Gender robustness of overconfidence and excess entry

Abstract

Camerer and Lovallo (1999; henceforth CL) present thought-provoking experimental evidence that overconfidence might lead to excess entry into markets. As their findings are based on the majority of sessions exclusively consisting of male participants, we conduct two experiments in an attempt to replicate their study while including both men and women in all of our sessions. Our Experiment 1 closely follows CL’s design whereas Experiment 2 employs a gender-neutral addition task and provides more control in assessing gender differences in overconfidence and excess entry. Using participants of both genders we are unable to replicate CL’s main finding that market entry decisions are driven by overconfidence. Contrary to CL, where self-selection increases the entry rate, in our Experiment 1 self-selection leads to less entry. This result is driven by self-selected females who rationally enter the market less often than self-selected males as their rank-determining performance on a sports and current events trivia quiz is weaker than the performance of self-selected males. In Experiment 2 we find no effect of self-selection on entry and no gender differences in entry rates or performance in the addition task. Our results point out that (i) the finding that overconfidence leads to excess entry is not robust to a population consisting of both genders; and (ii) the self-selection effect is sensitive to both gender and task that is used to determine the rank upon entry.