Despite gender convergence in many areas of the labor market, men and women still work in markedly different occupations. In this paper, I aim to understand whether preferences over the gender composition of one’s occupation can explain continuing occupational gender segregation. To assess the willingness to pay for gender composition of a job, I run a survey in which respondents choose between multiple hypothetical jobs characterized by pay, tasks, and demographics. I then pair the survey results with a structural model to assess the welfare effects of policies that could re-allocate workers across occupations by gender.
Intervention
The main “intervention” is a survey-based conjoint experiment, where respondents choose between multiple pairs of job offers, with the wages and demographic characteristics randomized. Thus the random intervention is in a sense the wage and demographics that a respondent sees for each job pair.
Experimental design
The main trial is a conjoint experiment varying wages and gender compositions of job offers.
Design details
Randomization method
Randomization will be done by computer through Qualtrics.
Randomization unit
Randomization is done at the individual respondent level.
Clusters
2000 individuals
Observations
2000 individuals
Arms
no treatment arms per se
Primary outcomes
The primary outcome of interest is the profile of average willingness to pay for different gender compositions, estimated separately for men and women.
Secondary outcomes
I will consider several secondary outcomes of interest, listed below:
• willingness to pay for age and education composition
• heterogeneous preferences within gender, estimated using a latent-class logit model
• heterogeneity in gender willingness to pay by individual covariates, including correlating these
with latent classes as well as estimating single-type logit separately by group, with covariates
including
– true female share of occupation, including both reported female share of firm and cowork-
ers with same job at firm, and female share of reported occupation calculated using CPS
and Census/ACS data
– expectations of jobs with different female shares, including satisfaction with coworker
interactions, task content, schedule, work environment, earnings, promotion, and family
preference
– reported gender attitudes from questions on whether women should stay at home and
men work, whether men or women have it easier in the US these days, and stance on
affirmative action for women in the workplace
– demographics including age, education, income, number of children, race, marital status,
employment status, and industry
• occupation choice given choice between two real occupations with and without gender share
shown (secondary experiment, used here only for robustness)