Extended Modeling of Banks’ Credit Ratings

The aim of this paper is to construct a reliable model based on public information for the practical usage of interested agents, regulators and banks themselves. During the work, a table of representative variables that have potential influence on ratings was constructed. The research is based on the Bankscope database that contains financial information about international banks for the period from 1996 to 2011. A Matlab code was created in order to fill in the gaps in a database. As a result, an ordered probit model provides the following conclusions. First, macro variables improve explanatory power of the model. Second, the regional affiliation effect is significant. Furthermore, rating grades are adjusted to economy's business cycle. Moreover, Standard&Poor's and Moody's are predicted to be the most and the least conservative rating agencies, respectively. The final model was classified as practically useful as predicting of rating grades gave 31% of precise results and up to 70% forecasts with an error within one rating grade, while predicting of rating classes resulted in 62% and 95% respectively.