Date of Award

Spring 5-10-2019

Level of Access Assigned by Author

Open-Access Thesis

Degree Name

Master of Science in Economics (MSECO)




Philip Trostel

Second Committee Member

Angela Daley

Third Committee Member

Elizabeth Allan


Cumulative student loan debt in the United States has now surpassed $1.5 trillion. Moreover, since the turn of the century, cohort default rates of these loans have steadily risen across all types of institutions. The latest data from the U.S. Department of Education shows 10.8 percent of borrowers who entered repayment in fiscal year 2015 have defaulted within three years. In turn, the first chapter of this paper summarizes student loan policies as well as trends in debt and default. Furthermore, it highlights the consequences of high student debt and default for individual borrowers and the economy. Results show evidence to support significant decreases in purchasing homes, having children, and getting married as results of high student loan debt burdens. Other significant findings include many individuals reporting working in jobs outside their fields of study, as well as working more than desired. Continuing, this paper evaluates higher education-related policies and how such policies have impacted default rates and debt in recent years. The general conclusion is that such policies have done little to decrease student loan debt and default.

The second chapter consists of a micro-level Probit regression analysis of student loan default, using institutional and individual level characteristics as explanatory variables. The goal is to highlight which factors, if any, are more related to high likelihood of defaulting on student loan payments. Data from the Beginning Postsecondary Students Longitudinal Survey is used for the analysis. Results are mostly consistent with previous literature. Degree completion and/or the level of degree completion has the greatest impact on repayment behavior, although there are other factors also associated with default. Furthermore, institutional characteristics have little bearing in predicting default once individual level characteristics are added to the model.