Income Inequality: the Tip of the Iceberg?

Income Inequality: the Tip of the Iceberg?

05.04.2016, by
Discrimination against women in the job market begins as early as the recruitment stage. CNRS researcher David Masclet explains this phenomenon highlighted by experimental economics, and discusses the vicious circle of sexism that takes root during childhood.

Your research concerns experimental economicsFermerResearch method used for testing conventional economic models with participants, in the laboratory. The objective is to be able to observe whether or not a model prediction is true, if such prediction cannot be validated using conventional methods—such as surveys., a relatively recent discipline that involves testing the reactions of participants in simulation games. The idea is to compare theoretical economic models with data observed in “real” situations. What does this new aspect of economics reveal about discrimination against women in the job market?
David Masclet:1 First of all, it has enabled us to observe that discriminatory behavior towards women does indeed exist during the recruitment process.2 Secondly, it makes it possible to study the root cause of this discrimination, which is usually explained by two opposing theories. One suggests that it is the result of “intra-group” favoritismFermerA behavior that favors members of one’s own group against others who do not belong to it. For example, it prompts a man to prefer recruiting another man. This type of discrimination is already present in early childhood, according to tests on children who were split in two groups (red and blue, for example)., which influences employers to hire someone who “resembles” them. The second postulates that, having no reliable information on a candidate’s skills, employers are reduced to interpreting signals, the main one being educational level. The problem is that recruiters often try to draw conclusions from other parameters for which individuals cannot be held accountable, such as age, gender, skin color, etc. And these interpretations are influenced by stereotypes—women supposedly performing less well than men in the same job, for example. Conventional data derived from employment statistics or surveys is insufficient to determine which theory, whether intra-group or statistical discrimination, is actually at play.

Why can’t conventional survey results be used to identify the origin of the discrimination?
D.M.: The hiring process is a real “black box.” Statistics can tell which candidates were recruited, but not what was going on in the employer’s mind during the decision process. And if recruiters are asked about it in a survey, they are not likely to tell the truth if they indeed practice any kind of discrimination. Our experiments, on the other hand, allow us to observe the participants’ decision-making process in the laboratory: they are asked to rank, from most to least favorite, various job candidates based on files that include their educational level and areas of study, and an avatar that makes it possible to determine the gender of each applicant. Discrimination is shown when the educational level and all other variables being equal, a participant ranks a woman lower than a man. Of course, the educational levels and disciplines were homogenous among the 72 women and 72 men of our 144-strong experimental panel.

Women marching against discrimination and family violence and demanding equality for women on International Women's Day on March 8th, 2016 in Melbourne (Australia).
Women marching against discrimination and family violence and demanding equality for women on International Women's Day on March 8th, 2016 in Melbourne (Australia).

And what were your results? Why is there discrimination against women in hiring?
D.M.: In our experiments, the discrimination is essentially statistical, in other words based on prejudices and lack of information. Indeed, the “employers” who discriminate against women in the first phase of the experiment no longer do so in the second phase, when they are given information on the candidates’ performance. So the good news is that employers, or at least the ones on our panel, are not prejudiced!

We observed that women discriminate against women as much as men do.

In addition, we noted that women (43 of the 72 on the panel) discriminate against women as much as men do (40 out of 72). This shows that it is not a case of intra-group preference, and that women are also influenced by negative stereotypes about their own professional skills.

I’m not sure I understand why the participants in your experiments would be more honest and forthcoming than those in a survey. Since they are making choices that have no actual consequences for them, how can this be considered a “real” situation?
D.M.: But their choices do have consequences! The participants actually “work” during the experiment. Once “hired,” they are asked to solve logic problems, play code breaking games, etc., for their “employer.” At the end of the trial, employers and employees are paid based on their performance. It prompts the employers to reveal their true choices, instead of trying to appear as ethical as possible. This is, of course, assuming that the individuals are trying to maximize their gain, which seems reasonable in an experiment that takes place in the context of work, business and professional achievement.

