Notices

On December 13, 2012, the Canadian Securities Administrators (CSA) published CSA Discussion Paper and Request for Comment 81-407 Mutual Fund Fees (Paper) which identified potential investor protection and fairness issues that may arise from Canada’s current mutual fund fee structure.  In particular, the Paper examined how embedded advisor compensation and other forms of tied compensation could give rise to actual or perceived conflicts of interests.

 

Following comments received on the Paper, the CSA conducted extensive stakeholder consultations, including a public roundtable on June 7, 2013, and discussion forums in the summer and fall of 2013.  As a result of feedback received during the consultations, the CSA decided to obtain and publish independent research to assist in determining whether regulatory action is needed.

 

Specifically, the CSA commissioned independent third party research in two main areas:

 

1.      Researchers, using data sourced directly from Canadian investment fund managers, would evaluate the extent, if any, to which sales and trailing commissions influence mutual fund sales.

 

2.      Researchers would conduct a literature review to evaluate the extent, if any, to which the use of fee-based vs. commission-based compensation changes the nature of advice and investment outcomes over the long term.

 

Following a request for proposals by the CSA, the above pieces of research were awarded to Professor Douglas Cumming and the Brondesbury Group, respectively.  Professor Douglas Cumming has now completed his data focused study and his findings are set out in the enclosed report entitled “A Dissection of Mutual Fund Fees, Flows and Performance”. 

 

This research, together with the Brondesbury Group’s previously published “Mutual Fund Fee Research” report and the comments received during the previous consultation period, are intended to be among the inputs that will be factored into the CSA’s determination of whether to effect certain policy changes.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[This page intentionally left blank]

 

A Dissection of Mutual Fund Fees, Flows, and Performance

 

 

Douglas Cumming

Professor and Ontario Research Chair

York University - Schulich School of Business

4700 Keele Street

Toronto, Ontario M3J 1P3

Canada

http://ssrn.com/author=75390 

dcumming@schulich.yorku.ca  

 

 

Sofia Johan

York University - Schulich School of Business

4700 Keele Street

Toronto, Ontario M3J 1P3

Canada

http://www.schulich.yorku.ca/ 

and

University of Tilburg

Extramural Research Fellow

Tilburg Law and Economics Centre (TILEC)

Postbus 90153

5000 LE Tilburg

The Netherlands

http://ssrn.com/author=370203

sjohan@schulich.yorku.ca

 

 

Yelin Zhang

York University - Schulich School of Business

4700 Keele Street

Toronto, Ontario M3J 1P3

Canada

http://www.schulich.yorku.ca/

ZhangY12@schulich.yorku.ca   

 

 

October 19, 2015

 

We are indebted to the mutual funds for providing access to their detailed data.  Also, we are indebted to the Ontario Securities Commission, and in particular Minami Ganaha, Rhonda Goldberg, Chantal Mainville, and Dennis Yanchus, for their help with facilitating data collection.  

A Dissection of Mutual Fund Fees, Flows, and Performance

Executive Summary

            This paper analyses Canadian mutual fund data on fund fee structures, fund flows and fund performance.  The data in this study has been gathered with a view to better informing academics, practitioners and policy makers about the relationship between specific types of mutual fund fees and flows, and how fees and flows are related in conjunction to fund performance.  As such, the paper provides an empirical assessment of the possible and often debated conflicts of interest in respect of different types of fund fee structures in the Canadian mutual fund industry.  For easier readability, we have included an executive summary glossary on pages 10-11.

