Predicting Bank CAMELS and S&P Ratings: The Case of the Czech Republic

In this paper we investigate the determinants of the movements in the long-term Standard & Poors and CAMELS bank ratings in the Czech Republic during the period when the three biggest banks, representing approximately 60% of the Czech banking sector's total assets, were privatized (i.e., the time span 1998-2001). The same list of explanatory variables corresponding to the CAMELS rating inputs employed by the Czech National Bank's banking sector regulators was examined for both ratings in order to select significant predictors among them. We employed an ordered response logit model to analyze the monthly long-run S&P rating and a panel data framework for the analysis of the quarterly CAMELS rating. The predictors for which we found significant explanatory power are: Capital Adequacy, Credit Spread, the ratio of Total Loans to Total Assets, and the Total Asset Value at Risk. Models based on these predictors exhibited a predictive accuracy of 70%. Additionally, we found that the verified variables satisfactorily predict the S&P rating one month ahead.