Communication in Economic Mechanisms

This chapter considers the problem of …nding allocations that satisfy certain social goals when economic agents have private information about their preferences. While economists have traditionally considered the problem of providing incentives for agents to fully reveal their preferences, such full revelation is often impractical or undesirable, for several reasons: (1) it may require a prohibitive amount of communication as measured in bits or real numbers, (2) it may be costly for agents to evaluate their complete preferences, and (3) the revealed information may be exploited by the designer or other agents. Thus, we consider the question: What is the minimal information that must be elicited from the agents in order to achieve the goals? Note that the question arises even if agents can be counted on to report truthfully. Segal (2005) shows that for a large class of social problems, any minimally informative way to verify that a given alternative is desirable is equivalent to giving each agent This chapter was prepared for the 9th World Congress of the Econometric Society, London, August 19-24, 2005. I gratefully acknowledge …nancial support of the National Science Foundation (grants SES 0214500, 0427770). I also thank my discussant, Roger Myerson, and the participants of too many seminars and conference presentations where some of the research surveyed in the chapter was presented. In particular, I am grateful to Susan Athey, Jonathan Levin, Eric Maskin, Paul Milgrom, Andy Postlewaite, Thomas Sjostrom, and James Jordan for discussions related to this topic. Last but not least, I am indebted to my coauthors Noam Nisan and Ronald Fadel for helping clarify my understanding of the topics discussed in this chapter. yDepartment of Economics, Stanford University, Stanford, CA 94305. Email: ilya.segal@stanford.edu