July 28, 2003

Mr. Jonathan G. Katz, Secretary
Securities and Exchange Commission
450 Fifth Street, N.W.
Washington, DC 20549-0609

Re: Concept Release: Rating Agencies and the Use of Credit Ratings Under the Federal Securities Laws, Release Nos. 33-8236, 34-47972, IC-26066, File No. S7-12-03 (Jun. 4, 2003)

Dear Mr. Katz:

Below are my comments on the Concept Release. I am a law professor at the University of San Diego, and have written several articles on the use of credit ratings in regulation, including: The Paradox of Credit Ratings, in The Role of Credit Reporting Systems in the International Economy (Kluwer Academic Publishers 2002, Richard M. Levitch, Giovanni Majnoni, and Carmen Reinhart, eds.) and The Siskel and Ebert of Financial Markets: Two Thumbs Down for the Credit Rating Agencies, 77 Washington University Law Quarterly 619 (1999), reprinted at 33 Securities Law Review 161 (2001). Instead of repeating arguments I previously have made, I simply list those two articles for your reference. They are largely responsive to the first set of questions in the Concept Release.

I believe this Concept Release is an important step in the Commission's review of the role of credit ratings in regulation. Section 702(b) of the Sarbanes-Oxley Act of 2002 directed the Commission to address these issues, and both the hearings on credit rating agencies and the Commission's January 2003 Report on the Role and Function of Credit Rating Agencies in the Operation of the Securities Markets were useful responses to this directive. However, the participants in those hearings did not focus on what I regard to be the most important issue related to the use of credit ratings under the federal securities laws: alternatives to the NRSRO designation.

Alternatives to the NRSRO Designation

I was delighted to see that the Concept Release begins with a discussion of alternatives to the NRSRO designation, as well as ten key questions on this topic. I am encouraged to learn that the Commission is quite serious about exploring alternatives to the NRSRO designation, and I believe members of Congress also are impressed by the evident shift in focus in this Concept Release. In particular, I believe the first ten questions in the Concept Release are well put; considered answers to these questions are crucial to the direction of proposed rules.

To further discussion, the Commission staff has identified three possible alternatives to the NRSRO designation for a few significant Commission rules that utilize that concept. To the extent these alternatives draw from market-based measures as alternatives to the NRSRO designation, I believe they deserve careful consideration. I address each alternative in turn.

First, I believe the Commission should allow broker-dealers (subject to certain restrictions, discussed below) to use market-based measures for purposes of determining the capital charges on different grades of debt securities under the Net Capital Rule. The use of NRSRO ratings under Rule 15c3-1 is problematic, and has been so for three decades. Rule 15c3-1 was the first rule to incorporate NRSRO ratings, and the tangle of NRSRO-based rules grew from this one mislaid acorn. Therefore, it seems appropriate for the Commission to begin its consideration of NRSRO alternatives with proposed rules amending the use of credit ratings under Rule 15c3-1.

However, there are risks associated with allowing broker-dealers to use the alternative of internally-developed credit ratings alone. Firewalls between employees at broker-dealers are notoriously unreliable, and the voluminous evidence from recent scandals should make the Commission nervous and skeptical about relying on such firewalls in any rule. The Commission could minimize the opportunities for strategic behavior by broker-dealers by requiring that internally-developed credit ratings be consistent with - or constrained by - market information or market-based measures, such as the rolling average of credit spreads for particular instruments.

Market-based information and market-based credit rating methodologies have the characteristics of public goods. Accordingly, encouraging numerous broker-dealers to develop their own information and methodologies would be inefficient, as each broker-dealer would be needlessly duplicating costs without benefit. A broker-dealer could generate additional value from this exercise only if its ratings were somehow skewed to reduce its net capital requirements, an unacceptable result from the Commission's perspective. Otherwise, all broker-dealers would be better off if the information and methodologies were provided by - or at least constrained by - the Commission; then broker-dealers could share costs while representing credibly to investors that that their ratings were reliable and accurate.