The Data Set

The mutual fund data for this research was obtained further to a data request sent to all Canadian fund managers offering conventional mutual funds to the public under prospectus.  Canadian fund managers’ participation in the research was voluntary.  The data request questionnaire is attached as Appendix III.  The sample is from 2003 to 2014 and includes:

•          43 fund families of 113 in total in Canada at the end of 2014 or 38.1% of the market

•         $746 billion of Assets Under Management (AUM) for stand-alone funds (of an estimated total of $1.1 trillion from Investor Economics) or 66.7% of total AUM, comprising 18,012 FundSERV codes

•         $152 billion in AUM for fund-of-funds (FOF) or 51.5% of the estimated total $295 billion market, comprising 4,018 FundSERV codes

The data set has 22,077 FundSERV codes in total, compared to the 16,752 FundSERV codes from fundlibrary.com and 12,300 FundSERV codes from fundata.com, both of which state they have the most comprehensive mutual fund datasets in Canada.[1]  In total, the sample comprises more than 1 million monthly FundSERV code observations on fees, flows, and performance. 

Fund fee structure

We analyzed the data according to the following fund fee structures:

By whether Fund-Of-Fund and by whether sold by manager.  We divide the sample into four groups depending on whether they are fund-of-funds or stand-alone funds and whether or not they can be purchased directly from the fund manager.

By Purchase Option. Mutual funds in Canada can be purchased in one of four primary options: 

•         No Load (NL) purchase option does not include a front end commission nor can it attract deferred sales charges but it does pay a trailer fee to the fund dealer. 

•         Deferred Sales Charge (DSC) purchase option requires the investor to pay a redemption fee if the units are sold before a predetermined period has elapsed, and also pays an upfront commission and a trailer fee to the fund dealers. 

•         Front End (FE) or initial sales charge purchase option allows the fund dealer the option of charging a negotiable front end commission directly upon initial sale, and pays a trailer fee to dealers. 

•         Fee Based (FB) purchase option does not allow for a front end commission to be charged, and it cannot attract deferred sales charges and there are no trailer fees paid to the dealer. High net worth or institutional series/purchase options where there is no front end commission, deferred sales charge or trailer fee paid are also reported as fee-based.

By Fund Affiliation with Dealer.  Investors may purchase mutual fund securities through a dealer affiliated with the fund or from an independent dealer.  We divided fund flows into whether or not they flowed through an affiliated dealer. 

Methodology

The analysis is twofold. 

1         How does past performance affect fund flows?

First, we examine the relationship between risk-adjusted performance (“alpha”) and future fund flows (“flow-performance slope”), and fund flows that are obtained regardless of past alpha (“flow-performance intercept”), and consider whether or not flow-performance intercept and slope are influenced by the fund fee structure. 

Higher flow-performance slope means disproportionately more investor money flows into the fund when past performance is strong and disproportionately more investor money flows out of the fund when past performance is weak.  Conversely, lower flow-performance slope means that disproportionately less investor money flows into the fund when past performance is strong and disproportionately less investor money flows out of the fund when past performance is weak (see “possible effect 1” on chart).  Higher flow-performance slope incentivizes fund managers to achieve higher alpha because the fund manager will receive disproportionately more flows into the fund for each increase in performance.

Large positive values of flow-performance intercept means that fund managers obtain new capital flows each month regardless of past performance and therefore they don’t have strong incentives to achieve higher alpha (see “possible effect 2” on chart). Positive values of flow-performance intercept means that fund managers achieve flow even if there is no positive risk-adjusted performance. 

 

2         Do fees and fund flows have any effect on future fund performance?

Second, we examine whether or not fund fee structures and flow-performance intercept and slope have any systematic effect on future alphas.

 

For both analyses, we use fixed-effects regressions by FundSERV code, which enables an isolation of the impact of fees separate from fund characteristics that may be correlated with flow. 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Results of the Analysis

1.      How does past performance affect fund flows?

For stand-alone funds that cannot be purchased directly from the manager, regression analyses comparing across funds and over the sample period indicate that:

•         Funds that perform better attract more flows…   Funds with higher alpha are generally rewarded with higher future flows.  For example, all other things being equal, a fund ranked at the bottom quartile (25^th percentile) of all funds for risk adjusted performance loses 0.02% of its AUM in the next month, while a fund performing at the median gains 0.08% of new AUM in the next month, and a top quartile fund gains 0.21% of new AUM in the next month. 

