Credit Ratings are a Poor Predictor of Corporate Failure, Finds Research

Credit ratings agencies (CRAs) have played a high-profile and pivotal role in financial markets since the downturn, with some of their decisions making headlines for all the wrong reasons, according to research by Saïd Business School, University of Oxford. Regulators and industry experts have called into question the reliability and usefulness of credit rating agencies and, as research shows, for good reason.

“Our research proves what many critics of credit ratings agencies have been arguing for years – that the accuracy and informational value of corporate credit ratings is dishearteningly low. Ratings are not an optimal predictor of default probability. They explain little of the variation in default probability across firms and they fail to capture the considerable variation in default probabilities and empirical failure rate over the business cycle,” said Dr Mungo Wilson, lecturer in financial economics at the Saïd Business School, University of Oxford.

Credit ratings remain the most common and widely used measure of corporate credit quality, a position challenged by a new paper from Wilson and Jens Hilscher, assistant professor of finance at Brandeis University.

The paper examines the informational content of corporate credit ratings and demonstrates that ratings are poor measures of raw default probability. In fact, a basic measure containing accounting and market-based measures of financial distress easily outperforms credit ratings. The researchers demonstrate this by creating a simple model to predict default risk based on publicly available financial information – referred to as a ‘failure score’ – and comparing its ability to predict actual defaults to that of Standard & Poor’s (S&P) corporate credit ratings. Using data from 1986 to 2008, Hilscher and Wilson find that the ‘failure score’ is almost twice as reliable as rating.

The data also show that ratings are strongly related to systematic default risk – the tendency of firms to default in bad times – and offer investors insight into how corporations will fare during an economic downturn. This finding helps rationalise agencies’ practice of ‘rating through the cycle’ – the practice of rating agencies to disregard the effect of business cycle fluctuations on default probability – and may help explain why investors pay so much attention to ratings, even though they are not optimal predictors of default.

The researchers conclude that, given the nature of credit risk, a narrow focus on only one measure of credit quality reduces the accuracy of default prediction and, therefore, cannot be an optimal measure of credit quality. Instead, a more accurate and useful measure would be to separate default prediction from the measurement of systematic risk. Default prediction data could update frequently and rapidly and respond to firm-specific news, while measures of systemic risk could be a combination of current credit ratings and aggregate credit conditions.


Related reading