One striking phenomenon in the U.S. labor market is the reversal of the gender gap in college attainment. In 1980, 57% of young men aged 25-34, compared with 46% of young women, had some college education by age 34. By 1996, however, female college attainment had reached 64%, 5 percentage points higher than that of the males in the same cohort. In tact, females overtook males in college attainment in 1987 and have led ever since.
A large body of empirical research emphasizes the role of the earnings premium as a key explanatory variable for the determination of education outcomes (see, e.g., Becket 1967; Mincer 1974; Willis and Rosen 1979). In addition, an extensive literature shows that family background is an important determinant of the schooling decision (see, among others, Cameron and Heckman 1998, 2001; Eckstein and Wolpin 1999; Ge 2008; Kane 1994). Recently several papers have argued empirically and theoretically that expected marriage is important in determining the schooling decision (e.g., Chiappori, Iyigun, and Weiss 2006; Ge 2008; Iyigun and Walsh 2007).
On the basis of this literature, we construct a life-cycle model that includes potential costs and benefits from the labor market and marriage market which determine individual college decisions. In our model, individuals with differing learning abilities first decide whether or not to enter college. Then they might get married and have children. Parents are altruistic and value their children's learning ability, which increases with the parents' education. Forward-looking individuals take into account the impact of their own schooling on their children's learning ability. Other factors that affect an individual's decision on whether to pursue higher education include the expected direct labor market returns to college over one's lifetime, the expected marriage market returns to college, and the financial and effort costs of attending college. These costs and benefits can differ by gender.
We calculate from Panel Study of Income Dynamics (PSID) and Current Population Survey (CPS) parents' education distributions; the life-cycle profiles of single, marriage, and divorce probabilities by education; and the lifecycle profiles of earnings by education and marital status as exogenous inputs of the model. We observe that the number of college-educated parents increases over time. In the marriage market, a substantial increase of single probabilities and an increase of divorce probabilities has occurred for both genders, regardless of college attainment status. Lifetime earnings by cohort are decreasing slowly for males of all marital statuses, especially for married males. Lifetime earnings for married and divorced females are increasing gradually, whereas those for single females are decreasing slightly. To formally endogenize those changes is beyond the scope of our paper. We instead focus on the mechanism in which, under perfect foresight, these changes affect education decisions.
We estimate the parameters of the model by matching data on college attainment by gender from the PSID. We present evidence on how well the model fits the data. We then use the parameter estimates to simulate counterfactual experiments, which break down the sources of changes in college attainment into the effects of changes in relative earnings, changes in parental education, and changes in the marriage market.
What accounts for the increase in college attainment over the past few decades? We find that the increasing gap in earnings between college and high school graduates has important effects on the increase in college attainment for both genders. When earnings are fixed at 1946 cohort levels, attainment rates in 1996 drop by 15.5 and 14.2 percentage points for males and females, respectively. We also emphasize the importance of intergenerational persistence in schooling on the increase in college attainment for both genders. …