March 2, 2004
Mr. Jonathan G. Katz
U.S. Securities and Exchange Commission
450 5th Street, N.W.
Washington, DC 20549-0609

Re: Concept Release on Measures to Improve Disclosure of Mutual Fund Transaction Costs (the "Concept Release"), Release No. 33-8349; 34-48952; IC-26313; File No. S7-29-03.

Dear Mr. Katz:

Fidelity Investments applauds the Commission for issuing its Concept Release and seeking public input on how mutual funds can improve their disclosure of transaction costs associated with the management of their investment portfolios. We welcome the opportunity to share our views, which can be summarized as follows:

  • The Commission should require mutual funds to quantify and separately report soft dollar research expenses in dollars and as a percent of assets. This data should be reported in or near the expense table. The Commission should consider whether these expenses should be included within funds' expense ratios as well, but it may be preferable to defer inclusion in the expense ratio to allow the Commission an opportunity to evaluate the comparability of this data across different funds and fund complexes.

  • The Commission should require comparable disclosure of soft dollar research expenditures by investment managers of other types of managed accounts and should coordinate with other regulatory agencies where necessary.

  • The Commission should require broker-dealers to assign a value to proprietary soft dollar research they provide to advisers in connection with the execution of portfolio trades, thereby assisting advisers in carrying out their duties to account for soft dollar research expenditures. The Commission should consider whether to prescribe a methodology that broker-dealers must follow in computing the value of proprietary soft dollar research.

  • The Commission should enhance disclosure for mutual fund investors by requiring funds to set forth total commissions as a percentage of net assets in their prospectuses, but outside of the fee table and expense ratio.

  • The Commission should not require funds to disclose other measures of transaction costs other than brokerage commissions. Methodologies that focus upon market impact, opportunity costs or other measurements may be useful to advisers as management tools and may also be relevant to funds' boards of directors in evaluating overall portfolio management performance of fund advisers. These methodologies, however, have distinct limitations, differ widely in their approach, and can produce significantly inconsistent results when applied to the same trading data. The Commission may wish to revisit the utility of these methodologies in the future, but at this time the Commission should refrain from mandating the use of any methodology or the disclosure of any results produced when an adviser chooses to use any particular methodology.

  • The Commission should adopt rules to require that exchange specialists disclose data, with an appropriate lag period, identifying their proprietary trading in the securities for which they have been granted an exclusive right to engage in dealing on the floor of the exchange. This will enable investment managers and the investing public to evaluate the impact of a specialist's proprietary trading on the market for a particular security and on the ability of investors' orders to meet without the interpositioning of specialists trading for their own account.

We emphasize that soft dollar disclosure reforms should apply to all investment managers who use brokerage commissions to acquire investment research, because all investors will benefit from greater disclosure.

In the following discussion, we (i) discuss our recommendations regarding disclosure of soft dollar research expenses, (ii) respond to specific questions posed in the Release and (iii) address the need for disclosure of specialists' trading activities.

I. Disclosure of Soft Dollar Research Expenses

Fidelity believes that the time has come for the Commission to introduce dramatic change in disclosure rules relating to soft dollars. Investors, including mutual fund investors, will be better served and better informed if the Commission requires investment managers to disclose the amounts of brokerage commissions they pay for investment research in reliance upon the safe harbor of Section 28(e) (the "Safe Harbor") of the Securities and Exchange Act of 1934 (the "1934 Act"). We urge the Commission to use its powers to adopt such rules for mutual funds to require disclosure in dollar terms and as a percent of funds' net assets.

Our litmus test for evaluating the need for disclosure rules is whether the information to be produced will be useful to investors in making informed investment choices, evaluating the economic performance of their investments, or understanding the costs of an investment, including any mutual fund. Rules to require disclosure of soft dollar research costs pass this test.

In urging the Commission to act, we are confident not only that investors will benefit, but that the mutual fund industry will emerge better able to serve its investors. Notwithstanding unsettling recent events, Fidelity firmly believes that mutual funds, having withstood the tests of the marketplace for over 60 years, will continue to serve investors well in the future. This is so because of their intrinsic advantages -- professional investment management, diversification, liquidity and reasonable expenses, a sound regulatory scheme and robust competition.

We understand that some may question the need for disclosure of soft dollar research expenses in light of the extensive disclosure of expenses already required of mutual funds, including the Commission's new rules requiring actual dollar expenses incurred by shareholders.1 The answer is that soft dollar research expenses are the least transparent of fund expenses, because they are "bundled" with payments for a distinctly separate service -- the execution of securities trades -- in brokerage commissions paid by investment managers to brokers. As noted by the Commission in 1995:

"Soft dollar practices ... diminish the ability of a client to evaluate the expenses it incurs in obtaining portfolio management services and may hinder the ability of the client to negotiate fee agreements, because the costs of soft dollar services are "hidden" from investors in brokerage commissions." 2

New disclosure rules are needed to promote "unbundled" disclosure of these bundled commissions. We note that Congress clearly contemplated that special disclosure rules may be called for when it adopted the Safe Harbor in 1975.3

Industry-wide disclosure of soft dollar expenses will not only enhance individual investors' understanding of their mutual funds' expense structures and performance. It will also help the Commission, Congress, the industry and others in their extensive efforts to identify and analyze trends in fund expenses.

We believe that it is of crucial importance that the Commission draw no distinction between research prepared by an executing broker (so-called "proprietary" research) and research by a broker uninvolved in the execution of the fund's trades (so-called "third party" research). Requiring disclosure of only third party research expenses would not advance sound public policy. It would discriminate against independent research. It would also provide unhealthy incentives for advisers to send trades to brokers for reasons other than achieving best execution for fund trades.4

This, of course, is not the first time that the SEC has been asked to discriminate against third party research. In 1995, the SEC considered arguments that would have favored proprietary research and rightly rejected them.5 The most recent proposal to favor proprietary research has now been made by the Investment Company Institute. The ICI contends, as advocates did ten years ago, that valuation of research provided by brokers that are involved in the execution of fund trades (sometimes only in a nominal way) is too elusive and uncertain. The ICI therefore urges a partial repeal of the Safe Harbor, preserving its protections only for proprietary research.

