Are successive investments in education equally worthwhile? Endogenous schooling decisions and non-linearities in the earnings–schooling relationship

Most of the studies that account for the endogeneity bias when estimating the returns to schooling assume that the relationship between education and earnings is linear. Studies that assume the latter relationship to be non-linear simply ignore the endogeneity bias. Moreover, they either assume an ad-hoc non-linear relationship or argue that non-linearities are due to sheepskin effects. The approach I adopt in this paper allows estimation of the returns to years of schooling without assuming any explicit form of non-linearity while accounting for the endogeneity of education. The results suggest (i) endogeneity is indeed a crucial issue, (ii) there are sharp non-linearities which do not seem to be only due to certification effects and (iii) it is important to account for the impact of education on the returns to other observable characteristics such as age and seniority.