Home » JEL Codes » I » I2 » I26

Category Archives: I26

Predictors of Student Loan Repayments: A Comparison Between Public, Private For-Profit and Private Nonprofit Schools

by Mannat Bakshi and Arjun Ahluwalia

Using a sample of over 3,500 colleges from the College Scorecard Dataset , we investigate the association of average federal student loan repayment rates with institutional, regional, and student demographic characteristics of colleges. We consider educational cohorts from 2010 to 2016 at public, private for-profit, and private non-profit institutions. Our data do not allow us to see individual student characteristics, hence we control for traits of the average student in each college and focus on institutional traits that impact repayment rates. Our controls for demographics are consistent with prior research on student loan repayment rates (Lochner and Monge-Naranjo, 2014; Kelchen and Li, 2017).

We ran a Random Effects panel regression to determine how institutional, regional, and student demographic characteristics impact repayment rates. We see an important influence of the institution attended. Institution selectivity (lower admission and withdrawal rates) is associated with higher average repayment. Furthermore, the highest degree awarded is a more significant variable when it comes to describing the variation in repayment rates for public schools; private for-profit schools exhibit lower repayment rates and private nonprofit schools exhibit higher repayment rates regardless of the highest degree awarded. This could be due to a combination of signaling and screening effects. Local income and unemployment impact repayment for the average student in public and for-profit schools, but not in private non-profit schools.

A noticeable institutional finding is that, even after controlling for average school demographics, for-profit schools exhibit lower repayment rates across all types of degree-granting programs. Attending a for-profit school may be a negative signal of ability or value to potential employers. Median family income positively affects repayment twice as much for for-profit schools compared to other school types. These finding on for-profit institutions help explain Obama’s “crack down on for-profit career training colleges” (Simon & Emma, 2014).

View Thesis

Advisor: Professor Genna Miller | JEL Codes: I2, I22, I26

Bridging the Persistence Gap: An Investigation of the Underrepresentation of Female and Minority Students in STEM Fields

By Aaditya Jain and Bailey Kaston

Prior literature on mismatch theory has concentrated primarily on minority students, whose lower average levels of pre-enrollment preparedness tend to discourage them from persisting in STEM fields as often as their non-minority counterparts at selective universities. Our study shifts the focus to the persistence gap between men and women, invoking social cognitive career theory to investigate how factors beyond preparedness – such as self-confidence – cause women to switch out of selective STEM programs at higher rates than men. Using the High School Longitudinal Study of 2009, we investigate the drivers of STEM persistence for all students and arrive at two main conclusions. First, higher levels of STEM preparedness are more beneficial to STEM persistence at selective universities, confirming mismatch theory in the sample. We then simulate the counterfactual scenario and find that 33% of students at selective schools would have been more likely to persist in STEM had they attended less selective schools, a figure that reaches 50% for underconfident female students. This observation ties to our second conclusion – that underconfidence in math relative to one’s true performance decreases the likelihood of STEM persistence for all students at selective universities, and that female students at selective schools are more likely to be underconfident than their male counterparts. Our findings suggest that the appropriate policy solution to reduce STEM attrition rates among women should then become a two-pronged approach: (1) more selective universities should better support the STEM self-confidence levels of female students, and (2) home environments should ideally cultivate that self-confidence long before women even reach college. In our final set of analyses, we thus explore the factors that drive math overconfidence in the first place, and conclude that both student and parental biases against female STEM ability are detrimental to the STEM self-confidence of female students.

View Thesis

View Data (Email for Access)

Advisors: Professor Peter Arcidiacono, Professor Michelle Connolly | JEL Codes: I2, I24, I26


Undergraduate Program Assistant
Jennifer Becker

Director of the Honors Program
Michelle P. Connolly