We disagree with assumptions that reasonable, reliable and comparable estimates of the value of proprietary research cannot be made. To begin, we note that an investment manager is entitled to pay a brokerage commission to acquire research in reliance upon the Safe Harbor only if it determines "in good faith that such amount of commission was reasonable in relation to the value of the brokerage and research services provided by such ... broker." It is difficult to understand how an investment manager can make a good faith determination of the value of a broker's combined provision of execution services and research unless it is able to reach a reasonable estimate of the value of each of those services viewed separately.

Further, a mutual fund's board has a duty under Section 15(c) of the Investment Company Act of 1940 to request and evaluate all relevant information in connection with the approval or renewal of a fund adviser's management contract. The fund adviser has a corresponding duty under Section 15(c) to provide such information. As the Commission has noted, the adviser's use of commissions to acquire research is relevant to the fund board's evaluation of the terms of the fund's management contract.6 Requiring a fund adviser to make a good faith, reasonable estimate in dollar terms of the value of research acquired through commissions will support a fund board's ability to evaluate the management contract, particularly the management fees to be paid to the adviser. If a fund's adviser is able to quantify the value of soft dollar research for the fund's board, we see little reason to withhold this information from fund shareholders.

For its part, Fidelity is able to arrive at reasonable estimates of the dollar value of soft dollar research, including proprietary research, and we believe that we are certainly not alone. Our ability to place a dollar value on soft dollar research is fostered by the approach we take in placing portfolio trades for the Fidelity Funds. The Fidelity Equity Trading Desk, responsible for placing orders on behalf of the Fidelity Funds, relies upon a relatively small group of "core brokers" selected solely for their ability to obtain best execution for the Funds' trades, taking into account speed and certainty of execution, willingness to commit capital, preserving anonymity for pending orders, price, and transaction costs.

Because Fidelity relies on a small group of core brokers to execute fund trades, we have negotiated arrangements with these brokers, in reliance on the Safe Harbor, for their use of commissions to obtain for the Fidelity Funds investment research from third party brokers. Precise dollar amounts, of course, are allocable to such research. Many of these brokers may also be involved in the execution of trades for other investment managers, including other mutual funds. Thus, the same research that is "third party" research for the Fidelity Funds is "proprietary" research for other investment managers. Brokers preparing this research should therefore be in a position to assist investment managers, including fund advisers, in ascertaining its dollar value, whether or not they are involved in the execution of fund trades.

Even with regard to brokers that tie the furnishing of internally prepared research to their execution of customers' securities trades, it is possible to derive reasonably accurate valuations of this research. At Fidelity, based upon our experience with the pricing and usefulness of third party investment research, we can extrapolate to arrive at very disciplined estimates of the "price" of proprietary research. We are able to estimate the aggregate value of all proprietary research received within a given period. We then allocate such value among the Funds' executing brokers through a "point system" by which fund managers evaluate the relative value of research provided by any particular executing broker.

The approach we take to value proprietary research, of course, is not the only approach fund groups might take to arrive at reasonable estimates of the value of proprietary research. The possibility that fund advisers might take somewhat different valuation approaches, however, is not a compelling reason for the Commission to forego any effort, even on a pilot basis, to require funds (and other investment advisers) to value research acquired with soft dollars. To address concerns over differences in valuation approaches, the Commission could require funds to explain their methodology for valuing proprietary research alongside disclosure of their soft dollar research costs. The Commission need not, at the outset, require inclusion of soft dollar research expense in a fund's expense ratio. Instead, this information can be set forth separately in a fund's prospectus, perhaps immediately following the fee table.

We encourage the Commission to consider the use of its rulemaking powers to foster comparability in the disclosure of soft dollar research expenses by adopting recordkeeping rules requiring brokers to account separately for their expenses in preparing internal research. The Commission could prescribe the methodology for such accounting, which could include allocation of salaries, dedicated systems and other operating costs, as well as an imputed profit margin. Alternatively, the Commission might consider requiring brokers to adopt their own accounting methodologies and publicly disclose their methodologies. Even though brokers would be required to attach a dollar value to soft dollar research provided to investment managers, the final determination of the value of soft dollar research to be disclosed in a fund's prospectus must be left to the fund.

The Commission may also wish to consider whether to provide guidance on the proper allocation by an investment manager of the cost of soft dollar research to mutual funds (and other accounts) over which the investment manager exercises investment discretion. This allocation could be based upon an estimate of the consumption or use of research for different accounts or upon the respective levels of soft dollar-eligible commissions paid by mutual funds and other accounts under common investment management.

The Commission and the self-regulatory organizations can conduct periodic inspections involving a broker's accounting for its expenses in producing proprietary research, and can require brokers to furnish their valuations of proprietary research to investment managers, including fund advisers, who rely on the Safe Harbor.

With the adoption of rules prescribing a uniform methodology to value proprietary research, the Commission could thereafter require inclusion of soft dollar research expenses into fund expense ratios. We emphasize, however, that requiring disclosure of soft dollar research expenses in fund prospectuses and elsewhere, outside of expense ratios, need not await adoption by the Commission of a uniform methodology. We also stress that the usefulness of disclosure of soft dollar research expense depends ultimately on the universality of disclosure rules applicable to all investment managers who rely on the Safe Harbor. Congress recognized this in 1975 by giving broad rulemaking power over soft dollar disclosure not only to the Commission but to regulators of other types of investment managers as well.

We believe that the Commission need not reach a decision now on whether to ask Congress to repeal the Safe Harbor. Rather, we recommend that the Commission require mutual funds and other users of soft dollar research to disclose the costs of that research. The Commission should allow the markets a period of time, perhaps one or two years, to operate within these new disclosure rules. Thereafter, the Commission, and the mutual fund and brokerage industries, can evaluate the benefits to investors made possible through disclosure of research costs. It is reasonable to anticipate that disclosure not only may lead to a lowering of the costs of investment research but may also lead to heightened competition by brokers in offering execution-only pricing.

