Prediction of student loan default rates

Abstract

This study focuses on the modeling of the probability of student loan default using certain characteristics of individual students. A generalized linear model is adopted in the rating of students' characteristics to parameterize their significance. Factors to be rated include the type of university attended, the type of loan taken by the student and the period of the course. By determining the probability of default, the amount of loss expected can be detem1ined and thus create reserves to counter the .adverse effects of high student loan defaults.