|Thursday, February 20|
|PS1 Poster Session I & Opening Mixer||
Thu, Feb 20, 5:15 PM - 6:45 PM
Modeling Auto Loan Delinquency: Zero-Inflated Models (302821)*Emiliana Inez Patlan, Firstmark Credit Union
Keywords: zero-inflated model, hurdle model, delinquency
Many mid-sized regional banks and credit unions lack internal automated loan decisioning systems that allow for minimizing delinquency rates while increasing the loan portfolio. Two types of regression models for count data are used to model the occurrence of auto loan delinquency for Firstmark Credit Union. Since the dependent variable, the number of times delinquent, requires special consideration, zero-inflated and hurdle models are utilized and compared. The implications for the credit union are two-fold: a more accurate framework for decisioning is provided, and it lays the groundwork for a future algorithm that will allow for automatic decisioning of a large majority of loan applications. Zero-inflated and hurdle models are shown to give similar results. Several criteria that have long been a part of loan decisioning are shown to be insignificant in explaining and predicting delinquency. Implementations that result from this analysis reduce credit, operational, compliance, and strategic risks for the credit union.