Disclosure of soft dollar research costs may help discipline investment managers in their use of research and may lead the brokerage industry to improve the quality of proprietary research, reducing the redundancy of research focusing on large cap companies. Further, disclosure of soft dollars spent on research can help constrain investment managers from purchasing services that do not constitute bona fide research -- although we note that the Commission staff's study in 1998 suggests that the mutual fund industry has a good compliance record in this regard.7

There remains a possibility that disclosure of soft dollar expenses might not lead to the improvements noted above and that repeal of the Safe Harbor may be necessary. The Commission ultimately may come to this conclusion, in which event Congressional action would be called for. We urge the Commission to defer consideration of this question until it has had an opportunity to assess the results of the disclosure approach that we recommend. If repeal is necessary, then the Safe Harbor should be repealed in its entirety, without regard to the nature or source of investment research or the type of investment manager who uses commissions to acquire research. To discriminate among different types of research or investment managers will lead only to unfair competitive advantages for some and to a dysfunctional market for investment research.

Further, we note that management contracts approved by mutual fund boards and independent directors under Section 15(c) of the Investment Company Act of 1940 typically contain explicit authorization for fund advisers to use commissions to acquire investment research under the Safe Harbor. Repeal of the Safe Harbor, standing alone, would thus effectively result in a governmentally-mandated decrease of management fees charged by fund advisers, without any involvement on the part of fund boards. To avoid this result, we urge that any repeal of the Safe Harbor involving mutual funds should be accompanied by transitional rules that provide mutual fund advisers, fund boards and fund shareholders time to consider what, if any, adjustments to management fees may be appropriate if fund advisers are to be required to bear all investment research expense in the management of mutual funds.

II. Specific Questions in the Release about Quantifying Transaction Costs

1. Is investor decision-making harmed because investors lack numerical information about mutual fund transaction costs?

We doubt it. In our view, securities transaction costs (other than soft dollars) should be understood solely as contributors to trading gains and losses, which in turn are meaningful only in the context of overall investment results. We believe investors are very aware of net performance results already. Decomposing net performance results into different components may perhaps provide insights into how they were achieved, but will not change the net results.

While trading costs contribute to performance, they are not the most important factor. As we discuss below, the same trading activity may be viewed as adding to or detracting from results, depending on the methodology used to calculate trading cost.

2. What would be the best way to provide investors with additional numerical information about the amount of transaction costs that mutual funds incur? Would the information most appropriately be located in the prospectus, the SAI, or in another disclosure document?

Simple, objective figures like the amount of commissions paid by a fund and a fund's portfolio turnover rate can be included in fund prospectuses without unduly long explanatory material to accompany them. Some of the more complex measures of transaction costs, if practical to calculate at all, would likely require too much explanatory disclosure for the prospectus, which needs to be a short, easy to use document.

We expect that most of the use of transaction cost data will be made by fund commentators and academics rather than individual investors, and these parties already have access to all fund disclosure documents (including SAIs and shareholder reports as well as prospectuses). Indeed, the main purpose of disclosing commissions in percentage terms, in our view, would be to elevate the debate as to the significance of portfolio transaction costs and their relevance to fund investors. At present, there is a presumption that transaction costs simply reduce returns. Greater attention to actual commissions paid should reveal a more complex relationship between fund returns and fund turnover, which may also help develop a more standardized view of transaction costs in general. That, in turn, could lead to a greater consensus as to the proper standard cost calculations for funds or other investors to use in the future.

Questions about Quantifying Commissions and Spreads

3. Would a requirement to quantify (expressed as a percentage) and disclose brokerage commissions, but not other transaction costs provide useful information to fund investors? If funds are required to quantify and disclose their brokerage commissions, should the number be included in fund expense ratios and fee tables?

Quantifying brokerage commissions as a percentage of assets would be a useful way of presenting them to investors. Commissions are currently disclosed in dollar terms, so this number can easily be calculated. But disclosing them in percentage terms will relate them more readily to fund returns.

Commissions quantified as a percent of assets have the advantages of being objective, easily measurable and straightforward to calculate (indeed, it is easy to calculate them based on disclosures made by funds today). Their chief advantages are (1) that they can be disclosed by funds at little additional cost, and (2) the components of the calculation are already public knowledge, so disclosing them is unlikely to induce managers to engage in artificial trading behavior. These advantages make disclosure of brokerage commissions in percentage terms a relatively harmless, and probably useful, additional disclosure requirement.

Adding brokerage commissions to expense ratios, however, would be both theoretically inaccurate and, in our view, unhelpful to investors: theoretically inaccurate because commissions are properly considered capital costs rather than expenses, and unhelpful because including commissions in expense ratios or fee tables would make those expense figures more volatile, less meaningful, and less useful as a way of choosing funds.

Commissions are far more variable than other expenses. Fund management fees, service charges and distribution fees are relatively predictable and stable from year to year. This is part of what makes the expense table and the expense ratio so useful for investors: the expense figures can generally be counted on to be representative of what investors can expect in the future. Commissions, on the other hand, can and often do vary significantly from year to year. For instance, turnover may be abnormally high in one year because of a change in management, a change in fund size, volatile cash flows, a temporary increase in market timing by shareholders, or simply a change in strategy by the manager. It would confuse investors and would dilute the usefulness of the expense ratio or the expense table to include such transitory events. Investors should not be encouraged to give the same weight to a 5 or 10 basis point change in brokerage commissions from year to year as to a 5 or 10 basis point change in advisory or 12b-1 fees. Commissions are interesting for analyzing performance, but they lose their transparency and will detract from investors' understanding of fund expenses if they are included in the expense ratio or fee table.

Commissions also may have limited comparability among funds. Although they are objectively measurable, commissions can be artificially changed by advisers or brokers if there is an incentive to do so. Most commissioned trades could easily be recharacterized as principal trades without commission, or vice versa, without affecting the total price. Too much emphasis on commissions would create an inappropriate financial incentive to make commissions appear small, without affecting overall return.

Finally, one should not assume that lower commissions are necessarily associated with higher returns (or lower overall transaction costs). A fund with high turnover and high commissions may perform better than a fund with low turnover and low commissions. Adding commissions to fund expense ratios would give them the wrong kind of emphasis, and reduce the usefulness of the information that is already provided.

