Division of Investment Management. 2000. Report on Mutual ...

A Comparison of the Performance of Actively Marketed Mutual Fundsand Directly Marketed Funds Using Panel Data

by

Albert E. DePrince, Jr.Professor of Economics and Finance

Box 27Middle Tennessee State University

Murfreesboro, TN [email protected]

615-898-5995

presented to

Southwestern Finance Association46th Annual Meeting

San Diego, CAMarch 13-17, 2007

Abstract

Mutual funds are marketed in two broad ways: direct marketing which involves no active sales force and (2) active marketing through a sales force. The distinguishing feature between the two methods is a distribution fee known as a 12b-1 fee and a sales commission or load paid by the buyer of actively marketed funds. Some argue that there should be no difference in the asset management between actively marketed and passively marketed funds. However, this is an assertion with little empirical proof. At the same time, third-party marketing channels will not be anxious to handle funds that have sub-par performance. Thus, it is possible that there is a difference in portfolio management between actively marketed and directly marketed funds. To investigate this possibility, this study compares the total returns of funds that rely upon third-party sales forces with funds that rely upon direct marketing. Simply put, the research questions are the following. First, is performance of actively marketed funds comparable to the performance of directly marketed funds? If not, what is the difference and how sizeable is the difference? Methodologically, to avoid problems with multiple share classes, Class A share is used as the representative funds with active distribution. A filtered version of the Morningstar classification for no-load funds will be used as the directly marketed funds. No-load funds that are a share class within a family of share classes that include actively marketed funds are excluded from the universe of no-load funds. To assure the no distortions may be introduced through different mixes of actively managed and index funds in either marketing approaches, index funds are removed as part of the data filter. Thus, only actively managed and actively marketed funds will be compared with actively managed and directly marketed funds. Finally, funds will be stratified by Morningstar categories so that the Morningstar categories serve as the panels in this study.

Primary Conference Code: 200Secondary Conference Code: 021

JEL Code: G2, G20, G23Key Words: mutual fund performance, 12b-1 fee, actively marketed funds

A Comparison of the Performance of Actively Marketedand Directly Marketed Funds Using Panel Data

Introduction

This study compares the total returns for actively and passively marketed mutual funds, in order to assess whether the portfolio management of actively marketed funds differs from the management of directly marketed funds. In the following paragraphs, the rationale for this study is presented, followed by a review of the differences in fees between actively and directly marketed funds and likely effects on differences in performance, details on the research hypothesis, the research model used to test the hypothesis, data and data filters, the empirical findings, and an epilogue on the importance of the study.

Mutual funds are marketed in two broad ways: (1) direct marketing, which involves no active sales force, and (2) active marketing through a sales force. Typically, a distribution fee known as a 12b-1 fee and a sales commission or load paid by the buyer of actively marketed funds is the distinguishing feature between the two marketing approaches. Funds marketed directly to the investing public are typically termed no-load funds. These funds have no 12b-1 fee or a very minimal fee of no more than 25 basis points and no commissions. Thus, if management of the funds is independent of the marketing method, the 12b-1 fee should be a deadweight loss to the actively marketed funds.

Given the different marketing arrangements and their associated fees, some have argued that load funds are “sold” while no-load funds are “bought” (Grove, 2002). Here, there is the supposition that returns on the two types of funds probably differ systematically, since the active marketing of funds puts them in more intense competition with other actively marketed funds compared with the competition among funds that are passively marketed. In other words, relative returns are an important ingredient in the sales effort. Third-party marketing channels will not be anxious to handle funds that have subpar performance.

Thus, the implication is that comparative return is of greater importance to investors in load funds compared with no-load funds. This suggests the possibility that managers of actively marketed funds may behave differently than managers of passively (directly) marketed funds. Some argue that there should be no difference in the asset management between actively marketed and passively marketed funds (Ferris and Chance, 1987). However, the Ferris and Chance study left this as an assertion and went on to examine expense ratios.

