Panel Paper: Pathways through Repayment: A Typology of Student Loan Borrowers and Their Repayment Patterns

Thursday, November 2, 2017
Soldier Field (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Austin Lacy, RTI International


In this paper, “Pathways through Repayment: A Typology of Student Loan Borrowers and Their Repayment Patterns,” researchers from RTI International use large, nationally representative student loan repayment data to classify the different pathways student borrowers take through repayment. Historically, research treats student loans as binary outcomes (i.e., loans payed in full or in default) and ignores that many behaviors can be exhibited (i.e., students can enter into and out of deferment, forbearance, have different month-to-month payments, and so forth). This paper will examine these different trajectories using machine learning techniques to induce a typology to examine actual student repayment behaviors and patterns across time. This new typology will provide policymakers and researchers with the language and understanding needed to address the $1.3 trillion dollars in outstanding student debt and its effects on individual borrowers and reveal possible avenues for intervention. As a result, this work will explore the following research questions: (1) What are the various repayment pathways students take after exiting postsecondary education, including moving in and out of repayment, deferment, and the various repayment options available? and (2) What are the characteristics and financial situations of students in each of these repayment pathways? The data used come from a forthcoming supplemental datasets created by RTI International on behalf of the U.S. Department of Education’s National Center for Education Statistics (NCES); they include records from the National Student Loan Data System (NSLDS) allowing RTI to follow these borrowers through their years in repayment. Specifically, these data matched NCES’ 1996 and 2004 Beginning Postsecondary Student cohorts to the NSLDS in 2016, providing detailed repayment, default, and deferment information for 20 and 12 years. Machine learning techniques are then used to inductively generate a categories of repayment patterns across this large dataset, resulting in the latent states of borrowers’ pathways en route to repayment or default.