Comments on the Concept Release
"Rating Agencies and the Use of Credit Ratings
under the Federal Securities Laws"
Securities and Exchange Commission
File Number S7-12-03

Cate Long
July 28, 2003

We appreciate an opportunity to forward our comments on the Concept Release. In line with the review of other significant parts of the regulated financial system the review of the Nationally Recognized Statistical Rating Organizations will serve to strengthen and enhance an important part of the US financial markets.

The work of our firm is the development of tools and systems for the visualization of credit risk and return for retail investors, registered representatives and financial advisors. These tools and systems transform the complex and varied data of the fixed income markets into simple charts and graphs.

Due to the focus of our work we wanted to provide our views to the Commission related to the effect of proposals within the Concept release on the retail market for fixed income. The majority of our comments will not relate to regulatory issues before the Commission but rather provide a view on the use and limitations of ratings within the retail marketplace and the value of the NRSRO designation for retail investors.

To provide a quantitative framework for our comments we have included the latest data from the "Survey of Consumer Finance" published triennially by the Federal Reserve. This data, current for 2001, outlines retail ownership of fixed income securities.

Household Income Percentile

Owning Bonds

40-60 %

2 %


4 %


4 %


13 %

Proportionally, fixed income securities are held by a small number of American households relative to other financial assets. Within the highest income bracket only 13% of households own fixed income securities. Relative to the institutional fixed income market the retail market has limited access to information flow, market data and sophisticated analytics. Rather the retail market has come to rely in a significant way on the ratings systems maintained by NRSROs. Our comments will respond to the Concept Release from this perspective.

We see the role of the Commission relating primarily to the creation of requirements for designation as NRSROs, approving a rating agency as a NRSRO, and monitoring and disclosing the ongoing adherence to the requirements for designation as a NRSRO.

For designation requirements that could relate to staff proficiency, financial solvency and statistical data for default statistics we support the use of disclosure by the NRSRO rather than specific regulatory requirements from the Commission. In the area of "conflict of interests" we also prefer the use of disclosure rather than regulatory constraints on rating agencies activities. This view extends the general regulatory approach of the Commission.

The use of quantitative statistical modeling or spreads on credit default swaps would not currently serve as a viable substitute for ratings issued by NRSROs for the retail market. There is limited access to these systems and data and no viable systems exist to present the data to retail investors in a meaningful way.

The rating systems of the NRSROs are useful to investors due to their ability to signal the relative creditworthiness of individual issuers and securities. These systems have particular value to retail investors due to the "long-term" nature of the rating. The assignment of a rating is meant to reflect the ability of the issuer to make principal and coupon payments through all parts of the credit cycle. This stability of outlook is useful for retail investors, as they tend to hold securities to maturity and generally have relatively illiquid markets with which to trade fixed income.

In contrast, some quantitative modeling methods can be relatively volatile in their ability to signal risk. They tend to respond to short-term fluctuations in other capital markets such as the equity markets for those systems derived from the Merton model. This "signal volatility" poses particular hurdles for the retail investing community, as they generally do not have a liquid market with which they can trade fixed income securities. If a quantitative model signals a big spike in default probability due to a sell off in the equity of the issuer, the retail investor interpreting that signal as a "sell" might not find any liquidity for the security and be faced with selling at a significant discount. As the equity market for the issuer improves the quantitative default probability for the issuer declines. This situation could be seen for Ford Motor Company this year. Ford is a widely held retail credit. The qualitative or traditional, fundamental analysis continued to give a relatively steady signal to investors about the long-term health of Ford. This steadiness was beneficial for retail investors.

Responses to questions: (Many questions do not have implications for the retail fixed income market. Those questions will not appear.)

1. We believe that the Commission should continue the designation and oversight of NRSROs.

11. Yes, the criteria currently used by the Commission staff to determine whether a rating agency qualifies as a NRSRO is appropriate. The high standards imposed by the Commission are particularly useful for the retail market as the designation serves as a "mark" of quality. This designation signals that proscribed quality standards for the rating agency are being meet.

12. It is generally appropriate for the Commission to seek attestations from the predominant users of securities ratings relative to the designation of NRSROs. Although this could prove a barrier to access for a rating agency that intended to solely rate securities that were widely held within the retail fixed income marketplace (i.e. municipal securities).

