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.