【主讲】Rongzhu Ke 助教授 (香港中文大学)
【主题】A Max-min-max Approach for General Moral Hazard Problem
【时间】2012年10月23日 (周二) 15：30-17：00
【地点】上海财经大学经济学院楼801室
【语言】英文
【摘要】This paper develops a unified solution method for principal-agent problems with moral hazard under a general setting (i.e., multi-task and multi-signal). Our approach utilizes a key feature of the principal-agent model, namely, the contact of interest between two parties when the output and effort are given. This feature allows us to establish a max-min-max representation of the original problem. Thus, for any implementable action, an optimal contract for it is a stationary point of the Lagrangian that contains three constraints: the individual rationality constraint, the corresponding first-order condition for the incentive compatibility, and one additional inequality constraint that allows the agent to gain utility under targeted action no less than another alternative action (it is called no-jumping constraint). This contract is called the augmented Mirrlees-Holmstrom (AMH) contract, which is used to characterize the optimal contract and solve the problem, regardless of the validity of the first-order approach (FOA). We apply the new characterization to extend the existing criteria for ranking the efficiency of information systems without the FOA. We show that Holmstrom's sufficient statistic criterion remains valid. However, not even Blackwell's condition is sufficient for the ranking of signals. We thus propose several new criteria without the FOA.
