Performance comparison of empirical and theoretical approaches to market-based default prediction models

Sammanfattning: The Black Scholes Merton (BSM) contingent claims approach to modeling corporate default risk entails mapping a distance to default (DD) to a probability of default (PD) in application. To accomplish this, the research community typically assumes a normal distribution. The authors question the practical relevancy of such research, since the BSM contingent claims approach most commonly used in practice, Moody's KMV, uses an empirical expected default frequency (EDF) for this purpose. In this study, the authors test the assumption implied in prior research that PD calculated under a normal distribution can serve as a reasonable proxy for PD calculated with EDF. Without access to Moody KMV's proprietary database, however, the authors use an empirical EDF distribution based on an approximated simulation. The authors find that information content does differ between the two approaches. Furthermore, the authors findings imply that, given a sufficiently large sample, the empirical EDF approach can provide a higher quality forecast of default probability than under a normal distribution.