Second-order limitation: Uncovering latent effects of board network ties - Statistical Data Included

Administrative Science Quarterly,  Dec, 2001  by James D. Westphal,  Marc-David L. Seidel,  Katherine J. Stewart

• E-mail
• Print
• Link

**********

A long tradition of research in organization theory has examined the diffusion of technology, policy, and strategy through social networks (Burt, 1987; Mizruchi, 1996). One of the most important propositions in this literature is that, under conditions of uncertainty, social influence processes will lead firms to imitate the individual policy decisions of other firms to which they are connected by social network ties (DiMaggio and Powell, 1983; Galaskiewicz, 1985). Empirical studies in the board interlocks literature, in particular, have examined how overlapping board memberships between firms may facilitate the imitation of particular organizational structures, such as the multidivisional form, or individual policy decisions, such as the adoption of poison pills (e.g., Davis, 1991; Palmer, Jennings, and Zhou, 1993; Haunschild, 1993; Westphal and Zajac, 1997). Prior research may have underestimated the magnitude of network effects, however, by restricting its focus to first-order imitation, or the act of imit ating the content of a particular policy decision, such as the level of spending on research and development (R&D). Second-order imitation, or the imitation of an underlying decision process or script that can be adapted to multiple policy domains (e.g., business strategy, compensation policy, acquisition activity, etc.), has been ignored in the interlocks literature, despite a body of literature on second-order effects.

In general, second-order phenomena are characterized by an underlying process mechanism that can explain multiple discrete first-order effects (e.g., Bartunek and Moch, 1987; Farmer et al., 1997). In the organizational behavior literature, for instance, researchers have invoked the notion of second-order change to describe cases in which a firm or unit adopts a decision-making routine such as participative management that leads to organizational changes in multiple policy domains (e.g., information systems, customer service, reporting practices, etc.): the underlying process represents a second-order factor that can explain multiple discrete first-order changes in policy content (Manz and Sims, 1987; Cohen, Chang, and Ledford, 1997; Farmer et al., 1997).

Although prior research has found some evidence for the network diffusion of structures and policies that are widely applicable in varying industry environments, such as takeover defenses and the multidivisional form of organization, several researchers have noted that there is less evidence for the diffusion of strategic policy decisions (e.g., the level of spending on R&D or advertising) through board network ties (Davis, 1994; Fligstein, 1995; Westphal and Zajac, 1997; Palmer and Barber, 2001). Most board ties, including ties examined in this study, link firms in different industries (Zajac, 1988), which in turn present different strategic threats and opportunities; thus, the extent of first-order imitation of strategic policy decisions (i.e., in which firms imitate the specific content of strategic decisions made by their interlock partners) may be limited. But decision-making processes may spread through network ties that bridge industries. For instance, the propensity for a firm to formulate strategic p olicies by imitating its competitors is a more generalizable component of the decision-making process that can be applied in many different industries, such that interlock ties may spread underlying mimetic processes, without necessarily diffusing the content of individual policy decisions. Thus, for instance, a pharmaceuticals firm may not imitate the level of R&D at a tied-to firm in a less technology-driven industry, such as cosmetics. Yet the tendency for the tied-to cosmetics firm to determine its expenditures by imitating its competitors may influence the propensity for the pharmaceuticals firm to imitate the R&D spending of its pharmaceutical competitors. In this way, network ties may facilitate second-order imitation in which a focal firm imitates the mimetic decision process of a tied-to firm, without imitating the policy content of that firm.

Second-order imitation is related to, but distinct from, second-order learning or double-loop learning. Argyris and Schon (1978: 2-3) described second-order learning as "modification of an organization's underlying norms or objectives," which then leads to a variety of (first-order) policy changes in the organization. Similarly, second-order imitation involves changes in an underlying decision-making process that can lead to multiple changes in individual policy domains. Moreover, Tushman and Romanelli (1985; Virany, Tushman, and Romanelli, 1992: 73) described second-order learning as fundamental organizational change that creates a "new relationship of the firm to its environment." Similarly, second-order imitation results in fundamental change in a firm's relationship to its environment by increasing similarity between a focal firm and its competitors across different policy domains. The difference between the two is that while second-order imitation involves change in underlying mimetic processes that resu lt from network ties to firms that have made similar changes in their decision processes, secondorder learning is typically conceived strictly as change in underlying processes; that is, the source of the change is not addressed or is assumed to result from factors internal to the organization (Weick, 1979; Miner and Mezias, 1996). In effect, therefore, second-order imitation can be viewed as a particular type of second-order learning.