The winner's curse: experiments with buyers and with sellers

Abstract

This paper explores the winner's curse phenomena as it was studied experimentally by Kagel and Levin. Experiments with the winner's curse are complicated by the fact that subjects can lose money and the experimenter has only a limited means of collecting it from them. Thus subjects enjoy only limited liability which has theoretical implications for behavior. In the Kagel and Levin experiments subjects were removed from the bidders' competition after losses reached a predetermined value. This experimental procedure has unknown implications for the results so ambiguity exists about whether the winner's curse was actually observed. In this study their results were replicated in an environment in which subjects were not removed. The case in which competitors are sellers is also studied. Bankruptcy cannot be a problem in sellers' competition. In both cases the winner's curse is observed. Thus the limited liability cannot be an explanation for the phenomenon reported by Kagel and Levin. In addition the paper examines the bidding behavior of all individuals and shows that this behavior does not fit any of the tested theories either on the aggregate or individual level. The "winner's curse" did not disappear over time during the conduct of the research.