We put ourselves into one of the worst situations for observing discrimination, and we see it anyway! 

But given the small sums at stake (the participants earned an average of about €20), one could doubt the reliability of the process for ensuring that they are being honest…
D.M.: You are right in the sense that these laboratory studies have their limits and are, after all, simulations. But they are the best way today to get people to reveal their preferences. In fact, there is another flaw: our panels are mostly made up of university students—who are statistically better educated than the average, and less inclined to discriminate. It is therefore probable that the results of our experiments underestimate the extent of discrimination against women. It also means that, although we are putting ourselves into one of the worst situations for observing discrimination, we see it anyway! That says it all…

However, if women are discriminated against in hiring, why is their unemployment rate in France the same as men's (about 10%)?3
D.M.: Because you need to take the employment rate into account: among people aged 15 to 64, 60.2% of the women work, compared with 68.1% of the men.4 Other variables also need to be considered: most part-timers are women.5 This is a “choice” that is often imposed, due to traditional family roles and the salary differences between women and men:6 these two factors prompt mothers, rather than their partners, to spend part of their time taking care of the children, with a minimum loss of household income. Lastly, education plays a role in the vicious circle of gender discrimination. For example, a study7 has shown that, without knowing it, teachers encourage competition among boys rather than girls. This tends to build up an overconfidence biasFermerA tendency to overestimate one's knowledge and abilities, and have too much confidence in one’s judgment. in boys, paving the way for future professional inequalities. Not to mention, outside of school, the weight of social norms in general (advertising, toys, etc.). This certainly explains young people’s educational choices, and in particular the under-representation of girls in scientific programs, despite their excellent reputation. Girls “integrate” the biases dictated by society from an early age. It is therefore hardly surprising that they should end up taking most of the short-term, insecure jobs.8 There is much talk about the salary gap between women and men, but it is just the tip of the iceberg! Everything that happens upstream, in education, skill acquisition and the job recruitment process, is of vital importance.

Do your results have any concrete effect on decision-making among managers and elected officials? I’m also thinking of the ABCD of EqualityFermerThe objective of this education program put forth by Najat Vallaud-Belkacem, then French Minister of Women's Rights, was to fight against sexism and gender stereotypes. Tested experimentally in 2013, it was discontinued after a violent disinformation campaign accusing her of wanting to remove gender differences and teaching "gender theory." Unlike gender "studies," there is no such theory ( Click here for more on this story. educational program, which sparked a flood of absurd rumors of unprecedented vehemence. Has your work inspired schools to fight sexism?

Even before salary gaps, what occurs in education and during the recruitment process is vital.

D.M.: All I can tell you is that our article from 2012 on gender discrimination in hiring was read by French legislators and quoted in the National Assembly information report of April 24th, 2013. It is too soon to talk about a real decision-making aid, but our research at least puts forth a solid base from which to broach the subject, in particular through the media. As you know, media coverage can help bring about very rapid change for the better. In 2002, a study in the USObjet inconnu showed that basketball referees called fouls on black players more often than whites. Five years later, the results were widely reported in the media, which led to the almost total disappearance of this type of prejudice.

Part of a TV advertisement for a French store that has for years offered a non-sexist toy catalog, which features both girls and boys playing with vacuum cleaners, stoves, or cars.
Part of a TV advertisement for a French store that has for years offered a non-sexist toy catalog, which features both girls and boys playing with vacuum cleaners, stoves, or cars.

I suppose that knowing the exact nature of the bias, as revealed in your study, would be extremely valuable for the formulation of future government policy…
D.M.: Absolutely. When the origin of discrimination is known, it is easier to take specific action to combat it. To further our investigation in this area, I’m now working with a PhD student, Guillaume Beaurain, on quota systems, and conducting experiments similar to those mentioned above. Once again, discrimination against women in hiring was observed in the first phase of the “game,” but disappeared when the participants who didn’t achieve recruitment parity had to pay a fine. Our results also show that "recruiters" comply with the quota system even when the fines are low.


