In the paper, Who Killed the Inner Circle? The End of the Era of the Corporate Interlock Network, which was recently made publicly available on SSRN, I show that the American board interlock network has changed in fundamental ways.
Throughout the 20th century, the American board interlock network—in which companies are linked by shared board directors—exhibited a stable cohesiveness, characterized by short path lengths between companies and the existence of an “inner circle” of well-connected directors. This enduring cohesiveness of the interlock network was both the result of elite social cohesion and a key mechanism for maintaining this elite cohesion (e.g., Mills, 1956; Useem, 1984). The characteristics of the interlock network were so stable that Davis, Yoo, and Baker (2003) asserted that short path lengths and an inner circle were inevitable properties of “networks qua networks”.
However, I find that during the first decade of the 21st century, this hitherto rock-like stability crumbled: The average path length between the largest American companies increased from 3.21 steps to 4.23 steps and the number of directors on six or more boards dropped from 17 to zero. This de-cohesion was driven by changes in the types of directors who gained multiple board appointments. In 1997, minority directors were 1.52 times more likely than non-minority directors to gain an additional board seat during the next year. By 2010, as the supply of minority individuals who served as directors rose from 385 individuals to 779, being a member of an ethnic minority group no longer led to a higher propensity to gain new board seats. Also, in 1997 white men with Harvard, Yale, Princeton or Stanford educations were 1.42 times more likely than directors with other backgrounds to gain additional board seats, but this difference in propensities was gone by 2010. Most strikingly, in 1997 directors who sat on two boards were 1.87 times more likely to gain additional board seats compared to directors who sat on only one board, but this difference disappeared during the early 2000s.
These findings suggest that the rapid de-cohesion of the interlock network was driven by a reversal of social construction. Previously, directors were sought after because they were sought after, much like celebutantes are famous because they are famous. The financial scandals of the late 1990s and early 2000s and the attendant public outcry reversed this social construction and transformed directors who served on many boards—“corporate diplomats” (Useem, 1984),”shiny ‘ornaments’ on the corporate Christmas tree” (Mace, 1971: 90)—into “busy directors” (e.g., Ferris, Jagannathan, and Pritchard, 2003; Fich and Shivdasani, 2006). Companies became leery of hiring multi-board directors lest they be chastised by analysts and investors. This rapid social de-construction of prestige was possible because by the late 1990s, multiple-board membership was decoupled from social elite membership and corporate power: Directors who served on the most boards were more likely to be African-American, non-executive directors rather than white, pedigreed and “clubbable” elites who ran companies.
The paper proposes implications for the study of corporate board ties, the American elite, social networks and the democratization of previously-exclusive fields:
- 1. Studies of American board interlocks have proliferated over the last 30 years. For the interlock researcher of today, our results are both good news and bad news. On one hand, everything old is new again. The foundational laws of the network have changed. Directors sitting on multiple boards are shunned. The interlock network is no longer a synecdoche of elite ties. Social distances are no longer reliably shortened by board ties. My results imply that the findings of classic studies—for example, the spread of corporate bylaw provisions (Davis, 1991)—may no longer hold. Every empirical “fact” previously discovered needs reconfirmation. On the other hand, the interlock network is now less interesting. The interlock network no longer tells us much about who holds power in American society. Nor does it provide a distance-shortening substrate for isomorphism. Future studies will have to establish new reasons for studying the changed interlock network.
More generally, this study highlights the historical contingency of the “laws” governing human and group behavior. Gergen made an argument for “social psychology as history”, asserting that “observed regularities, and thus the major theoretical principles, are firmly wedded to historical circumstances” (1973: 315). This study provides a case in point, where attitudes and regularities documented even a few years ago may no longer hold.
- 2. The differential findings for social elite versus corporate executives imply that they are not one and the same. If the social elite do not dominate the public corporation board landscape, and if board directors and corporate executives have become bound agents of shareholders and can no longer act together for their class interests, then boards and the people on them may matter less in the power structure of American society. Assuming that a cohesive social elite still exists, where do they now interact with each other? Do they have their own enclaves that are not open to mere managers but require social status, connections and money to enter? Understanding where the elite now congregate may help us to understand the power dynamics in American society after the “twilight of the public corporation” (Davis, 2011).
- 3. This study also has implications for research on social networks. Durkheim (1951) famously studied social network breakdown and its effects—anomie and suicide. However, with few exceptions (e.g., Burt, 2000; 2002) recent network theory has tended to focus on network growth, and does not consider how networks are maintained in steady state and how they decay. This one-sided focus is understandable as the paradigm for network evolution studies has been the expanding World Wide Web. But at some point even the WWW will stop growing. Looking beyond growth and isomorphism, we need to better understand how systems maintain stability and how such homeostasis is broken.
- 4. The empirical trends found in this study mirror those found in studies of phenomena such as occupational tipping and urban flight. Consider the former phenomenon. Studies of gendered occupational tipping find a common pattern (e.g., Reskin and Roos, 1990). Men leave an occupation as the work becomes less rewarding. Women enter the occupation because it is still a step up from their other options. The status of the occupation becomes lower and the conditions of employment deteriorate. Sometimes, however, the status and conditions of an occupation do not deteriorate even as the occupation is feminized (e.g., court reporting [Jacobsen, 2007]; veterinary medicine [Irvine and Vermilya, 2010]). The current study suggests that such occupations may be susceptible to rapid status degradation in the future as they encounter exogenous shocks, unless there are strong non-gendered reasons for their continued high status.
These similar patterns across settings suggest a generalized paradox of democratization. The increasing number of minority directors in American boardrooms is a testament to the progress towards racial equality made in the United States over the latter half of the 20th century. However, this silver lining has a cloud. Minorities may only have been able to become corporate directors after the social elite started eschewing board service (because it became harder work for less reward). Then, shortly after minorities gained entry into the field, the field became stigmatized and working conditions degraded further. It seems that non-elites can be trapped in a cycle of struggling to gain entry into elite fields, then when they gain entry finding that being in the field no longer provides the same benefits. The elite on the other hand, who appropriate more than their share of society’s rewards, have already left and gone on to greener pastures or bluer oceans, where they continue to gain disproportionate rewards for less effort.
The full paper is available for download here.