Optimal Contracts for Teams

Abstract

In a team subject to both adverse selection (each member's ability is known only to himself) and moral hazard (effort cannot be observed), optimal contracts are, under certain conditions, linear in the team's output. The outcome is the same whether the principal observes just the total output or each individual's contribution. Thus monitoring is not needed to prevent shirking by team members; instead, the role of monitoring is to discipline the monitor. Production is a collective enterprise: by workers in a firm or firms in a joint venture. The synergy that is the reason for the team's existence may mean that an individual's contribution to the team's output is not distinguishable, so that it is impossible to pay him according to his own productivity. How should the principal remunerate the team members so as to maximize his own profit? In this paper we analyze incentives in teams with asymmetric information: both adverse selection (each member's ability is known only to himself) and moral hazard (effort cannot be observed directly). Holmstrom (1982) showed that, in a team model with moral hazard, the principal can ensure an outcome arbitrarily close to the full-information ideal by using a contract that punishes each team member arbitrarily severely whenever team output falls below some target. However, this seems to be an unrealistically drastic way of solving the moral-hazard problem. In our model, we in effect prevent the principal from using such a contract by introducing adverse selection in addition to moral hazard, and by assuming that ability and effort interact in such a way that the principal is unable to disentangle an agent's effort from his ability. Under the assumption that the principal and the team members are risk neutral, we shall show that the principal can implement his information-constrained optimum simply by offering each agent a payment linear in the team's output. Although the essential feature of our analysis is the interaction between adverse selection and moral hazard, this result can be understood by first considering the special case in which the agents' abilities are common knowledge. In this case the moral-hazard problem can be completely solved: the principal can do as well as he