College on Credit: A Multilevel Analysis of Student Loan Default

This study updates and expands the literature on student loan default. By applying multilevel regression to the Beginning Postsecondary Students survey, four key findings emerge. First, attending proprietary institutions is strongly associated with default, even after accounting for students' socioeconomic and academic backgrounds. Second, cumulative loan debt has a non-linear relationship to defaulting. Third, minoritized and students from low-income families default at disproportionately high rates; and fourth, unemployment and degree completion are strongly associated with greater default rates. These findings counter the argument that default is a "preexisting condition," and the discussion is framed around implications for federal financial aid policy.