could craft exit counseling to the particular needs of a given college or major when borrowers are leaving the university.
While it might be desirable to target first-time freshmen borrowers based upon information from their admissions applications, the variables from the College Preparedness section, which describes the borrower’s coursework and success in high school, will not serve as very strong predictors of future default trouble. This fact is somewhat surprising, considering that variables describing success in college are so important in the study. For example, college GPA has the strongest relationship to the occurrence of default of any variable in the study, but the borrower’s high school class rank has only a moderate relationship to default at best. It appears that as borrowers become more removed in time from past performance, the performance has less influence on college outcomes and student loan default risk.
Using the study results to target defaulters will incur a certain amount of misclassification. If, for example, the university administration provides supplemental loan counseling to borrowers whose GPAs fall below 2.0, it will be addressing the default risk of over half of the borrowers who would otherwise default after leaving TAMU. If the counseling is effective, then at least some of these borrowers will avoid default and TAMU’s default rate will be lower than it otherwise would have been. This appears to be an effective strategy. But because the default rate of borrowers with GPAs less than 2.0 is around 20 percent, 80 percent of the borrowers who receive the supplemental counseling would not default even if they did not receive it. Essentially, supplemental counseling will have been provided unnecessarily to 80 percent of the targeted group.
Nevertheless, the inefficiency of a targeted intervention strategy might be acceptable for two main reasons. First, the benefits of preventing some defaults might simply outweigh the costs associated with “errantly” targeting other borrowers. Second, a given targeting strategy might incur less error than the next best alternative approach. For example, if a commitment already exists to provide supplemental loan counseling to everyone who begins their sophomore year, then targeting borrowers based upon GPA might very well incur less error than, and have many of the benefits of, the more sweeping strategy.
Additional research could help increase the efficiency of default aversion strategies and decrease the costs. For example, research might reveal that a combination of borrower characteristics identifies most of the defaulters at TAMU but is associated with very few non-defaulters. Targeting that is based upon such research would be very efficient and could justify more costly and more intensive intervention strategies, depending upon their perceived benefits. A multivariate analytical approach would be a possible way to build upon the research presented in the present study.