Increasing Returns to Education and Progress towards a College Degree

Returns to college have increased, but graduation rates have changed relatively little. Modifying a human capital model of college enrollment to endogenize time-to-graduation, we predict that higher returns to education will both speed graduation and increase enrollment. Some of those new entrants may, however, take longer to graduate. Using the 1989 and 1995 Beginning Postsecondary Studies, we employ a multinomial logit to model the association between individual and family characteristics, and five-year college outcomes: graduation, continued enrollment, and non-enrollment. Between cohort differences arise either because the characteristics of those entering college are different or because the relations between characteristics and outcomes have changed. We utilize a Oaxaca-Blinder style decomposition to distinguish between these two alternatives, attributing differences in characteristics to newly attracted students and differences in the relations between characteristics and outcomes to historically attracted students behaving differently. It is changes in behavior that explain the increased progress we observe.