A long-standing issue in the literature on education is whether marginal returns to education fall as education rises. If the population differs in its rate of return, a closely related question is whether marginal returns to higher education fall as a greater fraction of the population enrolls. This paper proposes a nonparametric method of estimating marginal treatment effects in heterogeneous populations, and applies it to this question, examining returns to higher education in the UK. The results indicate that marginal returns to higher education fall as the proportion of the population with higher education rises, consistent with the Becker Woytinsky Lecture hypothesis.