1

Background on the 12b-1 Fee

While the purpose of this paper is not an evaluation of 12b-1 fees, they are the distinguishing feature of actively marketed funds. Hence some understanding of the fee might be useful before presenting the research hypothesis. Actively marketed funds carry a two-part fee to incent sellers. One part is a one-time charge or commission known as a load; the other is a perpetual fee expressed as a percent of net assets and known as a 12b-1 fee. The 12b-1 fee was approved by the SEC in 1980 to help mutual funds finance the distribution of new shares. In doing so, the SEC used the exemptive rule authority granted by Congress in the Investment Companies Act of 1940. This act permits the SEC to exempt the prohibition against prohibited activities “… when the exemptions are in the public interest and consistent with the protection of investors…” (SEC, 2005). Exemptive Rule 12b-1 permitted funds to use their own assets to pay for the sale of new shares (SEC, 2005) and is thus expressed as a percent of fund net assets. Though initially viewed as a temporary measure (Hillman, 2004; Walsh, 2004), the fee has, in effect, become a permanent feature in the distribution process.

Current SEC rules limit the 12b-1 fee to a maximum of one percentage point (100 basis points) of net assets held by investors subject to the fee. Since 1993, the National Association of Securities Dealers (NASD, 2005; Hillman, 2004) has further limited the portion of the fee for distribution to 75 basis points and allowed up to another 25 basis points for administrative costs faced by the third-party marketers of the funds. Also, current rules allow a fund with a 12b-1 fee of 25 basis points or less and no load or commission to be classified as a no-load fund (Division of Investment Management, 2000).

The fee is paid to the fund distributor, and most of the fee is passed on to the intermediary that sold the funds. In addition, the fee can be used cover to the following items: advertising, and printing and mailing of prospectuses, and sales literature (Hillman, 2004). The SEC prohibits the inclusion of distribution expenses in excess of the 12b-1 fee as an expense in evaluating the profitability of the fund’s advisor. However, the fund advisor can make additional payments to the distributor (Hillman, 2004) but cannot include them as expenses for purposes of determining advisor profitability. Finally, the exemptive rule requires that the independent directors must determine that the fee is in the best interest of the shareholders.

The 12b-1 fee and the commission are related. A larger commission is associated with a smaller 12b-1 and vice versa. Actively marketed funds come in various share classes, with the size of the distribution fee and the commission being the distinguishing feature among them. At year-end 2005, there were $2.8 trillion in assets reported as either A, B, C, D, M, or N shares.

With over 25-years since its authorization, questions have arisen as to the purpose of the fee from the investor’s perspective. Several justifications are reported in the literature. Of these, the most important, from the investor’s perspective, is the services provided by the seller of the fund. This includes both investment advice and other types of assistance (ICI, 2005; Collins, 2004). Others have noted that it affords investors an opportunity to pay the sales charge for the acquisition of their shares through a number of alternative mixes of loads or commissions and 12b-1 fees (Collins, 2004; Reid and Rea, 2003). In this sense, some have compared the 12b-1 fee

2

to the payment of sales commissions on the installment plan (ICI, 2004); however, unlike commercial installment plans, the 12b-1 fee extends over an infinite horizon.

So far, investigations have focused on the 12b-1 fee from the perspective of a cost to the shareholder which needs to be measured against the benefit. They have not addressed the possibility that the more critical difference could be the difference in the management of funds actively marketed or directly marketed due to the different clientel effect of investors in the two types of funds.

Thus, it is not surprising that there is far from universal agreement on the merits of 12b-1 fees, with critics focusing on benefits of the fee to shareholders. Ferris and Chance (1987) compared 12b-1 plans with no-load funds. They concluded that the 12b-1 fee is a deadweight loss and found no evidence of systematic differences in expense ratios. This study involved a relatively small number of funds and covered only two years. Thus, it is hard to generalize from it.

