Occupational gender composition is related to occupational wages: Causal evidence from a survey experiment investigating occupational devaluation

Occupations with a higher percent of women pay less than occupations with a lower percent of women. One explanation for this difference is the devaluation hypothesis: the idea that occupations with more women are devalued because of the relatively high percentage of women working in the occupations. Decades of research, using observational data from national surveys, find support for the hypothesis. A limitation of this previous research is the potential for omitted variable bias. That is, observational data cannot rule out the possibility that there might be a third factor which causes both a higher percent of women in occupations and lower pay for occupations. In this paper, we employ an experimental design, using nationally representative survey data, to address this limitation. We use random assignment to condition because it can be used to specify a causal mechanism—i.e., to investigate whether percent women in an occupation can cause lower wages. We find that when presented with an occupation with a high percent of women, respondents suggest lower pay for that occupation, as compared to when respondents are presented with a mixed occupation or an occupation with a high percent of men. We also test whether random assignment to gender composition of occupation can affect perceptions of workers’ characteristics in that occupation. We find little evidence of such an effect.