委托-代理问题解的等价性与近似等价性

摘要

For a broad class of principal-agent problems, and for “almost all” specifications of the distribution of the agent’s type, any two optimal mechanisms are equivalent: conditional on each type of the agent, neither the principal nor the agent cares which mechanism is in force. Nearly-optimal mechanisms are nearly equivalent. This provides substantial reassurance about numerical analysis of mechanisms in applied work, and has important theoretical implications as well. The model allows for both adverse selection and moral hazard. The substantive assumptions are that the set of types is finite, and a separability assumption on the agent’s utility.