Mismatch or Convergence? Affirmative Action and Students' Selection in an Elite French University

Abstract

I study an affirmative action program in an elite French university. This program is original because it focuses on social background rather than race or ethnic origin. I examine the ability of the affirmative action beneficiaries to catch up, using the presence of a common core in first year, and random allocation to tutorials to avoid bias from self-selection into courses. I find that some convergence occurs between entrance and graduation, but that there is still an important gap at graduation and that affirmative action students are more likely to fail to graduate. However, there is no evidence of sorting across majors based on difficulty, nor of mismatch, as has been observed in many American universities. It seems that the university does not use all the information available on the applicants to the affirmative action program, some pre-entrance characteristics being strong predictors of failure. These findings have implications for the design of effective affirmative action programs.