Unraveling in a repeated moral hazard model with multiple agents
研究一个委托人雇佣多个代理人的无限期重复道德风险问题，设计了一类防合谋契约，使得当贴现因子趋近1时，委托人和代理人的均衡收益接近最优基准。
Abstract: This paper studies an infinite horizon repeated moral hazard problem where a single principal employs several agents. We assume that the principal cannot observe the agents ’ effort choices; however, agents can observe each other and can be contractually required to make observation reports to the principal. Observation reports, if truthful, can serve as a monitoring instrument to discipline the agents. However, reports are cheap talk so that it is also possible for agents to collude, i.e. where they shirk, earn rents, and report otherwise to the principal. The main result of the paper constructs a class of collusion-proof contracts with two properties. First, equilibrium payoffs to both the principal and the agents approach their first-best benchmarks as the discount factor tends to unity. These payoff bounds apply to all subgame perfect equilibria in the game induced by the contract. Second, while equilibria themselves depend on the discount factor, the contract which induces these equilibria is independent of the discount factor.