Latzko (1998) expanded studies such as Ferris and Chance. This study also focuses on expense ratios and examined 2,610 funds encompassing 22 different investment objectives. The study finds evidence of economies of scale in fund administrative costs. However, the economies of scale were generally exhausted at about $3.5 billion in assets. The paper discusses 12b-1 fees and at least creates the impression that these economies of scale at least partly derived from the ability for funds to more easily increase assets.

Martin, Malhotra, and McLeod (2001) also find that 12b-1 fees add to expense ratios, but the mix between 12b-1 fees and loads complicates the results. Collins (2004) argues that studies such as those noted above are flawed, since 12b-1 fees must be evaluated in conjunction with the load or sales commission. It is noteworthy that Martin, Malhotra, and McLeod acknowledged the effect of the mix between the two distribution costs but did not conclude that it flawed studies of expense ratios. The Walsh (2004) results confirmed many of the earlier studies’ findings. The fee did contribute to asset growth, but there is no evidence that it contributed to either higher returns or lower expense ratios.

Thus, there seems to be a sense that the 12b-1 fee is a deadweight loss against returns, but net of the 12b-1 fee most conclude that there is no systematic difference between funds with and without a 12b-1 fee or between actively marketed and directly marketed funds. That is, the is no clientele effect that affects the management of the funds. However, this latter conclusion has not been rigorously tested. It is the purpose of this paper to fill this void. That is, is a difference between the portfolio management characteristics of actively and directly marketed funds as measured by total returns. The paper now turns to the development of the model to test this question.

The Hypothesis

Admittedly, 12b-1 fees add to expense ratios, and any addition to expense ratios reduces total returns by a like amount. However, unanswered in the various studies is whether portfolio management between actively marketed and passively (directly) marketed funds differ, and if

3

there is a difference it either offsets or amplifies effects of the 12b-1 fee. That is, are 12b-1 fees simply a deadweight loss and nothing more, or are portfolio management differences evident?

To answer this question, this study compares the total returns of funds that rely upon third-party sales forces with funds that rely upon direct marketing. To avoid problems with multiple share classes, Class A share will be used to represent funds with active distribution. A filtered version of the Morningstar classification for no-load funds will be used to represent directly marketed funds. To assure the elimination of distortions that might be introduced through different mixes of actively managed and index funds in either of the two marketing approaches, index funds will be excluded as part of the data stratification. Thus, only actively managed and actively marketed funds will be compared with actively managed and directly marketed funds.

There is widespread evidence that return differs by investment objective. As a result, this study will stratify funds by an investment objective. Since the Morningstar mutual funds database is used, the study will stratify mutual funds by the investment objective assigned by Morningstar.

The working assumption is that performance within a fund category is more homogeneous than performance among fund categories. As a result, the study will form panels of funds for each quarter where each panel represents a different Morningstar fund category. Each panel of funds with active sales programs will then be compared against a comparable panel of directly marketed funds. Since size may also play a role in performance, each panel will be constructed with simple averages of fund performance and asset-weighted averages of fund performance.

The Basic Model

At the risk of introducing too technical a description in this proposal, the following model is proposed as a vehicle to test the basic hypothesis:

(1)

RET_A = total return for actively marketed funds,RET_D = total return for directly marketed funds,TO_A = turnover rate for actively marketed funds,TO_D = turnover rate for directly marketed funds,j = 1, … , n and each j represents a separate Morningstar category, i.e., a

separate panel, andt = time subscript, in this study running from 1 through 24.

