Modelling credit rating transition probability matrices

Increased risk of credit related exposures and contagion effect of the recent Global financial crisis have led to stringent regulations and the need for accurate credit risk models. This thesis investigates and addresses various issues related to rating migration and default probabilities in the context of credit risk. The present study contributes to the literature on transition probability matrices in three different aspects. In the first part, conditional models are presented for transition probability matrices using macroeconomic variables. The model is applied for sovereign entities considering short term horizon and incorporates asymmetric distribution assumption. This is found to provide superior results compared to the traditional symmetric models. The second part proposes a simulation based methodology for the stressed transition probability matrices. The model targets accuracy of credit loss estimations and uses conditional correlation dynamics. This is analyzed under different portfolio structures and the model outputs capture economic contraction periods. The third and the final part introduces a smoothing methodology on the transition probability matrices using a risk sensitive loss function. The model results are consistent with empirical observations and the desired theoretical properties. Portfolio credit risk simulations are performed to estimate the impact of smoothing.