Subprime Lender Short Term Loan Models for Credit Default and Exposure

The subprime borrower market of “short-term loans” (aka cash advance loans) consists of individuals and households who cannot qualify for prime financing terms, because of their low credit scores (FICO credit scores in the neighborhood of 620). A borrower gets quick cash from the lender, e.g., an online bank, or a small “payday” loan operation in a street mall. This loan must be paid back over a period of seven to 21 days, typically, before finance charges kick in and raise the cost of the borrowing even higher. Because customers’ profiles of credit and exposure are not static, banks need to understand the latter profiles, which change over time, in order to deploy capital efficiently. Accordingly, the two key management objectives are: 1) to identify the likelihood of first-payment defaults (via a First-payment Default Model), and 2) to determine the dollar amount to loan, such that in the event of a default the outstanding obligation is minimized (via a Credit Exposure Model). The subprime borrowers’ range of their credit scores is rather “tightly-knit,” rendering the subprime borrower market a uniquely homogenous segment. The homogeneity of the subprime segment causes the loan-decision factors – e.g., risk of default characteristics, the absence of collateral, charge-off rates, purposes of loan, property types, and current market conditions – to be “tightly knotted” (highly correlated), a condition that is not favorable for building any scoring model. Statistical methods are virtually unproductive at untying the knotty relationships among loan-decision factors. The purpose of this article is to present the GenIQ Model©, an alternative machine learning method, that has the ability to unlace (data mine) the knotty relationship among loan-decision factors, producing impressively predictive subprime lender short term loan models: First-payment Default Model, and a Credit Exposure Model.