Brad J. Hershbein

The Distribution of College Graduate Debt, 1990 to 2008: A Decomposition Approach

Despite tremendous recent interest in the subject of student debt by both researchers and policy makers, little is known about how the distribution of college graduate debt has been evolving and what factors can explain it. We use National Postsecondary Student Aid Study data from 1990 through 2008 to document the evolution of college graduate debt profiles. We find that growth in debt over the 1990s was rapid and occurred throughout the distribution; during the 2000s, in contrast, debt grew appreciably only for the top quartile. Employing several decomposition techniques, we exploit the richness of the data to explain these shifts. Over the entire horizon, observable characteristics of students and institutions explain about one-third of the debt increase, though this share tends to be higher around the extensive margin and the median and lower in the right tail. While observables — largely costs — explain a majority of the increase between 1990 and 1996 and again from 2000 to 2008, they explain nothing over the late 1990s. We offer suggestive evidence that this “unobservable�? share was supply-side driven, owing to the advent of both federal unsubsidized Stafford loans and private loans.

How Much Does High School Matter? High School Classes and Subsequent College Performance

We investigate how high school instruction in certain subjects is related to academic performance in college in the same courses. We find that high school exposure has only a mild association with college grades after controlling for academic, demographic, and individual-level variables. We examine most of the popular courses taken in college that are typically not required in most high schools, and find that a year of exposure in high school to physics, psychology, economics, and sociology is associated with changes in grades for the same college course of between 0.003 and 0.2 points (on a four-point scale), none of which is statistically significant. Calculus, the exception, has a moderate association, with about a half- letter-grade improvement (0.49 points). Our results are consistent with decades of smaller scale research based on individual high schools or college classes. We discuss reasons why high school course-taking can, in many instances, have a small impact on college performance. Finally, our results call into question the effectiveness of policy interventions that do not link high school outcomes to college performance.

The Impact of Subsidized Birth Control for College Women: Evidence from the Deficit Reduction Act

This paper uses a unique natural experiment to investigate the sensitivity of American college women’s contraceptive choice to the price of oral birth control and the importance of its use on educational and health outcomes. With the passage of the Deficit Reduction Act of 2005, Congress inadvertently and unexpectedly increased the effective price of birth control pills (“the Pill”) at college health centers more than three-fold, from

Playing the Admissions Game: Student Reactions to Increasing College Competition

5 to

The Opt-In Revolution? Contraception and the Gender Gap in Wages

10 a month to between

Graduating High School in a Recession: Work, Education, and Home Production

30 to

Worker Signals Among New College Graduates: The Role of Selectivity and GPA

50 a month. Using quasi-difference-in-difference and fixed effects methodologies and an intention-to-treat (ITT) design with two different data sets, we find that this policy change reduced use of the Pill by at least 1 to 1.8 percentage points, or 2 to 4 percent, among college women, on average. For college women who lacked health insurance or carried large credit card balances, the decline was two to three times as large. Women who lack insurance and have sex infrequently appear to substitute toward emergency contraception; uninsured women who are frequent sex participants appear to substitute toward non-prescription forms of birth control. Additionally, we find small but significant decreases in frequency of intercourse and the number of sex partners, suggesting that some women may be substituting away from sexual behavior in general.