November 8, 2002
STATEMENT OF STEVEN L. SCHWARCZ, PROFESSOR OF LAW, DUKE UNIVERSITY
Re: File No. 4-467
To: U.S. Securities and Exchange Commission, att: Jonathan G. Katz, Secretary
This statement is submitted in response to the November 1, 2002 request from Annette L. Nazareth, Director, and relates to the SEC's scheduled November 21, 2002 Hearing on Credit Rating Agencies. As a supplement to and background for this statement, the SEC and other interested parties may wish to also review my recent comprehensive article on the subject of rating agency regulation, "Private Ordering of Public Markets: The Rating Agency Paradox," published in Volume 2002, Issue No. 1, pp. 1-27, of the University of Illinois Law Review.
There are two major conceptual issues facing rating agencies today: (1) are rating agencies still relevant to today's capital markets; (2) if so, should rating agencies be regulated? The answers to these questions are intertwined.
Consider the issue of regulation. To analyze regulation, one must first understand its normative rationale. In an economic context where health and safety are not at issue, this rationale is fostering improvements in efficiency.1 An exception might arise, however, where society has objectives in addition to economic efficiency.
Are there any such other objectives in a rating agency context? This is not merely a rhetorical question; even commercial regulation might have other objectives based on indirect social consequences. Consider, for example, the dispute over the proper goals of bankruptcy reorganization law. Some argue that the only goal should be economic efficiency; others argue that there are also distributional objectives, such as rehabilitating troubled debtors and ensuring equality of distribution to creditors.2
One can imagine this same type of dispute over the goals of rating agency regulation. Ratings affect a company's ability to raise funds and the cost thereof, which in turn can affect the company's ability to hire and retain employees. There are, however, cogent reasons why economic efficiency alone should prevail as the standard.
The regulatory scheme most analogous to rating agency regulation-securities law-focuses primarily on the goal of economic efficiency in lieu of distributional objectives. Rating agencies perform the same function as securities law: reducing the information asymmetry between issuers of securities and investors.3 Thus, economic efficiency alone should be the standard.
There are two ways that regulation could increase efficiency: (A) improving rating-agency performance; (B) limiting negative consequences of rating-agency actions. Consider each in turn.
A. Regulation to Improve Performance
Regulation could improve performance only by reducing overall costs or by improving ratings reliability.
Presently, there is little reason to believe that rating agency costs are excessive. The fee charged by a rating agency typically is market-driven.4 Nonetheless, I later argue that increased NRSRO (nationally recognized statistical rating organization) designation to other entities would increase competition. That, in turn, might drive down fee-based rating agency costs. Otherwise, I do not see how regulation could reduce costs, and indeed government regulation often increases costs.5
Could regulation improve ratings reliability? Rating agencies presently have strong incentives to issue accurate ratings. Inaccurate ratings will impair, if not destroy, their reputation. Even more than for accountants and lawyers, rating agencies must trade on their reputations. If, for example, bond investors lose faith in the integrity of rating agencies' judgments, they will no longer pay attention to their ratings; if rating agencies' opinions cease to affect the price that borrowers pay for capital, issuers will not pay their fees. So market forces should make rating agencies careful of their good names.6
Nonetheless, the economic downturn has reduced the perceived reliability of ratings by highlighting their relative instability. Stability, I think, is an important key to continued rating agency relevancy. Rating agencies therefore should be careful to examine the reasons for any downgrading, and should be especially sensitive to defaults on investment grade debt.
To some extent, however, instability may be inherent in the way ratings are presently interpreted. Say, for example, that a AAA rating is assigned to bonds issued at a given time. That rating may not be intended to be a predictor of the safety and soundness of those bonds five years hence. Nonetheless, so long as the bonds are rated AAA, which they might be four years hence, investors will assume there is still long-term ratings stability.
Therefore, why not consider giving each rating a set life, and update the rating prior to the end of that life (unless, of course, there needs to be an unanticipated prior ratings downgrade). For example, if a AAA rating with a three-year life were assigned to a particular bond on January 1, 2002, investors buying that bond in April 2004 would understand that the rating is not intended to be an opinion on the safety and soundness of the bond after January 2005. This would give investors greater certainty of what the rating means at any given time.7
There is also a perceived problem with ratings reliability. To some extent, ratings are seen as a rational counter-force to market sentiment. Rating agencies are thus conservative, and arguably reasonably so, in downgrading ratings because downgrading the rating of bonds of an otherwise healthy company could become a self-fulfilling prophecy. This can be analogized to the American approach to criminal law, in which it is better to let ten guilty persons go free than convict one innocent person. Nonetheless, a rating agency's failure to downgrade, like the failure to do so in Enron, can significantly impair rating agency credibility. This suggests that rating agencies and government should consider to whom rating agencies should have primary responsibility: to issuers and existing investors, on the one hand, who want topreserve the existing rating and thus favor the conservative approach; or to potential new investors, who want the most up-to-date market information and thus would favor more hair-trigger downgrades?
