This dissertation consists of three chapters regarding labor economics. The first chapter studies the relative preference men and women have for working with coworkers of the same or opposite sex. The second chapter develops a conceptual framework for estimating the distribution of perceived returns to investments conditional on observed characteristics. The third chapter applies the methods described in the second chapter to estimate perceived returns to college and discusses policy implications.
The first chapter analyzes the effect of occupational gender composition on job-specific labor supply for workers of each gender. I construct a static model of job selection wherein preferences regarding coworker gender composition produce gender-specific compensating differentials. I estimate the model to identify the underlying coworker gender preference parameters. Based on estimated compensating differentials, men's preference is highest for occupations that are 60% female and lowest for female-dominated occupations. Women prefer jobs that are female-dominated, and are least satisfied with jobs that are 25% male all else equal.
The second chapter describes a conceptual framework for inferring agents' perceived returns to college by exploiting the dollar-for-dollar relationship between perceived returns and tuition costs in a binary choice model of college attendance. This approach has four attractive features. First, it provides estimates of perceived returns in terms of compensating variation, which directly inform financial policies that seek to (dis)incentivize the investment. Second, it provides very fine continuously-heterogeneous estimates conditional on a large set of observed characteristics, allowing for differential predictions for how selective, well-publicized policies are likely to affect different types of individuals. Third, because it obtains type-specific perceived returns distributions instead of point elasticities, it provides differential predictions for the effects of type-specific financial interventions depending on the magnitude of the intervention. Finally, the estimates are obtained assuming rational expectations only on prices (one component of returns) rather than on returns as a whole.
The third chapter applies the method described in the second chapter to estimate perceived returns to college using NLSY79 data. Estimating the model using both maximum likelihood and moment inequalities, I find that the scale of the distribution of perceived returns is an order of magnitude lower than past work has found when assuming rational expectations on income returns. The low variance I find in perceived returns implies high responses to financial aid. I predict a 2.6 percentage point increase in college attendance from a $1,000 universal annual tuition subsidy, which is consistent with quasi-experimental estimates of the effects of tuition assistance on college attendance. Adapting the difference-in-difference estimation performed by Dynarski (2003) on the effect of the Social Security Student Benefit to the current setting, I find that the policy increased perceived returns to college by $23,800, compared to an average aid amount of $6,700 per year ($26,800 per four years) (year 2000 dollars). Using the estimated distribution of perceived returns, I perform a counterfactual policy experiment that induces a set percentage of the population to attend college at minimal cost to the government.