Board of Directors: The Role of Networks

Abstract

Board composition and role in corporate governance have been under close scrutiny both in the academic and "civil" worlds. Independence has been advocated as a way to reinforce the board's power over the managers. However, the empirical literature does not find convincing results to support this view. The first chapter of this dissertation offers a survey of these results and of two other strands of the literature that could be used to explain conflicting results. On the one hand, it presents findings on how social connections can affect corporate governance. On the other hand, it outlines the main results of information aggregation and conformity effects in committees. The object of this dissertation is to study how these play a part on boards and more specifically how they affect a CEO's election. Indeed, people commonly think that CEOs get appointed not only because of their talent or ability but also because of their social connections.

The second chapter explores this intuition from a theoretical perspective, using a static Bayesian game. Directors want to elect the best candidate but they also want to vote for the winner. In that context, results show that, when no candidate is part of the network, boards with a network perform better in electing the right candidate. On the other hand, it becomes detrimental for stockholders if one candidate is part of the network. The network has power over the results of the election and limits the power of the future CEO.

The third chapter explores the role of network in the election process from an empirical perspective using a unique database FIBEN of over 12000 firms, built at the Banque de France, and alumni directories for the three most prestigious Grandes Ecoles in France (HEC, Polytechnique and ENA). Results show that indeed, alumni tend to favor their peers in the voting process but also that there is rivalry between the three Ecoles. This last finding is original and supports the idea of a network effect.