•         …but this effect is less strong for funds sold through affiliated dealers.  Affiliated dealer flows show a less robust statistical relation between past alpha and future flows, meaning that affiliated dealer flows are often not sensitive to past risk-adjusted performance.

•         …and this effect is less strong for funds with trailers.  Trailer fees reduce flow-performance slope such that for any given improvement in performance, net capital inflows are 15% less than what they would have been in the absence of a trailer fee and likewise for any given reduction in performance, net capital outflows are 15% less than what they would have been in the absence of a trailer fee.   

•         In fact, funds with trailers increase flows regardless of performance.  Trailer fees increase new flows regardless of past performance.  Generally, the greater the trailer fee, the greater the level of net flows that has no relationship to past performance.  For example, a 1.5% trailer fee increases the average monthly flows by 0.3% of AUM each month regardless of past performance.

•         Fee-based fund flows are much more sensitive to performance.  Fee-based purchase options exhibit significantly higher flow-performance slope.  A fund that moves from the top quartile to the bottom performance quartile, for example, experiences a drop in flow relative to AUM by 0.32% under fee based purchase options, but only 0.26% under front end purchase options, 0.19% under no load purchase options, and 0.13% under deferred sales charge purchase options, controlling for other things being equal.

Similar effects were seen for funds that can be purchased directly from the manager and for Fund-of-Funds.

2          Do fees and fund flows have any effect on future fund performance?

In the second part of the analysis, we found the following relationships between flow-performance sensitivity, fee structures, and future performance alpha:

•         An increase in trailers corresponds to a decrease in performance…  For the full sample of all funds, the data indicate that 2.5% of funds permanently increased their trailer fees in the sample period (from an average of 0.39% to 0.78%).  Comparison tests of the alphas for these funds from the 6-month prior fee change period to the 24 month post fee change period show alpha dropped by 32.4% on average. 

•         …and a decrease in trailers corresponds to an increase in performance.  For the full sample of all funds, the data indicate that 0.6% of funds permanently decreased their trailer fees in the sample period (from an average of 0.42% to 0.27%). Comparison tests of the alphas for these funds from the 6-month prior fee change period to the 24 month post fee change period show alpha increased by 88.4% on average.

•         Funds that have flows more sensitive to past performance tend to have better future performance.  Regression analyses comparing across funds and over time indicate that funds for which flows are more sensitive to past performance have better risk adjusted performance on average.  For stand-alone funds that cannot be purchased directly, an increase in trailer fees and deferred sales charges by 1% is indirectly associated with a reduction in future alpha by 1.4% and 0.6% relative to the average monthly alpha, respectively, due to the effects on the flow-performance intercept and slope.  The data indicate similar findings for stand-alone funds that can be purchased directly, and fund-of-funds that can and cannot be purchased directly, and the findings are robust over time.

•         Funds that sell more through affiliated dealers tend to perform worse.  Regression analyses comparing across funds and over the sample period indicate that funds which receive higher levels of affiliated dealer flows experience lower future alpha on average.  Funds that were in the top quartile in terms of receiving affiliated dealer flows on average experienced a reduction in future monthly alpha by 0.2% relative to those funds that did not receive any affiliated dealer flows.   The regression analyses indicate similar findings for stand-alone funds that can be purchased directly, and for fund-of-funds that can and cannot be purchased directly, but there were some differences in these effects at different points in time.

Specific details are described in the body of this report.

 

Executive Summary Glossary

 

Affiliated Dealer: A mutual fund dealer that is an affiliate of an investment fund manager by virtue of one of them being the subsidiary of the other or both of them being subsidiaries of the same company or each of them being controlled by the same person or company.

 

Alpha: The risk adjusted excess performance of the fund in each month. Alpha is calculated based on monthly gross return and Fama-French North America 4 factors.

 

Fund flow (net retail flow):  The fund flows net of pre-authorized contributions plan (PAC), systematic withdrawal plans (SWP), switches in/out, reinvested distributions, affiliated dealer flows, and affiliated investment fund flows [hereafter referred to as ‘net retail flow’].  The net retail flow variable is measured over each fund series/purchase option  – monthly observation in the data, and as a percentage of the prior month’s AUM, in order to make comparative assessments of flow across funds and over time. 