13. It is very beneficial for the Commission to condition NRSRO recognition on a rating agency developing and implementing procedures designed to ensure credible and current ratings. We believe though that disclosure of these procedures and the NRSRO's adherence to them is more beneficial than the Commission establishing specific guidelines. For example, the Commission could require that NRSROs report on the number of firms covered by each analyst, the frequency of their formal review, the number of new issues rated on a quarterly and/or annual basis, the number of firms added or removed from "Creditwatch" or other similar watch list, the number of "withdrawn" ratings and the number of unsolicited ratings. These and other quantitative performance indicators, primarily default statistics, could be publicly disseminated, by the Commission, in a manner similar to "EDGAR" for corporate financial filings. This would enhance transparency of the ratings process while allowing rating agencies to operate their business in the most efficient manner given their focus and resources.

The proposal that rating agencies be required to adopt uniform rating symbols would be an advantage for the retail fixed income market. The work of Multiple-Markets is centered on this effort of simplifying and visualizing the symbology of the various rating agencies and designing systems that make fixed income information useful to the retail investor, registered representatives and financial advisors.

As for the rating agency's procedures in obtaining information on which to base rating actions we would propose that agencies be required to disclose their general procedures but that the Commission or other entity not proscribe minimum due diligence requirements.

15. If a rating is assigned through the use of computerized statistical model it does not reflect a broader context within which a rating is applied such as the condition and outlook for the industry of the issuer or the general interest rate environment. For the retail investor having the industry or interest outlook factored into the rating is a benefit. We would not recommend exempting a rating agency from designation that uses statistical methods but would recommend disclosure of the method.

16. We would not recommend that the Commission set minimum staff size and quality requirements. We would recommend that the staff qualifications and size be disclosed on an annual basis. This would allow retail investors to understand the relative size and background of the staff of rating agencies.

17. We would recommend that the Commission require disclosure of the number of analysts and the average number of firms covered by the analysts. We would also recommend the general disclosure of the qualifications of the agencies analysts.

19. It is useful for retail investors to have some understanding of the financial independence and stability of an NRSRO. Disclosure of the resources and concentration of revenue for issuers would be useful.

20. It could be useful if there were NRSROs that specialized in rating either municipal issues or retail securities that are issued in $ 25 denominations for example.

22. The NRSROs provide signals to the market in two forms. First is the relative creditworthiness of a security as indicated by its rating. This could be thought of as a qualitative signal. The second is the general quantitative performance as shown by the annual cumulative default rate for that particular rating. If the Commission where to develop supplemental criteria to evaluate traditional fundamental analysis and analysis that relies on statistical models, the most useful criteria would be the general quantitative outcome of a particular rating or signal.

38. The model adopted on the equity side of the market regarding "conflict of interest" for analysts might be useful for analysts of rating agencies. This model generally embraces disclosure rather than regulatory constraint. The retail market generally perceives rating agencies analysts to be "conflict free".

41. Subscribers to rating agencies are almost entirely institutions. It seems useful that they have access to analysts concerning a general view of an issuer. It could be detrimental if they have advance knowledge of the exact timing of a rating change prior to the general investing community. Generally, as shown by the data from the Survey of Consumer Finance cited earlier in this document, the retail investing community does not broadly own fixed income securities and is not currently accessing information related to ratings.

44. We believe that general disclosure, through a system similar to EDGAR, would address in the most useful way conflict of interest for rating agencies.

49. There are several areas that we would recommend the Commission consider for review that might enhance the information flow for retail investors. It would be useful, for example, for NRSROs to include the CUSIP identifier or other agreed upon market data identifier when they publicly disseminate information relative to a rating or ratings change. The equity markets have stock symbols that allow investors to quickly gather information relative to individual securities. The inclusion of a security identifier would be beneficial.

It would also be useful for NRSROs to have a standard method of reporting default statistics. For example, although rated municipal securities utilize the same rating symbology as rated corporate securities the default statistics vary between the two asset classes.

Generally we would recommend that the Commission consider the disclosure of default statistics and other requirements be done through an online system such as EDGAR. This would make access to the information much easier and clearer for retail investors.

50. We would recommend that the Commission require the disclosure of the key bases and assumptions in a general way that highlights the type of analysis (statistical modeling or fundamental, qualitative). We do not feel that it is necessary to require disclosure of the specific assumptions made on a particular issuer.

51. We believe that the most useful performance data for retail investors would be the disclosure of annual cumulative default rates for a rating agency. This provides a quantitative signal to the investor of the risk of owning a security with a specific rating. Again we would recommend that these disclosures be accessible on a system similar to EDGAR.

54. We would recommend that the Commission require public notification when an NRSRO ceases following an issuer. We would encourage the Commission to require the inclusion of a CUSIP number or other market data identifier.

We again thank the Commission for the opportunity to forward our comments.