The Barra Credit Series: Market Implied Ratings

Jan 1, 2003

In recent years, the growth of the global credit market has been spectacular. From an investor perspective, this has created many new opportunities for higher returns and diversification, but a careful management of risk is more necessary than ever. In this context, measures of credit quality are becoming an increasingly important reference. Agency ratings are a standard measure of credit quality. The question of capital  requirements and the recent collapse of several high-profile large corporations have fueled a vigorous debate about agency ratings. Although not necessarily timely indicators of a company's ability to repay its debt, they are still likely to remain a convenient, widely used reference that can be used across markets and financial institutions (see for instance Basel Committee on Banking Supervision, 1999). Agency credit ratings are designed to indicate the probability that a given borrower will fail to service its debt, but they serve other purposes. In particular, they are useful as an ingredient of multi-factor risk models. Such models assume that returns of bonds with similar sectors and ratings will be highly correlated, and use agency ratings to build common factors. In the last thirty years, most attempts to improve on agency ratings have come from a mathematical approach (e.g. Merton (1974), Black-Cox (1976)) that generates equity market-based default probabilities, a statistical approach based around conditional default rates (e.g. Duffie-Singleton (1995), Jarrow-Turnbull (1995)), and more recently an hybrid approach (Giesecke (2001), Giesecke-Goldberg (2003)). These approaches result in arguably better credit risk forecasts but do not incorporate agency rating information. This study differs from most in the literature in two ways. First, we derive price-based equivalent ratings, rather than default probabilities based on firm fundamentals. For many practitioners in the credit area, improving forecasts of default risk has so far been the central concern. This study, in contrast, looks at a market-based approach to classifying bonds, improving spread risk forecasts, and detecting rating anomalies. Finally, in contrast again to typical fundamental models of default, our data come directly from the bond market rather than the equity market.