March 10, 2004
Mr. Jonathan G. Katz
Re: File No. S7-03-04, Investment Company Governance, Release No. IC-26323 ("Release")
Dear Mr. Katz:
I am writing on behalf of Fidelity Investments to offer our views on the Commission's proposed mutual fund governance rules set forth in the Release. Our letter responds, in particular, to the Commission's proposal to require all mutual funds to be chaired by an independent director, regardless of the choice that independent directors, constituting a supermajority of a fund's board, would make if permitted to exercise their business judgment on behalf of fund shareholders. The Commission's proposal would compel a single result for all mutual funds, without regard to their origins, history, performance or the adviser's record of fealty and service to shareholders.
Fidelity, in the strongest terms, urges the Commission to refrain from adopting an independent chairman requirement that would reduce the very discretion and authority of independent directors that the Investment Company Act is founded upon and that the Commission has sought to strengthen through its rulemaking in recent years. Our reasons can be summarized as follows:
Enhancing the Board Process
Fidelity fully supports the principle that independent directors, assigned key roles in mutual fund governance, must be empowered to carry out their responsibilities free from undue pressure or influence from the fund's adviser. This principle extends not only to the conduct of board meetings but to all aspects of the independent directors' oversight of the performance and services provided by a fund's adviser and affiliates.
With regard to the independent chairman proposal -- a proposal that would compel a fundamental departure from prevailing governance practices in the mutual fund industry -- we respectfully submit that the rationale put forward by the Commission in the Release is quite cursory. In a 31-page release, the Commission (after an introductory paragraph) devotes two brief paragraphs to explain its reasons in support of the proposal. The reasons advanced are that (i) an interested chairman can control the board's agenda, have a substantial influence on the "culture" of the boardroom and chill meaningful dialogue and (ii) an independent chairman may promote a boardroom culture more conducive to decisions that favor long-term interests of fund shareholders and may lead to more effective negotiations with the adviser.
This rationale, in our view, borders on speculation and paints with an overly broad brush. It is a truism that some boards may be better served if chaired by an individual who happens to be an independent director. It is far from clear, however, that an independent chair is preferable for all fund boards or even for most fund boards. In any event, the Commission need not impose this procrustean governance rule on all mutual funds throughout the industry. Strengthening the board process can be readily achieved through more focused measures. We commend the SEC for identifying these in the Release, namely:
(1) Requiring independent directors to appoint a "lead" director for their group to chair separate meetings of the independent directors, act as their spokesperson and supervise independent fund counsel;
(2) Requiring some or all board committees to be chaired by an independent director; and
(3) Requiring the board and the independent directors, by separate vote, to elect the board chairman annually.
These measures are far preferable to a "one size fits all" independent chairman requirement for the mutual fund industry. We would also suggest that the Commission, in lieu of such rule, consider requiring funds to comply with the following:
(1) Independent directors should have sole authority to set their own compensation;
(2) Independent directors should separately vote to approve the agenda for every board meeting, should have the authority to add any matter to the agenda; and
(3) Independent directors should have authority to direct the adviser to provide for their consideration at a board meeting any information that is relevant to their decision-making.
These measures, in our view, ensure that independent directors will exercise separate authority over those aspects of board governance - the setting of agendas, the creation of board committees, and assignment to board committees and controlling the flow of information to the board - that bear directly upon the effective functioning of fund boards. They are certainly sufficient to ensure that individuals serving as independent directors, selected and nominated solely by incumbent independent directors, will be able to discharge their oversight role on behalf of shareholders without undue influence from the advisor, regardless of whom the independent directors select to serve as board chairman. Indeed, if the independent directors of a fund board fall short of their oversight responsibilities notwithstanding all of these safeguards, it is extremely difficult to understand how that shortcoming could have been avoided if only they had been required to choose one of their own to serve as board chairman.
Our proposed approach is, in fact, entirely consistent with that taken by the 1940 Act and by the Commission in its rulemaking over the years. There has been much heated rhetoric over the asserted conflicting roles of an interested chairman with regard to a fund board's annual approval of the adviser's management contract. The question is typically posed "How can an interested chairman effectively negotiate with himself?" The short answer is, of course, that the 1940 Act already resolves this issue. The Act requires affirmative approval not only by a fund's board but also the separate approval by the fund's independent directors. No interested director, whether or not serving as board chairman, can take part in that second vote. The same is true of other key decisions under the 1940 Act or Commission rules, including, for example, approval of principal underwriting agreements, Rule 12b-1 plans, multi-class plans under Rule 18f-3, procedures permitting an adviser's brokerage affiliate to place a fund's portfolio trades, and procedures permitting a fund to purchase securities in a public offering from a syndicate in which the adviser's affiliate is a member. With all the responsibility that the 1940 Act and rules already entrust to independent directors to act in the best interests of fund shareholders, why should the Commission curtail the exercise of judgment by these very directors in choosing a chairman of the fund board?
