The diverse range of problems faced by decision makers almost precludes the possibility of their possessing the expertise needed to make informed judgements in all areas. As a result, the power to make decisions is often divorced from the relevant specialized knowledge. It is the `division of labor` between decision makers and experts that is the focus of this research. Specifically, its goal is to analyze the strategic interplay between a single decision maker and a number of experts from whom he might choose to seek advice. Employing tools from the economics of information and game theory, the project will undertake a theoretical analysis of a model broad enough to be consistent with a variety of institutional settings. Of particular concern are situations in which the decision maker is able to consult multiple experts and, thus, is the recipient of multiple, sometimes conflicting, pieces of advice. Moreover, this advice is not given by completely disinterested parties: the experts may have interests which, even if not at odds with, are not coincident with those of the decision maker. The latter's task is then to aggregate the information received in such an environment effectively. Three broad sets of issues will be addressed. The first concerns the tradeoff between expertise and loyalty, the latter determining how closely an expert's interests are aligned with those of the decision maker. Some questions addressed are: How should a decision maker integrate the opinions of experts who are equally proficient but have differing loyalties? Is it possible to get the experts to disclose more information by playing one off against the other? Should the advice of `inferior` experts necessarily be ignored? How many experts should be consulted? The second relates to the twin issues of compensation and delegation. How should experts be compensated for their advice? Can optimal compensation schemes lead to full disclosure? When should a decision maker delegate the authority to make decisions to an expert? The third part of the project will explore some wider implications of the research summarized above to the general question of information elicitation from imperfectly informed agents. The fact that the decision maker cannot credibly commit to a course of action, a mechanism, distinguishes this problem from that traditionally considered in mechanism design theory. In particular, contrary to the lessons of the traditional theory, achieving full-disclosure and first-best outcomes is difficult.
