Friday, October 5, 2012: 4:35 PM
To test for discrimination against women and immigrants in recruitment, we exploit a natural experiment in the hiring process of higher management staff in the French public sector, which is based on anonymous and non-anonymous nationwide entrance exams. Subsequently, we also observe exam results and performance evaluations after the newly hired employees have passed a one-year trainee program.
The job applicants first have to pass an anonymous written exam and, if successful, an oral exam, which is not anonymous. The written test scores must by regulation not be known to the oral examiners, but they are known to us. The hiring contest exams turn out to be procedurally fair, that is, we do not find systematically worse oral exam results for women or immigrants once we hold the written test scores constant. Immigrants might even be positively discriminated.
However, at the end of the trainee program, that is the first year of employment, women outperform men, both conditionally and unconditionally on the previous entrance exam results. Hence, the entrance exams have a disparate impact on women and men. They put more emphasis on skills that men have a comparative advantage in but that are less relevant for the job. We conclude that although there is procedural fairness in the hiring process, there is functional discrimination against women through the design of the hiring contest. We find no such disadvantage for immigrants.
The job applicants first have to pass an anonymous written exam and, if successful, an oral exam, which is not anonymous. The written test scores must by regulation not be known to the oral examiners, but they are known to us. The hiring contest exams turn out to be procedurally fair, that is, we do not find systematically worse oral exam results for women or immigrants once we hold the written test scores constant. Immigrants might even be positively discriminated.
However, at the end of the trainee program, that is the first year of employment, women outperform men, both conditionally and unconditionally on the previous entrance exam results. Hence, the entrance exams have a disparate impact on women and men. They put more emphasis on skills that men have a comparative advantage in but that are less relevant for the job. We conclude that although there is procedural fairness in the hiring process, there is functional discrimination against women through the design of the hiring contest. We find no such disadvantage for immigrants.