 

Flow Performance Intercept: The monthly fund flow (defined above) that results when the prior month’s alpha is zero.

 

Flow Performance Slope: The additional monthly fund flow (defined above) that results when there is a change in prior month’s alpha.

 

FundSERV Code: The full FundSERV fund code assigned to each fund series/purchase option covered in the sample. If the fund manager did not process fund purchase and sale transactions via FundSERV, then the fund series/purchase option code that they used internally was used. .

 

Funds that can be purchased directly: Funds that include all those FundSERV codes for which the fund manager answered “yes” to the following question in the data request questionnaire: “Can this series/purchase option combination be purchased directly from the fund company (i.e. the fund manager is also registered as a fund dealer or has received exemptive relief to allow purchases directly from investors).”

 

Fund-of-Fund: A mutual fund that invests in other investment funds and is typically sold as a total portfolio solution.

 

Stand-Alone Fund: A mutual fund that typically invests directly in stocks, bonds, commodities, and other non-investment fund securities. It is not sold as a total portfolio solution and usually makes up one of several holdings in the investor’s mutual fund portfolio.

 

Trailer fee:  A commission paid by a fund manager to a dealer on an ongoing basis for as long as the dealer’s clients hold investments in the fund manager’s mutual funds.  It is calculated based on a percentage of the average daily value of units of each fund held by the dealer’s clients and is typically paid to the dealer on a quarterly basis.  The rate of the commission typically varies depending on the asset class of the fund and the purchase option.  This type of fee is referred to a 12-b1 in the United States.

 

 

 

 

 

 

Dissection of Mutual Fund Fees, Flows, and Performance

 

 

Abstract

                                               

This paper provides a dissection of both mutual fund fees and flows into several categories, and presents evidence that relates specific components of fees to flows, and fees and flows to performance.  For stand-alone funds that cannot be purchased directly from fund managers, fees that compensate fund advisors when investors maintain their portfolio positions, and fees that penalize investors for early withdrawal, have a much flatter flow-performance relationship (“flow-performance slope”), and higher flows regardless of past performance (“flow-performance intercept”).  Further, the data indicate that flow-performance intercept and slope are significantly negatively and positively, respectively, related to future risk-adjusted performance, which is consistent with the view that flow-performance provides a strong incentive to generate future returns.  These findings are quite stable over time, and robust to numerous sensitivity checks.  We find some consistency in the evidence but less robust statistical significance amongst the subsamples of direct purchased funds, and among fund-of-funds.

 

Keywords: Mutual Funds, Management Expense Ratio (MER), Trailer Fees, 12b-1s

JEL Codes: G23, G30

 

1. Introduction

Compensation in the mutual fund industry has long been a popular topic rife with controversy.  John Bogle, the founder of the Vanguard index mutual fund, has long argued that a majority of mutual fund managers in the United States (U.S.) do not earn returns that account for their fees (see, e.g., Bogle, 2006).  Other countries like Australia and the United Kingdom have recently engaged in mutual fund fee reforms, including a ban on embedded commissions in 2012.  These reforms have inspired other jurisdictions such as Canada to consider whether or not there are any conflicts of interest in the ways in which mutual fund managers are compensated (CSA, 2012), and if so, exactly how do these conflicts affect the mutual fund industry and its investors.

One potential conflict of interest in mutual fund compensation structures is that investment advisors may have incentives to recommend products to investors for reasons that are not strictly based on the expected risk-adjusted performance (“alpha”) of the investment.  This conflict may arise when fees compensate advisors and managers regardless of performance.  If such a conflict of interest exists, we would observe a market in which flows of investor capital into mutual funds were less sensitive to past performance, and possibly towards funds that have lower risk adjusted performance.  To test this proposition, in the first part of this paper we examine the relationship between past alpha and future fund flows (which we label the “flow-performance slope”), and examine fund flows that are obtained regardless of past alpha (which we label the “flow-performance intercept”).  We empirically consider whether or not flow-performance intercept and slope are moderated by the structure of fund fees.  Higher flow-performance slope means that fund managers are rewarded with additional AUM when past performance is strong, while lower flow-performance slope means that fund managers are relatively more insulated from losses of AUM when past performance is weak.  Large positive values of flow-performance intercept means that fund managers obtain new capital flows each month regardless of past performance.  Higher flow-performance slope incentivizes fund managers to achieve higher alpha, while large positive values of flow-performance intercept mitigate strong incentives to achieve higher alpha.

