Mortgage Risk and Disparate Impact Associated With Student Debt
415
outcomes. Studies have found, for example, that student loan debt is associated with delayed
marriage (Bozick and Estacion, 2014; Gicheva, 2011; Stone, Van Horn, and Zukin, 2012) and
childrearing (Nau, Dwyer, and Hodson, 2015; Sieg and Wang, 2017).
Another relevant strand of the literature looks at the relationship between student loan debt
and the financial health of borrowers post-schooling, such as repayment difficulties. Using the
2007–2009 Survey of Consumer Finances, Elliott and Nam (2013) find lower net worth for those
with high student loan debt, and Thompson and Bricker (2014) find families with student loans
more likely to be 60 or more days late paying bills. In addition, the research consistently finds high
student loan debt is not a strong predictor of repayment difficulties (Baum and Johnson, 2016;
Dynarski and Kreisman, 2013). Instead, high student loan debt is associated with higher levels of
degree attainment and completion. A recent study from Baum and Looney (2020) found that those
with professional and doctorate degrees, only 3 percent of the population sampled, held 20 percent
of the outstanding student loan debt.
In the context of homeownership, the relationship between student loan debt and homeownership
is also well examined. However, the findings are mixed between studies finding no relationship
between student loan debt and homeownership (Velez, Cominole, and Bentz, 2019; Zhang,
2013), and others finding a negative relationship between student loan debt and homeownership
(Bleemer et al., 2017; Mezza et al., 2016; Miller and Nikaj, 2018). The conflicting results are likely
explained by two factors. First, student loan debt is not randomly assigned, and selection into
student loan debt and homeownership are correlated. Studies have addressed this concern through
instrumental variables (Houle and Berger, 2015; Mezza et al., 2020; Velez, Cominole, and Bentz,
2019). The second concern is omitted variables. Dynarski (2016) and Miller and Nikaj (2018) find
degree completion to be an important consideration.
The literature on student loan debt is extensive. Prior studies find a direct relationship between
student loan debt and adult milestones such as marriage, childrearing, and homeownership.
Although the student loan literature is informative, this article may be the first to examine the
relationship between student loan debt and mortgage performance.
Data
To conduct the analysis, the authors obtained information on borrower characteristics and loan
performance from the National Mortgage Database (NMDB). The NMDB program is administered
by the Federal Housing Finance Agency (FHFA) and combines credit attributes and performance
data from a 1-in-20 sample of residential first lien mortgages from one of the three primary credit
bureaus with administrative records and information from the Home Mortgage Disclosure Act.
The authors examined owner-occupied home purchase mortgages originated between 2014
and 2018 and observed performance through 2019, ending before the COVID-19 pandemic.
Observations are restricted to loans with at least one borrower with a credit score and where
the reported back-end DTI ratio used in underwriting is greater than or equal to the sum of the
mortgage and any student debt payments reported to the credit bureau. Borrowers with student
debt are defined as any nonzero student debt balances when the mortgage was originated. The