Response to Competition: Gender, Domain, STEM choice, and Persistence
Abstract: Women’s lower performance in competitive environments has been advanced as an explanation for gender inequality in the labor market. We explore the domain specificity of gender differences in response to competition (RC) and test whether it can predict a subsequent decision to study STEM, an important contributor to the gender pay gap. Using administrative data from the universe of four cohorts of compulsory education students, and defining RC as the performance difference on the High School Entrance Exam versus a mock exam, we find a female disadvantage in RC in STEM subjects, but a female advantage in non-STEM subjects. Principal component analysis reveals the presence of both a domain-general component (favoring boys) and a domain-specific component (favoring girls) of RC. Both components are significant predictors of subsequent STEM track choice, after controlling for STEM ability, comparative advantage in STEM, peer or role model effects, age, and SES. Only the domain-specific component contributes to the gender gap in STEM choice, accounting for 19.4% of the remaining gap. Using RC measured three years later, we find evidence of intertemporal stability and persistence in domain-specific gender gaps.