Earnings, Schooling, and Ability Revisited

This paper presents a survey and interpretation of recent research on the return to education. The empirical findings in a series of current papers suggest that the causal effect of education on earnings is understated by standard estimation methods. Using a simple model of optimal schooling developed by Gary Becker (1967), I derive an explicit formula for the conventional estimate of the return to schooling and for alternative instrumental variables and fixed- effects estimators. The analysis suggests that instrumental variables estimates based on 'interventions' that affect the schooling choices of children from relatively disadvantaged family backgrounds will tend to exceed the corresponding OLS estimates.