Each Morningstar categories (j) is a cross-sectional identifier in this model. They also serve as the panels. So, for 2,j through 4,j there will be separate coefficients calculated for each of the n categories, if there are cross-sectional effects of the returns of actively marketed funds compared with directly or passively marketed funds. Therefore, there would be n different coefficients on each of the variables. The assumption here is that there are cross-sectional effects. If there are no cross-sectional effects, the n coefficients on each variable would collapse into a single coefficient for that variable. That is, the general model allows for the independent variables to exhibit either cross-sectional effects or common effects. The cross-sectional effects manifest themselves when

4

the impact of a variable differs across investment categories. If the impact does not vary by investment categories, it is said to have common effect. In this case, the coefficient would be the same for all n Morningstar categories. For example, if an RET_D has only a common and not a cross-sectional effect, 2,1 = 2,2 = … = ,n, and the five coefficients would collapse into one common coefficient. All independent variables are evaluated on both a common effect basis and a cross-sectional effect basis.

The constant term (1,j) captures the fixed effects. This is a panel model in which fixed effects estimation has been used. The fixed effects estimation uses a dummy variable approach in which a dummy variable is created for all but one of the investment categories. The remaining category is then picked up by the constant term. When using this technique in the regression analysis, a constant term is estimated for all of the investment categories giving each of the categories a unique basis on which the effects of the independent variables are compounded in order to obtain a predicted value for the dependent variable. In this analysis, the fixed effects model is used because there seems to be an unobserved, time constant factor, 1,j, unique to each investment category that affects the performance of actively marketed funds. The idiosyncratic error term, ui,t,, in the fixed effects model then captures the unobserved factors which change over time and affect the performance (Wooldridge, 2000).

On the general assumption that returns are related for the same category, there should be a relationship between the actively and directly marketed funds. If there is no difference between the performances of the two, the fixed effects coefficients (1,j) should be zero and the 2,j

coefficients should be equal to one. However, there is the expectation of a difference, and the1,j

coefficients capture the additive portion of effects of 12b-1 fees, which in this case are expected to be negative. The 2,j coefficient would be less than one if efforts to recoup effects of distribution costs on investor returns have a systemic adverse effect. They would be greater than one if portfolio management more than recouped effects of the 12b-1 fee. There is no a priori reason to expect the slope coefficient to be greater or less than one.

Thus, the basic assumption is that1,j is equal to zero and that 2,j is equal to 1.0. That is, there is no systematic difference in performance results between directly and actively marketed funds which is related to the level of performance of the directly marketed funds. Thus, the null hypothesis is that the slope coefficient on performance would be equal to 1.0. Unfortunately, if it differs from 1.0, it would affect the magnitude of the constant term. Thus, the constant term could differ from zero simply because the slope coefficient on the directly marketed performance differs from 1.0. Because of that, a joint test is needed to test for 1,j = 0 and 2,j = 1 for each of the j categories.

While this test is straightforward, the constant term would not offer a measure of the deadweight loss imposed by the 12b-1 fees, even if the joint test rejected the hypothesis that 1,j=0 and 2,j=1. There is, however, a simple solution. As shown in DePrince (2006), removing the mean value from each of the independent variables and subtracting the mean of the RET_Dj,t from RET_Aj,t, produces a constant term whose value would be the difference between the mean return on actively marketed and directly marketed funds. This is illustrated below. Therefore, it would provide a direct measure of the deadweight loss of 12b-1 fees regardless of the size of the coefficient on the performance measure of directly marketed funds.

5

(2a)

The transformation above has no effect on the values of 2,j, 3,j and 4,j, and the constant term for the jth category becomes

. (2b)

Finally, the coefficients on the turnover rates (3,j and 4,j) capture the effect of differential turnover rates in the performance of the actively managed funds. It is expected that 3,j is negative and 4,j is positive. If they are approximately equal, but of opposite signs, the two terms could be combined and expressed as the difference in the turnover rates of actively and directly marketed funds.

As noted above, four alternative measures of RET_A will be used. The effects of loads are distributed over the holding period used in calculating the return. Thus, the longer the holding period, the less is the effect of the load on the load-adjusted return. A three-year return seemed to be a reasonable compromise between the short-term and long-term holding periods.

Data and Data Filters

Mutual fund data are from the Principia Pro database supplied by Morningstar. The study will involve 24 quarters (2000:3 through 2006:2). This represents the extent of the mutual fund data acquired by the author. With 63 Morningstar categories, this produces a sample set with 1606 observations (24x64).

