On Wednesday, August 5, 2009, the New York Times printed an advertisement from Standard & Poor’s entitled “Standard & Poor’s Commitment: Quality and Independence.” The advertisement was a partial response by the firm to the mountain of criticism that all of the rating agencies have received for both structured products and corporate ratings actions during the current credit crisis. This post comments on key points in the S&P advertisement and speculates on the evolving role of the rating agencies going forward.
The August 5 Standard & Poor’s advertisement in the New York Times appeared on page A21 and is available via the S&P website on the following link:
We think that this kind of response is necessary and helpful for setting the investment community’s expectations for the future behavior of the rating agencies. In what follows we comment on the key points in the S&P document. As always, comments are welcome at firstname.lastname@example.org.
“At Standard & Poor’s, we know the performance of our ratings in the area of residential mortgage-related securities has been disappointing.”
We appreciate the candor in this statement, but we don’t think it is broad enough. We believe that the rating agencies’ performance in rating corporations has been less accurate than modern reduced form quantitative default models for many years, as documented in van Deventer, Li and Wang (“Advanced Credit Model Testing to Meet Basel Requirements: How Things Have Changed,” The Basel Handbook: A Guide for Financial Practitioners, Michael Ong, editor, second edition, Risk Publications, 2007). Recent performance has been even less accurate, with most analysts’ criticism focused on names like FNMA, FHLMC, Lehman Brothers, and Bear Stearns. Analysts have attributed the “over rating” of these names to both the inherent conflict of being paid by the firm being rated and the reluctance to downgrade financial firms when tens of billions of dollars of margin calls on derivatives-related collateral can result.
We’ve addressed this issue in three blog posts. The first, “The Ratings Chernobyl,” argues that the very franchise of the rating agencies would be seriously damaged by an accurate accounting of rating agency performance on corporate ratings through the credit crisis:
The second blog post on this topic was our July 22, 2009 post entitled “It’s Time for an Independent Audit of Rating Agency Corporate Bond Rating Performance,”
In this post we pointed out that S&P and Moody’s have both omitted the failures of FNMA and FHLMC from their 2008 year end historical assessment of corporate bond rating performance, in spite of the fact that the conservatorship for both FNMA and FHLMC triggered an event of default as defined by the standard ISDA provisions for credit default swaps. We recently also received an e-mail from a very respected credit analyst at one of the five largest banks in Europe who asserted that Lehman Brothers and Bear Stearns failures have also been omitted from the performance statistics, but we have not confirmed this independently.
Finally, we pointed out in this post that the willingness to rate any structure, no matter how complex or ridiculous, was simply a game of “gotcha” played by Wall Street, where only mis-rated structured products would be brought to market (“Taylor Swift and Rating New Issues of Structured Products: A Great Business or a Fairy Tale?”, May 1, 2009):
“…despite its potential conflicts, the issuer pay model—with conflicts properly managed—is the best available model because it enables the provision of ratings simultaneously to all investors.”
Our view on this is simple. The issuer pay model is fatally flawed and the phrase “with conflicts properly managed” is an oxymoron. In other areas of credit risk assessment, the issuer pay model is inconceivable. Would any lender use a credit score that is available only if the “issuer” (the consumer “borrower” agreed to pay for it? Of course not. Obviously, consumers would only pay for credit assessments if they were either accurate or too good. This would lead to very large losses for lenders who are unable to assess credit independently. Within the existing rating agency environment, one very senior official in corporate ratings told me in 2003 that he believed that the entire rating agency franchise was in danger. Another very well-known analyst from a rival rating agency agreed, but he argued that human inertia was so great that it would take 50 years before the rating agencies ceased to play a significant role in financial markets. We argued in this blog post (May 12, “A Ratings Neutral Investment Policy) that sophisticated investors were already full capable of managing investments with no reference at all to agency ratings:
The reason this is true is because of the rise of quantitative rating approaches, including the following examples:
- 1990: Launch of KMV LLC by Steve Kealhofer, Mack McQuown and Oldrich Vasicek using the Merton to default probability estimation for public firms
- 2002: Launch of the Kamakura Risk Information Services public firm default service, using a multiple models approach to credit assessment. The multiple models approaches include the Merton approach and the more modern reduced form approach, complete with a hybrid model than includes the Merton default probability as an explanatory variable in the reduced form framework
- 2008: Launch of the Kamakura Risk Information Services sovereign default service using the reduced form approach.
The three quantitative default services listed above are NOT “issuer pay” services. They are subscription services paid by the investor, and this subscription pay service has continued at KMV a full seven years after the firm’s acquisition by Moody’s for $210 million. All three of these “alternative rating agencies” have successfully proven that the “investor pay” model is commercially viable.
The S&P advertisement argues that the “issuer pay” model is essential because ratings are free to all investors. We beg to differ. I am an enthusiastic user of “google alerts” and every mention of the phrase “CDO” on web sites tracked by google. Only Fitch Ratings has consistently put its CDO ratings into the public domain. On the corporate rating side, I registered on www.sandp.com today to see what information is available on Lehman Brothers ratings. The site informed me that Lehman’s rating was changed to “NR” on September 25, 2008. In order to assess the accuracy of S&P’s ratings, this is useless. A full history of the rating over time and an independent default data base is necessary. The full history of ratings on either a monthly or daily basis is not free from Standard & Poor’s. Only the current rating is currently available on www.sandp.com that I could find.
“In contrast to the issuer-pay model, the subscriber pay model tends to be a ‘black box’ with limited disclosure to ordinary investors.”
This statement could not be more misleading. Kamakura provides a 140 page technical guide to all subscribers containing complete disclosure of the default probability models behind its default service. All inputs to the model are made available via www.kris-online.com to subscribers daily. By contrast, the ratings process is so opaque that U.S. banking regulators caused wide-spread consternation when they required stress testing of the market value of portfolios with respect to real gross domestic product, unemployment rate, and home prices in February 2009. Banks relying on ratings had no idea how ratings would change with respect to these macro factors. With the Kamakura models, these links are explicit.
“Reinforced our policies and procedures so there is no advising or consulting in the ratings process of structured securities”
This “reform” listed in the advertisement does not go far enough. Many in Congress and in the investment community have been brutally accurate in describing the rating agencies as just as conflicted as accounting firms used to be when consulting services were provided to an audit client. The rating fee on a corporate debt issue pales in comparison to other fees paid by clients, particularly by financial institutions, to the rating agencies. This practice can only be brought to a halt in one two ways: by legislation or by avoiding the use of ratings altogether because of the overwhelming and inescapable conflict of interest between the consulting side and ratings side of the agencies.
“From where we sit, the more open and transparent the process of credit ratings is the more we can earn back the confidence of investors in credit markets.”
We totally agree. That is why all regulatory agencies world-wide have been extended an offer to audit the accuracy of Kamakura’s default models subject only to two restrictions: that it must be done subject to a confidentiality agreement and that it must been done at Kamakura’s headquarters office. We believe that the first step in restoring confidence in the traditional rating agencies is for each of the agencies to commission and publish a third party report on the accuracy of both their corporate ratings and their structured products ratings. As pointed out in the blog citation above, the omission of FNMA, FHLMC and potentially many other corporate failures from performance statistics means that an “in-house” self-assessment does not have enough credibility to be convincing.
Donald R. van Deventer
Honolulu, August 19, 2009