Benchmarking Credit Rating Systems

The validation of credit rating systems has recently attracted particular interest both from banks and their supervisors as well as from academic research. Whereas the main interest has been focused on backtesting methods, alternative approaches such as benchmarking are of growing importance. Benchmarking methods make use of available multi-rater information, i.e. rating assignments about identical obligors stemming from different rating sources. Employing a unique data set provided by the Austrian central bank with rating information on European corporate obligors by nine major Austrian banks, we conduct a benchmarking analysis based on the framework suggested by Hornik et al. (2005) on both market specific and bank specific levels. It turns out that overall similarity among rating systems is remarkably lower for foreign markets compared to the domestic market. Transition economies and markets with a generally low involvement of Austrian banks show particularly dissimilar results. Differences in the overall similarity of rating systems are partly explained by different performance of some individual rating systems in the domestic and foreign markets.