Abstract:
In dynamic principal-agent relationships, it is sometimes observed that the agent's reward depends only on the final outcome. For example, a student's grade in a course quite often depends only on the final exam score, where the performance in the problem sets and the mid-term exam is ignored. The present paper shows that such an arrangement can be optimal if the agent's effort in each period has strong persistent effects. It is shown that the optimality of such a simple payment scheme crucially depends on the first order stochastic dominance of the final outcome under various effort sequences.