As discussed above, we do believe, however, that the soft dollar component of commissions should be unbundled, expressed as a percentage, and either added to the fee table (perhaps as a separately disclosed component of the management fee) or discussed near to it. Soft dollars are not like execution-related transaction costs: they represent a cost to the fund and a benefit to the adviser. This is so even though broker-supplied research may ultimately benefit investors by contributing to better investment performance. There is little difference, economically, between paying a management fee to the adviser from fund assets and using fund commissions to defray research expense that the adviser otherwise would directly incur.

We understand that commissions paid by equity mutual funds across the industry average approximately 0.20% of fund assets (slightly more as a simple average, and slightly less as a weighted average). Opinions differ as to the percent of this total that represents soft dollars. At Fidelity, approximately 20-25% of commissions is paid for investment research. If this estimate is reasonably accurate for the fund industry as a whole, then approximately 0.04-0.05% (4 to 5 basis points) of equity mutual fund assets, on average, is being used to provide research services in addition to research generated by the fund's adviser that funds pay for in their management fees.

4. Does the increased use of riskless principal trades on NASDAQ make it easier to quantify the cost of NASDAQ trades? What proportion of NASDAQ trades are subject to commission-equivalent fees?

Riskless principal trades account for a very small percentage of total trades by the Fidelity Funds in NASDAQ securities. It is unclear what types of trades the Commission intends to reach by referring to "commission-equivalent" fees. Beyond riskless principal trades, the Fidelity Funds will also frequently pay an imputed commission (which we refer to as a "commission-equivalent" fee) on trades, both in NASDAQ securities and exchange-listed securities, in which the broker-dealer commits capital. Capital committed trades constitute an increasingly higher portion of trades of the Fidelity Funds in recent years. Consistent with the provisions of the Safe Harbor, we do not treat imputed commissions as eligible for use in our soft dollar research program.

5. Would quantifying commissions mislead investors because it would result in a number that includes some transaction costs and excludes others? Please explain the reasons for your answer.

Like many numbers that are quoted with respect to investment portfolios, disclosing commissions as a percent of assets risks misleading investors if investors treat the figure too simplistically or give it too much weight. This is one reason why we urge that commissions not be added to the expense ratio or fee table. As the Concept Release notes, commissions are only a partial measure of transaction costs, and commissioned trades in listed securities can easily be negotiated into principal trades (and vice versa).

6. If the answer to question 5 is yes, would the concern be alleviated if funds were required to quantify commissions and provide investors with disclosure that details the portion of trades that are performed on a commission basis; spread basis; or some other basis (e.g., directly from an issuer)?

This would be helpful additional disclosure alongside commission disclosure. We note, however, that the need for explanation and exceptions argues placing undue emphasis on disclosure of commissions or including them in a fund's expense ratio.

7. What effect, if any, would a requirement to quantify commissions have on the incentives of fund managers with respect to (1) use of principal versus agency transactions; and (2) use of soft dollar transactions?

Commissions are already disclosed by funds and have been, in a variety of formats, for many years. We would not expect more prominent disclosure of commissions to affect Fidelity's commission practices one way or another. We would note, however, that most fund advisers probably could change their ratio of agency vs. principal trading (reducing commissions) without affecting the amount of soft dollar research the adviser received. As noted above, at Fidelity, soft dollars (including amounts spent for third-party research, market data services, and research bundled together with execution services) represents roughly 20-25% of total commissions. There would be ample room to increase this percentage if some trades were recharacterized as principal transactions, allowing for a reduction in commissions without a reduction in soft dollar expenditures. This is one reason why we caution against over-emphasizing commissions as a cost of the fund; we believe the SEC should not create a financial incentive to favor principal transactions over agency transactions.

8. Could any possible adverse effects identified in questions 5 and 6 be mitigated or eliminated by requiring funds, in addition to reporting their commission costs, to estimate the spread cost of their principal trades (for example, by imputing to principal trades the fund's average commission rate on agency trades)? If yes, should this number be included in fund expense ratios and fee tables?

We are skeptical of spread costs as a means of quantifying transaction costs. Unlike commissions, which are objectively measurable (if subject to recharacterization), spread costs are estimates based on a hypothetical transaction — a purchase at the offer price and/or a sale at the bid price — that may or may not occur. Spread cost estimates typically assume that all investors buy at the quoted ask and/or sell at the quoted bid. We generally have not found spread estimates to be useful in the investment process and would not recommend using them for general disclosure purposes.

9. Alternatively, can the portion of spread cost that represents payment for executing a trade be measured separately from the portion of the spread that represents the market impact cost associated with that trade? If yes, should this number be included in fund expense ratios and fee tables?

Spread measures in general seem to us to involve too much estimating and take too little of the variations between investors into account. Most spread measures do not take into account skill in trading, or assume equivalent skill by all parties. A hypothetical trader who always bought at the bid price and sold at the ask price would have the same numbers as one who never bettered the bid/offer spread. This makes spreads less useful in comparing execution quality across different funds or advisers.

Questions about Quantifying All Transaction Costs

10. Would a requirement to quantify all transaction costs provide useful information to fund investors? Would a requirement to quantify all transaction costs, except opportunity costs be a better alternative? If you advocate that we mandate either of these alternatives, please explain as specifically as possible, how the alternative should be implemented. Please discuss the specific algorithms, formulas, definitions, recordkeeping requirements, and internal control requirements that should be used. Commenters are encouraged to address the following specific topics:

  1. How should funds measure their spread costs?

  2. How should funds measure their market impact costs?

  3. How should funds measure their opportunity costs?

  4. Should spread, market impact and opportunity costs be measured trade-by-trade or for all transactions?

  5. Should spread, market impact and opportunity costs be measured absolutely or relative to a benchmark?

  6. Should this number be included in fund expense ratios and fee tables?

In managing the Fidelity equity funds, we generally focus on transaction costs measured by comparing actual execution prices for a given day to the available volume-weighted average price of the stock for the day (abbreviated as "AVWAP"),8 though we have the ability to measure equity execution costs a variety of ways and have also experimented with implementation shortfall and opportunity cost methodologies. When reviewing execution quality with the Brokerage Committee of the funds' Board of Trustees (a separate committee of the Board made up solely of disinterested board members), we compare our equity trading against AVWAP and also against VWAP calculated by an independent third party. We generally do not focus on spread costs or opportunity costs in the normal course of trading for equity funds.

