Coordination through Reputations: A Laboratory Experiment

One of the functions of reputations in societies is to enable coordination, and some of the most familiar instances of this occur when reputations are based on past actions. For example, a history of aggressive play is often interpreted as predicting more aggressive play in the future. This permits a pair of individuals with unequal reputations for aggressiveness to avoid wasteful conßict that would result from aggressive play by both. This paper describes a laboratory experiment designed to explore whether, in a population of subjects, such coordination through reputation formation will emerge spontaneously, what form it will take, and how reliably it will be observed by the individuals in the population. Reputation-based coordination of the sort illustrated above did arise in all experimental sessions, but the rate of adherence to the social rule implicitly adopted by the group varied across the experimental populations as a function of treatment variables. ∗We are grateful to a referee, to participants at seminars at M.I.T. and Boston University for helpful suggestions, to the National Science Foundation for Þnancial support, and to Sumon Majumdar for research assistance. Data from these experiments is downloadable (in Stata 6.0 format) at www.wws.princeton.edu/~rjmorgan. †Department of Economics, Princeton University ‡Department of Economics and the Woodrow Wilson School, Princeton University §Department of Economics, Boston University