Women have made major inroads in labour markets throughout the past century. As a result, there has been a clear convergence in their levels of human capital investment and their employment prospects and outcomes relative to those of men. But while in most rich countries the gender gap in education has closed – and even reversed – there remain considerable gender differences in pay and employment levels, as well as in the types of activities that men and women perform in the labour market.
Labour economists have long tried to understand these differences and three key channels have been identified as potential explanations:
- Labour-market discrimination;
- Gender differences in preferences; and
- Productivity (Altonji and Blank 1999).
However, analysing these hypotheses with traditional tools and data in economics is not straightforward. The study of discrimination, for example, is often complicated by the presence of unmeasured confounding factors, while extracting clean information on people’s psychological traits from naturally-occurring data is often difficult, if not impossible.
By providing data explicitly suited to address the questions of interest and to allow tight control over the environment, the experimental approach provides a valuable source of evidence on these and other gender issues.1
Early economic work on discrimination extensively used the traditional approach of regression analysis on observational data. But increased awareness of the limitations of this approach has gradually shifted the emphasis of empirical work on this topic towards field experiments, such as audit and correspondence studies, which aim to compare outcomes in the same job for two individuals who are identical in all respects other than gender.
These experiments are widely viewed as the most compelling way of testing for discrimination. Audit studies compare interview call-back rates and/or job offer rates on a given job opening for pairs of applicants – one male and one female – with identical resumes. Correspondence studies compare call-back rates for fictitious applications instead of real-life auditors.
While it is not trivial to extrapolate a clear consensus view from experimental research on gender discrimination, the conclusions of this body of work can be broadly summarised as follows:
- There is evidence of significant discrimination against women in high-status or male-dominated jobs as well as discrimination against men in female-dominated jobs.
Compared with the regression approach, however, the experimental approach tends to find far more limited evidence of discrimination against women in the marketplace. Different results from the two approaches may be driven by systematic gaps in unobservables in favour of men, which would clarify the unexplained gap in wages.
Despite recent advances, several aspects of discrimination have yet to be understood. In particular, disentangling the nature of discrimination has proved to be challenging, namely whether employers have a ‘taste for discrimination’, or whether they use gender to extrapolate a signal of unobserved components of productivity (‘statistical’ discrimination). Existing evidence on gender discrimination in the workplace is not suited to parsing its nature and typically expresses no claim to it. Progress in this direction might be made possible by enriching typical field experiment set-ups with tighter researcher control on elements directly linked to the nature of discrimination.
Finally, while research has focused on discrimination in hiring practices, experiments have offered little insight into on-the-job discrimination, such as gender discrimination in earnings, and on how anticipated discrimination in the labour might feed back into individuals’ choices in human capital accumulation or job search.
The traditional economic approach to understanding gender differences in labour-market outcomes has focused on demand-side explanations, such as employer discrimination, as well as on supply-side constraints that are based on educational differences or family responsibility.
More recently, economists have considered alternative supply-side explanations for gender differences in outcomes (Croson and Gneezy 2009, Bertrand 2011). For example, potential gender differences in psychological attributes – including preferences for risk and competition, as well as concerns about other people – might offer insights into gender gaps. Experiments offer a useful methodology for studying behaviour and strategic interaction in a controlled environment – and they can be adapted to elicit gender differences in preferences in spheres potentially associated with labour-market success.
Occupational and industry segregation of men and women is one of the leading components of gender gaps in earnings – and these have been widely documented. As jobs in different sectors offer different arrays of job security, earnings stability and working conditions, systematic gender differences in preferences for risk and competition have the potential to shape gaps in earnings through job sorting behaviour.
Job-related risks (such as job-loss or earning volatility) are typically rewarded by higher mean earnings, and high-risk sectors tend to be male-dominated. If women are more risk-averse than men, they end up being overrepresented in jobs with lower mean and variance salaries.
- Lab experiments have often found that men are more tolerant of risk than women. While some attribute these differences to the emotional reaction to uncertain situations, others report that they are related to confidence.
High-profile careers typically develop in highly competitive settings, in which rewards are linked to relative, rather than absolute, performance.
- Lab experiments have detected significant gender differences in attitudes towards competition.
For example, men thrive in competitive environments and they have a greater tendency than women to self-select into these environments.
- Another hypothesis for why earnings of men and women differ, even on identical jobs, is that men and women bargain salary differently.
In the psychology literature, it has been suggested that women earn less than their male counterparts because they avoid competitive negotiation. The experimental results on gender differences in negotiation and social preferences are somewhat mixed and depend strongly on context. For example, women’s performance is more strongly affected by the gender of whom they work with or whom they compete against than men’s performance.
