Abstract
We study whether gender enters beliefs about the economic returns associated with occupations. Utilising the fact that the German language distinguishes female and male forms of all occupational titles (like waiter and waitress in English), adults and adolescents in Germany were randomly assigned to estimate the earnings of five occupations presented in either masculine or feminine grammatical form. The masculine form raises perceived pay by 11.3% among adults and 7.9% among adolescents. Among adults the effect is similar for male and female respondents; among adolescents it is approximately zero for boys and 16.1% for girls. The size of the treatment effect also varies sharply across occupations. We benchmark this heterogeneity against occupation-specific male-female pay differences from administrative earnings data and a DAX CEO compensation study. Actual gender wage gaps can only partially account for the occupation-specific treatment effects: occupations with nearly identical observed gender gaps generate radically different experimental responses. The results imply that gender is part of the prior people attach to the economic value of an occupation rather than merely a reflection of known wage differences. They also have a measurement implication: the linguistic form used to elicit labour-market beliefs can materially change the beliefs researchers observe.