Credit rating agencies were at the center of the financial crisis. Without high ratings, complex collateralized debt obligations (CDOs) and structured investment vehicles (SIVs) that did most of the damage could not, and would not, have been created or sold. Yet there have been few lawsuits filed against credit rating agencies. Why?
The answer goes much deeper than Stoneridge Investment Partners v. Scientific Atlanta,(1) the Supreme Court case limiting scheme liability. Even after Stoneridge, the case for rating agency liability should be strong. Several rating agency employees essentially have admitted culpability (one said a deal "could be structured by cows and we would rate it").(2) And they have deep pockets (the market value of Moody's Corp. shares is more than $5 billion, even after the 2008 declines). So why not sue them?
One reason is that, incredibly, the agencies have a statutory exemption from Section 11 liability. Hopefully, the new Congress will correct that error. But the more troubling explanation is that the rating agencies have persuaded some courts that ratings are merely "opinions" entitled to First Amendment protection.(3) The agencies have used free speech as a shield to disclaim liability from investor lawsuits. Moody's says it is The New York Times; Standard and Poor's says it is the Wall Street Journal.
Plaintiffs are understandably reluctant to sue into the teeth of bad precedent. But there are numerous grounds for judges to distinguish, or reverse, earlier cases when presented with lawsuits arising from complex subprime mortgage-backed securitizations. Moreover, judges know more about the problematic role of rating agencies, and the dubious nature of their free speech defense. The shroud of credibility that once protected the agencies has been lifted.
The key distinction between earlier cases and today's potential litigation is between "plain vanilla" corporate bond ratings and what I call "second-level" securitization ratings. "Second-level" securitizations involve the repackaging of assets, such as subprime mortgage loans, which already have been repackaged through "first-level" securitizations. They often involve the use of credit default swaps to create synthetic CDOs. The key point is that the credit rating of such a "second-level" deal applies not to the obligations of an extant operating company, or even the obligations of a newly-created company that purchases assets, but to the obligations of a newly-created company that purchases already-securitized assets that are then combined in a "second-level" mixing process. A "second-level" securitization is like making a cake, not from ingredients such as eggs, milk, and sugar, but from the remixed remains of some already-eaten half-baked cakes.
Ratings of "second-level" securitizations are not protected speech, and rating agencies are doing much more than merely speaking when they rate such complex securitizations. They have a high level of initial and ongoing involvement in these deals, at early and later stages, and receive significantly higher fees for them. Rating agencies determine the capital cushions that are required for particular tranches; they provide capital matrix parameters that govern the operation of special purpose entity issuers; they are involved in the operation of the issuers on an ongoing basis; they instruct the asset managers regarding the kinds of assets the issuers can acquire, both initially and over time; and the deal documentation for these transactions typically includes descriptions of the simulation models the rating agencies use to
determine the relative proportions of an entity's capital structure, as well as the necessary over-collateralization ratios and triggers, both initially and over time. Moreover, CDOs and SIVs require a much more in depth analysis than corporate bond transactions by the rating agencies, including the use of their mathematical models and assumptions. For these reasons, judges should reject the claim that ratings of "second-level" securitizations are merely "opinions."
The History Behind Credit Rating Agencies
Credit rating agencies do not engage in the kinds of activities that warrant First Amendment protection - at least not any more - if they ever did. In the early debt markets, credit rating agencies played the role of information intermediary, from the nineteenth century mercantile credit rating agencies through John Moody's application to bonds.(4) Moody's insight was that he could profit by selling to the public a synthesis of complex bond data into a single letter rating.
For the most part, credit rating agencies fit this reputational investor-pay model until the mid-1970s, when, not coincidentally, the Securities and Exchange Commission began relying substantively on credit rating agencies for regulatory purposes and the agencies shifted to an issuer-pay model.(5) As the regulatory dependence on ratings increased, rating agencies became more profitable and also began providing ratings of transactions designed to achieve particular ratings.
Bankers and issuers created a range of highly-rated asset-backed transactions and collateralized bond obligations, as well as derivative product companies, financial guarantor transactions, and AAA-rated arbitrage vehicles. The first SIVs and asset-backed CDOs were created during this period. Ultimately, these parties worked with credit rating agencies to create more than a trillion dollars of "second-level" securitizations.
If the credit rating agencies had used reasonable and accurate models and assumptions, then these transactions might not have been problematic. However, the agencies faced financial incentives to use unreasonable and inaccurate models and assumptions to complete deals and earn greater fees. These incentives were especially strong given the expected absence of any consequence or cost. After all, the rating agency managers thought that ratings were protected free speech.
