Do we need a (large) committee?

Abstract

This paper studies committee design when a homogeneous population is uncertain about which alternative is correct, individuals can acquire costly information about the state of the world, and the decision must be taken via voting with the majority rule. We assume verifiability of costs and cost sharing, which are realistic assumptions in many applications, e.g. if the time spent in learning is recorded, or if reports have to be produced. We show that the optimal committee size depends on the properties of the cost function and on the size of the population. If either the marginal cost function satisfies a standard single-crossing condition or the population is sufficiently large, it is optimal to delegate voting to a small committee and often to a single individual, in which case no committee is needed at all. If information acquisition costs increase slowly for small values of information, but the magnitude of such costs is large, the optimal committee might comprise many members. Our results add a rationale for the widespread use of small committees for decision-making.