ABSTRACT
In this paper we have designed a model to act as an early warning system to monitor changes in the credit quality of corporate obligors. The structure of the model is hybrid in that it combines the two credit risk modeling approaches: (a) a structural model based on Merton’s contingent claim view of firms, and (b) a statistical model determined through empirical analysis of historical data. Specifically, we extend the standard Merton approach to estimate a new risk-neutral distance to default metric, allowing liabilities and the corresponding default point to be stochastic. Then, using financial ratios, other accounting based measures and the risk-neutral distance to default metric from our structural model as explanatory variables we calibrate the hybrid model with an ordered – probit regression method. Using the same econometric method, we calibrate a model using our risk-neutral distance to default metric as unique explanatory variable. Then, using cumulative accuracy plots we have test the classification power of those models to predict default events out of sample. Finally, we use the optimized model to predict expected default probabilities for industrial companies listed in the Athens Stock Exchange.