In the second part of this paper, we empirically examine whether or not fund fee structures and flow-performance intercept and slope have any systematic effect on future alphas.  That is, if fund managers have less incentive to generate alpha to attract capital inflows and mitigate capital outflows then they may in fact achieve lower alphas.  Such effects would benefit some mutual funds at the expense of other mutual funds, and have a detrimental impact on the mutual fund industry overall, and a detrimental effect on its investors.  While there has been much discussion of these possibilities in news media,[2] whether or not such effects actually exist is unknown without examining data.

A major challenge in the analysis of mutual fund fees and their effect on investors involves access to data.  Publicly available data from Morningstar, Bloomberg and other related sources have estimated information on fund flows based on reported net asset values (NAVs), and coarse information on fees that does not enable an accurate analysis of fee structures.  Mutual fund fee structures in some countries like Canada are extremely detailed and complicated.  For example, apart from the publicly available management expense ratios (MERs), there are trading expense ratios (TERs), trailer fees (equivalent to 12b-1s in the U.S.), deferred sales charges, front end commissions, switch fees, performance fees, negotiated management fees, and other payments to broker dealers (other payments that do not qualify as trailer fees, DSC sales commissions or referral fees and that are tied to fund series/purchase type inflows or AUM held in each fund series/purchase type), and some of these items may change over the life of a fund.  Similarly, fund flows cannot be accurately assessed from an examination of NAVs, since total inflows and outflows are comprised of not merely new flows from retail investors but also pre-authorized contributions (PAC), systematic withdrawal plans (SWP), switches, reinvestments, distributions, affiliated dealer flows, and affiliated investment fund flows.  Furthermore, publicly available fund flow information is not available at the FundSERV fund code level.[3]

This paper presents proprietary data obtained directly from mutual funds in Canada that relate detailed mutual fund fee structures to specific types of fund flows, and to performance.[4]  The Canadian context is particularly interesting as there is a very wide array of different components of fees, effectively no competition from foreign funds, and the overall magnitude of fees in Canada has been noted to be rather high relative to other parts of the world (Ruckman, 2003; Khorna et al., 2014).  We examine data that comprises the period 2003-2014, covering 43 fund families, 18,102 stand-alone funds FundSERV codes and 4018 fund-of-funds FundSERV codes, and $746 billion of AUM (of an estimated total of $1.1 trillion) or 66.7% of the market, as well as $152 billion in AUM or 48.6% of the market for fund-of-funds.

The data exhibit a number of insights into flow-performance sensitivity, and how this relationship between prior alpha and future fund flows is moderated by fund fees.  For stand-alone funds that cannot be purchased directly, the data indicate that funds with higher prior alpha experience higher flows in the next month, such that a 1-standard deviation improvement in past alpha increases subsequent flow by 18.6% relative to the average monthly flow in the sample. Fee-based fund purchase options exhibit significantly higher flow-performance sensitivity, while front-end purchase options, deferred sales charge purchase options, and no load purchase options exhibit significantly lower flow performance slope.  The data indicate that a 1-standard deviation increase in trailer fees reduces flow-performance sensitivity by 15.4%.  The data indicate that a 1-standard deviation increase in deferred sales charges reduces flow-performance sensitivity by 14.6%.

