Can gender differences in distributional preferences explain gender gaps in competition?

Highlights

• Our experiment evaluates effects of distributional preferences on competitiveness.

• Selection into competition is negatively associated with behindness aversion.

• Preferences for favorable inequality increase selection into competition.

• Significant gender differences in the impact of distributional preferences.

Abstract

We design a lab experiment to specifically examine whether a preference for favorable inequality and behindness aversion, as well as egalitarian preferences, affect competitive choices differently among males and females. Using data on approximately 2,000 subjects, we find that selection into competitive environments is negatively related to egalitarian preferences, with smaller negative impacts of being egalitarian on females’ choice to compete. Further, behindness aversion and preference for favorable inequality affect willingness to compete in opposite ways. The willingness to compete is negatively affected by behindness aversion, while a preference for favorable inequality positively influences willingness to compete. Interestingly, when we disaggregate behavior along gender lines, we find that compared to behindness averse males, behindness averse females are more likely to enter the competitive environment. In contrast, there is no significant gender difference in the impact of preference for favorable inequality on competition. Our results suggest that the observed gender difference in competitiveness can stem from male-female differences in distributional preferences and selected personality traits developed during one’s lifetime.

Introduction

Winning a competition has obvious economic benefits. However, competition is inherently inequality inducing, creating a divided society of winners and losers (Frank, 1996). Consequently, a distaste for inequality might encourage individuals to stay away from competitive environments, apprehensive of the disutility from an unequal outcome. Fehr and Schmidt (1999) point out however that individuals might have different preferences towards advantageous and disadvantageous inequalities. Consequently, attitudinal differences towards advantageous and disadvantageous inequalities can influence choices differently (Beranek et al., 2015, Teyssier, 2008).

The relationship between competitiveness and distributional preferences assumes further importance in retrospect with the evidence from the experimental literature. First, experimental results consistently find females to be less inclined to compete than males (Buser et al., 2014, Flory et al., 2015, Niederle and Vesterlund, 2007). Second, the literature generally finds women to be more egalitarian than men (Andreoni and Vesterlund, 2001, Dufwenberg and Muren, 2006, Engel, 2011, Sharma, 2015).

The existing literature has largely focused on examining gender gaps in competitiveness, controlling for variations in characteristics such as risk preferences, confidence, personality traits, family background, and session composition (e.g., Croson and Gneezy, 2009, Niederle, 2016, Niederle and Vesterlund, 2011). However, the extent to which gender gaps in competitiveness are explained by gender differences in distributional preferences remains understudied. A few papers explore distributional preferences as plausible mechanisms for explaining gender differences in competitiveness, but do not delve into gender-differentiated effects of distributional preferences. For instance, Balafoutas, Kerschbamer and Sutter (2012) classify a small sample of Austrian university subjects into inequality averters, efficiency seekers, inequality lovers, and spiteful agents, and find that spiteful and inequality-averse subjects avoid tournaments when given a choice. Additionally, upon controlling for these distributional preferences, risk attitudes, overconfidence and past performance, the gender difference in competitive behavior disappears. Kamas and Preston (2015) explore behavior in a battery of economic games as a function of distributional preferences measured using a three-person dictator game with US university students. They examine the choice among three compensation schemes (egalitarian, piece rate, and competitive payments) as a function of their social categorizations as self-interested, inequity averting, or efficiency maximizing decision-makers.¹ They find women to be no more likely to choose the piece rate scheme over the competitive scheme; women appear to be significantly more likely to choose the equal pay over competitive pay than men, but that gender difference is explained away upon adding controls for confidence. Using data from a real-effort task played over multiple rounds, Gill and Prowse (2014) find that women decrease work effort after winning a large prize, and they suggest that this response to a competitive outcome may be explained by their distributional preferences (influenced by guilt or through egalitarian preferences).

In this paper, we contribute to this relatively small literature on distributional preferences and competitiveness by focusing on the following three issues: first, we are interested in understanding the relationship between distributional preferences and competitive choices. To that end, we categorize distributional preferences into preference for favorable inequality and aversion to unfavorable inequality (or behindness aversion). Note that an observed distaste for competition need not necessarily stem from distaste towards unequal outcomes; instead it can also be due to a lack of confidence and/or risk aversion (Bartling, Fehr, Maréchal, & Schunk, 2009). We hypothesize that after controlling for confidence, risk preferences and other observable characteristics, a preference for favorable inequality should positively affect willingness to compete; behindness aversion in contrast, might be negatively related to willingness to compete. That is, subjects who prefer to be ahead of others would self-select into competition that allows them the opportunity to get ahead. In contrast, a subject who does not like to be left behind might like to avoid competitive environments that can lead to such an outcome.²

Second, we find that the existing literature have not typically delved into gendered differences in family background, behavior (including distributional preferences), and personality in explaining gender differences in competitive choices. Typically, the binary variable for gender in regressions reflects the male-female gap in competitiveness after controlling for other characteristics. But such regressions assume that the influence of these right-hand side (RHS) characteristics on one’s willingness to compete remains identical for males and females, which is often not realistic. It is important to control for this heterogeneity, especially in the presence of multi-dimensional gender inequalities prevailing in many developing countries. For example, in the case of India, which constitutes the setting of our study, there is consistent evidence of skewed sex ratios disfavoring females at birth. Even among surviving children, females are more likely to be neglected in health and education related investments (e.g., Jayachandran, 2015). Further, high levels of gender inequality may cause males and females to internalize societal norms and expectations differently. This may lead women to believe that competitiveness is an undesirable trait, or that their decision to compete will not be positively rewarded by society (Andersen et al., 2013, Barry, 2016). In such a situation, it is important to include gender-differentiated effects of characteristics to evaluate nurture-based differential investments that lead to male-female gaps in competitiveness. To allow for the effects of these characteristics to vary by gender, we interact the female dummy with distributional preferences and the full vector of controls that have been introduced in the previous literature to further explain gender gaps in competitive choices in our framework.

Third, we measure and explore the role of personality traits (Big Five and Locus of Control) as potential determinants of competitive choices. Studies on occupational choice find that one’s preferences and traits such as risk-taking propensity, Big Five, innovativeness, and Locus of Control determine entry into inherently uncertain activities such as entrepreneurship (Caliendo et al., 2014, Cobb-Clark and Tan, 2011, Mueller and Thomas, 2000). More recently, experimental evidence also shows that Big Five personality traits determine behavior in a variety of games in the lab (Bartling et al., 2009, Cubel et al., 2016, Gill and Prowse, 2016, Müller and Schwieren, 2012). To the best of our knowledge, other studies have not explored the importance of locus of control as a determinant of competitive preferences.

Using a large sample of approximately 2000 subjects at colleges in University of Delhi in India, we observe at first blush a significantly negative relationship between egalitarian distributional preferences and willingness to compete. Upon disentangling these egalitarian distributional choices into preference for favorable inequality and behindness aversion, we find that the choice to compete is negatively related to behindness aversion, and preference for favorable inequality positively determines the choice of the tournament wage scheme.³ Interestingly, when we disaggregate behavior along gender lines, we find behindness averse females are almost 13 percentage points more willing to compete than behindness averse males. In contrast, there is no gender difference in the relationship between preference for favorable inequality and willingness to compete. We follow up with standardized regressions to assess the relative importance of various factors in explaining gender gaps in competitiveness, and find that gender differences in distributional preferences and Big Five measures of conscientiousness and emotional stability are the most economically and statistically significant effects in our sample. Our results suggest that the gender gap in competitiveness observed in the literature could be driven by gendered differences in distributional preferences and selected personality traits that are developed during one’s lifetime.