How Costly is Diversity? Affirmative Action in Light of Gender Differences in Competitiveness

Affirmative action improves women’s willingness to compete.


Opponents of affirmative action argue that it results in reverse discrimination and lowers the qualifications of selectees. However, affirmative action policies might be helpful if the initial selection process is flawed. For example, if minority candidates choose not to apply for certain competitive positions, then affirmative action may encourage them to do so. The increase in qualified candidates may reduce or even eliminate societal and organizational costs. Experimental evidence shows that women and men significantly diverge in their preferences to enter into a competitive environment. In particular, research shows that high-performing women choose not to enter into lucrative competitions, hurting their payoffs. This paper examines whether affirmative action policies can eliminate or reduce the gender gap in propensity to compete. Specifically, it examines whether high-performing women can be encouraged to enter into competitions.


After gender quotas were introduced, women’s propensity to enter into potentially more lucrative competitive environments increased.

  • Women, including high-performing women, were less likely to choose to participate in competitive environments. When given a choice, 31% of women selected the competitive tournament compensation scheme instead of the non-competitive piece-rate scheme, while 74% of men selected the competitive scheme.
  • With affirmative action—reserving one of the two winner positions for the highest performing woman—women’s willingness to compete increased by 25 percentage points while men’s willingness to compete decreased by 9 percentage points. This nearly closes the gender gap, with 61% of men opting in and 56% of women.
  • The increase in women’s willingness to compete was more than what can be explained by the increased probability of winning (due to affirmative action policies). The authors suggest that the gender gap in entering competition could be due to different confidence levels, especially in mixed sex environments. Women tend to be more averse to participating in mixed sex environments while men tend to express overconfidence in their ability in mixed sex versus single sex environments.  
  • Results show that affirmative action policies need not be costly when they encourage qualified applicants to enter into competitions  -- qualified candidates apply to competitive positions that they otherwise would not apply for.

In short, affirmative action increases women’s participation in competitive environments. They express more confidence and are more willing to compete when gender quotas are introduced.


42 male and 42 female students were recruited from the standing participant pool of Harvard Business School’s laboratory for experimental research. During the experiment, participants were placed into groups of three women and three men, and asked to complete six rounds of an addition task.  In each round of the addition task, participants added sets of five two-digit numbers for five minutes. At the end of the study, participants were randomly paid for only one round of the task. Participants learned their own performance at the end of each round but did not know how they performed relative to others until the end of the study.

In Round 1, participants performed the addition task under a, non-competitive, piece rate compensation scheme. If randomly selected for payment, this scheme awarded 50 cents per correct answer.  

In Round 2, participants performed the same task under a competitive, tournament compensation scheme. Under this payment scenario, the two highest-scoring participants were paid $1.50 per correct answer, while everyone else received no payment.

In Round 3, having experienced both schemes, participants chose between non-competitive piece rate and competitive tournament incentive schemes. The chosen scheme was applied to the focal participant’s Round 3 performance. Participants were informed that if they chose the tournament scheme their Round 3 performance would be compared to the other members’ Round 2 performance. Therefore, focal participant’s decision to compete was an individual one because it did not affect the earnings of others nor did it depend on the entry decisions of others.

In Round 4, participants chose between a piece rate compensation scheme and an affirmative action tournament. Once again, the decision to enter the competition was an individual decision because focal participants round 4 performance was compared to other members’ past (Round 2) performance. Under affirmative action, two winners received $1.50 per correct answer, while others received no payment. Of the two winners, one was the highest performing woman and the second winner was the highest-performer of the remaining five participants,(regardless of gender).

In Round 5, participants did not have to perform, but chose between piece rate and tournament compensation scheme to apply to their past task-1 performance. In Round 6, participants did not have to perform either, but had to choose between piece rate and affirmative action tournament to be applied to their task-1 performance. Therefore, Round 5 and Round 6 entry decisions were individual as well. Furthermore, by not requiring participants to actually perform in a competition, Round 5 and Round 6 removed the possibility that individuals’ decisions’ would be affected by thrill, anxiety or fear of performing in a competition.

MLA: Niederle, Muriel, Carmit Segal, and Lise Vesterlund. "How costly is diversity? Affirmative action in light of gender differences in competitiveness." Management Science 59.1 (2013): 1-16.

APA: Niederle, M., Segal, C., & Vesterlund, L. (2013). How costly is diversity? Affirmative action in light of gender differences in competitiveness. Management Science, 59(1), 1-16.

Chicago: Niederle, Muriel, Carmit Segal, and Lise Vesterlund. "How costly is diversity? Affirmative action in light of gender differences in competitiveness." Management Science 59, no. 1 (2013): 1-16.