文摘
This paper studies the accuracy of the Roll Rate model -- the most commonly used technique in forecasting economic losses due to credit card charge-offs -- by taking 1278 banks and credit card issuing institutions in the USA within the period of 4th quarter of 2007 to 2 nd quarter of 2008. It also examines the augmentation of the Roll Rate model by incorporating some bank specific variables -- bank size, interest fees, the share of credit card loans in total assets, and aggregate amount of credit card loans. A significant association between charge-off rate and severe delinquency is observed in both the methodologies -- the standard Roll Rate model and the Augmented Roll Rate model. Results show better performing (measured by profitability) banks are associated with lower charge-off rate and charge-off rate increases with asset size of banks. A positive association between the interest rate and fees charged by the banks and charge-off is found. I also find a positive association between the amount of credit card loan share in its total asset and the charge off rate.