The foregoing distinction also helps to highlight the difference between ratings, which are intended to be conservatively stable, and more market-based tests such as credit spreads, which may be more hair trigger. Which is "better" is a question of judgment, not of numbers.
A further perceived problem with ratings reliability is the problem of fraud. Ratings do not cover the risk of fraud, and indeed rating agencies make their rating determinations based primarily on information provided by the issuer.8 It does not appear to be economically viable for rating agencies to act as guarantors of fraud. Nonetheless, this creates a public outcry in cases, like Enron, where the rating agencies may have received incomplete or misleading financial information from the company whose securities are to be rated. In computer talk, this failure is called "garbage in, garbage out": recognizing that even the finest computers will produce bad results if they start with bad input.
It is unclear how further regulation can help the fraud problem. Companies, after all, are already required to produce accurate financial information, and fraudulent corporate officers can be fined and go to jail. Rather, I believe the solution is one of public education because, in my experience, few investors even recognize the existence of the fraud exception to ratings even though rating agencies publicly disclose it.
B. Regulation to Limit Negative Consequences
There are various negatives associated with rating agency actions. First is the perception - related to a central question of private ordering, the extent to which it undercuts democratic authority9 - that rating agencies are not accountable because they are not officially subject to public scrutiny. I nonetheless have argued that rating agencies, like administrative agency officials, derive democratic legitimacy through their reputational constraints.10
A second potential negative is the conflict of interest inherent in the way that rating agencies are paid. They are virtually always paid their fee by the issuer of securities applying for the rating.11 This raises the possibility that the issuer will use, or the rating agency will perceive, monetary pressure to improve the rating. Nonetheless, there appears to be little alternative to this arrangement because of the collective action problem in coordinating potential investors to pay this fee.12
The lack of alternatives does not, however, eliminate the potential conflict of interest, that an issuer might be tempted to use the fee to strategically bargain for a higher rating. It is therefore important that the amount of the fee be independent of the rating, as it presently is.13 Regulation still might be appropriate, however, to ensure that ratings remain independent of the fee.
In discussing rating agency compensation, it is useful to also reflect briefly on why rating agencies are, or are not, different from accounting firms and the temptations that, for example, may have misled certain Arthur Andersen auditors. One difference is that rating agencies, unlike accounting firms, have no disproportionately large concentrations of customer fees, which might influence ratings decisions. A second difference is that, unlike accounting firms, rating agencies do not have internal conflicts of interest based on fees - such as combining both auditing and consulting businesses with the same entity. This does not, of course, mean that rating agencies will never have such conflicting businesses. It therefore may well be useful to examine whether regulation could restrict any such conflicts from developing.
Another potential negative, which Moody's has been accused of, is rating without request, or misbehaving by issuing artificially low unsolicited ratings in private transactions.14 The rationale is that this forces issuers to pay for Moody's services in the hopes of getting a better rating.
To the extent such misbehavior has actually occurred,15 targeted remedies, such as requiring disclosure of the fact that a rating is unsolicited, would appear appropriate. Recently, in fact, Moody's voluntarily instituted this disclosure for certain unsolicited ratings in recognition that market participants have shown an interest in knowing which ratings lack the issuer's participation, and to increase the credibility andutility of Moody's ratings in the capital markets.16 Reputational costs therefore appear to have been sufficient to correct this alleged misbehavior.
The final negative is that rating agencies are presently conservatively biased against innovation. This is because the negative reputational consequences of providing a rating that, in retrospect, turns out to be incorrect far outweighs the fee a rating agency can charge for providing that rating.17 Regulation might help solve this by increasing the number of NRSRO rating agencies and thereby increasing competition.18 I therefore next discuss the issue of NRSRO designation.
C. NRSRO Designation
The theory of NRSRO designation is that some form of regulatory approval is appropriate so long as the applicability of law, as it does, turns on ratings.19 The concern, however, is that only three ratings agencies, Standard & Poor's, Moody's, and Fitch, are presently designated NRSROs.20 This suggests that the government should consider balancing the need for a rigorous standard for NRSRO designation against the need to ensure that a sufficient number of rating agencies receive NRSRO designation to assure competition.
My recommendation is to provisionally award NRSRO-designation to foreign-recognized rating agencies, and also to subsidiaries of high-reputation U.S. firms active in evaluating business and securities of companies.21 It is important, however, to ensure that any such newly designated NRSRO be of the highest standards, in order to avoid the perverse and unintended result of encouraging ratings shopping. There should be relatively little risk if these entities are well-capitalized, have reputations for quality financialanalysis in the investment community, and have acceptable business plans to rate securities.22 The risk could be further minimized by making any de novo applicant's NRSRO-status provisional for some trial period. In this way, the potential anti-competitive effect of NRSRO designation can, consistent with the integrity of that designation, be reduced.
Steven L. Schwarcz