The Origin of the Winner's Curse: A Laboratory Study

Abstract

The Winner's Curse (WC) is one of the most robust and persistent deviations from theoretical predictions that has been established in experimental economics and claimed to exist in many field environments. There have been many attempts to explain the winner's curse, such as ignoring the cognition process of other agents, having a cursed system of beliefs, the presence of level-k heterogeneity and beliefs, and/or misunderstanding the game. In order to capture the underlying roots of this behavior, we design and implement in the laboratory a simplified version of the Acquiring a Company game. Our transformation reduces the game to an individual-choice problem, but one that still retains the key adverse-selection issue that results in the WC. We also conduct a treatment in which one can vastly simplify the problem using ordinal reasoning rather than cardinal reasoning.

Our main results find that the WC is alive and well in all of these environments where equilibrium theories, based on relaxed belief structures, are mute. Our results also suggest that the WC is better explained by bounded rationality of the form that people have difficulties either performing Bayesian updating or performing contingent reasoning on future events. To delve more deeply into the issue, we added a treatment that presents the game as a choice among simple lotteries that are equivalent to the bidding choices available, but which circumvents the need to perform contingent reasoning on future events.

Keywords: Winner's curse, auction, experiment, take-over game, acquire-a-company game, self-cognition bias, cursed (equilibrium) beliefs, level-k reasoning, bounded rationality

JEL Classification: A12, B49, C91, D81