Interactions of Incentive and Information Structures.

摘要

We have argued previously on the importance of information structure in multiperson optimization problems. More recently, incentive problems have gained attention as a particularly interesting class of multiperson optimization problems with non-tested information structure. Briefly, ''incentive'' is concerned with the ability of one decision maker, the leader, in influencing the decision of another, the follower, by appropriate modification of the latter's payoff. There are several somewhat different versions of the incentive problem. The principal-agent problem has its main difficulty in the fact that the leader or principal cannot observe directly and exactly the decisions of the follower or agent. On the other hand, the incentive-compatibility problem is concerned mainly with the fact that the follower have private information not known to the leader; hence the leader cannot ascertain independently whether or not the decision of the follower which he observes is in fact the correct one. A third version of the incentive problem gives the leader the ability to modify the information structure rather than the payoff structure of the follower. It is through the manipulation of the information structure that the leader hopes to induce the appropriate decision from the follower. The purpose of this paper is primarily concerned with the second version of the problem, but with the added feature of partial information feedback and multistage considerations. 14 refs., 9 figs.