Internal ratings systems, implied credit risk and the consistency of banks' risk classification policies

Counterpart risk rating is at the heart of the banking business. In the new Basel II regulation, internal ratings have been given a central role. Although much research has been done on external ratings, much less is known about banks´ internal ratings. This paper presents new quantitative evidence on the consistency of internal ratings based on panel data from the complete business loan portfolios of two Swedish banks and a credit bureau over the period 1997-2000. We study rating class distributions, - transitions and default behavior and compute the credit loss distributions that each rating system implies by means of a semi-parametric Monte Carlo re-sampling method following Carey (1998). Our results reveal, for a portfolio with identical counterparts, substantial differences in the implied riskiness between banks. Such differences could translate into different amounts of required economic capital and create (new) incentives to securitize part of their loan portfolios or increase the riskiness of loans in certain rating classes. We also shed light on the quantitative importance of portfolio composition, portfolio size and the forecast horizon for loss distributions. For example, with common portfolio parameters, credit risk can be reduced by up to 40 percent by doubling the loan portfolio size. We also discuss the relation between loss distributions and the desirable level of insolvency risk.