Social Comparisons and Optimal Information Revelation: Theory and Experiments

I analyze a principal-multiple agent model in which agents have imperfect information about their abilities. When performance is affected by shocks that are common to everyone (such as task difficulty), performance comparisons with others are useful in forming beliefs about own ability. Beliefs, in turn, affect effort and hence subsequent performance. In this context, I explore the principal’s organizational design problem where the amount of interim information disclosed to agents about each other’s performances is a choice variable. I find that the optimal disclosure policy depends on: (1) the degree of substitutability of the agents’ performances in the principal’s payoff function; and (2) the amount of discretion the principal has over manipulating contracts. With exogenous contracts, if agents’ performances are sufficiently complementary, withholding social comparison information may be optimal. However, when the principal can choose the wage scheme in addition to the information policy, full information revelation, coupled with a “cooperative” incentive scheme, is universally optimal. The paper also presents findings from a laboratory experiment, which confirm many of the theoretical predictions. The results are potentially applicable to many real-world situations, ranging from the revelation of grade distributions in classrooms to interim performance evaluations and team formation policies in firms. JEL Codes: D83, J33, M54, C91. ∗I would like to thank David K. Levine, Leeat Yariv and Bill Zame for their invaluable advice and support through all stages of this project. I would also like to thank Michael Chwe, Hongbin Cai, Ichiro Obara, Jozsef Molnar, Joon Song and Yossi Yakhin for helpful conversations. This paper greatly benefited from comments of the participants of the theory seminar at UCLA and the California Institue of Technology, and the Budapest Workshops in Behavioral Economics. Financial support from the Russell Sage Foundation for the experiments is gratefully acknowledged. All errors are mine. †Department of Economics, UCLA. Email: sertac@ucla.edu. An online version of the paper and an online appendix for the experiments can be found at http://www.bol.ucla.edu/~sertac/research.html