarding are likely to suffer. Nonetheless, as the research agenda we outline above suggests, we still have much to learn about network governance. Our conceptual framework provides an enhanced understanding of and guides needed empirical research on network governance.