While we certainly do not suggest that every board be chaired by an interested chairman, it seems manifestly clear to us that in many instances a fund's independent directors would be exercising sound business judgment if they were to select, from among all the directors on the board, a member of the adviser's management to serve as board chairman. Independent directors may well conclude that an interested chair can promote informed, efficient and deliberative board meetings by drawing upon his or her broad expertise and experience in the mutual fund industry. This need not foreclose, in the least, independent directors from acting effectively as a separate decision-making group or from performing their oversight responsibilities in a rigorous and thorough manner.
Indeed, we believe that independent directors, at least at certain fund groups, can have a preference for an interested chairman and can reasonably conclude that the best interests of fund shareholders are promoted by having an interested chairman. This is so when independent directors rightly perceive that an individual brings considerable expertise to the performance of his or her responsibilities, has a strong commitment to serving fund shareholders, and has demonstrated the high ethical standards required of any fiduciary. The Commission should not, by rule, prevent independent directors from exercising their own judgment on these issues.
Other Proposed Measures to Strengthen the Effectiveness of Independent Directors Make the Independent Chairman Rule Unnecessary
The Release asks whether the proposed requirement for a supermajority of independent directors on every fund board renders the independent chairman proposal unnecessary. In our view, the answer, quite clearly, is yes. With a supermajority, independent directors will certainly have the authority to select an independent chairman if they believe this would best serve the interests of a fund's shareholders. More broadly, independent directors, acting as a supermajority, will be in a position to determine the outcome of any matter put to a board vote and can effectively control all aspects of the board process, including the scheduling and duration of meetings, the setting of the agenda, the flow of information prior to and during board meetings, and the structure and membership of board committees. The autonomy of independent directors will be reinforced by the proposed requirement that they meet in executive session at least quarterly.
Of course, the measures proposed in the Release to strengthen the role of independent directors build upon a foundation of current safeguards. These include the statutory responsibility of independent directors, voting as a separate group, to approve the adviser's management contract and the fund's principal underwriting agreement. The existing authority of independent directors to select and nominate their successors provides an additional safeguard to allow independent directors to perform their oversight role without undue influence from the adviser. Recent rule changes ensuring the authority of independent directors to retain independent legal counsel and to approve the appointment of a fund's chief compliance officer who reports directly to the board further reinforce the autonomy of independent directors.
In sum, we see no purpose being served by depriving independent directors of the ability to reach informed judgments on the selection of a board chairman from among any member of the board, whether the individual happens to be an independent director or an interested director.
Empirical Data Strongly Indicate No Positive Correlation Between Independent Chair Funds and Either Better Investment Performance or Lower Expenses
At the Commission's hearing to consider the pending fund governance proposals, two members of the Commission expressed an interest in considering empirical data that might indicate whether shareholders of funds with independent chairs have fared better than shareholders of funds led by interested chairs. The staff indicated at the meeting that they had not reviewed any such data and were unaware that any studies had been undertaken.1
Following the Commission's meeting, Fidelity engaged two respected fund industry consultants to conduct a study. We include their report, the Bobroff-Mack Report, as an attachment to this comment letter.2 The results strongly indicate that there is no positive correlation between independently chaired funds and either better investment performance or lower expenses when compared to funds with interested chairs. Indeed, the conclusions reached in the report are quite to the contrary, namely:
(1) With regard to investment performance, "[o]n each of several historical performance measures, independent chair funds have not performed as well as those having management chairs. For example, using Morningstar's fund rankings within style-based peer groups, independent chair funds on average rank in the 53rd percentile (100=best) over the past three years, while management chair funds on average rank in the 58th percentile. Over ten years the ranking difference is more pronounced, with the independent chair funds averaging in the 48th percentile versus the 59th percentile for the management chair funds. For these and the other performance comparisons included in this study, the differences were statistically significant."3
(2) With regard to expenses, the Bobroff-Mack Report found that "[t]he expense examination showed no significant positive correlation between independent chair funds and lower expenses. Independent chair funds were found to have competitive to high expense levels, depending on the way expenses are measured and aggregated. For example, when distribution-related charges are excluded and equal-weighted averaging is used, independent chair funds have annual expense ratios which average 0.01% per year lower than comparable management chair funds, which is not a statistically significant difference. When expenses are asset-weighted, independent chair funds have 0.16% higher expenses than management chair funds."4
In our view, the results of this study raise serious questions regarding the wisdom of forcing independent chairs upon funds that have delivered superior performance and lower expenses to their shareholders. At the Commission's public meeting in January, the Commission staff acknowledged that they were not aware of any empirical data attempting to compare independent chaired funds and interested chair funds from the standpoint of performance or expenses. There was also some discussion that the Commission could consider "anecdotal evidence" in reaching its final position on the independent chairman proposal.