We measure our trading costs against AVWAP on an absolute basis, since this is an internally generated measurement that it is difficult for third parties to replicate for other clients (because it requires accurate time-stamping of each fund trade, among other reasons). The third-party analysis based on VWAP does include benchmarking relative to a selection of clients considered comparable to Fidelity by the third party. We have also made use of third-party services that measure execution costs based on implementation shortfall and opportunity cost concepts in the past, but have not found them satisfactory and with the Brokerage Committee's concurrence have discontinued using them.

As of our most recent report to the Brokerage Committee, we measured our execution cost vs. VWAP at approximately 12 bp and our execution cost vs. AVWAP at approximately -10 bp (the latter figure does not include commissions). These are aggregate figures across a large number of funds, and fund-by-fund figures would vary greatly. However, it is important to recognize that transaction "costs" which take market impact into account may be negative — i.e., they may represent a short-term trading profit even after all costs rather then a drag on results. We feel this strongly argues against putting any sort of transaction cost estimate into fund expense ratios. Although it may be commendable for a fund, through skillful trading, to realize a negative transaction cost, we do not believe it would be appropriate for funds to use such negative costs to offset fund expenses in an expense ratio. Funds should not be able to use trading profits to reduce their reported expenses; and conversely, trading losses should not increase such expenses.

11. Would the trade effect measure provide useful information to investors, and if so, should we require its disclosure? If the Commission mandated trade effect disclosure, should trade effect be measured with respect to same day closing prices or next day closing prices?

As stated above, we normally measure transaction costs using a VWAP or AVWAP measure rather than a trade effect measure. Trade effect measures will show a different number, but it is far from clear that it would be a superior measurement or useful disclosure for investors. To illustrate, we calculated the trading cost for domestic equities for a sample of six of our domestic equity funds using a variety of measures. The following table provides the result, in basis points of net assets.

2003 Transaction Costs for US Equity Trades, in Basis Points
Based on Actual Execution Prices vs. Various Reference Prices

Reference Price

T-1 Close





T+1 Close








Notes: Based on domestic equity trades during calendar 2003, excluding commissions. Costs for buys are calculated as (traded price — reference price) / reference price; flip the sign for sells. All costs are expressed in basis points of average net assets. Negative costs are gains (desirable). Reference prices are defined as follows:

T-1 Close: The close price of each stock the day before the trade

Open: the price of each stock at the open of the trade date

VWAP: the volume-weighted average price for each stock for each day

AVWAP: the volume-weighted average price for each stock available from the time of the first order by the fund to the end of that trading day

Close: the price of each stock as of the close of each trading day

T+1 close: the close price of each stock the day after each trading day

Trade effect measures — "Close" and "T+1 Close" in the table — generally show a more negative trading cost (i.e., a greater trading gain) than measures based on VWAP in this sample, and a significantly better result than measures comparing to open or prior day prices. But it is difficult to say which number is more meaningful. Did these funds have negative transaction costs of 25 bp, based on their trade prices compared to the close each day? Or did they have positive transaction costs of 26 bp, based on their transaction prices compared to the open price each day? Academic literature will point out that each figure has advantages and drawbacks. We have chosen measures vs. AVWAP as most helpful for our own internal trading measurements, but we assume others would find them less useful for their purposes. And it should be borne in mind that public disclosure by a fund of transaction cost measurements under multiple methodologies carries its own risks for the fund's shareholders. For a professional trader, these disclosures could give insights into the fund's trading patterns that could be used to trade against the funds.

It should also be noted that calculating these figures requires a substantial amount of data, including not only complete trade information for each of our funds but also tick-by-tick trade data for each stock in which the funds invest. We doubt that small fund groups would be able to perform these calculations in-house today.

12. More generally, if the Commission were to choose to require disclosure of only one transaction cost measure, which measure should it be?

We don't believe it is possible at this time to define a measure that would be agreed upon by all parties. And since all measures have biases and can be "gamed" to some extent — and more generally, because any measure may change behavior — it would be dangerous to adopt a measurement without considering the potential impact on trading and investing behavior by funds. We expect that fund groups would be able to adjust their trading methods to any measure the Commission chooses.

One caution for any measurement that may be mandated by the Commission: Although we strive to make our cost measurements as accurate as possible, there are significant data quality issues in calculating measures that require intraday tick-by-tick trades and/or spreads, and there are numerous types of securities (privately placed securities or convertible bonds, for example) that we generally exclude from transaction cost calculations. When dealing with third-party data providers and with evaluators of transaction costs, we generally must spend significant time resolving these data quality issues. When calculating cost estimates for internal use or for discussion with the funds' Brokerage Committee we are confident in our ability to provide reasonable margins of error and to explain the limitations of any data we provide. We would be more hesitant to guarantee the accuracy of these figures at a level satisfactory for mandatory public disclosures.

Questions about Accounting Issues

13. Would it be appropriate to include some or all transaction costs in fund expense ratios and fee tables without accounting for these items as an expense in fund financial statements?

As we discussed above, that is certainly an option with respect to soft dollars, which we believe should be more clearly reported and identified as a fund cost. Before requiring inclusion of soft dollars in a fund's expense ratio, the Commission, we suggest, should consider developing uniform disclosure requirements for mutual funds and for other users of soft dollars (including other investment advisory products, whether or not organized as investment companies) since the safe harbor provided by Section 28(e) is not limited to mutual fund advisers.

14. Would it be feasible to account for some or all transaction costs as an expense in fund financial statements? If it is not feasible to reliably measure market impact and opportunity costs, should we still require that commission costs be expensed? If yes, should the requirement apply to all commission costs or only those commission and spread costs that do not relate to the execution and clearing of a portfolio transaction (i.e., soft dollars)? If it is not feasible to reliably measure all research costs, should we still expense those costs that can be reliably measured (i.e., payments to third parties for research)?

