Glass ceiling in research: evidence from a national program in Uruguay

We present evidence that female researchers have 7.1 percentage points lower probability of being accepted into the largest national research support program in Uruguay compared to males. They also have lower research productivity than their male counterparts. Differences in observable characteristics explain 4.9 of the 7.1 percentage points gap. We show that the gender gap is wider at the higher ranks of the program consistent with the existence of a glass ceiling. The results are robust to issues of bidirectionality (impact of research productivity on the probability of accessing the program and impact of the program on research productivity), joint determination and correlation of variables (e.g. having a Ph.D., publishing and tutoring) and initial productivity effects (positive results at early stages may have long term effects on career development). We present three hypotheses for the gender gap (an original sin in the organization of the system, biases in the composition of evaluation committees, and differences in field of concentration) and find some evidence for each. Glass ceilings are stronger in the fields where females are overrepresented among the applicants to the system: Medical sciences, Natural sciences and Humanities. Finally, we present a counterfactual distribution of the program in the absence of discriminatory treatment of women and discuss the economic costs of the gender gap.