Association for Public Policy Analysis & Management

Many students from low socioeconomic backgrounds do not earn their bachelor’s degree until two or three years later than the expected degree completion date (Chetty et al., 2017). Yet voucher impacts on college attainment have thus far been estimated experimentally before program participants have reached this point in time (Foreman, 2017). We provide the first experimental estimates of impacts on college graduation rates of an offer of a voucher to families from low-socioeconomic backgrounds that adequately allows for delayed graduation rates. We link data from a privately sponsored, lottery-based voucher initiative in New York City to degree attainment information maintained by the National Student Clearinghouse. A match is made for 99 percent of the sample of about 2,500 students.

We first compare postsecondary attainment outcomes measured five and eight years after high school completion. Consistent with delays in postsecondary attainment among students from low-socioeconomic backgrounds, we find postsecondary enrollment and degree completion rates each increase by five percentage points throughout this time period. These changes are substantively large, especially for the latter outcome. Enrollment rates increase from 47 to 52 percent. Degree completion rates increase from 10 to 15 percent. These increases, however, do not vary by treatment condition. That is, students who were offered a voucher via the lotteries experienced small, statistically insignificant increases in postsecondary enrollment and degree completion relative to students in the control group.

Yet comparisons based on the full sample mask important effect heterogeneity. We find evidence of larger, positive, and statistically significant impacts for children from racial minority backgrounds and non-immigrant families. Effect sizes range from 3 to 5 percentage points based upon intent-to-treat estimates.

Given the combined role that different socioeconomic factors play in educational attainment, we conduct additional analyses of effect heterogeneity that consider interactions between multiple indicators of socioeconomic status (Park & Hossler, 2014). We find that among participants from racial minority backgrounds and non-immigrant families, program effects are most pronounced for children who are relatively more advantaged. Specifically, children with parents who have at least attended college and are relatively wealthier benefited most from the program. Effect sizes for these subgroups range from 8 to 10 percentage points. It is important to note, however, that the voucher program is already targeted to families from the lowest socioeconomic backgrounds. Nearly half of participants have parents who did not attend college and 90 percent of participants had annual household incomes that were less than $30,000. Thus, with postsecondary enrollment and degree attainment rates for control group students of 50 and 15 percent, respectively, the voucher program effectively increased college enrollment rates by one-fifth and raised degree attainment rates by two-thirds for students at the lower end of the socioeconomic distribution but not for those at the very bottom.