Peer Gender Composition and Choice of College Major

In this paper we analyze whether the gender composition of classmates in high school affects the choice of college-major by shifting it towards those majors preferred by the prevalent gender in the class. We use a novel dataset of 30,000 Italian students graduated from high school between 1985 and 2005 and followed through college and in the labor market. We exploit the fact that the gender composition of the graduating high school class, from one year to the next, within School-Teacher assignment group, shows large variation that we document to be as good as random. We find that male students who attended a high school class with at least 90% of male classmates were significantly more likely to choose "prevalently male" college majors (i.e. Economics, Business and Engineering). However, in the long-run, the higher propensity to enroll in "prevalently male" majors (that are more academically demanding) did not translate in higher probability of graduating in them. In fact, male students from high school classes with >90% males ended up with lower probability of graduating altogether, and they exhibited worse university performance. The peer-pressure towards prevalently male majors may generate mismatches that are counterproductive for college performance and graduation probability of male students. We do not observe these effects on female