 A quota policy may be a crutch, but it’s a useful one.

Does that mean that the penalty for failing to meet the quota serves more as a reminder than an incentive?
D.M.: Yes—it’s a kind of “warning” that no doubt prompts the participants to avoid appearing in a bad light. Extrapolating to the scale of a company, we can imagine that the reputational cost, i.e. the risk of being stigmatized in the media, is more important than any sum of money they might have to pay. Let’s take the case of L’Oreal, which had very few women on its board of directors until 2009. This was pointed out in the media in response to the actions of a feminist group , and the multinational took action to correct the situation.9

You have also shown that quota systems do not harm the companies that apply them. What does this mean exactly?
D.M.: It means that, according to our experiments, the overall performance of recruited employees is not affected by the adoption of a quota system. Here again, we assessed their performance by asking the participants to carry out actual tasks, alone or in teams, and we observed none of the negative effects usually cited by the critics of quota systems.

What are the usual criticisms leveled against quotas? Aren’t they ultimately just crutches for a system that remains fundamentally unequal?
D.M.: Regarding the criticisms of positive discriminationFermerAct of temporarily giving preference to groups who are victims of systematic discrimination (on the grounds of age, gender, ethnic or social origin, disability, religion, etc.), in order to restore equal opportunities. Quotas are an example. Widely applied in the US during the 1960s (in particular towards African Americans), positive discrimination in France is often deemed to be contrary to the country's republican principle of equality., some studies conducted in the US claim, for example, that it leads to the rejection of qualified candidates and thus hinders the company’s performance. This is called reverse discrimination. We have indeed seen this in our work, but it does not outweigh the positive effects of quotas. Another criticism is that employees might feel as though they have been recruited for the wrong reasons (in order to fill a quota rather than for their skills), which lowers their motivation and dedication, leading to a potential drop in productivity. But in our study, the women who were hired through positive discrimination have, on the contrary, tended to “go the extra mile,” perhaps to prove that they are qualified and show the employer that hiring them was a good decision. A quota policy may be a crutch, but it’s a useful one. In France, it has enabled more women to access careers in politics. Similarly, the law of January 27th, 2011 stipulating that the administrative boards of medium and large companies must include at least 40% staff of each gender is a positive step. We can’t deny that there is still a “glass ceiling,” but it seems to me that our public policies are headed in the right direction.

  • 1. Researcher at the Centre de Recherche en Economie et Management(CNRS / Université Caen Normandie / Université Rennes I) and winner of the CNRS Bronze Medal in 2012.
  • 2. Studies in experimental economics had already revealed discrimination in hiring against women and persons of color: David L. Dickinson & Ronald L. Oaxaca (2009) and Marco Castillo & Ragan Petrie (2010) in the US; Magnus Rödin & Gülay Özcan (2011) in Sweden.
  • 3. See:
  • 4. See:
  • 5. According to the Centre d’Observation de la Société, 30% of women hold part-time jobs, compared with 6.7% of men:
  • 6. The salary gap between women and men is about 20 points. Even when the figures are adjusted to “full-time equivalent” (to account for the larger proportion of women in part-time jobs) and compared only for equal work, there is still an unexplained gap of about 10 points—an indication of “pure” discrimination according to
  • 7. Duru-Bellat and Mingat, 1993.
  • 8. (Le, Jan. 29, 2014).
  • 9. On April 16, 2009, activists from La Barbe pointed out that only 3 of the 14 members of the L’Oreal board of directors were women.
Go further


Charline Zeitoun

Science journalist, author of chilren's literature, and collections director for over 15 years, Charline Zeitoun is currently Sections editor at CNRS Lejournal/News. Her subjects of choice revolve around societal issues, especially when they interesect with other scientific disciplines. She was an editor at Science & Vie Junior and Ciel & Espace, then...


0 comment
To comment on this article,
Log in, join the CNRS News community