“A” shares are used as the representative share class of actively marketed funds. However, there are funds that have the pricing characteristics of Class A share but are sold as share classes with different designations. Therefore, to be as inclusive as possible, a filter will be used to identify the share classes that have distribution cost characteristics comparable to those of “A” shares. In it, any funds with a 12b-1 fee of between 25 and 50 basis points and a commission or sales load of 3.0 percent or more will be considered a Class A share.

The funds used as directly marketed funds will be generated from a filter of the funds listed as no-load funds by Morningstar. In the filter, no-load funds that are part of a multi-class fund structure will be excluded from the directly marketed universe. This is because such funds have Class A shares with 12b-1 fees and loads or other classes of actively marketed funds as part of the overall fund family. The inclusion of no-load funds that are part of a multi-class structure within the universe of no-load funds would bias the results in favor of the basic hypothesis laid out in the next section. By excluding no-load funds that are part of a multi-class structure, the selected universe will be truly those that are exclusively directly marketed and provide an unbiased test of the basic hypothesis.

Four performance measures of actively marketed funds will be used: (1) 1-year returns including effects of 12b-1 fees but excluding effects of the sales load, (2) 3-year returns including effects

6

of 12b-1 fees but excluding effects of the sales load, (3) 1-year returns including effects of 12b-1 fees and the sales load, and (4) 3-year returns including effects of 12b-1 fees and the sales load. The effects of loads are distributed over the holding period used in calculating the return. Thus, the longer the holding period, the less is the effect of the load. A three-year return seemed to be a reasonable compromise between the short-term and long-term holding periods.

Both 1-year returns [(1) and (3) above] on actively marketed funds will be compared with the one-year returns on directly marketed funds, and both 3-year returns [(2) and (4) above] will be compared with the 3-year return on directly marketed funds. It is expected that the 12b-1 fee will have an adverse effect on the returns of actively marketed funds compared with directly marketed funds and that underperformance will be exacerbated with the inclusion of the effects of the load. The research questions are the following. First, is underperformance of actively marketed funds evident? If so, second, how sizeable is the underperformance?

Other factors may also contribute to differential performance. Turnover rates are often high for actively sold compared with directly marketed funds. Effects of turnover rates will also be incorporated into the empirical tests.

Empirical Findings

General Findings

The relevant returns data and turnover data have been consolidated by Morningstar category for each quarter. Summary measures comparing actively marketed and directly marketed funds, drawn from the pooled database constructed for the model’s estimation, are reported in Table 1. First, there were 1,197 funds with 12b-1 fees within the screening criterion and 837 no-load funds with a zero 12b-1 fee common to all 24 quarters. Second, there were 64 Morning star categories common to all 24 quarters.

Next, the one-year return for the actively marketed funds averaged 7.27 percent over the 24 quarter sample period, and the one-year return for the directly marketed funds averaged 7.85 percent. The difference is 58 basis points, a bit over twice the average 12b-1 fee (27 basis points) for the funds used in this study. Total returns as reported by Morningstar include effects of 12b-1 fees for A shares. Excluding 12b-1 fees, the total return for actively marketed funds average 7.54 percent, and the difference is now 31 basis points. Thus, on there is at least weak statistical evidence that 12b-1 fees are not recouped for actively marketed funds, and there is an additional deadweight loss of 31 basis points that might be attributed to differences in portfolio management between actively and directly marketed funds on average. This is consistent with most of the findings noted earlier in the study.

Looking to three-year returns, overall comparative results are similar to the one-year returns. Actively marketed funds posted an average return across categories of 6.04 percent versus 6.71 percent for directly marketed funds, for a difference of 68 basis points.

