Credit ratings and credit risk: Is one measure enough?

摘要

This paper investigates the information in corporate credit ratings. If ratings are to be informative indicators of credit risk they must reflect what a risk-averse investor cares about: both raw default probability and systematic risk. We find that ratings are relatively inaccurate measures of raw default probability - they are dominated as predictors of failure by a simple model based on publicly available financial information. However, ratings do contain relevant information since they are related to a measure of exposure to common (and undiversifiable) variation in default probability ('failure beta'). Systematic risk is shown to be related to joint default probabilities in the context of the Merton (1974) model. Empirically, it is related to CDS spreads and risk premia. Given the multidimensional nature of credit risk, it is not possible for one measure to capture all the relevant information.