Consensus in Diverse Corporate Boards

Abstract

A spatial model of preferences over actions illuminates decision-making by boards of directors. Directors have diverse preferences characterized by circular indifference curves around their ideal actions. Board decisions are given by a new model of consensus, which is motivated by the idea that the outcome must be consistent with recourse to majority voting for resolving any disagreements. Intuitively, the consensus balances the local directions in which agents would like to move the outcomes, without regard to intensity of preference, but allowing agents to have different bargaining power. The consensus is always Pareto optimal; if there is a majority of directors with the same target, the consensus is that target; if all directors' targets are on a line, then the consensus is the target of the weighted median director (similar to the median voter theorem). The consensus also handles reasonably a director with extreme preferences: such a director cannot enforce a large deviation from what the other directors want but does have some influence. The model can be used to address a number of policy questions. Perhaps most notably we develop the idea that the information a new director brings to a large board is probably more important than the new director's impact on voting, especially when existing directors have similar preferences. This implies, for example, that although an executive from a supplier might have preferences far from value maximization for this firm, the executive may still be a valuable director because of the information the executive brings to the board. There is also a question of whether a simple majority of outsiders is good enough. If insiders can dictate the agenda or more generally have higher bargaining weights than the outsiders, they may be able to obtain their preferred outcome even if in the minority in numbers. And, a minority with concentrated preferences may prevail over a majority with conflicting interests, since the directors in the majority might spend much of their persuasive efforts fighting each other. Another policy question relates to the stiffening of penalties for deviations from measured value-maximization as seems to be the intent of parts of Sarbanes-Oxley. The ability to improve matters using such a penalty depends a lot on being able to measure board performance ex post. If only part of performance can be measured, directors facing the penalty will be driven to optimize the measurable part and neglect the rest.