Overall effects of loads on “A” shares are also noted in Table 1. For all “A” shares used in this study, loads across categories averaged 4.73 percent. Spread over one year, this reduced the total

7

return to 3.17 percent. Spread over three years, it reduces the return from 6.04 percent to 7.85 percent. Readers should keep in mind that the filter identifying “A” shares drew only those with a load of three percent or more. The 4.73 percent is a bit more than the midpoint of the 3.0-5.5 percent range for the loads of “A” shares chosen in this study. Readers should note that this average load of funds in this study is greater than the average load of all “A” share since the selection process excludes funds with loads less than three percent As noted earlier, this band on the load was established to assure some degree of homogeneity among the “A” shares used in this study.

Panel Model Results

All Variables Have Common Effects. The first step in the estimation phase is the estimation of Equation – with all 64 panels having common effects on the dependent variable. Several screening equation were run and it was determined that turnover rates do not have a statistically meaningful effect and were thus dropped from the model. Final results for this stage are reported in Table 2. Model 1 reports results for one-year return; Model 2 reports results for 3-year returns. In Model 1, the constant term suggests that roughly 30 percent of the deadweight loss of the average 12b-1 fee by the portfolio management techniques of the actively marketed funds. However, that gain comes at a cost.

The slope coefficient on directly marketed funds is less than unity which suggests that actively marketed funds under performs directly markets by roughly 2.75 percent of the industry wide average return on directly marketed funds. While quantitatively a small difference from unity, the 0.9725 does differ from 1.00 at least at the 95 percent level of confidence. Thus, there is at least some evidence that management of directly marketed funds do differ from directly marketed funds.

There is evidence of auto-correlated residuals, and Model 1.b adjusts for that through a first order autocorrelation corrections. As can be seen, the constant term dips a bit, but is still different from zero as well as 0.27 (the average 12b-1 fee). The slope coefficient rises to 0.99, but still differs at least at the 5 percent level. In any event, it suggests that there may be a slight difference in the managers of actively versus directly marketed funds with evidence of a slight underperformance, however, that underperformance is far less than suggested by the raw data in Table 1.

The same exercise was repeated for the three-year returns, and a somewhat similar outcome was observed. As can be seen in Model 2.a, the constant term is now roughly the same size as the average 12b.1 fee of the selected funds over the 24 quarters period. In other words, management techniques fail to recoup the fee when measured over a three-year period. Interesting, the slope coefficient did drift a bit lower, and is still statistically different from unity.

Here, the case for auto-correlated residuals is probably more persuasive than for the one-year returns. Each three-year return carries effects of the one year return in the previous two years. These returns are also parts of the three-year return for the previous two years. Thus, outcomes, and probably residuals, are auto-correlated. Model 2.b adjusts Model 2.a for auto-correlated, and as can be seen, the constant term is now indistinguishable from zero, suggesting that managers of

8

actively marketed funds offset effects of the 12b-1 fee almost in their entirety over a three-year period, on average. The slope coefficient rises to 0.99, but still differs at least at the 5 percent level.

In any event, it suggests that there may be, at worst, a slight penalty in terms of performance for managers of actively marketed versus directly marketed fund.

Panel Results. Next, the papels were used to estimate cross-sectional coefficients on the independent variable and fixed effects coefficients. Table 3 reports results for the one-year returns. Again, several screening equations were estimated which indicated that (1) account should be taken of cross-sectional effects of panels, (2) fix effects should be included, and (3) including the turnover rates again adds no information.

Interestingly, the cross-sectional slope coefficients average 0.963, almost indistinguishable from the results for Model 1.a in Table 2. However, the cross-sectional slope coefficients differ across Morningstar categories. Separately, 27 are greater than unity which implies that the actively markets funds outperform directly marketed funds by a multiple of the return on the directly marketed funds. Unfortunately, 37 categories have a slope coefficient less than unity, implying that they under perform, on average, directly marketed funds in their respective categories. Interestingly, the outcome in terms of statistically significance is mixed. Of the 64 coefficients, 39 coefficients differ from unity at least at the five percent level of confidence. For the other 29 categories, the slope coefficient can be considered equal to unity. Of those statistically different from unity, -- were greater than 1 and – were less than unity.

