Modelling Sovereign Ratings and Transition Probabilities: New Evidence (original) (raw)

In this paper, we employ sovereign ratings data for 129 countries, spanning the period 1990 to 2006, to investigate rating transition probabilities in the sovereign context. Our key goal is to suggest and evaluate a practical approach to constructing conditional sovereign default probabilities, given the limited amount of sovereign rating data. Our modelling approach conditions transition probabilities upon market and obligor-specific factors. While our models of rating levels perform well, sovereign rating transitions, particularly upgrades, prove more difficult to model. However, our approach performs well out-of-sample when predicting deteriorations in credit quality and as such would be useful to banks aiming to calculate the VaR of portfolios held within a particular sovereign market. Moreover, we compare the determinants of sovereign ratings and rating transitions of the three major rating agencies. We find some variation in the factors which affect rating levels and rating transitions across the three agencies.