Abstract:
This thesis provides new empirical evidence on the credit risk literature. Rating agencies regularly measure the probabilities of default on current and historical data, they are not forward looking. Merton models, on the other hand, can provide forward-looking risk neutral probabilities of default. Changes in these risk neutral probabilities of default might provide leading information about changes in credit quality of debt issuer, and thus about either credit rating changes or default. This thesis attempts to bridge this gap. There are three main results are found in this study. First, risk neutral probabilities of default changed significantly before credit rating changes event. Moreover, after credit rating changes event, there is asymmetry of statistically significant changes in risk neutral probabilities of default between credit rating downgrades and credit rating upgrades, as same as study of impact of credit rating changes on stock and bond return. Second, the relationship between risk neutral probabilities of default and credit rating categories is significantly negative relation as expected. Finally, including risk neutral probabilities of default into rating changes prediction model can improve predictive power of the model. Therefore, changes in risk neutral probabilities of default can be used to predict for credit rating changes in the future.