Further, the data provide a number of insights into the relationship between flow-performance sensitivity, fee structures, and future performance as measured by gross-of-fees alpha. For stand-alone funds that cannot be purchased directly, the data indicate that 2.5% of funds increased their trailer fees in the sample period, and the comparison of gross alphas pre and post trailer fee change shows that alpha dropped by 32.4% on average.  Similarly, the data indicate that 0.6% of funds decreased their trailer fees in the sample period, with the comparison of gross alphas pre and post trailer fee change showing an increase of 87.9% on average.  In the full sample comparing across both funds and time, the data indicate that flow-performance sensitivity is significantly positively related to future risk-adjusted performance, whereby a 1-standard deviation increase in flow-performance sensitivity is associated with an increase in alpha of 4.9% on average.  A 1-standard deviation increase in trailer fees and deferred sales charges is associated with a reduction in future alpha of 5.2% and 2.4%, respectively, on average due to the effect on the flow-performance intercept and slope. 

For stand-alone funds that can be purchased directly[5], and for fund-of-funds, the results are in part consistent with the above results, but with some differences in terms of the statistical significance and the size of the effects.  Those details, among others, are not summarized here but are reported in the body of this paper.

This paper is related to a number of other papers in the academic literature on mutual fund flows in general (Del Guercio and Tkac, 2002; Christofferson et al., 2014), and in particular flows that are related to fee structures.[6]  Prior work has shown that option-like incentive contracts can exacerbate risk taking by fund managers (Starks, 1987).  By contrast, fulcrum fees, or fees that encourage managers to just beat a benchmark lead managers to just buy the benchmark (Admati & Pfleiderer 1997), reduce the reward for good performance or ‘flow-performance sensitivity’ (Heinkel & Stoughton 1994).  Hence, fulcrum fees are rare (Golec, 1992; Elton et al., 2003).  More generally, prior work is consistent with the view that fixed fees and incentive fees significantly vary with fund flows and performance (Warner and Wu, 2011; Deli, 2002; Kuhnen, 2004; Cumming et al., 2015).  The present paper adds to the important stream of literature by providing specific information on different components of fee structures that has not previously been possible with prior datasets, and relating fee structure details to specific types of fund flows that are not estimated from publicly available sources.  To this end, we provide an empirical assessment of the possible and often debated conflicts of interest in respect of different types of fee structures in the mutual fund industry.

This paper is organized as follows.  Section 2 briefly describes the institutional structure of the Canadian mutual fund industry in terms of fees and flows.  Section 3 introduces the data and provides summary statistics.  Section 4 presents empirical tests.  Conclusions follow in Section 5.

2. Institutional Structure of Mutual Funds Purchase Options, Fees, and Flows in Canada

Mutual funds in Canada comprise fund families, which are groups of funds, and within each fund there are numerous series and purchase option types available, each identified by their FundSERV fund code.  Much publicly available information on mutual fund data from Canada is not available at the series/purchase option type level.  Likewise, specific details on fees and flows are not publicly available or are not publicly available in an easily downloadable form from data vendors.

Mutual funds can be purchased in one of four primary options.  The No Load (NL) purchase option does not include a front end commission nor can it attract deferred sales charges but it does pay a trailer fee to the fund dealer.  The Deferred Sales Charge (DSC) purchase option (which includes low load purchase options) requires the investor to pay a redemption fee to the fund company if the units are sold before a predetermined period has elapsed. The fund dealer receives an upfront commission directly from the fund company under this option as well as a trailer fee.  The Front End (FE) or initial sales charge purchase option allows the fund dealer the option of charging a negotiable front end commission directly upon initial sale and the fund dealer also receives a trailer fee.  The Fee Based (FB) purchase option does not allow for a front end commission to be charged, and it cannot attract deferred sales charges and there are no trailer fees paid to the dealer. Instead, the dealer charges fees directly to the investor’s account. High net worth or institutional series/purchase options where there is no front end commission, deferred sales charge or trailer fee paid are also reported as fee-based.