The efficient regime would be for the Commission - with the input of the broker-dealer community and others - to institute a methodology for Rule 15c3-1 purposes. The Commission could set market based (i.e., credit spread) parameters and permit broker-dealers to use internally-developed credit ratings so long as the internal ratings fell within a specified range of these market parameters. Put another way, broker-dealers would not be permitted to use internal ratings if those ratings were inconsistent with market measures. For example, broker-dealers could be permitted to use an internally-generated rating for a particular debt security provided that the 90-day rolling average of that security's credit spread was within a specified range for that rating category (e.g., a range of one percent).

Finally, if the Commission permits broker-dealers to use internally-developed credit ratings in any way, constrained or not, it should require that broker-dealers using such ratings disclose the particular information and methodology used in notes to their financial statements, along with examples sufficient to allow investors to assess a particular broker-dealer's safety and soundness based on the net capital it provides for particular securities. Otherwise, the internally-developed credit ratings will disappear into a "black hole," not unlike the types of information investors would have found relevant in recent scandals. Such lack of transparency was precisely what Congress was seeking to avoid in Sarbanes-Oxley, particularly in Section 702(b).

As a second idea, the Commission has suggested eliminating the objective test from Rule 2a-7 and relying solely on the subjective test. I think this suggestion is an excellent one, with two caveats. First, I would recommend constraining the discretion of investment companies with market-based parameters, in the same way I have just suggested for broker-dealers. In other words, investment companies should be permitted to assess their own securities in a subjective way, so long as the assessments fit within a range of market parameters. If an investment company claimed a particular instrument was of the highest quality, when the market credit spread (i.e., the spread between that instrument's yield to maturity and the yield to maturity of a U.S. government obligation of equivalent maturity and structure) was, say, five percent, the investment company's assessment of quality should be deemed presumptively unreasonable. The Commission should solicit comments about which credit spreads are appropriate for particular categories.

Second, I would recommend requiring that investment companies publicly disclose their methodologies for arriving at subjective determinations and include in the notes to their financial statements a list of several examples of such determinations (e.g., the determinations made for the twenty largest positions held in a particular fund). Having investment companies make subjective determinations of credit quality could create incentives for them to take the optimal level of care, but only if market participants have access to information about these determinations.

Finally, as a third idea, the Commission has suggested allowing registrants to use Form S-3 for certain nonconvertible securities and asset-backed securities. It was not clear to me whether the Commission was somehow linking this suggestion to the NRSRO designation. If the proposal is that the ability to use Form S-3 should not depend on a particular NRSRO designation, then I agree with it, although I would urge the Commission to consider a market-based approach and additional required disclosure in this instance, too.

Objections to Market-Based Alternatives

Opponents of market-based alternatives to NRSROs have argued that such alternatives might be problematic because they either would be too volatile or not accurate. The volatility objection is easily overcome. First, if the securities are more volatile, that volatility should be incorporated into regulation. Indeed, the primary purpose of credit-rating-based regulation is to constrain the investment decisions of regulated entities for safety and soundness (and perhaps other regulatory) purposes. Second, it is a straightforward exercise to reduce the volatility of a market-based measure by lengthening the relevant historical period (e.g., one could employ the closing price each day or instead use a rolling average of any length of time, from a week to a year). If the Commission wishes to preserve the lagging characteristic of NRSRO ratings, it could do so by specifying its lag time of choice. Instead, of relying on the current NRSRO ratings, which lag the market in an indeterminate way, the Commission could precisely select a measure with, say, a 90-day lag.

The second objection, that market-based measures would not be accurate, is perhaps a more serious one. According to this argument, for certain illiquid or new issues, it might be more difficult to assess a market measure, such as a credit spread. Some commenters, particular the NRSROs, even have claimed that NRSRO ratings are more accurate than market measures, a claim that is belied both by abundant evidence that NRSRO ratings generate little informational value, and by the more general logical point that market measures reflect market participants' best assessment of available information, including information available to NRSROs and NRSRO ratings.