We respectfully suggest that, to the extent that anecdotal evidence is considered to have any relevance at all, it certainly presents an unpersuasive case for the Commission's proposal, given the number of fund groups with independent chairmen that have been embroiled or implicated in the wrongdoings that have been exposed over the last six months. Second, empirical data is generally preferable to anecdotal evidence in the formulation of public policy. The empirical data that we submit for the Commission's careful consideration strongly caution against adoption of a proposal to require that all fund boards be chaired by an independent director without regard to the particular facts and circumstances that might apply to any given fund or fund group. Indeed, in some instances (and perhaps in many instances), independent directors could rightly come to the view that fund shareholders can be better served by a board chaired by an individual who happens to be an interested director.
At the Commission's open meeting, Commissioner Glassman asked the staff to develop empirical data comparing independent chair funds and interested chair funds to assist the Commission in reaching a final decision on whether to adopt the independent chairman requirement. Other Commissioners expressed an interest in evaluating this data. If such a review has been undertaken, then we respectfully suggest that the Commission should publish this data and invite public comment. Indeed, this would appear in keeping with the Administrative Procedure Act's requirements to afford interested persons adequate notice and a meaningful opportunity to comment on an agency's proposed rules.5
Broader Implications of an Independent Chairman Requirement
It is noteworthy that the Directors' Committee of the ICI has endorsed an approach very similar to ours, and by its letter of December 31, 2003, has urged the Commission to preserve the ability of fund boards to exercise the full range of their business judgment in choosing a board chairman. Members of the Directors' Committee who are independent directors far outnumber those who are interested directors.
Our views, in fact, are fully consistent with the Commission's own position, set forth in the testimony of Paul Roye, Director of the Division of Investment Management in testimony before the House Financial Services Subcommittee on Capital Markets on June 18, 2003. Speaking for the Commission, Mr. Roye pointedly did not endorse the provision of H.R. 2420 which would have deprived fund boards of authority to select an interested director to serve as chairman. In light of the bill's requirement that at least two-thirds of a board consist of independent directors, he noted that this would empower independent directors to select one of their own as board chairman if they so desire.6 We also believe Mr. Roye was correct in noting that a supermajority requirement for independent directors would ensure that those directors would be free to exercise their business judgment regarding the selection of any director to serve as board chairman. We submit that none of the well-publicized problems in the mutual fund industry that have been brought to light in the last six months negates the correctness of the Commission's position expressed last June.
We also urge the Commission to consider the implications for the rest of Corporate America that would flow from a governmental fiat that mutual fund boards, without exception, be chaired by an independent director -- regardless of whether a fund's directors, left to their own business judgment, might conclude that the interests of shareholders would be best served by an interested director serving as chairman. Corporate governance experts express widely differing views on the merits (and demerits) of an independent director chairing corporate boards.
Professor Charles Elson, Director of the Center for Corporate Governance at the University of Delaware, College of Business and Economics, has expressed serious reservations over the "non-executive" chairman (even in the absence of a governmental rule that would require this result):
We are aware that mutual funds are unique in that their operations are externally managed by investment advisers and other service providers, and that this structure gives rise to potential and actual conflicts of interest. Mutual fund boards, and the independent directors of those boards, have singular responsibilities under the Investment Company Act to act for the benefit of fund shareholders in resolving these conflicts of interest.
To assume, however, that conflicts of interest arising within operating companies somehow pose lesser risks to shareholders or are less acute than those faced in the fund industry, simply because operating companies have "internal" management is to ignore the history of corporate governance in this country -- a history punctuated by the debacles of recent years, such as Enron, WorldCom and Tyco, that have imposed heavy losses on mutual funds and other investors.
We submit that there is no principled distinction between mutual funds and other companies on the question of whether board directors should be free to exercise their informed business judgment to select any member of the board to serve as its chairman. It is unwise public policy for the mutual fund industry and unwise for every other U.S. industry as well to impose a "one size fits all" independent chairman requirement. If the Commission deprives fund boards of the discretion to choose their chairmen, this will have unmistakable implications for the corporate boards of all other American companies.
We hasten to acknowledge that fund directors might well decide upon an independent director to serve as board chairman in particular cases, as they have already done in a number of fund complexes. We also acknowledge the possibility that the selection of independent directors to serve as board chairmen may emerge over time as the norm in the fund industry. If this reflects the informed business judgment of fund boards, and their independent directors, this is as it should be. Fund boards, not the government, should make these decisions.