As noted above, we believe that total commissions should not be expensed, although expensing the soft dollar component of commission should be explored. It would ultimately be feasible to account for soft dollars as an expense in financial statements, if the Commission requires brokers providing research to assist funds and their advisers in estimating the value of such research. However, further guidance from the Commission may be required to assure sufficient uniformity in expense estimation and allocation practices across funds before soft dollars can be fully integrated into the expense ratio.

We strongly urge the Commission to refrain from any rulemaking or policymaking that would include third party research but exclude proprietary research. It is critical that an evaluation of research bundled with execution must be included in any soft dollar expense estimate. The bundling of research and execution is the least transparent aspect of transaction costs, and the Commission should not do anything to favor it over other uses of commissions to obtain legitimate benefits for the adviser. At Fidelity, we estimate that research bundled with execution represents a larger share of commissions than third-party research, and we expect this would be true at other firms as well. Thus, in addition to giving rise to competitive unfairness for third party research providers, quantifying third party research without quantifying bundled research would significantly understate total soft dollar use and could be highly misleading to investors.

15. Are mutual funds and their managers better able than they were in the past to track the portion of commission costs that purchase research services from brokers? Has the improvement been sufficient to make it feasible for us to require funds to expense these items in their financial statements? Since soft dollars are earned based on complex-wide trading activity, how should research and other non-execution costs be allocated among funds? Can soft dollars be traced to individual portfolio transactions? (This would entail adjusting the basis of the securities purchased in those transactions for the portion of the commission cost that was used to purchase research services.) Alternatively, should an aggregate adjustment (not specified to a particular portfolio transaction) be made to realized and unrealized gain or loss? If funds and their managers are not yet capable of tracking the portion of commission costs that purchase research services from brokers, what factors continue to prevent funds and managers from developing this capability?

Fund managers and other investment managers are more able to separate costs of execution from costs of research than they were in the past, in part because brokers and dealers have increasingly been willing to trade, and quote commissions, on an execution-only basis. While we are readily able to estimate the value of research bundled with execution, without assistance from the brokers we do business with we will not be able to calculate it with the precision necessary for inclusion in expense ratios or fund financial statements. However, we are confident that industry-standard methodologies for separating commissions into execution and non-execution components (probably based on a broker's full allocation of expenses associated with the production of investment research) can quickly be developed with the Commission's encouragement, creating the basis for investors to draw comparisons across different funds and fund groups.

As for allocating research among funds and other accounts under common management, conceptually there are two types of methods that could be used: allocating among funds according to their presumed "consumption" of research, or allocating among funds based on their commission and trading levels.

The former — allocating by utilization or consumption of research — could be approximated by allocating research received by the adviser as a constant percent of assets across similar funds, or as a constant percent of the management fee. The latter — allocating according to commissions — could be implemented by identifying a percentage of soft-dollar-eligible commissions across all funds and clients as representing soft dollars for a given period, or identifying different percentages of commissions for each broker.

We believe either an asset-based or commission-based method can reasonably be used to allocate soft dollar benefits across funds or clients. As a matter of public policy and advancing investors' interests, we firmly believe that either method is far preferable to leaving soft dollars bundled into commissions without any additional disclosure.

If soft dollars are treated as an expense we would recommend that an aggregate adjustment be made to unrealized gain or loss rather than attempting to allocate the expense to any individual security or transaction.

Questions about Improving Disclosure Related to the Level of Transaction Costs

16. Are there ways to provide a rough estimate of transaction costs, or develop a scheme to categorize these costs (for example, "very high," "high," "average," "low," or "very low") under general guidelines set by the Commission that would mitigate the difficulties involved in coming up with a more precise measure, and yet still provide useful information to investors? Could such an approach produce results that are consistent enough to permit meaningful comparison among funds? If yes, please provide specific suggestions.

To assign costs to the high, average and low categories, it would still be necessary to calculate costs with some precision (to decide a close call between "low" and "average", for instance). It is not clear why this would be better than disclosing a specific figure.

17. In general, do the current disclosure requirements relating to transaction costs described in this section of the release provide investors with adequate information? If not, what additional information should funds provide? Would one or more of the alternatives described in this section provide useful information to investors, or would the alternatives lengthen the prospectus while providing no real benefit? If one or more of these alternatives would provide meaningful information, would the information most appropriately be located in the prospectus, the SAI, the report to shareholders, or in another disclosure document?

We have no objection to giving greater prominence to portfolio turnover rate, although it is a number that already is widely quoted and generally well-known. It is perhaps fair to consider how great the marginal benefits are likely to be by giving greater prominence to a figure that is so widely available already. Additional narrative discussions of portfolio turnover, if required at all, seem more appropriate in the context of a shareholder report, where turnover can be related to historical investment activities of the fund. By contrast, the benefits of "force feeding" the discussion of portfolio turnover into the prospectus, are uncertain, because policies regarding portfolio turnover are likely to be vague and future turnover is difficult to predict. We would also caution against an instruction always to address portfolio turnover, as such instructions tend to lead to checklist approaches to disclosure and boilerplate discussions of data points that may have no great importance.

18. Does existing portfolio turnover disclosure provide useful information about transaction costs? If additional narrative disclosure concerning portfolio turnover and its relationship to transaction cost is needed, what information should be required?

The turnover rate is a useful number for investors. While it is limited, it is also simple, objective, and relatively well understood. Turnover rates can be augmented by associating them with commissions paid as a percent of assets. Those two figures are probably the best available proxies for transaction costs at this time, and are preferable to more speculative numbers.

19. Does the existing requirement to disclose the dollar amount of commissions paid provide investors with meaningful information about transaction costs? How can the existing requirement be improved?

We believe commissions quantified as a percent of assets, notwithstanding its limitations, would be a more useful figure to highlight for investors. While that measure has drawbacks, and may be gamed if too much emphasis is placed upon it, further study and public discussion of commission rates in percentage terms would be likely to help advance debate towards more widely acceptable measures of overall transaction costs.

