April 18, 2002

Jonathan G. Katz
United States Securities and Exchange Commission
450 Fifth Street, N.W.
Washington, D.C. 20549-0609

Re: National Association of Securities Dealers and The New York Stock Exchange Proposals Relating to Research Analyst Conflicts of Interest, Release No. 34-45526; File Nos. SR-NASD-2002-21;SR-NYSE-2002-09.

Dear Mr. Katz:

We appreciate the opportunity to comment on the above-referenced proposals.

At the outset, it is important to acknowledge that there has been a loss of confidence in the independence of research analysts and that restoration of that confidence is necessary to insure the continued vitality of our capital markets. The SIA's Best Practices were an important first step. We recognize, however, that more is needed. Therefore, we underscore our unqualified support both for the overall intent of the proposals, as well as most of the specific provisions. We also would note that regardless of the final form of the proposals, we are continuing to evaluate our existing policies and procedures in this area with a view toward identifying and voluntarily implementing any appropriate changes that we believe will help to assure our investing clients of Goldman Sachs' commitment to the quality, utility and integrity of our research product. Finally, in assessing the above-referenced proposals and what types of changes may be needed, we should all be mindful of the fact that the U.S. capital markets are the envy of the world - great care should be taken to insure that any mandated modifications represent well thought out improvement, not merely change, which may carry with it unintended and adverse consequences for our financial system.

Before commenting on the specific provisions, we have three general points. First, the two proposals are not identical, and they should be. There is no reason why the operative language of the proposals should vary, thereby creating additional compliance burdens, unnecessary costs and confusion.

Second, while some of the provisions can be implemented very quickly, others would require the expenditure of many millions of dollars and likely require a very significant time period to implement fully. The new systems required to comply with the compensation disclosure provisions of the proposals would be by far the most expensive and time-consuming to design and build. The "chaperoning" and record retention aspects of the provisions relating to the interaction between Research and IBD and Research and the subject company will also require building new systems and hiring and training many new Compliance officers. Building systems to deliver "real-time" information about firm ownership of stock of subject companies for inclusion in reports and disclosure by research analysts as a part of public appearances similarly will require considerable time and expense, as will modifying or building systems to produce the required charts and other statistical disclosures and then integrating them into research reports. We direct you to the Securities Industry Association's comment letter for a discussion of the minimum time periods that the industry believes will be necessary to comply with the proposed rules, if adopted as currently written.

Third, after the proposals have been revised, we would strongly urge the SEC to republish them, with an abbreviated final period for comment. These rules are significant and will be costly and time consuming to put into effect -- every effort should be made to get them "right." They also are complicated and have been drafted in a relatively short period of time, and a number of interrelated amendments no doubt will be made to deal with the comments being raised from a variety of sources. A final review is likely worthwhile.

1. Disclosure by firms of compensation from the subject company and other material conflicts.

The proposals would require firms to disclose if they or their affiliates have received compensation from the subject company during the twelve months prior to the publication of a research report or if they reasonably expect to do so over the next three months. The NASD proposal uses "or;" the NYSE proposal says that the firm "must also disclose." The NASD formulation would not appear to require firms to specify whether compensation has been received or it is expected that it will be received. Both of the proposals raise a series of issues.

Informing an investor that the firm that publishes a research report or its affiliates may have earned from the subject company at least one dollar somewhere in the world in the last year does nothing to identify or quantify potential conflicts. Moreover, a firm may have done no business with a subject company, ever, but be actively soliciting such business, so the fact that no compensation has been received or is expected to be received in the next few months may not be evidence that no potential conflict exists.

Even the limited disclosure that is currently proposed has the potential for "tipping" savvy market professionals, particularly for firms that are smaller and less diversified. The SROs recognized this risk and attempted to mitigate it by requiring disclosure from anywhere in the firm or from any affiliates worldwide. The risk still exists, particularly in the NYSE proposal, if disclosure about expected compensation is broken out. Arbitragers and other market speculators will be closely watching for any changes in this disclosure. Ordinary investors will derive little or no value, while other market professionals may receive an unintended and entirely undesirable benefit.

For the same reasons that such disclosure may tip market professionals outside of a firm, it may also serve to tip internal constituencies that should otherwise not have access to such information by virtue of the Chinese Wall.