On the other hand, for over half a century, the Fidelity Funds Board has reached the informed judgment that the Funds' shareholders have been well-served through the strong leadership and vision of the Johnson family, the founders of the Fidelity Funds. In recognition of this, the Board chose as its chairman, Edward C. Johnson 2nd and has chosen to be led by its current chairman, Edward C. Johnson 3rd. It is open to question whether all of the innovations that have advanced the interests of Fidelity Funds' shareholders over so many years, including the enormous commitment to the use of technology, could have been achieved if the Trustees of the Fidelity Funds had been prohibited from exercising their judgment in choosing the Board's chairman. We respectfully suggest to the Commission that it not deprive the Fidelity Funds Board - or the board of any other fund complex - of the authority, and responsibility, of choosing its chair.
* * * *
We appreciate the opportunity to comment on the Commission's most recent fund governance proposals. In summary, we are generally supportive of those proposals but strongly urge the Commission to reconsider its proposal to curtail independent directors from exercising the full range of their business judgment in selecting who shall serve as chairman of a fund's board.
cc: Hon. William H. Donaldson
Hon. Paul Atkins
Hon. Roel Campos
Hon. Cynthia A. Glassman
Hon. Harvey Goldschmid
Division of Investment Management
Assessing the Significance of Mutual Fund Board Independent Chairs
A study for Fidelity Investments by
Fidelity Investments commissioned this study to see whether mutual funds which have independent, disinterested directors serving as board chairs have provided shareholders with better performance or lower expenses than funds with interested or management-affiliated board chairs. The study encompassed all retail-oriented fund complexes with $10 billion or more in long-term assets, which account for 83% of industry long-term fund assets and included such analysis as was appropriate and practical given the time available. Data from Morningstar, Lipper and Strategic Insight were utilized, which are widely used and respected industry data sources. The key findings were as follows:
On each of several historical performance measures, independent chair funds have not performed as well as those having management chairs. For example, using Morningstar's fund rankings within style-based peer groups, independent chair funds on average rank in the 53rd percentile (100=best) over the past three years, while management chair funds on average rank in the 58th percentile. Over ten years the ranking difference is more pronounced, with the independent chair funds averaging in the 48th percentile versus the 59th percentile for the management chair funds. For these and the other performance comparisons included in this study, the differences were statistically significant.
The expense examination showed no significant positive correlation between independent chair funds and lower expenses. Independent chair funds were found to have competitive to high expense levels, depending on the way expenses are measured and aggregated. For example, when distribution-related charges are excluded and equal-weighted averaging is used, independent chair funds have annual expense ratios which average 0.01% per year lower than comparable management chair funds, which is not a statistically significant difference. When expenses are asset-weighted, independent chair funds have 0.16% higher expenses than management chair funds.
While there are differences other than chair type between the independent chair and management chair funds that may have played a role in these results, we did not find evidence of better performance or lower expenses for the independent chair funds. Rather, we found that independent chair funds have not performed as well as management chair funds and that independent chair funds' expenses are competitive to high depending on how expenses are measured and compared.
The following sections summarize the key performance and expense results, discuss the analytical approach and methodology, and then provide detailed results in the Appendix.
The purpose of this study was to examine whether mutual funds with independent board chairs have provided their shareholders with lower fund expenses or higher fund performance than other comparable funds. To accomplish the study, expense and performance data for the retail-oriented fund families with more than $10 billion in long-term fund assets was assembled and used to compare results for funds having an independent director as the board chair (independent chair funds) with those for other funds (management chair funds).
The study focused on a set of larger fund families rather than the total universe of funds in order to facilitate determining which of the subject funds had independent chairs, as well as to allow a more detailed review of fund data and classifications than would be practical working with all funds. In addition, this approach provides for reviewing results family-by-family, so that the distribution approaches and other unique characteristics of fund families can be considered in interpreting the results. Other key elements of the methodology are discussed beginning on page 6.
The performance analysis results are summarized in Exhibits 1 and 2, which show average Morningstar performance rankings and star ratings for independent and management chair funds. As shown in these exhibits, there is no positive correlation between independent chair funds and better performance. Rather, funds with independent chairs have underperformed those with management chairs, and the differences are statistically significant.
Over 3, 5 and 10 year periods, the independent chair funds averaged (equal-weighted basis) 5-11 percentage points lower rankings within Morningstar categories than management chair funds. Statistical tests, which are discussed in the appendix, show that these differences are significant. While equal-weighting is the more usual way of assessing aggregate performance, the asset-weighted figures do express an important "bottom line" of what was experienced by the average dollar invested. When the rankings are asset-weighted, the differences are even larger, with management chair funds having average rankings 13-19 percentage points better than for independent chair funds. For both the equal- and asset-weighted figures, the largest differences were noted for the longest (10-year) time period examined.
Similarly, management chair funds have higher average Morningstar star ratings. On a 1-5 scale, management chair funds average 0.5 more Morningstar "stars" than the independent chair funds, and again the differences are statistically significant. And the difference is larger if the data are asset-weighted, with the management chair funds averaging 0.8 stars more than for the independent chair funds.
