Stochastic Monitoring and Moral Hazard
研究在单期委托代理模型中，如何同时最优选择合同和随机监控策略，以缓解因委托人无法无成本观察代理人行动和环境状态而产生的道德风险问题。
The phenomenon of monitoring subordinates' actions and local environments, so commonly observed in large firms, is an attempt to alleviate the moral hazard problem induced by central managers (principals) being unable to costlessly observe subordinates' (agents') actions and the state of their environment. These monitoring activities take many forms and are generally performed in a stochastic manner. For example, supervisors make surprise checks on workers, and investigations are often triggered by standard cost variance reports. The information gathered from monitoring activities is used to penalize or reward subordinates (agents), thus altering their incentives and behavior. It is intuitively clear that the monitoring policy that will be employed depends upon the contract to be enforced. In a sense, every type of contract has both benefits and associated monitoring costs. The purpose of this paper is to study the simultaneous optimal choice of contracts and stochastic monitoring policies in a single-period principal-agent paradigm. Townsend [1979] first studied monitoring (or verification) in the context of a simple trading economy where the endowments of consumers were stochastic and known only to themselves. However, in Townsend's analysis verification policies were constrained to be deterministic, that is, states were verified with probability one or not at all. With deterministic verification Townsend found that, in the nondegenerate case, unfavorable outcomes were verified and more favorable outcomes were