Moreover, market participants provide prices and quotations for nearly all debt issues, including illiquid and new issues, as part of their standard business practices. It is a straightforward exercise to obtain valuations for individual securities, whatever their liquidity, and most well-run institutions obtain such valuations periodically for their own reporting and risk management. For new issues, bankers and market participants typically engage in "price talk" to assess the value of a security before it is issued (the same is true even more explicitly in "when issued" markets), and there frequently are comparable securities already traded in the market. Even if it were impossible to obtain the same degree of accuracy for illiquid or new issues as for liquid, seasoned issues, market measures might nevertheless be preferable to NRSRO ratings for regulatory purposes. The point of using market-based measures in regulation is not to create a range of narrow trigger points. Indeed, the NRSRO ranges for regulatory purposes are quite wide (e.g., the difference between an investment grade and a non-investment grade rating is considerable). Similarly, market participants easily could use market-based measures to determine whether a particular security falls into one or the other of such broad categories. But market-based measures also could be used to make finer distinctions for regulatory purposes, something the current NRSRO-based categories cannot do.

Additional Areas of Inquiry

Finally, I believe the Commission should study the use of the NRSRO designation in other areas of regulation, in addition to the federal securities laws. In response to Question 55, I also believe the Commission should require substantial additional disclosure of facts related to credit ratings and the use of the NRSRO designation. The Commission began the use of the NRSRO designation by instituting requirements related to the Net Capital Rule. Since then, the proliferation of NRSRO-based rules, and the abuse of such rules, has affected a range of regulators and registrants. I understand that the decision to begin using the NRSRO designation was made more than thirty years ago, by a different generation of experts at the Commission. Still, I believe the Commission today should claim some responsibility for the proliferation of these NRSRO-based regulations in a range of industries.

As part of the Commission's mandate to assess the role of credit ratings in regulation, the Commission could examine the impact of the NRSRO designation in other areas, particularly in the insurance industry. Much insurance company regulation takes place at the state level, and much is now based on NRSRO ratings. The insurance industry is subject to more dysfunction use of credit ratings in regulation than virtually any other industry. The investment grade/non-investment grade split is especially stark in the insurance industry.

The use of NRSRO ratings in insurance regulation would be an appropriate area for the Commission to consider. The lines between insurance and other areas of the financial markets are increasingly blurred; insurance companies are housed within parent companies the Commission and other federal regulators oversee. I believe the Commission should consider - and can consider as part of its regulatory ambit - whether it could play a role in assessing the use of the NRSRO designation in the insurance industry.

Although I am not submitting detailed comments on the other questions in the Concept Release, I will note that I believe that any other recommendations - including those related to the recognition, examination, and oversight of NRSROs - will not address the pervasive problems related to credit ratings that Congress pointed to in enacting Sarbanes-Oxley. In particular, opening the NRSRO designation to more competition will have little effect: in the recent past, there have been more than three or four NRSROs, yet market failures and dysfunctional behavior related to ratings persisted; in the future, absent changes in industry structure and economics, adding a few NRSROs inevitably would lead to the same kind of industry consolidation and ultimately a reduction in the number of NRSROs to current levels. I believe some of the entities seeking NRSRO designation would provide a superior service to that of the existing NRSROs - indeed, given the abysmal performance of NRSROs, a reasonable person could ask how they could not - but adding a few NRSROs will not resolve the endemic problem of NRSRO-based rules. The Commission began the practice of granting regulatory entitlements to NRSROs and then relying on their ratings in regulation. It should stop that practice, now.

I believe the most useful comments on the Commission's various proposals will come after the Commission has proposed specific rules, and I look forward to reading and commenting on those proposals. I encourage the Commission to examine closely the market-based alternatives to NRSRO designation before issuing proposed rules. If the Commission is serious about market-based alternatives, it could, under the mandate of Sarbanes-Oxley, empanel a group of experts to examine these alternatives and make recommendations before the Commission introduces proposed rules. Such an advisory group would not be part of the rulemaking process, but might be useful to the Commission as it continues down the path of questioning the NRSRO designation.

Whatever the Commission decides regarding specific rules, I am delighted by the recent attention to alternatives to NRSRO-based regulation. I look forward to the Commission's continued progress in this difficult area.


Frank Partnoy