Stress Testing Credit Card Non Performing Loans, A Financial Macroeconomic assessment

In this paper we propose a new comprehensive data set on credit card loans, issued by twenty-one of the biggest United States Bank Holding Companies (‘BHCs’), and on their default characteristics. Analyzing the data set, we observed that credit cards net charge off rates (‘NCOs’) and probabilities of defaults (‘PDs’) have a flat distributions, with picks related to business cycles’ characteristics and macroeconomic scenarios’ evolution. Given the importance of these loans, in terms of revenues generated for BHCs, we decided to study their evolution using a dynamic panel data model, with individual fixed effects. Rather than looking at these loans on a BHC individual base, we propose a system oriented approach. The study reveals three main outcomes, or contributions to the risk management literature: i) PDs and NCO rates of credit card loans are well explained by a set of Macroeconomic Variables and by the Institutions’ individual characteristics; ii) the forecasting scenario driven analysis leads to the conclusion that this category of loans generally follows business cycles and present huge potential losses at industry level; and (iii) highlights the characteristics of a new data set and methodology, leaving room to further research.