20. Would an average daily net flow measure provide useful information to investors?

Additional disclosure of shareholder flows could be useful to investors in identifying funds that have welcomed or have been targeted by active traders. Our preferred figure is a shareholder turnover rate calculated in the same fashion as portfolio turnover (the lesser of gross purchases or redemptions, divided by average net assets). Any figure based on gross activity has some drawbacks. Notably, the calculation of gross shareholder activity may differ because of variations in the amount of netting undertaken by fund transfer agents or omnibus account providers. Nonetheless, we have found shareholder turnover rates to be useful in identifying funds with unusually high or low shareholder activity relative to other funds with similar customer bases. The ability to compare shareholder turnover rates and portfolio turnover rates also provides this figure with an advantage over other calculations.

21. Should the Commission consider policies to encourage funds to charge purchasers and redeemers of fund shares a fee payable to the funds to compensate existing and remaining investors for the costs they bear when their funds accommodate the purchases and redemptions of other investors? If yes, should the Commission consider requiring funds to disclose how they compute these fees, if they require them; and why they do not require these fees, if they do not?

The Commission has solicited public comment on this issue in a recent release, and we anticipate sharing our views with the Commission in response to the questions raised in that release.

22. Should the requirement to disclose average commission rate per share be reinstated, in either its original form or in a revised form? If you advocate that it be reinstated in a revised form, please provide specific suggestions.

We believe this figure was not useful to investors and was appropriately abandoned by the Commission. While we would have no objection to disclosing it again as we did in the past, we question the importance of resuscitating it as a mandatory disclosure point.

23. Is "transaction costs" as described in this release a useful concept, or would it be more useful for investors to see the effect of all costs combined, for example, by showing the following:

  • Gross or "pure" portfolio return;

  • Net return to shareholders; and

  • The resulting difference?

In our view, the debate on the impact of expenses on investment results would be greatly improved if it were focused more on the difference between gross and net results and less on net results and fees only.

We strongly support the concept that funds should be permitted to report gross returns before costs. The current debate over fund fees largely misses an important point: as more elements of fund return are isolated and treated as fees, fund net returns do not change. In other words, the current debate is an exercise in recharacterizing dollars from return to expense. If we find that commission costs are an expense, and treat them as such, fund net returns do not go down by an equivalent amount — they are unchanged. To the contrary, fund gross returns increase, dollar for dollar, by every amount that is taken out and treated as an expense.

24. If it would be useful for investors to see the effect of all costs combined, could funds calculate and report the gross or "pure" portfolio return, net return to shareholders and the resulting difference on an annual basis?

There are several methods for calculating these figures, some more complex than others. The simplest methods would involve "grossing up" net returns for identified costs, such as the expense ratio and sales charges, plus, perhaps, transaction cost amounts. Such methods can be quite straightforward to calculate. A more complex method could start from the holdings and returns for each security held in the fund and from that produce a hypothetical return in the absence of expenses. This type of calculation may not be practical for all funds, depending on the amount they have invested in performance attribution systems.

25. Should the Commission require disclosure of gross returns? If so, what definition would be most useful? Of what benefit would these returns be to investors? How expensive would it be for funds to compute these returns?

We would recommend that disclosure of gross returns be required in at least one place in the prospectus. We would also recommend that funds be allowed to discuss gross returns in sales materials, always in the context of net returns as well. Those calculations need not be costly, and would require significantly less processing changes than disclosure of after-tax returns, for example. We expect that a methodology based on grossing up reported net returns for identified categories of expense would be most useful.

26. Would the disclosure of gross returns allow investors to better identify dilution due to market timers?

We expect it would be difficult to separate costs attributable to market timers from costs associated with normal shareholder activity, and that it would be difficult to separate "dilution" from other impacts of cash flow on fund performance.

27. If portfolio returns are to be disclosed, how should the returns be adjusted for fund flows into and out of the portfolio? Should they be computed using internal rate of return methods; time-weighted average methods; or should other methods be used?

We expect that grossing up time-weighted net returns, without adjusting for inflows and outflows, would be the most practical method of disclosing gross returns.

28. If portfolio returns are to be disclosed, should these returns only be disclosed, or should the differences between these returns and the shareholder returns be disclosed?

We expect that gross returns would be most useful to shareholders when presented together with net shareholder returns. To allow gross returns to be promoted without net returns opens up the possibility of abuses in the sales process.

29. Where should these returns or return differences be disclosed, and how should they be described?

This will depend on the specific methods chosen, but we would expect that required disclosure in the prospectus and permitted disclosure in sales literature would be the best combination.

Questions about Board Review of Transaction Costs

30. Are existing requirements for board review of transaction costs adequate? If they are not adequate, how can they be improved?

At Fidelity we provide the Funds' Board, through the Brokerage Committee, an annual review of transaction costs that includes internally generated and third-party estimates of transaction costs for equity securities. For fixed-income securities and money market instruments we do not provide the same level of data, in part because these instruments generally trade without commission and without soft dollars or directed brokerage practices. We cannot comment on practices at other fund groups.

31. Should boards be required to receive reports with mandated information regarding soft dollars and directed brokerage payments? Should investors be provided periodically with a summary of these reports?

We believe fund boards should be provided with transaction cost information for equity securities at least, on a regular basis. However, because detailed transaction cost data can reveal significant information to professional traders that could be used against the funds, and to protect proprietary trading techniques developed by fund advisers, funds should be allowed to give much more summary information to the general public than they provide to their boards.

32. One problem in evaluating execution cost measurements is in identifying a standard of comparison. It may be difficult for fund directors to assess the fund's execution performance statistics in a vacuum, without comparison with other funds' statistics. Should the Commission or other independent body collect these statistics from similar funds and make available aggregate statistics for comparison purposes?

Databases on transaction costs are already available to advisers commercially for a fee. It is hard to see how the Commission could maintain a database of execution costs for advisers without first agreeing on a standard for measurement, in which case disclosure to the general public would seem to be superior to the Commission's developing a new database standard. Fund trackers have generally had little difficulty ranking funds according to published statistics, once the standard statistics are agreed upon, and commercial entities may perform this function better than the government.

33. Should fund advisers be required to provide fund boards with an internal allocation of their uses of brokerage commissions, indicating the amounts and percentage used by the adviser to obtain execution services and soft dollar benefits, specifically detailing the types and amounts of the various kinds of benefits? Should there be separate allocations among types of research, such as research produced by underwriters, or other broker-dealer affiliates?

