An essential feature of schooling is not only that it occurs in a different site than most on-the-job training but also that it is more intensive. That is, a smaller proportion of gross potential earnings is sacrificed in on-the-job training than in schooling. In estimating human capital earnings functions it has generally been assumed that during schooling 100% of gross potential earnings are invested in all years, while in on-the-job training this percentage is smaller and is a declining function of age. This assumption has been quite useful since it allows the identification of an estimate of the rate of return on schooling from a regression of earnings on years of schooling. This paper argues that the percentage of gross earnings invested may fall below 100% well before schooling is ended, that this percentage is likely to be correlated with years of schooling, and thus this procedure yields only a biased estimate of the rate of return to schooling.
