Why are women underrepresented at the top of both academia and the corporate world? They represent only 25 per cent of professors in the UK and 5 per cent of Fortune 500 CEOs. Behavioural economists have recently offered various new explanations for the gender gap in education and labour market outcomes. Choices made in controlled experimental settings hint at differences in preferences, beliefs and stereotypes.

We highlight one such behavioural mechanism: “competitiveness”, the willingness to enter into contests with others. This personality trait has been shown to be a predictor of career choices and income. Experiments have consistently shown that women are more likely to shy away from competitions. Our experiments suggest that women are also less likely to try again after losing in competitions. Data from the Dutch Math Olympiad indicate that these results also hold outside the lab.

Results from lab experiments

Economists measure willingness to compete in the lab by offering individuals a choice between different rewards for their performance in a task. The choice is between a reward based solely on one’s own performance and potentially receiving a higher reward if one outperforms another participant (and nothing otherwise). When faced with this choice, women tend to choose the safer, non-competitive option. Although this effect has been observed in various settings and for diverse groups of subjects, previous studies have focused on one-shot games. We argue that in most real life settings, there are repeated opportunities to compete with others (think of chances for promotions and research funding). It is therefore important to understand how people respond to failure.

In our experiment students were asked to solve simple arithmetic problems. They also had to choose between either a fixed payment for each correct answer, or twice the payment if they managed to outperform a random “opponent” (and nothing otherwise). How did the choice for competition change over multiple rounds? Unsurprisingly, subjects who opted for the competitive incentive scheme and won almost all continued regardless of gender. A gender gap was found for those who opted for competition in the first round and lost: whereas male losers were 24 percentage points less likely to re-enter competition over the following rounds, this effect was more than twice as large for women, at 59 percentage points.

Interestingly, this gender difference in the reaction to losing was even more pronounced for top performers. High-performing men were hardly affected by a loss in the first round while high-performing women were 56 percentage points less likely to compete again following a loss.

In the analysis above we controlled for the absolute skill level of each subject, which means that any effect is solely driven by different reactions to losing. Being unlucky (i.e. matched to a very strong opponent in the first round) affected men to a much smaller degree. Not only are women less attracted to competing with others in the first place, even those who initially were willing to throw their hat in the ring seem less likely to persevere after a setback. Our results are striking because high performers who are (initially) willing to compete are exactly those whom we would expect to have the potential to reach the top of their careers. Our experiment indicates that high-potential women might be losing out the most by giving up too early.

Evidence from the Dutch Math Olympiad

To evaluate whether our results from the lab also carry over to the real world, we analysed data from the Dutch Math Olympiad. Thousands of high school students participate in this prestigious scholastic contest every year. The Olympiad consists of three successive tests, where only the top performers advance to the next round. We were interested in whether passing the first round affects the probability of re-entering in the following year.

This setting allows us to find the causal effect of passing the initial round, because only students with more than a specific threshold score, the “winners”, were invited to enter the second round. Intuitively, students who scored just below and just above the threshold are very similar in all regards except that some just reached the second round while others did not. They can be compared with each other to find out if losing has any influence.

We find that girls who just lost out were 10-20 percentage points less likely to return in the following year compared to girls who had just made it to the second round. On the other hand, boys who were unsuccessful in advancing were not deterred to participate again relative to those who had just succeeded. Thus the gender difference in perseverance after a competitive loss seems to also affect school-related choices by students, with potentially long-lasting impact.

 Final thoughts

Failure at some point is arguably an inevitable part of a long, successful career. In some careers, including academia or the arts, failure is arguably more common than success even for successful individuals (think about the extremely low acceptance rates at top scientific journals or for research grants). Our results suggest that bad luck early on could be especially costly for women who aim to pursue such careers. One possible way to encourage workplace diversity at the top would be to soften the focus on competition. For example, achievements could be evaluated on absolute levels rather than framed as contests with winners and losers. This means replacing incentive schemes that reward employees for outcompeting their direct colleagues with schemes that recognise absolute performance or with competition between teams rather than individuals. This could prevent discouraging employees and create a more collaborative environment that attracts and retains a more diverse set of talent without compromising on performance.



Thomas Buser is an associate professor of economics at the University of Amsterdam and a research fellow at the Tinbergen Institute. He is interested in empirical behavioural economics and uses lab and field experiments as well as registry and survey data.


Huaiping Yuan is a PhD student in economics at the University of Amsterdam. He is interested in the development of personality traits and the impact of personality on labour market outcomes.