Comparison of Default Probability Models : Russian Experience

Under the Basel II accord, improving probability of default models is a key risk-management priority. There are four main aspects of this research: suggesting the bank default classification; using a wide time horizon (quarterly Russian banking statistics from 1998 to 2011); investigating the macroeconomic and institutional characteristics of the banking sector environment and finally, testing the accuracy of the models developed.We have employed nonlinearity and automatic classification of the independent variables in our models, paying attention to the structure of the banking market as well as to the reliability of the models developed. We have compared several models for estimating default probabilities. From the results of this comparison, we have chosen the binary logit – regression with quasi panel data structure. Our key findings are:There is a quadratic relationship between bank's capital adequacy ratio and its probability of default.The “too big to fail” hypothesis does not hold for the Russian banking sector.There is a negative relationship between the Lerner index and bank's PD.Macroeconomic, institutional and time factors significantly improve the model quality.We believe that these results will be useful for the national financial regulatory authorities as well as for risk-management in commercial banks. Moreover, we think that these models will be valuable for other emerging economies