What's (Still) Wrong with Credit Ratings

INTRODUCTIONAcademics have long recognized the unique and important role of credit rating agencies in financial markets.1 During the 1980s and 1990s, the scholarly consensus about credit rating agencies centered on the "reputational capital" view-a theory, based on information economics,2 that credit rating agencies survived and prospered based on their ability to generate and aggregate credible information about debt issues.3 The essence of the reputational capital view is that credit rating agencies fill an important need arising from the information asymmetry between issuers and investors: credit rating agencies are reputational intermediaries that bridge the information gap, not unlike restaurant or movie reviewers, except that they use letters (such as AAA) instead of stars or tomatoes.4Beginning in 1999, I set forth a more critical alternative view of credit rating agencies, which I called the "regulatory license" view, based in large part on the empirical observation that regulators and market participants increasingly relied on credit ratings in substantive legal rules, and that this regulatory reliance distorted the market for credit ratings.5 The essence of the regulatory license view is that credit rating agencies are important, not because they provide valuable information, but because regulatory reliance on credit ratings effectively makes ratings valuable as a kind of financial license that unlocks access to the markets-even if the ratings themselves have little or no informational content.Before the introduction of regulation, the credit rating business was small and relatively unprofitable.6 But, regulatory reliance on credit rating agencies started increasing during the mid-1970s from references in statutes and rules to "Nationally Recognized Statistical Ratings Organizations," or NRSROs, particularly to the two most prominent rating agencies, Moody's Investors Service, Inc. and SP as regulatory licenses proliferated, NRSROs became both more profitable and less informative.7During the 2000s, the scholarly literature took multiple perspectives about the above two theoretical frameworks.8 On one hand was the argument that, notwithstanding some prominent miscues (such as Enron's investment grade credit ratings shortly before its bankruptcy in 2001), credit ratings overall were correlated with fixed income default experience and arguably reflected at least some information about issuers' creditworthiness.9 On the other hand was the argument that regulatory reliance on credit rating agencies continued to increase throughout this time, even while sophisticated market participants viewed credit ratings more skeptically.10Likewise, before 2007, regulators' views of credit rating agencies were mixed. Regulators were sympathetic to the important role of credit ratings in financial markets, but also were critical of potential problems related to that role, including perceptions of agency costs and conflicts of interest.11 In 2006, Congress adopted modest reforms to address perceived problems associated with credit rating agencies, even as regulators in a wide variety of areas continued to rely substantively on credit ratings.12Throughout this time, I offered the regulatory license view as a theory to explain the ongoing paradox in fixed income markets: that credit ratings were enormously important, yet possessed little informational value.13 I warned regulators and policy makers about the potentially toxic role of credit rating agencies in the use of credit default swaps and the creation of synthetic collateralized debt obligations, or CDOs.14 In particular, I criticized the reliance on crude mathematical models that did not adequately account for the correlation of CDO assets.15 This was all during the early- and mid-2000s, before 2007.16Then came the global financial crisis of 2007-08. The crisis occurred when it became apparent that major financial institutions had used complex and opaque transactions to take on substantial undisclosed exposure to subprime mortgage markets. …