Mutual fund fees in Canada encompass more than just the publicly available management expense ratios (MERs).  Fees also comprise trading expenses, trailer fees, deferred sales charges, front end commissions, switch fees, performance fees, negotiated management fees, and other payments to dealers.  Some of these items change over the life of a fund.  The stated purpose of trailer fees is to compensate the advisor’s dealer firm each year for the ongoing investment services and advice they provide to investors.[7]  Trailer fees are paid at the same rate in perpetuity among some funds, and sometimes either increase or decrease over time among different funds.  Switch fees are fees that apply to investors when they switch from one fund to another within the same fund family.

Publicly available data on Canadian mutual fund flows from Morningstar and Bloomberg are estimated based on changes in NAVs.  However, there are many components of flow that vary widely and have little to do with direct retail investor incentives.  In many cases, there are pre-authorized contributions (PAC) from investor and employer accounts, just as there are systematic withdrawal plans (SWP).  Many flows are merely switches within the same fund family.  Changes in NAV may likewise be attributable to fund reinvestments or distributions to unit holders.  Finally, there are flows that are specific to affiliated dealers, and affiliated investment funds. 

Some mutual funds in Canada must be purchased through an advisor or intermediary, while others may be purchased directly from the fund company (although when buying directly, the person the investor deals with would still be considered an advisor).  The advice provided may significantly influence flows.

3. Data and Summary Statistics

3.1 Description of the Data

            The data were collected on behalf of the Canadian Securities Administrators (CSA) with the assistance of the Ontario Securities Commission (OSC) in 2015.[8]  The sample comprises the period 2003-2014, including 43 fund families of 113 in total in Canada or 38.1% of the market, and covering $746 billion of AUM (of an estimated total of $1.1 trillion) or 66.7% of the market,[9] as well as $152 billion in AUM for fund-of-funds or 51.5% of the $295 billion market at the end of 2014 (estimated by Investor Economics). More specifically, the sample comprises 18,102 FundSERV codes[10] for stand-alone funds and 4018 fund-of-funds FundSERV codes, or 22,077 FundSERV codes in total, compared to the 16,752 FundSERV codes from fundlibrary.com[11] and 12,300 FundSERV codes from fundata.com,[12] both of which state they have the most comprehensive mutual fund datasets in Canada.  In total, the sample comprises more than 1 million monthly FundSERV code observations on fees, flows, and performance, where a FundSERV code is used to identify a fund series/purchase option combination.  We treat fund-of-funds separately, as it is possible that their flows, fee structures, management structures, operations and managerial incentives are significantly different relative to stand-alone mutual funds that make direct investments into companies.  We carried out some comparison tests with publicly available data for the representativeness of the respondent fund families by examining publicly available data such as location and ratings, and did not find any statistically significant differences.  For reasons of confidentiality and to make sure no fund identity can be reverse engineered, we do not present specific information on fund families.

            Table 1 defines the main variables from the data.  The first key variable in Table 1 is the fund flows net of pre-authorized contributions (PAC), systematic withdrawal plans (SWP), switches, reinvestments, distributions, affiliated dealer flows, and affiliated investment fund flows [hereafter referred to as ‘net retail flow’].  The net retail flow variable is measured over each FundSERV code – monthly observation in the data, and as a percentage of the prior month’s FundSERV AUM, in order to make comparative assessments of flow across funds and over time.  Table 1 indicates that the FundSERV code monthly observation shows an average net retail flow for stand-alone funds of -1.87%, with a median of 0.00%, and standard deviation of 8.64%, and an average net flow for fund-of-funds of -6.30% with a median of -0.35% and standard deviation of 13.49%.  Including all types of inflows and outflows, the average monthly flow for stand-alone funds was -0.50%, median -0.02%, and standard deviation 7.99%, and average fund-of-funds was -2.73%, median -0.14%, and standard deviation 11.08%.  The average flows attributable to PAC and SWP for stand-alone funds (fund-of-funds) were 0.03%