As discussed above, gathering this data on a real-time basis would be very expensive. No firm currently has a reason to collect it on a firm-wide basis and consequently we are aware of no systems which provide it. Tracking ownership of stocks of subject companies, which is mandated under Section 13 of the Securities Exchange Act of 1934, is significantly less difficult because each issuer's common stock has a distinct security ID number (CUSIP number).

We believe that the proposed disclosure set forth below is a better way to inform investors of the risk of potential conflicts. It would be just as effective, less likely to tip, less potentially misleading and vastly less expensive to implement. We would suggest that each research report very prominently (i.e., through placement, typeface and size) contain statements of the following type:

The proposals also would require an analyst to disclose in a research report or a public appearance "any other material conflict" of the firm if the analyst knows or has reason to know of such conflict. First, we do not know what "any other" refers to. Does it include the disclosure of compensation or ownership that is otherwise required to be disclosed? Even more important, in certain circumstances there is again a very serious potential for tipping associated with such proposal which we think outweighs any benefits with respect to disclosing such conflict, particularly in light of the alternatives described above.

2. The definition of "research report."

The proposals are intended to restore public confidence in the recommendations of equity research analysts. As drafted, however, the two proposals are different and unclear, particularly the NYSE proposal which grafts the new proposals into existing NYSE Rule 472. We understand that the NYSE proposal, like the NASD proposal, was intended to exclude research on fixed income securities. However, the definition has the potential to sweep in many forms of communication that should not be the subject of the proposals. The following types of communication should be excluded from the definition of research report:

In addition, while this comment does not strictly speaking relate to the definition of research report in the proposals, to the extent that certain disclosures are required to be made in research reports, such disclosures should be permitted (where appropriate) to be hyperlinked to the report, if it is in electronic form, or incorporated by reference, if it is in paper form. This is especially true in the case of "research reports" that take the form of short "notes" or "comments," as opposed to full-fledged reports. Thus, disclosures about, for example, the distribution of ratings should be hyperlinked to, or the URL or a telephone number should be provided to obtain, the information if the report is in hard copy. The rules should recognize the growing movement towards "access equals delivery" and should only require the inclusion in written reports of the most meaningful and essential disclosures.

3. The imposition of "quiet periods."

We do not understand the point of these proposals in the context of analyst conflicts of interest. The fact that a firm has recently underwritten the subject company's common stock is already fully disclosed. We are fully supportive of making that disclosure even more prominent if that would be helpful. Adding 15 days to the quiet period that already effectively exists under the Securities Act of 1933 rules in the case of an IPO and creating a 10-day quiet period for all follow-on offerings, will not limit the potential for conflicts. Nor will such periods eliminate the potential for post-IPO "hype." The proposals do not apply to public appearances by analysts nor do they apply to buy-side analysts or any member of the financial media.

These provisions will disadvantage retail investors who generally rely solely on published research and generally have no direct interaction with analysts. In cross-border offerings, foreign investors receive research before, during and after the offering. As a result of this provision, U.S. retail investors will be further disadvantaged. In addition, these provisions are inconsistent with the SEC's Rules 138 and 139 under the Securities Act and the general movement towards permitting more "free-writing" in connection with offerings.

The quiet period proposals only apply to the lead managers in an offering, which is a term of art not a legal distinction, and therefore have the potential of distorting or being manipulated in connection with underwriting syndicate arrangements. If these proposals are adopted, it might be preferable to apply them to the entire syndicate, at least in the context of IPO's.

We believe that the quiet period proposals should be eliminated for the reasons described above. If these provisions are retained for IPOs, they should certainly be eliminated for follow-on offerings. If retained for follow-on offerings they should not apply to shelf issuers or to issuers that are exempt from Regulation M. In no event should these provisions apply to offerings made solely off of a shelf registration statement by sellers other than the issuer.

4. Disclosure by firms of their ownership in the subject company.

The disclosure requirements with respect to the firm's ownership in a subject company should be modified in one of three ways:

The current proposal raises several issues.

First, it creates a selective disclosure regime in that investors who receive firm research reports obtain disclosure on a more current basis and based on a lower disclosure threshold than the general public.

