DISCRIMINATION IN THE LABOUR MARKET: The curse of competition between workers from different groups

Discrimination by employers against particular groups of workers, whether based on gender, ethnicity or some other characteristic, becomes more prevalent when groups are competing against each other for the same jobs. That is the central message of research by Thomas de Haan, Theo Offerman and Randolph Sloof, published in the August 2017 issue of the Economic Journal.

Using a framework for analysing labour market discrimination that lets workers from different groups compete for the same job, the study shows theoretically that a ''no-discrimination equilibrium'' becomes unstable. Testing that analysis with laboratory experiments indicates that labour market participants coordinate on a no-discrimination equilibrium if there is no competition. In contrast, persistent discrimination emerges once competition for a job is introduced.

These results are in line with empirical data on discrimination in the labour market. The OECD countries with the highest overall unemployment rates – Spain, Greece and Italy – are also those where the gap between female and male unemployment rates is largest (11.91%, 10.36% and 7.04%, respectively). Between 1960 and 2000, the path of unemployment rates within each of these countries also reveals that unequal treatment of men and women becomes larger as the overall unemployment rate rises.

The new study suggests potential caution about the effectiveness of affirmative action programmes. In the presence of unemployment, once the target of equal treatment is reached and the affirmative action programme is abandoned, one would predict on the basis of this theory and experiment that discrimination might soon flourish once groups again start competing for the same jobs.


Labour market discrimination is a complex issue. This study focuses on one potential driving mechanism: ''statistical discrimination'', a concept originally championed by Kenneth Arrow.

Theoretically, statistical discrimination may thrive if potential workers can only partly convince an employer of their individual quality. If their educational achievements are only a noisy signal of their true competence, then there is room for employers to include prior beliefs about the productivity of workers from certain groups into the decision to hire a worker.

This situation turns into one of (persistent) statistical discrimination if negative prior beliefs from employers regarding a particular group discourage people belonging to this group from investing in human capital skills, thereby confirming the beliefs of the employers.

To put this theory of statistical discrimination to the test, economists have run experiments in a laboratory setting, simulating a mini labour market with participants making choices based on incentives in the role of workers or employers.

Without competition for the same job, earlier experiments did not succeed in systematically generating statistical discrimination; typically, no persistent differences in hiring rates evolved between groups who were assigned an arbitrary identity (workers were randomly assigned to an ex-ante equal ''purple'' or ''green'' group).

So what are key elements in a labour market that could enable ''statistical discrimination'' to emerge when groups of workers are ex-ante equally skilled? The new study identifies competition between workers from different groups for a vacancy as a crucial element for statistical discrimination to emerge.

In a laboratory experiment with incentives, the researchers again find no evidence of worker group discrimination in a setting without worker competition. Yet they do find clear and persistent discrimination in a setting where workers from the purple group compete with workers from the green group for a vacancy.

The main intuition for the findings follows from a theoretical stability result. In case of worker competition, small differences in the perceptions of employers about worker productivity between groups can lead to sizeable hiring differences. This sparks a ''drift'' towards an unequal labour market outcome.

In contrast, without competition, small fluctuations in employer perceptions of worker competence from a certain group will barely affect hiring decisions.

''Discrimination in the Labour Market: The Curse of Competition between Workers'' by Thomas de Haan, Theo Offerman and Randolph Sloof is published in the August 2017 issue of the Economic Journal. Thomas de Haan is at the Norwegian School of Economics. Theo Offerman and Randolph Sloof are at the University of Amsterdam.