Director Capital and Interlock Formation
This chapter models an Australian corporate interlock data set with a bipartite exponential random graph model (ERGM) (see Chapter 10) using the BPNet software, a version of the PNet software (Wang, Robins, & Pattison, 2009) for bipartite data. By using bipartite modeling, we are able to directly examine the interdependence long talked about in the sociological literature. That is, we are able to study social tie formation (corporate interlocks) as a function of (1) the individuals’ attributes (directors), (2) the groups’ attributes (corporations), (3) the interaction of individual and group attributes, and (4) purely structural network effects (social ties that form without reference to actor attributes).
Our primary substantive objective is to study the effects of director characteristics on the pattern of corporate interlock formation. In particular, we are interested in the effects of three types of corporate director power on the formation of corporate interlocks: (1) physical or financial capital (i.e., wealth), (2) membership of exclusive businessmen’s clubs, and (3) attendance at elite private schools.
Traditional elite and corporate interlock studies have tended to emphasize the unifying role of director social capital and the convergence of the many dimensions of director, corporate, social, and economic power at the apex of the corporate community. We argue that there is considerable differentiation in the purpose and effects of the many different forms of social and economic power within the corporate community. In particular, we argue that the alienability of the benefits of physical capital (Coleman, 1990/1994) leads owners to place relatively low emphasis on their own social capital (e.g., interlocks). We also argue that businessmen’s clubs and elite private schools, traditionally viewed as markers of upper-class membership and facilitators of corporate unity, play different roles within the Australian corporate community: we hypothesize that businessmen’s clubs are bonding social capital (Putnam, 2000, 2224) closer to in-group social capital, binding those corporations with common interests and identities, whereas private schools act as bridging social capital, providing between-group social capital, drawing together the disparate parts of the corporate community, with little differentiation on the basis of interests and identities of corporate groups united by private school ties.
In addition, we make three methodological contributions. First, we demonstrate that there are purely structural network effects that operate on corporate interlocks, independent of the economic, political, and sociological attributes of directors and corporations. In this, we are looking for effects similar to the one-mode effects we call “path closure” (Robins, Pattison, & Wang, 2009) or “transitivity” (Granovetter, 1973; Holland & Leinhardt, 1976; Watts & Strogatz, 1998), and other effects such as “popularity” (Barabasi & Albert, 1999; Frank & Strauss, 1986; Wasserman & Pattison, 1996), in which ties form purely in response to the pattern of ties that comprise the local neighborhood. In the case of bipartite networks, we expect to find effects such as a tendency for 3- paths (L3) to close and become 4-cycles (C4), and a tendency for popular directors to become more popular (modeled with star configurations of various sizes). Second, we demonstrate the added benefits, both in terms of model fit and sociological explanatory power, of introducing director and corporation attributes into bipartite modeling. We model the effects of corporate attributes such as political donations, public listing, foreign ownership, regulated industries, and turnover. We model the effect of director attributes such as individual wealth, members of businessmen’s clubs, and education at elite private schools. Finally, we explore the benefit of the inclusion of interaction effects for the increased or decreased likelihood of directorship formation between particular types of directors and corporations.