Second, because the information includes firm merchant banking, venture capital, arbitrage, asset management, specialist, market making, underwriting and client discretionary account positions and only tracks long positions, the information is almost completely meaningless to a reader of research reports. Such positions are often hedged and may be extremely volatile. To the extent that they do provide any meaningful disclosure on occasion it will be the disclosure of proprietary information to market professionals who could use the information in a manner that is unintended and undesirable. Of course, one percent of one subject company may be material to a firm, while one percent of another subject company may be totally immaterial to the same firm.

Third, updating this information on a rolling five-day basis so that is available for all research reports and public appearances by analysts, while building on existing systems used for Section 13 reporting, will require extensive and expensive enhancements to those systems to provide the data within the time frames required and to the individuals that require the data on relatively short notice.

In light of these issues, we believe that any one of the three alternatives set forth above would be preferable. Each would provide as meaningful information. The first two would avoid selective disclosure issues and all three would avoid significant expenditures for very limited to no benefit to the readers of our research reports.

5. Certain aspects of the provisions relating to the interaction between research analysts and investment bankers and research analysts and the subject company.

The provisions that relate to the interaction between research analysts and investment bankers go to the heart of the public perception issues with respect to analyst independence and we only have some clarifying comments to make here.

The phrase "communication concerning [such] research report" should be clarified with respect to two points:

Similarly, the provisions that concern the interaction between the research analyst and the subject company are also central to the proposals and we again only have some minor clarifying points:

6. The definition of "immediate household" with respect to limitations on research analyst ownership of stock in subject companies.

The current definition would include employees who share a residence but are not financially dependent on one another as that concept is commonly applied in the context of other financial services regulation, for example roommates just out of college sharing an apartment and working at two different firms. This cannot be the right answer. We recognize that this is a difficult definition to draft; however, presumably the definition should include some concept of living at the same address coupled with financial dependence.

7. The requirements involving investment by research analysts in unregistered funds.

The requirements for permitting investments by analysts in unregistered funds are unworkable and unnecessary. There is no practical way to track the 1% and 20% requirements once the initial investment is made or to stop distributions in kind of subject company shares. The provision as drafted would in effect be a prohibition on such investments. This is unnecessary. There is already a requirement that the analyst be a passive investor. The investment should simply meet the 1% test at the time the investment is made and the fund should not be targeted at the analyst's coverage sector. In addition, there could be a general prohibition on any arrangement or scheme to circumvent these requirements on an ongoing basis or to cause an in kind distribution of the subject company shares.

8. The definition of "clients" for purposes of disclosure by research analysts making public appearances.

The provision requires disclosure if the analyst know or has reason to know that the subject company is a client of the member or its affiliates. There is no definition of "client." It is not limited to a client that has engaged in public transactions with the firm and therefore has the same potential for tipping as the "other material conflict" disclosure requirement discussed above. At a minimum, the definition of client should be narrowed to pick up subject companies that have received publicly disclosed investment banking or financial advisory services from the firm during the past twelve months. This would tie this requirement to our proposed revision to the firm compensation disclosure requirements and deal with both the tipping issue and the cost and confusion associated with different tests for determining if the firm has a relationship with subject companies depending on the context in which the disclosure is being made.

9. The definitions of "investment banking services" and "investment banking department."

The definition of "investment banking services" and "investment banking department" also need to be clarified. In fact, the definition of "investment banking services" is not really a definition at all, merely a non-exclusive list of services. "Investment banking services" should be defined to be underwriting or financial advisory services provided by the firm to corporations or partnerships (not governmental entities) and "investment banking department" should be a group (which may be only a sub-set of a particular division or department) that provides such services to corporations and partnerships with publicly traded common equity securities (thus excluding municipal and mortgage finance departments, for example).

* * * *

We again wish to commend the SROs and the SEC for their exhaustive efforts to restore confidence in the integrity of investment research and support both the broad objectives of the proposals, as well as most of the specific provisions. We believe, however, that our suggested modifications would strike a more appropriate balance between the costs and benefits associated with the proposals without diminishing in any way the positive impact that they will have on restoring public trust in research analysts.

If you have any questions with respect to any aspect of this letter, please contact either Kenneth L. Josselyn (212-902-3761) or me.

Very truly yours,

Gregory K. Palm