Research

This paper examines how differences in risk aversion influence occupational choices and the distribution of earnings. I first show that more risk-averse workers earn significantly less, and that this earnings gap widens over time. A key driver of this pattern is that risk-averse individuals tend to select occupations with more stable, but lower, earnings and slower earnings growth. To quantify the importance of this channel and distinguish it from sorting based on unobserved traits, I develop and estimate a structural model of occupation choice that accounts for heterogeneity in risk aversion and human capital accumulation. In the model, risk aversion is correlated with both observed and unobserved initial skills, and it influences skill accumulation through occupational choices. Using the estimated model on the non-college sample, I perform a decomposition analysis showing that 30 percent of the earnings gap between the most and least risk-averse workers (14 log points) can be explained by occupation choices. Of this, approximately 55 percent is due to lower pay in safer occupations, while the remaining 45 percent is attributable to slower human capital accumulation. In a counterfactual analysis, I find that social insurance, implemented as redistributive taxation, reduces the earnings gap between risk-tolerant and risk-averse workers by around 16 percent (2.2 log points), partly by enabling risk-averse workers to shift to relatively higher-return occupations. Similar patterns are observed in the college sample.

We estimate major-specific earnings profiles using matched American Community Survey (ACS) and Longitudinal Employer-Household Dynamics (LEHD) data. Our paper improves upon recent evidence provided by Deming and Noray (2020) about the relative age-earnings profiles of majors. We do this by leveraging a long panel of worker earnings that allow us to circumvent econometric challenges to estimating lifecycle earnings profiles using cross-sectional analyses. We find that engineering and computer science majors have similar or faster earnings growth relative to other majors, a category encompassing humanities, education, psychology, and similar fields. This is in contrast with Deming and Noray (2020), who estimate age-earning profiles using a cross-cohort approach and find that earnings for engineering and computer science majors decline rapidly over the lifecycle with respect to other majors.

We study how quantity-based labor regulations affect technology adoption using California’s nurse staffing mandate. Combining nationwide hospital data with a synthetic difference-in-differences design, we show that the mandate reduced adoption of Clinical Decision Support systems by up to 6.4 percentage points among hospitals with low baseline staffing conditions or tight financial constraints. Conversely, adoption of nurse staffing software increased by up to 9 percentage points among well-staffed, financially flexible hospitals. These patterns reflect the crowding out of quality-enhancing technologies and increased demand for compliance tools. Overall, staffing mandates shift the technology investment composition, with implications for mortality and clinical performance.