Heterogeneity in schooling, uncertainty, and the return to education.

• Author: Altonji, Joseph G.

There is an enormous empirical literature on human capital and earnings that has grown out of work by Jacob Mincer, Gary Becker, and other pioneers. Much of the research focuses on measuring the value of a year spent in school.(1) But in recent years, the research has begun to explore more fully the implications of the fact that schooling is heterogenous, both in quality and in subject matter, and that people make decisions about schooling without complete knowledge of their tastes and talents for different types of work. This article briefly summarizes three avenues that I have been pursuing: First, what are the implications of the fact that education is a sequential choice made under uncertainty? Second, can we get inside the black box of years-spent-in-school by examining the effects of actual courses taken on the payoff to school? And third, I use a new methodology to revisit an old and controversial question - do school inputs affect outcomes?

The Demand for and Return to Education When Education Outcomes Are Uncertain

The standard human capital model of schooling assumes that individuals are able to choose a future level of education with no uncertainty about actually completing that level. It also focuses on the number of years of school chosen, rather than on the specific major. In contrast, I model the demand for education as a sequence of choices made under uncertainty. I then examine how variables that influence tastes for school, ability to do college work, and the payoffs to particular college programs affect the expected return to a year of school.(2)

Three facts suggest that uncertainty is important, and they motivate this research. First, many individuals who plan to complete college instead drop out. Second, for some demographic groups the ex post returns to education are associated largely with completing high school or completing college. Such nonlinearities in the ex post returns, or a gap between the effect of education on wages and the borrowing rate, may produce substantial differences between ex ante and ex post returns to the first year or two of college. The return to the first year of college is not the earnings differential between individuals with 12 and 13 years of schooling. Rather, the return is the difference between the earnings of the person who stops at 12 years of education and the expected earnings net-of-education-costs of a person who attends the first year of college. Expected earnings depend on the earnings associated with 14, 15, 16, or more years of education. Expected earnings also depend on the probabilities of the various education outcomes for an individual who has completed year 13. Family background, aptitude, and other factors that affect the odds that an individual actually will...