A framework to assess portfolio credit risk using accelerated hazard rates

There exists empirical evidence suggesting that the default risk of a firm depends on certain systematic factors, specifically those that affect to some extent the default probabilities of all firms and give rise to correlated default risk. At the same time, there is empirical evidence indicating that firm-specific factors, such as, leverage ratios and volatilities, have a direct impact on the default risk of a firm. The correlation between different default probabilities is an important and significant portion of a portfolio's credit risk. The correlation between default probabilities also changes over time as the underlying systematic and firm-specific factors change. This dissertation proposes a framework for modeling default probabilities and the correlation between them by linking them to the underlying macroeconomic and firm-specific factors.; The proposed framework is based on a reduced form approach to credit risk. The default probabilities are modeled using hazard rates that become "accelerated" during times of stress. The hazard rate for a firm depends on an "acceleration factor", which in turn depends on underlying macroeconomic and firm-specific factors. Using this approach, one can estimate the default probabilities of firms in a portfolio as well as correlations between probabilities in order to determine how sensitive a portfolio's credit risk is to changes in the underlying macroeconomic and/or firm-specific factors.; Under the normality assumption for the acceleration factor, closed-form solutions for correlation among default probabilities are derived which provide some useful insights into the portfolio's credit risk. Closed-form solutions are also derived for the probability density function of the default probability of a single firm as well as for the joint probability density function of the default probabilities of two firms. Empirical illustration of the proposed modeling approach is done whereby distributions of the default probabilities are derived for a sample of firms. By using an appropriate functional form of the acceleration factor, the proposed approach can overcome the difficulty encountered by the currently used reduced form approaches that run into negative default probabilities while modeling hazard rates.