The marginal and average returns to schooling

The existing literature now features many examples where log wages are linear in years of schooling and which effectively attempt to correct for least squares bias using instruments based essentially on a single variable. Two recent developments, taken together, cast some doubt on the downward bias in least squares estimates of the return to schooling that have been an important feature of the recent literature. The first is the realization that instrumental variable (IV) only estimate the effects of some treatment if the effect is the same for everyone. The second is that IV may only estimate the effect of the treatment on the individuals whose choices are affected by the instrument in question [extract]

A hard copy is available in UCD Library at GEN 330.08 IR/UNI