The emergence of male leadership in competitive environments

Abstract

We present evidence from an experiment in which groups select a leader to compete against the leaders of other groups in a real-effort task that they have all performed in the past. We find that women are selected much less often as leaders than is suggested by their individual past performance. We study three potential explanations for the underrepresentation of women, namely, gender differences in overconfidence concerning past performance, in the willingness to exaggerate past performance to the group, and in the reaction to monetary incentives. We find that men's overconfidence is the driving force behind the observed prevalence of male representation.

Introduction

An important and often forgotten aspect of organizations is how efficient they are in selecting the right individual to perform a particular task. In competitive environments, an organization that is good at identifying and selecting its most talented individuals will be able to outperform organizations that might very well have more talented people but for some reason do not select them to perform the tasks they are good at. In this paper, we study experimentally whether organizations fail to select high-ability women to perform in competitive environments. Specifically, we investigate three potential gender differences that can lead groups to fail to select their high-ability women: first, gender differences in judging one's own abilities (e.g., due to overconfidence); second, gender differences in the willingness to exaggerate one's own abilities, and third, gender differences in how individuals respond to conflicts of interest between their own interests and their group's (i.e., agency problems). To do so, we ran an experiment where groups select a leader to represent their group and compete against the leaders of other groups in a real-effort task. Everyone in the group of the winning leader receives a prize. Barring any explicit discrimination against women – which would be unlikely in an experiment with university students – groups should aim to select their most talented individual irrespective of gender. However, we find that women are selected to represent the group 33.3% less often than their abilities would suggest. In other words, we find underrepresentation of women in a competitive environment that is not due to overt discrimination or to gender differences in preferences.

Our experimental design allows us to isolate the potential effect of each of the three aforementioned gender differences in order to explain why groups fail to select female leaders. The experiment was carried out with subjects who had already performed the real-effort task 15 months before participating in the current experiment. We took advantage of this fact to ask subjects, using an incentive compatible procedure, to recollect their past performance. If there is a gender difference in the recollection of past performance and groups are unaware of it, then low-ability but overconfident men will end up being chosen over high-ability but underconfident women.¹ In addition to remembering their past performance, we asked subjects to make known to their groups what their expected performance is, although their statements did not have to be truthful. For groups to make the best decision, individuals must truthfully reveal their true ability to others. Therefore, if there is a gender difference in the willingness of individuals to exaggerate their own ability (e.g., compared to women, men might dislike revealing they have a low ability, or they might be more willing to lie and overstate their own abilities) then groups could end up selecting low-ability men that claim to be high-ability over high-ability women that do not overstate (or even understate) their ability. Finally, we introduce an agency problem by giving an additional financial incentive to the individuals who are selected as leaders. If, compared to women, men react more to the monetary incentive (and are willing to lie about their ability), then we ought to see less women as chosen to represent their group in treatments where the additional payment is large.

We find that the underrepresentation of women is mainly driven by men being overconfident when they recall their own ability. Once subjects of both genders have recalled their performance, they do not overstate their performance differently in order to be picked as the group leader. Finally, although higher economic incentives do induce subjects to overstate their performance, they have a similar effect on both men and women.

There is a growing literature on gender differences in competitive environments. Gneezy et al. (2003) and Gneezy and Rustichini (2004) find that the relative performance of women vis-à-vis men worsens when they are competing against another individual, and in particular if the other individual is a man. Niederle and Vesterlund (2007) demonstrate that women can be absent from competitive environments because they ‘dislike’ being in competitive settings (compared to men). That is, women select themselves out of competitive environments even though they would be financially better off if they decided to compete. Dreber et al. (in press), Gneezy et al. (2009) and Günther et al. (2010) demonstrate that the negative effects of competition on performance and selection might be the result of cultural differences and stereotypes associated with the particular task used. Niederle et al. (2009) and Balafoutas and Sutter (2010) explore the effects of various policy interventions in lieu of gender differences in competitive environments. A survey of this literature is found in Booth (2009). This paper extends this line of research to the case where selection into competitive environments is the outcome of a group decision. Finally, our paper is related to the work of Babcock and Laschever (2003) and Bowles et al. (2007), who demonstrate that women tend to negotiate less than men and are more likely to be punished for negotiating. However, unlike in negotiation games, which tend to be zero-sum, in our setting incentives are more aligned and thus bias against selecting qualified women hurts not only these women but also the whole group.

The paper is organized as follows. In Section 2, we describe the experiment's design and procedures. In Section 3, we present the experiment's results. We provide a short discussion and conclusion in Section 4.