Women are under-represented in top management positions on both sides of the Atlantic. The academic literature suggests a number of explanations for this underrepresentation, including self-selection, investment in family and child bearing, lower female human capital investment, or gender discrimination. Some countries have responded by setting minimum quotas for female managers.1

A new strand of research considers another hypothesis – that the sexes perform differently under competitive pressures, even if these differences do not exist in non-competitive settings. While a number of papers find support for this argument in experimental set-ups (Gneezy, Niederle and Rustichini 2003; Niederle and Vesterlund 2007), the evidence from real life data that controls for other possible explanations of performance gender gap has been scarce. One exception is Paserman (2007) who studies Grand Slam tennis tournaments and finds evidence that women are significantly more likely to hit unforced errors at the crucial stages of the match, while men exhibit no such variation in performance.

Our recent work provides additional evidence from a series of natural experiments drawn from the French higher-education system where, contrary to tennis tournaments, men and women compete with each other, as is the case when they compete for top management positions (Örs, Palomino, and Peyrache 2008).

Competitive entrance exams

We study approximately 5,750 students who, over the 2005-2007 period, took the highly competitive entrance exam for admission to the HEC School of Management in Paris, a “grande école” and a top-ranked European business school.

The stakes for entry are high, as graduates of this elite French school do very well in the job market both in the short- and long-term. The average annual salary of HEC graduates 3 years after graduation is approximately €10,000 higher than that of the closest followers. Among the 32 French firms ranked in the Financial Time Global 500 in 2007, 22% had a CEO who graduated from HEC. The school’s alumni include the WTO Director Pascal Lamy and the IMF Director Dominique Strauss-Kahn.

The data sample

HEC entrance is based solely on overall rank obtained from written and oral exams, irrespective of educational or personal background. Only about 250 of approximately 2,000 applicants with a science background are admitted each year, following a series of written and oral exams covering a wide variety of subjects, from mathematics to general culture, including 2 foreign languages as well as history and geography, and taking place over a 3 week period. The applicant pool is larger, with candidates coming from science, economics, and literature backgrounds. We focus on candidates with science background, the largest cohort, in order to rule out the documented gender gap in high school concentration choices as a possible explanation for the performance gender gap that we observe.

We compare gender differences in entrance exam performance with the performance of the same population on the high school finishing exam (the French baccalauréat) two years prior and the subsequent first-year of their studies at HEC. Though the entrance exam is competitive in the sense that there is a preset quota of places available (known by all candidates), the baccalauréat and the HEC first year are non-competitive academic assessments. There is no mandated success rate. The broad scope of topics covered by each of these assessments should prove an even-handed comparison across genders.

Our analysis involves a series of univariate tests and multivariate regressions. In multivariate regressions, we control for the observable characteristics of the students that are available to us. For example, when examining performance differences in the oral part of the HEC entrance exam, we control for the year of the exam, the nationality of the candidate, the baccalauréat honours obtained, the grade obtained in the written part of the exam, and the preparatory school that the student attended.2

A gender gap in entrance exams

On average, men perform slightly better than women in both the written and oral exams despite evidence the female candidates are ‘better’ in the sense that:

  • in the same cohort of candidates, the females performed significantly better than men in the national baccalauréat exam two years prior to the sitting of the HEC admission exam; and
  • among the sub-sample of candidates admitted to the school, females outperform the males during the first year of their core curriculum classes at HEC.
Male performance has greater variance

The male performance distribution has greater variance – in the top quartile of examinees, men outperform women, while their written exam scores in the lowest quartile are worse than women. Female candidates’ performance is more concentrated around the median. Since only a small fraction of the initial candidates are admitted to the school, men are more likely to be admitted than women, even though roughly equal numbers of men and women apply.

Explanations: testable hypotheses

We posit two not mutually exclusive explanations:

  1. Male candidates may adopt riskier strategies, such as under-investing in some areas of study and focusing on fields given greater weight in the exam (such as mathematics).
  2. The competitive nature of the selection process may have a different impact on male and female candidates.

In our tests of the first explanation, we cannot reject the null hypothesis that male and female candidates' risk-taking behaviours are identical. We find no gender-based difference between individuals’ standard deviations of marks based on all of the written or oral grades that they have received during the examination process. We also find that candidates who do better in mathematics also do better in non-math parts of the exam.

To test the impact of the selection process on performance, we first compare the distribution of performances at the HEC admission contest with the distribution of the performances in other academic assessments.

At the baccalauréat exam, in the same sample of candidates, women perform better than men on average (with first-order stochastic dominance). Then, using only the subset of students who have been admitted to the school, we compare the performance distributions based on the grade point average that they obtain during their first year at HEC. We again show that, once admitted, women perform better than men in terms of grades (with first-order stochastic dominance).

An explanation for the gender gap in higher management?

Taken together, these results suggest that, within the same sample of subjects, the competitive pressure generated by relative performance evaluations has a different impact on men than on women. This provides an alternative (albeit, by no means exclusive) explanation for under-representation of women in top management positions, if one takes the perspective that these positions are the rewards of performances in a series of competitive tournaments during one’s professional career, as opposed to pass/fail type of evaluation processes.

Finally, if the structure of the selection process indeed impacts performance differently across genders, then the under-representation of women may start as early as the professional schools that select students based on competitive exams.

References

Gneezy, Uri, Muriel Niederle, and Aldo Rustichini (2003). “Performance in competitive environments: Gender differences.” Quarterly Journal of Economics 118: 1049-1074.
Niederle, Muriel, and, Lise Vesterlund (2007). “Do women shy away from competition? Do men compete too much?” Quarterly Journal of Economics 122, no. 3: 1067-1101.
Örs, Evren, Frédéric Palomino, and Eloïc Peyrache (2008). “Performance Gender-Gap: Does Competition Matter?” CEPR Discussion Paper 6891.
Paserman, M. Daniele (2007). “Gender Differences in Performance in Competitive Environments: Evidence from Professional Tennis Players”. CEPR DP6335

 


1 For example, as of January 1st 2008 Norwegian companies are required by law to have at least 40% female board members, as opposed to the prevailing average of 7% in 2003.

 

2 While all university applicants are guaranteed a spot after high school, grande école applicants attend non-degree conferring prep-schools for two years before taking the competitive entrance exams.

 

2,100 Reads