Ratings Reform, Redux

Policymakers are coming back to ratings reform, but the discussion remains skewed toward conflict of interest and liability. In doing so, however, policymakers are missing the opportunity to steer the debate in a different direction. Ratings predictability and stability comes from widespread applications that are used to rate well-understood financial securities. But recent discussions about ratings agency reform have attempted to avoid judging models in efforts to avoid commoditizing ratings. Dogmatically avoiding such commoditization, however, sets up a Type-I/Type-II error bias that is not, in fact, incentive compatible with the goal of producing high-quality NRSRO ratings.

Consider, for instance, applying a non-commoditized rating to a commoditized financial security. Ratings model innovation would be welcome in this context, since there is room for marginal improvement and the risks of mis-rating would be constrained by what is already known about the security and its rating process – few investors would be fooled by substantial deviations from existing ratings technology. It is this realm that ratings agencies seek to defend when advocating regulatory laxity.

Consider alternatively, however, a non-commoditized rating on a non-commoditized financial security. The application of unique ratings techniques to unique financial products allows for a substantial risk of mis-rating. Moreover, when little is known about the properties of the unique new financial security, it is difficult if not impossible, for investors or even a regulator to detect if the ratings agency models are wrong. (At the same time, such proprietary models are extremely valuable to the rating agencies, allowing them to distinguish their models from others in fundamentally unverifiable ways.) Hence, relying crucially on non-commoditized ratings models to rate non-commoditized financial securities – especially without ratings agency liability for error or omission or commission – nearly assures more of the kind of wrenching market adjustments we have recently experienced.

Instead, it might make sense to advocate applying a commoditized rating process to non-commoditized financial securities, so that at least one of those variables is known. Using commoditized rating models in this manner could help foster the sound and stable development of new financial securities within an incubator of known model limitations, before those instruments can be calibrated to fictional levels of accuracy on the basis of non-commodity models with unknown degrees of statistical error.

The hard questions of “ratability” – both within credit other ratings products – have to be tackled to enact meaningful ratings reform. Moreover, the “ratability” issue spills over to mark-to-market valuation: as Gerry Corrigan said at Federal Reserve Bank of Atlanta Financial Markets Conference a couple of years back, (I paraphrase) “every day Wall Street sells innovative products it cannot accurately value,” or rate. It is important to acknowledge, however, that the rating/valuation shortcoming is not a problem unless policymakers operate under a fictional belief that every product CAN be accurately valued and rated. As long as we continue building policy according to such beliefs, we will continue to be “surprised.” Issuer-pays conflict of interest and liability reform will be of little help to the financial system and policy if we keep blindly relying on unreliable models and valuation techniques, refusing to distinguish what we know from what we do not and failing to recognize the changing boundary between the elements of those sets.

† Hermann Moyse, Jr./Louisiana Bankers Association Professor of Finance, Louisiana State University, Senior Fellow at the Wharton School, and Partner, Empiris LLC. Contact information: [email protected]; (202) 683-8909 office. Copyright Joseph R. Mason, 2009. All rights reserved. Past commentaries and testimony are blogged on http://www.rgemonitor.com/financemarkets-monitor/bio/626/joseph_mason.