Results for the constant terms (fixed effects) show a similar theme. As can be seen, magnitudes vary from close to zero to several hundred basis points (percentage points) above and below zero. This suggests that management differences between the directly and actively marketed funds do exist, some enhance returns to investors in actively marketed funds compared with directly marketed funds, while other detract from return.

Interestingly, there is but a very weak relationship between the slope coefficient and the constant term. As a result, it does not appear that large fixed-effects coefficients are associates with large deviations of the slope coefficients from unity and vice versa. Thus, whatever effect management differences may have on the constant term, there is not an associated effect on the slope coefficient.

Results for the corrections of auto-correlated residuals are reported in Table 4.

Results the three-year returns corrected for auto-correlated residuals are retorted as Model 4 in Table 3. In examining the slope coefficients, 19 of the 64 coefficients differ from unity at roughly the five percent level of confidence, implying that 46 do not differ from unity. In other words, for over two thirds of the style categories, there is no difference between the managers of actively and directly marketed funds related to the level of performance.

9

In terms of the fixed effects coefficients (i.e., the difference between the three-year return on the actively marketed and the directly marketed funds), 27 differ from -0.27 (the average 12b-1 fee over the 24 quarters for the sample of actively marketed funds), implying that 37 did not differ from the 27 basis points. In other words, actively marketed funds in nearly 60 percent of the Morningstar categories had a difference in performance from the corresponding directly marketed funds that differed by roughly the average 12b-1 fee.

Of 27 categories where the performance of the actively marketed funds differed statistically from the 12b-1 fee, 13 had a fixed effect of -0.27 or less. And 14 had a fixed effect coefficient of -0,27 or more. That is, roughly half had a shortfall worse than the average 12b-1 fee and 14 had a performance that more than offset the 12 fee. This slight edge to the over performers should not be surprising since model 2.b had a constant term of roughly zero and differed from from -0.27 with virtual certainty.

Overview of findings

Results on average suggest that when measured over a three year period, managers of actively marketed funds, either by chance or design offset the effects of the distribution costs (27 basis points) with virtual certainty. Unfortunately, this success appears to come at a price. There is a slight, roughly systematic underperformance of actively marketed from directly marketed funds based on the fact that the slope coefficient differs from unity at least at the five percent level of confidence.

At the same time, there is a difference among investment categories.

The Importance of the Study

Most funds have active distribution programs. The rationale from a funds investment advisor is quite simple. Active distribution programs are presumed to provide the ability to grow assets, and hence the advisors’ revenue stream is faster than the asset growth of directly marketed funds. However, the advantage to the investor is less obvious. Aside from intellectual curiosity, the most important reason for this study stems from the Securities and Exchange Commission’s review of the concept of 12b-1 fees. Criticism of these fees has intensified in recent years, and while their rationale from the asset managers’ perspective is clear, some question the benefits to the investor.

At the same time, the marketers using 12b-1 fees do perform a function for the investor. They provide search services that are too complicated or too confusing for many individuals. Such service is not free. Presently, this search expense is born by the investor in the form of the 12b-1 fee. If the 12b-1 fee structure should be changed, another fee of some kind would likely appear to cover the search effort provided by the marketing of funds bearing 12b-1 fees. In the end, the investor would pay for the service that is being provided.

Behind the debate is the presumption that marketing and asset management are independent of each other. However, this assumption has not been empirically verified. If they are not independent, portfolio management differences could offset or exacerbate the effect of higher

10

expense ratios on returns. Thus, investors who require services funded through 2b-1 plans need to understand the interaction between the method distribution costs and portfolio management.

AED-V.3.1-022307

11

References

Chalmers, John M.R., Roger M. Edelen, and Gregory B. Kadlec. 1999. “Transaction Expenditures and the Relative Performance of Mutual Funds.” The Wharton Financial Institution Center. #00-02.

