Predicting payment and nonpayment of direct mail obligations: Profiling good and bad credit risks

Abstract

Profiles of prospect groups are developed in terms of their likelihood of fulfilling financial obligations to direct mail offers. An extensive database of individuals responding to a variety of direct mail offers is analyzed, and the variables that best differentiate the likelihood of defaulting or not defaulting on their financial commitments are identified. Two specific data-analytic models are presented. The first tests hypotheses pertaining to a limited set of demographic and credit-related variables. The second begins with a set of 271 variables and identifies the 11 that best predict default likelihood.