Abstract: We analyze whether financing higher education through student loans or grants affects the college major choices of students in Chile, where either type of financing is allocated based on a standardized test. Students who are marginally eligible for grants are more likely to enrol in high-paying fields such as STEM. Complementing the reduced form result, we estimate a discrete choice model on data for narrowly defined higher education programs. The results indicate that, holding other program characteristics constant, grant recipients place a higher value on fields with high earnings growth potential, while being less concerned about a lower graduation probability.

Abstract: This paper provides causal microevidence on how idiosyncratic cost-of-living shocks are transmitted to labor market outcomes. Using the case of energy price changes, I combine 20 years of German employer-employee data with a representative household expenditure survey and rely on regional differences in energy expenditure patterns to identify the pass-through of energy cost shocks into earnings. A one-standard-deviation shock raises annual earnings growth by 0.4 percentage points, offsetting 30-35% of local cost increases within a year and compensating for 80% over five years. Job-to-job mobility is a key mechanism: higher exposure to energy price changes both increases the likelihood of switching and makes switches more profitable. Through the lens of an equilibrium model of the labor market, I clarify that while labor market responses substantially mitigate the distributional impact of relative price changes, they impose additional non-pecuniary costs on workers.

Abstract: We embed an experiment in a large-scale representative survey to investigate how relative wealth affects risk-taking and how this effect varies as a function of perceived control over life outcomes. Our results contest the common prediction of higher risk-taking in the middle of the distribution. Instead, we find that respondents who are induced to perceive their relative wealth as low display more tolerance towards risk in a subsequent incentivized risk-taking task. This effect is not uniform but is mainly driven by individuals who more firmly believe that life outcomes are beyond their control.

Abstract: Using data from the Health and Retirement Study (HRS), we examine the effect of automation on older U.S. workers. On average, a one-standard-deviation increase in robot adoption reduces the probability of working by 7 percentage points and increases both early retirement and the uptake of social security and disability payments. A key contribution of our paper is to show empirically that the average effect masks strong treatment effect heterogeneity. Among high-ability workers, automation has little impact, whereas it significantly reduces employment rates among lower-ability workers. We exploit the availability of genetic information in the HRS to proxy ability types based on a person's genetic predisposition for educational success. In line with models of endogenous skill investments, we show that genetic predictors of education are important moderators of automation even conditional on educational attainment.