Morningstar's star rating methodology factors in the impact of sales loads on returns for funds with loads. While this may be helpful to investors that pay loads, it results in a potential bias in this study against independent chair funds, since a higher percentage of these funds are sold with loads than management chair funds. However, as discussed in the Appendix, tests indicate that the direction and significance of the star rating differences reported above have not been affected by this bias.
For convenience and ease of understanding, we have presented average values for the Morningstar rankings and ratings; the appendix presents the actual distributions of the rankings and ratings in categories and presents appropriate statistical tests on those distributions.
An additional test, independent of Morningstar rankings and star ratings, confirms the results above. A returns-based style analysis was performed on each fund having 5 years of monthly returns available and was used to determine for each fund a custom blend of market indexes that best matches the fund's style. Then, for each fund, a risk-adjusted excess return was calculated. Risk-adjusted excess return, also known as alpha, measures the extent to which a fund performed better or worse than the benchmark, after adjusting for risk. The risk-adjusted excess returns show a higher average return for management chair funds than independent chair funds, and once again the results are statistically significant. Please see the appendix for more details.
Of course expenses have a direct impact on performance, but this does not appear to be a significant factor here. The equally-weighted alpha average, which is an annual performance percentage figure, is 0.76% higher for the management chair funds than the independent chair funds, while as discussed in the next section the equally-weighted total expense averages for these two groups differ by only 0.05% -- a small figure compared with the performance difference. The degree of disparity allows us to conclude that the difference in returns is not attributable to the differences in expenses.
Why independent chair funds have performed less well than management chair funds is an interesting and challenging question. Apart from having different types of board chairs, the two groups of funds have other important differences that may have impacted performance results:
The independent chair fund groups are mostly bank-based, sales force oriented fund groups, which distribute their funds importantly through the banks' own trust departments and brokerage arms. In contrast, the management chair fund groups are mostly so-called "wholesale" firms: sales force oriented groups that sell mainly through third-party broker-dealers and other distributors. It is possible that differences in the types of clients served (for example, being more or less conservative) or other distribution-related factors could have influenced performance results. We do note that both the Morningstar star ratings and the alpha figures take into account investment risk and therefore should fairly compare riskier vs. more conservative investment approaches.
For the reasons discussed earlier we have focused on the larger fund groups, and the independent chair funds and fund groups tend to be smaller than the other firms. While there are independent chair groups dispersed across the size range in the study, their average size is about half that of the others ($38 billion vs. $80 billion). Therefore, as it happens the independent chair firms are being compared against mostly larger firms, which by definition have been more successful in asset gathering, which may be because they have produced particularly good investment performance.
Exhibit 3 summarizes the expense analysis results, showing equal- and asset-weighted expense averages, including and excluding ongoing distribution related expenses. The results are expressed as the number of percentage points that annual expenses are on average better or worse than peers. The expense analysis comparisons tend in favor of the management chair funds, but the results are less clear than with the performance results, and they differ considerably depending on what expense measure is used.
Using the total expense ratio, the independent chair funds' expenses vs. peers have an equal-weighted average of about .05% more than for the management chair funds, and this difference is statistically significant. The difference is much larger (0.24%) if the data are asset-weighted.
Using expenses excluding distribution expense, the independent chair funds have a 0.01% lower average expense ratio, which is not a statistically significant difference. However, when the data are asset-weighted, the independent chair funds have a considerably higher expense average, 0.16% above those of the management chair funds.
And so the interpretation of these results depends importantly on what measure is used:
Regarding total expenses vs. expenses less distribution, since load and no-load funds are being compared, expenses less distribution seems the better choice. This is because distribution-related fund expense charges (12b-1 fees and service fees) of load funds are used to compensate financial advisors and are in that sense the equivalent of sales charges, which of course do not generally exist for no-load funds. However, there are some no-load funds which have 12b-1 fees that are not used to compensate advisors but rather to defray other distribution expenses. In addition, it is increasingly common for load funds to use 12b-1 fees to pay for non-advisor-related distribution services. Therefore, while expenses less distribution may be the better single measure, total expenses are also presented.
Regarding equal- vs. asset-weighted figures, both deserve consideration. The asset-weighted figures do present the "bottom line" for the average dollar invested, and they also relate to the total dollars fund shareholders pay to fund sponsors and others for services. However, the asset-weighted averages tend to be dominated by a few of the largest, low-expense complexes and in turn their largest funds. The equal-weighted figures avoid this problem, but they can be distorted by small, high-cost funds inflating the averages.
Based on the discussion above, there is a mixed picture, with the independent chair funds having expense levels that are competitive to high vs. management chair funds, depending on how the comparisons are emphasized. To the extent that the independent chair funds have higher expense ratios, the fact that their fund families are among the smaller complexes included would likely be a factor; larger complexes generally have the advantage of lower expenses.