Finally, there has been a great deal of interest in whether men and women exhibit different degrees of social preferences and, in particular, whether other-regarding attitudes such as altruism, fairness or envy may play a stronger role in female than in male decisions.
- Gender differences in social preferences might help us understand why men and women select into certain sectors or jobs, for example, women tend to be overrepresented in the social sector.
Moreover, these differences may have an impact on gender outcomes in a given job, as individuals may take into account their social preferences when negotiating wages and collaborating with coworkers. Various lab studies based on simple sharing games conclude that women are more sensitive than men to equity considerations, but find no systematic evidence of gender differences in trust or in the willingness to contribute to public goods.
While evidence from various experimental settings suggests that women and men may differ in traits that are potentially related to labour-market success, the causes – nature, nurture, or the interaction of the two – and the economic consequences of such differences are not entirely understood. On the nature side, obvious differences between men and women related to, for example, childbirth and physical strength may have an impact on how well men and women fare in the labour market. On the nurture side, preferences may have roots in the education system, the household, or society at large. Understanding nature and nurture components of gender differences in behaviour is a key issue, with clear policy implications.
An important next step is to understand how findings from the lab on psychological attributes and preferences would map onto the labour market and whether there is scope for policy. While one may conjecture that disparities observed in the lab have implications for labour-market outcomes, more direct evidence from the workplace is needed to draw useful conclusions for gender gaps in real markets. Moreover, from a policy perspective, the prescriptions will differ depending on how traits are formed and how important they are in influencing outcomes.
A natural progression from the study of individual preferences has been to understand their role in group settings. If different psychological traits lead men and women to make different choices in similar contexts, the gender composition of teams becomes a relevant factor in collective decision-making. Higher female participation in the labour market has implied changing workplace demographics and more gender-diverse teams. In high-profile professions, such as politics or the corporate sector, these changes have been eased by the introduction of explicit gender quotas in a number of countries.
Despite a large body of lab evidence on individual preferences, experimental studies of gender and preferences at the team level is relatively scarce. One of the main problems with studying gender and groups is that groups are typically formed endogenously. Recent reforms that mandate certain levels of female representation on boards of directors offer a valuable, quasi-experimental setting to study the gender composition of teams and performance.
The first country to implement gender quota laws was Norway in 2003, followed by Spain, Finland, Iceland, and France. One study of the impact of female presence on boards on firm performance exploits the Norwegian reform, which requires listed companies to achieve 40% female board representation within two years. The research finds important effects of female board representation, notably that the constraints imposed by the quota implied a decline in stock prices and operating profits.
While quota reforms and other field and lab experiments offer valuable insights into the consequences of gender diversity, research on this issue is still very limited, not least because it is restricted to a small and selected group of women. Quota policies, as well as business games among MBA students, focus on women who may not be fully representative of the female workforce. The representation of women in decision-making at lower levels of responsibility can thus help to form a broader picture of the impact of gender diversity, and attenuate the stark selection of women at the top.
Experiments offer a novel and useful methodology that is being used widely in almost all areas of economics. In gender economics, the experimental approach offers a way to answer questions previously believed to be unanswerable because of data limitations, as well as new techniques to identify mechanisms and results in older topics traditionally studied by labour economists.
Despite recent advances, however, several important aspects of gender differences in labour-market success have to date been only partially explored experimentally. There is clear scope for further research in several directions concerning the nature of gender discrimination, the labour-market consequences of gender differences in preferences established in the lab, and the sources of such differences.
Altonji, Joseph and Rebecca Blank (1999), “Race and Gender in the Labor Market”, in O Ashenfelter and D Card (eds.), Handbook of Labor Economics, vol. 3C: 3143-3259. Amsterdam: Elsevier.
Bertrand, Marianne (2011), “New Perspectives on Gender”, in O Ashenfelter and D Card (eds.) Handbook of Labor Economics, vol. 4B: 1545-1592. Amsterdam: Elsevier.
Charness, Gary and Peter Kuhn (2011), “Lab and Labor: What Can Labor Economists Learn from the Lab?”, in O Ashenfelter and D Card (eds.) Handbook of Labor Economics, vol. 4A: 229-330.
Croson, Rachel, and Nancy R. Buchan (1999), “Gender and Culture: International Experimental Evidence from Trust Games.” The American Economic Review 89(2): 386–91.
List, John A and Imran Rasul (2011), “Field Experiments in Labor Economics”, in O Ashenfelter and D Card (eds.) Handbook of Labor Economics, vol. 4A: 104-228.
1 The growing usage of field and lab experiments in labour economics is discussed by List and Rasul (2011) and Charness and Kuhn (2011), respectively.