Working Paper: NBER ID: w3714
Authors: Joseph G. Altonji
Abstract: The vast literature on human capital and earnings assumes that individuals know in advance that they will complete a particular program of schooling. This paper treats education as a sequential choice that is made under uncertainty. A simple two period structural model is used to explore the effects of ability, high school preparation, preferences for schooling, the borrowing rate, and ex post payoffs to college on the probability of various post secondary college outcomes and the ex ante return to starting college. The model provides the basis for a simple empirical method of accounting for uncertainty about educational outcomes and for nonlinearity in the relationship between years of education and earnings when estimating the expected return to the first year of college. I present estimates of the effects of gender, aptitude, high school curriculum, family background characteristics, and other variables on the expected return to starting college.
Keywords: Human Capital; Education Returns; Labor Market Outcomes; Educational Choices
JEL Codes: I21; J24
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Ability (G53) | Decision to start college (I23) |
| Preferences (D11) | Decision to start college (I23) |
| Gender (J16) | Expected return to starting college (D29) |
| Aptitude (C12) | Expected return to starting college (D29) |
| High school curriculum (A21) | Expected return to starting college (D29) |
| Family background characteristics (J12) | Expected return to starting college (D29) |
| Aptitude (C12) | Educational outcomes (I21) |
| Educational outcomes (I21) | Expected return to starting college (D29) |
| Expected return to starting college (D29) | Decision to start college (I23) |
| Nonlinearity in returns to education (I26) | Differences between expected and realized returns (G17) |