Discussion of Methodology
The key elements of the study's methodology are summarized below; please see the Appendix for further details.
The Strategic Insight Simfund mutual fund database, a widely used source of fund industry competitive information, was the principal source for identifying and categorizing fund families and funds. Using this database, all fund complexes with $10 billion or more in long-term, open-end fund assets as of December 31, 2003 were identified - 68 in all. To allow meaningful expense comparisons, the 55 of these complexes which have a significant retail (sales force and/or direct marketed), active management fund business and that offer industry-standard retail load or no-load pricing were selected for the analysis. Two of these complexes (Fidelity and Dreyfus) were separated into sales force and direct market product lines, so there are a total of 57 fund families included, as listed in Exhibit 4.
To determine which of these mutual fund groups operates with an independent chair or a management chair, we reviewed various documents on file with the SEC or available from each fund group. We examined as a minimum the latest Statements of Additional Information and Annual Reports for sample funds. We were able to determine whether a group had a single board or multiple boards from filing documents, and where more than one board existed within a fund group we examined the documents covering each board. Based on this review, each of the 57 families has been identified as either an "independent chair" or "management chair" family, based on all or substantially all of its funds having an independent board chair, or not. On this basis, 14 of the 57 fund families have independent chairs, and 43 have management chairs. In making these determinations we did not consider recent chair type changes (e.g. Strong), which would not have impacted historical performance or expenses. While there has been some consolidation of fund boards in recent years, to the best of our knowledge the type of board chair for each of these complexes has been substantially the same as reported for at least ten years. However, a limitation of the study is that there will be funds included in a family's historical data which operated in a different board chair setting at an earlier time. For example, Wells Fargo (an independent chair complex) has acquired several small fund complexes over the years, which generally had management-affiliated chairs prior to their acquisition; by convention, industry data sources include this data under the new family, and it would be impractical to segregate this information without a great deal more time and research. Exhibit 4 also lists the chair designations.
Nine of the 14 groups with independent chairs are affiliated with a banking institution, reflecting the earlier banking laws (Glass-Steagall), which prohibited a banking institution from sponsoring a fund family. Most of these bank-based fund families were created through conversion of common trust funds or other pooled fiduciary accounts. The other five fund groups operating with independent chairs moved to that status in some cases occasioned by transactions involving the sale or control of the management companies. Two of these non-bank-affiliated groups have operated with independent chairman for over thirty years.
For each fund family, a single pricing format or share class was chosen for both the expense and performance analysis. This was the "A" or front load class for sales force funds and the principal retail no-load class for direct marketed funds, with a minimum investment of $10,000 or less required for inclusion. Thus, while a given fund might have a number of share classes, it is represented in the analysis by a single class, selected to be as comparable as possible across the funds included. Based on availability of data, a total of 2,101 funds were included in one or more performance comparisons, and 2,184 funds were included in the expense analysis.
For the expense analysis, each fund was classified into one of the 29 categories shown in Exhibit 5, which are based on a combination of Strategic Insight and Morningstar fund classifications and designed to make essential distinctions for expense analysis while keeping peer groups as large as possible. All the expense analysis is based on comparing each fund's expense figure with the average for all funds in its expense category, so that funds are compared only against relevant peers. Large-capitalization equity index funds are included and have their own category, but other index products are excluded. Funds in certain other specialized investment objectives (e.g. some single-country foreign equity funds) have been excluded due to small peer groups. The expense data draws on both Strategic Insight and Lipper Analytical to maximize coverage.
The expense analysis covers expenses with and without annual distribution-related (12b-1 and service) fees, but sales charges are not included. While load fund sales charges are important in assessing fund ownership costs over time, our main interest is in comparing costs across funds at a point in time, including both load and no-load funds, and including amortized sales charges would make load and no-load fund expenses less rather than more comparable. In addition, including sales charges would disfavor independent chair funds, nearly all of which are load funds. Finally, most major load fund suppliers indicate that a large but not-publicly-reported fraction of their A share sales are occurring in load-waived form - so that fund-level analysis including sales charge effects would be impractical.
Fund expenses were analyzed using the most recent data available. Mutual fund expenses are reported for each fund's fiscal year and are typically captured in databases with a 2-3 month delay from fiscal year-end. The actual expense figures were taken from the Lipper database and supplemented with Strategic Insight data where possible.
Performance analysis requires a finer set of categories than expense analysis to reflect investment style differences, and we have incorporated the widely used Morningstar performance rankings and risk-adjusted star ratings, which are based on comparisons across about 16,000 funds sorted into their 64 investment style-based categories. The performance rankings for the 3, 5 and 10 years ended December 31, 2003 are expressed as a simple percentile rank of a fund's performance within its category, presented here with 100 being best and 1 being worst (Note to Morningstar users: for presentation purposes, the ranking scale has been reversed from Morningstar's practice). The star ratings, which range from 5 (best) to 1 (worst), use a complex methodology that takes into account both risk and return.
