Human capital risk and portfolio choices: Evidence from university admission discontinuities

Theory suggests that increasing idiosyncratic, uninsurable labor income risk may cause individuals to reduce the risk in their financial assets. This relationship is confounded empirically by the tendency of risk tolerant people to choose riskier careers and hold riskier portfolios, leading to an upward-biased estimate of the effect of earnings risk on risky assets holdings. We overcome this identification problem by exploiting a discontinuity built into the Danish national university admissions system, which provides quasi-random assignment of similar applicants to programs with different earnings volatility profiles. Our methodology allows us to measure the causal impact of enrolling in a high-volatility program, holding fixed the average program earnings and human capital betas. We show that entering a program whose enrollees subsequently experience volatile earnings causes students to have more volatile earnings and, ceteris paribus, to hold fewer risky assets and be less likely to participate in the stock market. We calibrate our empirical results to a portfolio choice model with risky labor income that fits our empirical findings well with modest participation costs, myopic behavior, and reasonable levels of risk aversion.