Centre for Mathematical Sciences Mathematical Statistics

The recent European sovereign-debt crisis has made it clear that exposures towards sovereigns contain credit risk. However, according to the Basel frame-work’s standardized approach banks are not required to hold any regulatory cap-ital for highly rated sovereigns. In response, this thesis develops a shadow rating approach model for sovereign probability of default estimation, subsequently determining economic capital for sovereign exposures within a foundation in-ternal ratings-based framework. Furthermore, the empirical Bayes estimator is utilized for low-default portfolio probability of default calibration. The model is tested on five homogeneous sub-segments in addition to the entire dataset at hand. Empirical findings suggest that the full dataset performs adequately overall. Nonetheless, model performance is superior for accurately constructed sub-segments. In addition, economic, monetary and political indicators as well as banking sector health are found to best replicate S&P’s sovereign long-term issuer credit ratings