The Local Average Treatment Effect (LATE) interpretation of the IV estimates of the returns to schooling is becoming increasingly popular. Typically, researchers reporting LATE estimates do not provide systematic evidence that there is substantial heterogeneity across different ability levels in returns, and without such evidence, the LATE interpretation is short of being compelling. The recent abolition of Grade 13 in Ontario's secondary school system provides a unique opportunity to measure the benefits of an extra year of high school for high-ability students (those bound for college), rather than dropouts. I present a simple factor model which allows the value-added of Grade 13 (in terms of achievement) to be estimated, generalizing the standard difference-in-differences estimator to correct for heterogeneity in ability measurement across college subjects. The main finding is that the estimated return to an extra year of high school in terms of human capital is small for these high-ability students: students coming out of Grade 13 have a 2.2 point advantage (on a 100 point scale) over students from Grade 12, the estimated return to Grade 13 being around 2 percent. This evidence indicates that there is substantial heterogeneity in the return to an additional year of high school in the direction assumed in the prior literature.
Return to schooling factor model difference-in-differences