Skip to main content
Open Access Publications from the University of California

Essays on the Economics of College Access and Completion

  • Author(s): Lapid, Patrick Andrew
  • Advisor(s): Rothstein, Jesse
  • et al.

This thesis examines two uncommon topics in the economic literature regarding college access and completion: the role of college proximity in recent high school graduates’ enrollment decisions, and how structural changes to the student experience at a postsecondary institution can affect college completion.

The first chapter is a review of the current economic literature on college access and completion. I start by reviewing what we know about the factors and policies that influence college access, focusing on the literature on family resources, financial aid, and behavioral interventions. I follow this with a detailed discussion on the relationship of college proximity on both enroll- ment and educational attainment, first explaining how a nearby college can both lower potential students’ cost constraints of attendance in the local community and influence these students’ goals and expectations. I also review the empirical evidence, both the use of college proximity as a instrumental variable in estimating the returns to education and the direct effects of nearby colleges on application and enrollment behavior. I then transition toward college completion with reviewing what we know about college supply and how access and completion vary across institutional types. I close this review by looking at how financial aid programs can affect college completion as well as access, the effectiveness of various behavioral interventions on course grades and later performance in college, and institutional-level experiments and reforms to improve completion rates.

The second chapter concerns the role of distance in college access, by focusing on the opening of four new public universities in California from 1995 to 2005. I exploit these openings to test whether distance is a binding constraint on four-year college enrollment among new high school graduates. I show that distance is highly influential: Although California has dozens of public four-year colleges, 40 percent of enrollment from new graduates is at schools within 25 miles of home. Using event study and difference-in-difference models, I find that the opening of a new university nearby raises the four-year enrollment rate among recent high school graduates from local high schools by 1.6 percentage point (an 8 percent increase), with no effect on the share of local graduates who attend farther-away campuses. The extensive margin effect and lack of displacement show up across a range of subgroups, including under-represented minority students. My findings support the view that cost-of-living constraints are binding for many prospective college students.

The third chapter concerns the evaluation of college programs for causal impacts on graduation and other student outcomes, focusing on UC Berkeley Extension’s Fall Program for Freshmen (FPF), a first-year learning community for Spring-admit students to UC Berkeley. Participants choose from a subset of introductory courses and receive advising at facilities near UC Berkeley, while living and participating in activities with other Berkeley students. FPF participants then matriculate to the main campus in the Spring semester. I assess the treatment effect of FPF on college outcomes, using regression and propensity score methods to control for students’ backgrounds at admission and adjusting for differences between FPF participants and regular Fall enrollees at Berkeley. FPF participants are similar to Fall students in their admission characteristics and predicted graduation rates. I find that FPF participants are more likely to graduate and graduate on time from UC Berkeley compared to regular Fall admits in the College of Letters and Science, but do not have major differences in their college grade-point averages (GPAs) at graduation. Students with weaker academic backgrounds have larger program impacts. These findings are robust across a variety of model specifications.

Main Content
Current View