We provide this type of information to the board of the Fidelity funds for equity securities. We believe as a general matter that fund boards in their oversight of the investment performance of funds and their annual evaluation of the fund adviser's management contract should consider the relevance of brokerage commission uses.

III. Transparency for Exchange Specialists' Trading

The Commission has also solicited public comment on whether "sell side" disclosure of execution quality for large, institutional orders would be helpful to funds in evaluating execution costs. The Commission, for this purpose, includes exchange specialists as sell side participants. With regard to exchange specialists, especially those on the New York Stock Exchange, our answer is an unequivocal "yes." Notwithstanding the renewed emphasis by the Commission, Congress, other regulators, policy makers and academics on the need for transparency in the mutual fund industry and the markets generally, there remains one vestige of the securities industry shrouded in secrecy: trading by the NYSE specialist.

Simply put, it is impossible for any participant in the market (at least any participant not present on the NYSE floor) to know when any trade in an NYSE-listed security has been entered into by a specialist for its own account. This is unknown at the time when listed trades are reported on the consolidated tape and remains unknown thereafter. This is so, notwithstanding the exclusive privilege granted to the NYSE specialist to be the sole market maker on the floor. It remains so even in the aftermath of an unreleased report prepared by the Commission staff and the recent settlement of enforcement proceedings by the major NYSE specialist firms for allegedly widespread abuses arising from their privileged trading positions. We strongly encourage the Commission promptly to propose and adopt new disclosure requirements for exchange specialists that will enable investors independently to evaluate the trading activities of those specialists.

* * *

David Jones
Senior Vice President
Product Strategy & Communications


Eric D. Roiter
Senior Vice President and General Counsel
Fidelity Management & Research Company

Fidelity Management & Research Company

cc: Hon. William H. Donaldson

Chairman of the Securities and Exchange Commission

Hon. Paul Atkins

Hon. Roel Campos

Hon. Cynthia A. Glassman

Hon. Harvey Goldschmid

Division of Investment Management
Paul Roye

Cynthia Fornelli
Deputy Director

Robert Plaze
Associate Director

Douglas Scheidt
Chief Counsel

Paul Goldman
Assistant Director

Division of Market Regulation
Annette Nazareth

Robert Colby
Deputy Director

Josephine Tao
Special Counsel


1 On February 11, 2004, the Commission announced adoption of rules to require reports to fund shareholders to include the cost in dollars associated with a notional investment of $1,000, based on a fund's actual expenses for the relevant six-month period. To promote comparability, the newly adopted rules will also require disclosure of expenses based on a notional investment of $1,000 and an assumed return of 5 percent per year. Release 33-8393, 34-49333, IC-26372 (February 27, 2004)

The Commission has also proposed to require extensive disclosure to mutual fund investors at the point of sale and in confirmation statements, detailing the costs of distribution borne by the investor, whether paid directly by the investor or paid by the fund, and the sources of compensation received by dealers who distribute fund shares, including not only loads and fund-paid distribution fees but revenues sharing arrangements between the dealer and the fund's adviser or other affiliates. Release No. 40-26341 (January 29, 2004)

2 Disclosure by Investment Advisers Regarding Soft Dollar Practices, Investment Advisers Act Release No. 1469 (Feb. 14, 1995) ("Adviser Soft Dollar Release") at 9.

3 Section 28(e)(2) of the 1934 Act provides, "A person exercising investment discretion ... shall make such disclosure of his policies and practices with respect to commissions that will be paid for effecting securities transactions, at such times and in such manner, as the appropriate regulatory agency, by rule, may prescribe as necessary or appropriate in the public interest or for the protection of investors."

4 We are mindful of the Commission's guidance that:

Section 28(e) ... does not afford a safe harbor with respect to all conflicts of interest between the adviser and its clients that may arise from soft dollar arrangements. For example, soft dollar arrangements may cause an adviser, in order to obtain soft dollar services, to violate its best execution obligations by directing client transactions to brokers who could not adequately execute the transactions. Soft dollar arrangements also may give advisers incentives to trade client securities inappropriately to generate credits for soft dollar services." Adviser Soft Dollar Release at 8.

5 In 1995, the Commission considered and rejected a proposal put forward by Goldman Sachs and Morgan Stanley to require disclosure in Form ADV filings by investment advisers of the costs of third party research paid for with soft dollars, but not the cost of proprietary research bought with soft dollars. The Commission explained that:

"[U]nless the values of proprietary soft dollar services are also included ..., the information provided to the client would be incomplete and may distort client understanding about the benefits that advisers receive through client brokerage. Clients, for example, may incorrectly believe that soft dollar services are not a consideration in an adviser's direction of client brokerage to full service brokers or that third-party soft dollar services are of greater value (either to advisers or clients) than proprietary soft dollar services. Moreover, the Goldman/Morgan Proposal may provide an investment adviser an incentive to direct brokerage to a full service broker rather than a soft dollar broker for the same types of soft dollar services, simply because of differing client reporting requirements. This consequence may not be in the best interests of advisory clients and may be unfair to soft dollar brokers. Thus, ... the Commission is not proposing that only third-party soft dollar services be valued." Adviser Soft Dollar Release at 27-28."

6 Interpretive Release Concerning Scope of Section 28(e) of the Securities Exchange Act of 1934 and Related Matters, Exchange Act Release No. 23170 (April 23, 1986) at 9. See also, Office of Compliance, Inspections and Examinations, Inspection Report on the Soft Dollar Practices of Broker-Dealers, Investment Advisers and Mutual Funds (Sept. 22, 1998) (the "OCIE Report") at 12.

7 OCIE Report at n. 82 (93[A]dvisers of investment companies generally are prohibited from acquiring items outside of the safe harbor of Section 28(e), irrespective of disclosure. We did not observe any instances in which fund commissions were used to purchase non-research items which did not directly benefit the funds themselves.")

8 Available VWAP is a variation of VWAP, comparing only to prices occurring in the market after a trade is initiated. For example, if a trade is placed at 12:00 noon, a comparison to AVWAP would review all trades in the market from 12:00 until the market close, while a simple VWAP comparison would compare to all trades for the day, including trades from 9:30 to noon.