modeled using a discrete-time competing risks approach (see McCall, 1996, for example).
To allow for the possible endogeneity of the waiting time until college enrollment, the
unobservable determinants of the waiting duration until enrollment are allowed to be
correlated with the unobservable determinants of the dropout and graduation risks.
In the second model, we explicitly distinguish between those who first enroll in a
two-year institution and those who first enroll in a four-year institution using a
competing risks approach. Thus, we model both durations as competing risks. Moreover,
when analyzing the enrollment duration we estimate separate competing risks models for
those who enter two-year institutions and for those who enter four-year institutions.
Again, to account for the possibility of endogeneity of waiting time until enrollment, the
unobservable determinants of the risks are allowed to be correlated both within and across
the waiting time and college attendance durations.
After a brief (and selective) review of previous research findings in Section 2,
Section 3 describes the empirical methods employed in this paper. The 1979 National
Longitudinal Survey of Youth data are described in Section 4. Section 5 presents the
empirical findings. Even after controlling for potential endogeneity, waiting time until
college enrollment is found to be an important determinant of college completion. For
example, we find that those who delay enrollment into a four-year institution by one year
decrease their probability of graduating with a four year degree by up to 32 percent.