Recent studies in marketing have consistently shown that all customers are not equally profitable. In the credit card business, all customers are not equally risky. When a customer misses one payment on a credit card bill, a signal is sent to the credit card company. It is important for the card issuer to interpret the signal and to identify whether the customer is a low-risk one, who will eventually pay back the debt and contribute to the card issuer’s profits by paying interest on the overdue balance, or a high-risk one, who will not pay back the debt. The issuer can then customize its policies to deal with these different consumer types.
This article develops a model for debt repayment behavior of new customers in the credit card market that makes it possible to differentiate between low-risk delinquent customers and high-risk customers. Specifically, the authors treat the discrete variable of whether a consumer is delinquent each month and the continuous variable of the amount paid each month, conditional on not being delinquent, as two separate but possibly correlated observations. Furthermore, the model separates the repayment amount conditional on consumers being delinquent into two groups: an actual zero resulting from a consumer’s inability to payback to debt and a censored zero resulting from other factors, such as consumer oversight. The authors use a state space modeling approach to allow the model parameters to evolve over time to capture the possible evolution of new customers’ spending and repayment behaviors using the credit card because a new customer will become more familiar with the usage of the card over time and because the terms of conditions can change over time.
The authors apply the model to a data set of new consumers’ monthly spending and repayment records. The proposed model performs better in predicting consumer repayment behavior than static models, in which the parameters are assumed to be invariant over time, indicating that it is important to take into account the evolutionary aspect of cardholder behaviors over time to predict future cardholder behaviors. The proposed model also performs better in predicting a consumer’s risk type than the commonly used industry naive model, which uses only the delinquency data, a finding that highlights the importance of using the information on the repayment amount as well as the discrete delinquency outcome when identifying consumer risk type.
The authors conduct a simulation exercise based on the estimation results. The simulation outcome suggests that the proposed modeling approach benefits credit card companies by helping them identify high- versus low-risk delinquent consumers and also by developing customized policies according to the risk identification results.
Biography
Yi Zhao is a doctoral student at the Hong Kong University of Science and Technology and Fellow of the Center for Excellence in Brand & Customer Management in the J. Mack Robinson College of Business at Georgia State University. His research interests include empirical modeling of consumer choice behavior, dynamic models, and empirical modeling of competitive strategies.
Ying Zhao is Assistant Professor of Marketing at the Hong Kong University of Science and Technology, China. She received her PhD from University of California, Berkeley. Her research interests include empirical modeling of competitive strategies, pricing, consumer choice models, and consumer decision making. Her research has appeared in Journal of Marketing Research, Management Science, Marketing Science, Journal of Business, and Marketing Letters.
Inseong Song is Associate Professor of Marketing at the Hong Kong University of Science and Technology, China. He received his PhD from University of Chicago. His research interests include dynamics of the demand for new products, measuring and exploiting consumer heterogeneity, understanding consumer choice behaviors, and firms’ strategic and competitive behaviors. His research has appeared in Journal of Marketing Research, Management Science, and Quantitative Marketing and Economics.
Journal Marketing Research, Volume 46, Number 4, August 2009 View Table of Contents.