While all performance rating approaches have some limitations, discussed below, we believe the Morningstar rankings and star ratings provide a generally reasonable and unbiased performance picture, especially when averaged across many funds. In addition to the Morningstar figures, we have independently calculated and compared risk-adjusted returns against custom benchmarks for each fund.
It should be noted that the Morningstar rankings and star ratings are not necessarily true performance scales, in that for example having 5 stars compared with 4 may signify a greater or lesser performance benefit than having 4 stars compared with 3. However, they are widely used figures of merit which can be compared across types of funds, and we use them in averages for convenience of comparisons. The appendix includes more technical statistical tests that result in similar findings, thus validating the use of averages.
Both the expense and performance analyses use averages to summarize results. Averages are calculated across all of the funds of independent chair vs. management chair fund families, as well as for each fund family separately. The simple or equal-weighted averages indicate the expenses and performance of the average fund, while the asset-weighted averages are more appropriate when considering the expenses and performance experienced by the average dollar invested in the funds. The asset-weighted averages use portfolio-level assets (summed across all share classes), since the fund share class analyzed will generally reflect the performance and expense experience of the fund's other classes, except for distribution-related charges.
The averaging calculations are somewhat different for the performance and expense data. While the performance rankings and ratings can be directly combined across different types of funds, the expense data needs to be combined in a way that reflects the substantial expense level differences across the expense categories. That is, expense averages across different types of funds need to reflect that some types have higher expenses (such as international equity funds) than others (such as municipal bond funds). To deal with this, each fund's expense level is first expressed as a difference better or worse than the average expense level for its expense category. Then these differences are averaged to produce an overall better/worse-than-average figure for that fund family or group of funds.
This study was conducted under significant time constraints in order to meet a deadline for comments on rule changes proposed by the Securities Exchange Commission that would require all fund boards to be chaired by an independent director. These constraints limited the possibilities for further analysis beyond that described above, as well as the fund and fund complex research that could be accomplished and the range of complexes and funds that could be included. Within these constraints, we did not find that mutual funds with independent board chairs have provided their shareholders with lower fund expenses or higher fund performance than other comparable funds. Rather, we found that funds with independent board chair have not performed as well as those having management chairs and that independent chair funds have competitive to high expense levels, depending on the way expenses are measured and aggregated.
This Appendix presents the detailed analytical results and associated statistics, covering first the fund coverage statistics and then the performance and expense analysis details.
Fund and Asset Coverage
For both the performance and expense analysis, long-term open-end fund portfolios of the target complexes were eligible to be included if they offered an industry-standard load (A share) or no-load retail share class, with an investment minimum of $10,000 or less, as reported in the Strategic Insight Simfund database as of December 31, 2003. This excluded any pure institutional funds but allowed funds that have both institutional and retail classes. A total of 2,437 funds with $3,728 billion in portfolio-level assets were eligible by these criteria, or 83% of all industry long-term open end fund assets.
We thought it important to identify each complex in the analysis with a distribution and pricing approach, and in the case of Fidelity and Dreyfus we separated their product lines into their direct and sales force distribution components. With other complexes, we excluded funds which did not conform to the firm's main distribution and pricing format.
Exhibit A-1 reports the number of funds and assets eligible for each complex and the funds and assets included in each performance calculation. To be included in a Morningstar ranking or rating calculation, the fund's representative load or no-load share class would need the required length of track record and to be covered in the Morningstar Principia database. For the alpha calculations, it was required that the representative class have five years of monthly returns available in this database. Overall, 86% of the eligible funds, representing 97% of the eligible fund assets were included in 3-year rankings and Star rating calculations, both of which require a minimum 3-year track record. The most common reason for funds not being included in the analysis was lack of a sufficient track record.
Exhibit A-2 reports coverage statistics for the expense analysis, beginning with the same base of eligible funds as for the performance analysis. To be included in the expense analysis a fund needed to have expense information available from either Lipper or Strategic Insight and also to meet certain investment classification criteria. Funds were excluded where we were unable to identify an appropriate peer group across the funds included - generally at least ten funds considered to be reasonably comparable for expense analysis purposes. Examples of funds so excluded include prime rate funds and certain single-country foreign equity funds. Most importantly, while large-capitalization equity index funds (which encompasses S&P 500 index funds) were included as an expense peer group, other types of index funds were excluded.
Exhibit A-3 presents the key performance analysis results, where equal- and asset-weighted figures are presented for management and independent chair funds, further broken down by major asset class. The figures for Morningstar ranking and star rating averages are straightforward averages of information taken from the Morningstar Principia database as of December 31, 2003.
