The Debasement of Ratings: What's Wrong and How We Can Fix It

Executive Summary

This paper poses and answers three sets of questions that lie at the heart of the public controversy over rating agency performance and reform.

(1) What is the evidence that rating agencies have been performing badly in measuring credit risk on the debts that they rate? The evidence relates to two separate phenomena: inflated ratings and low-quality ratings. The inflation of ratings is defined as the purposeful over-rating (under-estimation of default risk) on rated debts. Low-quality ratings, defined as ratings based on flawed measures of underlying risk, are a related but logically distinct phenomenon. The recent collapse of subprime-related securitizations revealed both problems in the extreme, but these problems have been present in securitized debt instruments for decades.

(2) What are the causes of those deficiencies? Have rating agencies been suborned, and if so, by whom and to what purpose? Low-quality ratings could reflect innocent errors associated with learning about new products, but there is evidence that severe errors are often predictable. These severe, predictable errors in ratings reflect agency problems among buy-side investors that lead them to encourage rating agencies to purposely build inaccurate models that ignore or underestimate important risks. I review the evidence that suggests that important modeling errors in the recent crisis were predictable, and that they reflected buy-side investors’ demand for inaccurate and inflated ratings. Both of these phenomena (rating inflation and low-quality ratings) are legitimate targets of reform. Ratings inflation, while sometimes harmless, undermines the regulatory use of ratings to limit risks undertaken by institutional investors. Low-quality ratings, resulting from agency problems, can generate disastrously large system-wide losses for ultimate investors (the clients of institutional investors), and extreme disruption to markets, as in the recent crisis.

(3) What do policy makers propose doing to improve rating agency performance to eliminate ratings inflation and low-quality ratings? Do those reform proposals make sense, and if not, what would work better? Existing policy proposals would tinker with the rating process in ways that have little hope of fixing either the ratings-inflation or low-quality ratings problems. The regulatory solution that I propose – to create credible penalties that link the fees credit rating agencies receive to their actual performance in predicting defaults – would address both the incentive problem that gives rise to ratings inflation and the incentive problems that encouraged faulty rating methodologies.

Read the full report here.