Rating the "Second-Level" Deals
The simplest way to generate unwarranted high ratings was to use outdated and inapplicable historical assumptions with respect to the underlying mortgage-backed securities. The inputs to the relevant models were straightforward: expected default rate, recovery rate upon default, and, for portfolios of assets, the correlation of expected defaults. The rating agencies created models, with the assistance of bankers and arrangers, that generated tranche credit ratings for "second-level" deals based on these inputs. Those models, in turn, typically depended on assumptions with respect to the expected statistical distributions that returns on the underlying collateral would follow.
Rating agency assumptions and models did not accurately capture the risk associated with "second-level" securitizations. Default rate assumptions were derived from historical information, including default data about other asset categories as well as asset price correlations, rather than default correlations. Moreover, assumptions for "second-level" deals were based on ratings of mortgage-backed securities, even when both the rating agencies and other participants in the resecuritization transactions were aware that both that the credit quality of the underlying mortgages had declined and that the expected default correlations associated with those mortgages had increased. Nevertheless, the simulations the agencies ran to calculate tranche ratings were based on stale and inaccurate assumptions.
Rating agency correlation assumptions were particularly important. Inaccurate correlation assumptions based on incorrect statistical models enabled parties to structure deals with high ratings on senior tranches, given that the expected correlations of defaults of mortgage-backed securities were higher than the estimates used for the models. The rating agencies have struggled to understand the importance of correlation assumptions for CDOs and SIVs, even as those assumptions supported a sharply increasing number of "second-level" securitizations. Moody's conducted an in-depth study of corporate bond correlation, which led to a new Monte Carlo simulation-based market tool in 2004 for measuring the credit risk of synthetic transactions; it revised its methodology for structured finance asset correlations a year later.(6) S&P's inputs simply remained constant for years. By 2006 and 2007 at the latest, it was apparent that the relevant mortgage asset correlations underlying CDOs and SIVs were significantly higher than the rating agencies had assumed. By February 2008, Moody's had downgraded at least one tranche of 94.2% of subprime residential mortgage-backed deals it had rated in 2006.(7)
"Second-level" mortgage-backed securitizations, particularly CDOs and SIVs, explicitly depended on rating agency letter ratings. Without those ratings, the transactions could not, and would not, have happened. Without the ability to obtain high ratings for CDO and SIV tranches, there would have been little appetite for overpriced lower-rated mortgage collateral. Without that appetite, there would have been little pressure leading to the proliferation of sub-prime mortgages, because those mortgages could not have been offloaded through "second-level" securitizations. Without the proliferation of low quality mortgages, there would not have been a dramatic housing market rise and fall, with the attendant ripple effects.
Investors, regulators, and legislators need to understand the importance of holding credit rating agencies responsible for their role. One important way to help correct their dysfunctional role in the markets is to include them as defendants, and to challenge the specious First Amendment defense they have used to cloak their flawed business model from liability for decades.
1 128 S. Ct. 761 (2008).
2 See Securities and Exchange Commission, "Summary Report of Issues Identified in the Commission Staff's Examinations of Select Credit Rating Agencies," July 2008, at 12.
3 See Frank Partnoy, How and Why Credit Rating Agencies Are Not Like Other Gatekeepers, in Financial Gatekeepers: Can They Protect Investors? (Brookings Institution Press 2006, Yasuyuki Fuchita and Robert E. Litan, eds.).
4 Frank Partnoy, The Siskel and Ebert of Financial Markets: Two Thumbs Down for the Credit Rating Agencies, 77 Washington University Law Quarterly 619, at 637-38 (1999), reprinted at 33 Securities Law Review 161 (2001).
5 More precisely, the regulatory dependence on credit ratings began in 1973, when the SECSECSEC proposed amending broker-dealer "haircut" requirements, which set forth the percentage of a financial asset's market value a broker-dealer was required to deduct for the purpose of calculating its net capital requirement. Rule 15c3-1, promulgated two years later, required a different "haircut" based on the credit ratings assigned by Nationally Recognized Statistical Rating Organizations (NRSROs). See 17 C.F.R. 240.15c3-1. Since the mid-1970s, statutes and regulations increasingly have come to depend explicitly on NRSRO ratings. See Partnoy, 1999, at 690-703.
6 See Moody's Investors Service, "Collateralised Debt Obligations: A Moody's Primer," March 2005, at 5-6; Moody's Investors Service, "Moody's Revisits its Assumptions Regarding Structured Finance Default (and Asset) Correlations for CDOs," June 2005.
7 See Moody's Investors Service, "A Short Guide to Subprime," March 2008, at 3.