Reviewing the results in Exhibit A-3 shows the following:
Over three, five and ten-year periods, average percentile rankings against other funds in the same Morningstar investment category (e.g. large cap growth or short-term bond) are in favor of the management chair funds. In each case, a t-statistic is calculated for the equal-weighted averages; a value above about 2 indicates that, from a statistical perspective, the difference between the figures for management and independent chair funds is significant at the 95% level, or a 5% or less chance that the difference occurred by chance. When the data are asset-weighted, the differences are even larger.
The pattern of results is quite similar for the Morningstar star rating and alpha averages, with the management chair funds providing the stronger results and the differences being significant and larger for the asset-weighted data.
The alpha analysis was conducted as follows:
Each eligible fund with at least 60 months of monthly returns available was subjected to a returns-based style analysis, where the portfolio of six market indexes that most closely matches the fund's pattern of returns over the 60-month period is determined. This was accomplished with standard quadratic programming tools, where the set of indexes varied as shown below by type of fund. All international index returns are expressed in U.S. dollars, and all fund returns are returns to shareholders, net of expenses but not taking into account any up-front or deferred sales charges.
Then this custom benchmark for each fund was used to calculate a risk-adjusted excess return or alpha. This was done with a simple regression of fund excess returns on custom benchmark excess returns, excess here meaning the amount by which each out- or under-performed U.S. T-bills. The alpha is a statistical estimate of how much the fund has out- or under-performed the custom benchmark on an annual basis, after adjusting for the so-called beta effect. The beta effect is the tendency for fund returns to mirror the pattern of the benchmark returns, which can be accomplished by owning the benchmark and borrowing or lending cash and therefore is not considered part of the manager's value-added.
There are of course far more sophisticated ways to establish fund benchmarks and assess manager value added. However, we believe the approach used is a generally fair and reasonable way to assess manager value added adjusted for risk, especially when applied and averaged over large groups of funds.
As mentioned in the body of the report, the Morningstar star ratings utilize a complex methodology, and we refer the reader to Morningstar.com for the full details. The method does take into account both risk and return and is therefore appraises returns in the context of risk undertaken. There is one methodological factor of the star rating which presents an issue for this study: in assessing returns, Morningstar amortizes the maximum sales loads of load funds against performance. While sales loads are a reality for retail customers making traditional load purchases through brokers, as mentioned earlier more and more "load" funds are being purchased via mutual fund marketplaces, defined contribution plans, or mutual fund wrap programs, where the load is waived. This is important here since, as shown in Exhibit A-4, 97% of the eligible independent chair funds and 95% of the related assets are in load funds, compared with 70% of the funds and 54% of the assets for the management chair funds.
It is not practical to reverse-engineer the Morningstar star ratings to remove the load effect. However, we can gain some insight to the effect by recalculating our star averages looking only at load funds. This approach is somewhat unfair to the management chair group because some fund groups with particularly strong performance are excluded. However Exhibit A-5 shows the star rating averages still favor the management chair fund group (though to a lesser extent) and are still statistically significant.
Another potential issue is that load funds tend to have higher expenses than no-load funds, which can impact performance rankings. Exhibit A-5 also includes performance rankings including only load funds; all the comparisons continue to favor management chair funds, and the 3-, 5- and 10-year comparisons continue to be statistically significant. As with the star ratings above, these comparisons disadvantage the management chair group as some strong-performing groups are excluded.
As alluded to in the body of the report, the Morningstar rankings and ratings are formally what is known as rank-order or ordinal data, raising a question as to whether averages and related tests could be misleading. In Exhibit A-6, we present the actual distributions of the Morningstar star ratings, for both all funds and load funds only, by chair type, along with T-statistics associated with an appropriate measure for rank-order data (Somers' D). As shown, the T-statistics are actually slightly greater than those presented with the averaged data. In Exhibit A-7, the distributions of ranking data have been summarized into performance quartiles (e.g. rankings of 76-100 equate to the top quartile), and again the related T-statistics are somewhat higher than those presented earlier.
For reference, Exhibit A-8 reports performance analysis results by fund complex.
The principal expense analysis results are reported in Exhibit A-9, which shows equal- and asset-weighted expense averages by type of chair, both including and excluding distribution-related expenses and further broken down by major asset class.
The major comparisons and their significance were already discussed in the body of the report; additional comments and details are as follows:
The figures reported are averages of the difference between a fund's expense level and the equally-weighted average for its peer group (see earlier discussion for the peer grouping used). Therefore, a positive figure indicates that the funds in question have on average a higher expense ratio than their peers, and a negative figure indicates lower-than-peers average expenses. Figures for asset-weighted averages are generally negative (below average), since the larger funds that are more heavily weighted in these calculations generally have lower expenses than smaller funds.
Please see the prior section for a discussion of the t-statistics reported.
The breakdowns by asset class show some differences from the overall figures, which is to be expected.
For reference, Exhibit A-10 reports expense averages by complex.