Evidence on omitted variable bias in earnings equations (original) (raw)

In this paper we propose that family background variables are significant in earnings equations because they measure investments in children made by families in the home, over and above formal schooling investments and schooling quality. Together these variables account for a significant proportion of the difference between estimated rates of return to schooling across racial groups. Using data from the NLSY, we observe a convergence of rates of return across racial groups after accounting for differences in these variables. The estimated equations are used to predict that average minority earnings would be almost identical to white earnings if minorities experienced the same family background and school quality as whites. (JEL J13, J24, J31) 1. O'Neill (1990) provides a recent evaluation of the earnings differential between black and white males, but does not consider other racial groups.