A Dynamic Model for Repayment Behaviors of New Customers in the Credit Card Market

In this paper, we develop a dynamic model for debt repayment behaviors of new customers in the credit card market. We treat customer decisions of whether to be delinquent or not and of how much to pay conditional on deciding not to be delinquent as two separate but possibly correlated decisions and thus view the amount repaid by a delinquent consumer as a censored observation. We assume that the consumer specific parameters determining her repayment behaviors evolve over time as a new consumer would become more familiar with the usage of the card and also the terms of conditions can change over time. We use a state space modeling approach to capture the evolution of parameters. Our modeling approach enables us to distinguish between the consumers whose delinquency is due to high risk and the consumers whose delinquency is due to non-risk related factors such as oversight. We apply our model to a data set of new consumers’ monthly spending and repayment records for 12 months. The empirical results suggest that the parameters evolve over time. The proposed model performs better in predicting consumer repayment behavior than the static models in which the parameters are assumed to be invariant over time. We conduct a policy simulation based on the estimation result. The outcome suggests that the proposed modeling approach benefits credit card companies in helping them identifying high vs. low risk delinquent consumers and also in developing customized policy according to the risk identification results.