Disaggregating rates of return to education

Education is effectively used as a discontinuous variable in studies estimating the rates of return by level of education. We find that the normal procedure used for estimating the rate of return to broad aggregates such as secondary and high understates the returns to these levels and subsequently suggest a procedure for rectifying this bias. We also find, as expected, that broad aggregations that ignore sub-levels of education result in a loss of important insights about the interaction of education and the labor market.