Collins, Sean. 2004. “The Effect of 12b-1 Plans on Mutual Fund Investors, Revisited.” Investment Company Institute. March.

Blake, Christopher R., Edwin J. Elton, and Martin J. Gruber. 1993. “The Performance of Bond Mutual Funds.” Journal of Business. 66 (3): 371-401.

DePrince, Albert E., Jr. 2006. “A Technical Note on a Direct Estimate of the Significance of Bias in Forecasts.” Middle Tennessee State University. Working Paper. March.

DePrince, Albert E., Jr. 2003. “Active Portfolio Management and Performance: Domestic Equity Versus Bond Mutual Funds.” 56th International Atlantic Economic Conference, Quebec City, Canada, October 16-19.

Division of Investment Management. 2000. Report on Mutual Fund Fees and Expenses Securities and Exchange Commission. December.

Ferris, Stephen P., and Don M. Chance. 1987. “The Effect of 12b-1 Plans on Mutual Fund Expense Ratios: A Note.” The Journal of Finance, 42 (4): 1077-1082.

Grove, David L. 2002. The Critical Role of Independent Directors in the Governance of Mutual Funds. Management Practice, Inc. Stamford, CT.

Hillman, Richard J. 2004. “Additional Disclosure Could Increase the Transparency of Fees and Other Practices.” United States General Accounting Office. Testimony Before the SubCommittee on Financial Management, the Budget and International Security, Committee on Governmental Affairs, United States Senate. January 27.

Investment Company Institute (ICI). 2003. “Background Information about 12b-1 Fees.” http://www.ici.org/funds/abt/ref_12b1_fees.html. February.

Investment Company Institute (ICI). 2005. “How Mutual Funds Use 12b-1 Fees.” Fundamentals. 14:2, pp. 1-4.

.Investment Company Institute (ICI). 2004. “Frequently Asked Questions about Mutual Funds

Fees.” ICI Investment Awareness Series. http://www.ici.org/funds/inv/ bro_mf_fees_faq.html.

Latzka, David A. 1999. Economies of Scale in Mutual Fund Administration. Journal of Financial Research 22, (3): 331-339.

12

Lowenstein, Alice. 1997. “The Low Turnover Advantage.” www.morningstar.com. September 12.

Lucchetti, Arron. 2003. “Trade Happy Funds May Cost You.” Wall Street Journal. April 25: C1.

Martin, Rand, D.K. Malhotra, and Robert W. McLeod. 2001. “A Comparative Analysis of the Expense Ratios of Domestic and International Open-Ended and Closed Ended Equity Funds. Financial Counseling and Planning, 12 (2): 61-73.

National Association of Securities Dealers (NASD). Report of the Mutual Funds Task Force. http://www.nasd.com/web/groups/rules_regs/documents/rules_regs/nasdw_013690.pdf

Pozen, Robert C. 2002. The Mutual Fund Business. Houghton Mifflin Company. 2nd edition.

Rao, Spuma. 1996. “Does 12B-1 Plan Offer Economic Value to Shareholders of Mutual Funds?” Journal of Financial and Strategic Decisions. 9:3. 33-37.

Reid, Brian K, and John D. Rea. 2003. “Mutual Fund Distribution Channels and Distribution Costs.” Perspective. July. 9:3 1-19.

Securities and Exchange Commission. 2005. “Exemptive Rule Amendments of 2004: The Independent Chair Condition.” Staff Report of the United States Securities and Exchange Commission. April.

Walsh, Lori. 2004. “The Costs and Benefits to Fund Shareholders of 12b-1 Plans: An Examination of Fund Flows, Expenses and Returns.” Office of Economic Analysis. Securities and Exchange Commission. April 26.

Wooldridge, Jeffrey. 2000. Introductory Econometrics: A Modern Approach. Cincinnati, Ohio. Southwestern College Publishing. 420.

13

14

15

16

17