Empirical comparison of structural credit risk models for Bankruptcy Prediction of Financial Institutions in US.

Abstract

In this paper, we compare and analyse the expected default probabilities (EDPs) derived from the six structural models: Merton model, Leland and Toft model, Longstaff and Schwartz model, Briys de Varenne model and the recent Ericsson and Reneby model. In our comparison, we cover the aspects of EDP ranking, average EDP differences, relationship of EDP with key financial ratios and relative predictive power of each model compared with the other models. We found that Merton model tends to overestimate the default probabilities of the 532 non-bankrupt financial institutions used in our study. Its average EDP is found to be at least 16.3% higher than the other models. Also, it is discovered that financial ratios do have a strong explanatory power for the EDPs calculated by the various models. Furthermore, our analysis covers