University programs differ in their gender earnings gaps, the difference between the subsequent earnings of the program’s male and female enrollees. A program could have a positive gender earnings gap because the program attracts higher-ability men than women (a selection effect), or because the program increases the earnings of male enrollees more than female enrollees (a causal effect). To understand the source of cross-program differences in gender earnings gaps, we exploit a discontinuity built into the Danish national university admissions system, which provides a quasi-random assignment of similar applicants to different programs. Enrolling in a program with a $1 larger gender earnings gap, holding average earnings constant, does not affect male earnings but reduces female earnings by $0.45. This effect is small as women enter the labor market but increase over time, and cannot be explained by channels related to marriage or childbirth. Our results show that programs that appear worse for women– in the sense of having economically significant gender earnings gaps – are worse for women because they reduce female earnings more than programs with smaller gaps.
