Self-selection bias in estimated wage premiums for earnings risk

This note develops a simple occupational choice model to examine three types of selection biases that may occur in empirically estimating the premium for uncertain wages. Individuals may select themselves into risky (wage-uncertain) jobs because they have (1) lower risk aversion, or (2) lower income risks, or (3) higher individual ability. We show that (1) gives no bias, (2) biases the OLS estimate of the risk-premium in a wage regression upward, and (3) yields a bias that analytically may be positive or negative, but empirically is more likely to be negative if our occupational choice model is correct.