Table 1. Variable Definitions and Summary Statistics This table provides defines the main variables in the dataset and provides summary statistics for the number of fund-month observations (Obs.) between January 2003 and December 2014, and means, medians, and standard deviations (Std. Dev.).
			Stand-alone funds							Fund of Funds
Variable	Definition		Obs.	Mean		Median	Std. Dev.			Obs.	Mean		Median		Std. Dev.
Fund Flow and Performance Variables
Flows Net of PAC, SWP, Switches, Reinvestments, Distributions, and Affiliated Dealer and Affiliated Investment Funds	Total monthly flows net of total monthly pre-authorized contribution (PAC) inflows, total monthly systematic withdrawal plan (SWP) outflows, switches in and switches out, reinvested distributions and distributions to unit holders, and affiliated dealer and affiliated investment funds inflows and outflows) / assets under management at start of month		1102377	-0.0187		0	0.0864			294232	-0.0630		-0.0035		0.1349
All Inflows - All Outflows	(Total monthly inflows - total month outflows) / assets under management at start of month		1102377	-0.005		-0.0002	0.0799			294232	-0.0273		-0.0014		0.1108
PAC Inflows - SWP Outflows	(Total monthly pre-authorized contribution (PAC) inflows - total monthly systematic withdrawal plan (SWP) outflows) / assets under management at start of month		1102377	0.0003		0	0.001			294232	0.0002		0		0.0008
Switches In - Switches Out	(Total monthly switches in - total monthly switches out) / assets under management at start of the month		1102377	0.0022		0	0.0159			294232	0.0037		0		0.0181
Reinvested Distributions - Paid Distributions	(Total reinvested distributions - distributions to unit holders) / assets under management at start of month		1102377	-0.0002		0	0.0013			294232	0.0012		0		0.0033
Affiliated Dealer Inflows - Outflows	(Total affiliated dealer inflows - total affiliated dealer outflows) / assets under management at start of month		1102377	0.0018		0	0.0069			294232	0.0029		0.0001		0.0092
Affiliated Investment Funds Inflows - Outflows	(Total affiliated investment fund inflows - total affiliated investment fund outflows) / assets under management at start of month		1102377	0.0057		0	0.0398			294232	0.0159		0		0.0609
Alpha	Alpha is calculated based on monthly gross return and Fama-French North America 4 factors. Monthly gross returns are winsorized at 1% level; Fama-French 4 factors (market, SMB, HML and WML) and risk-free rate come from Kenneth R. French - Data Library.		1010575	0.2502		0.2492	0.7433			264199	0.1321		0.001		0.5666
Purchase Option Variables
Purchase Option Deferred Sales Charge	A dummy variable equal to 1 if the purchase option requires the investor to pay a redemption fee if the units are sold before a predetermined period has elapsed.		1209285	0.4579		0	0.4975			308336	0.3690		0		0.4567
Purchase Option Front End	A dummy variable equal to 1 for fund series that charge a front end (FE) or initial sales charge - the purchase option allows the fund dealer the option of charging a negotiable front end commission directly upon initial sale.		1209285	0.3956		0	0.4874			308336	0.2501		0		0.4007
Purchase Option Fee Based	A dummy variable equal to 1 for feed based (FB) if the purchase option does not allow for a front end commission to be charged, it cannot attract deferred sales charges and there are no trailer fees paid to the dealer. Fee based options where there is a default trailer fee triggered (also sometimes termed a service fee) if the investor does not opt out are reported as no load. High net worth or institutional series/purchase options where there is no front end commission, deferred sales charge or trailer fee paid are also reported as fee-based.		1209285	0.0835		0	0.2743			308336	0.2288		0		0.3873
Purchase Option No Load	A dummy variable equal to 1 for no load (NL) when the purchase option does not include a front end commission nor can it attract deferred sales charges but it does pay a trailer fee to the fund dealer. Do-it-yourself or discount brokerage ("D series") purchase options are also be reported as a no load purchase option if they pay a trailer fee but do not pay a front end commission nor charge a deferred sales charge.		1209285	0.0629		0	0.2408			308336	0.1521		0		0.3275
Table 1 (Continued)			Stand-alone funds							Fund of Funds
Variable	Definition		Obs.	Mean		Median	Std. Dev.			Obs.	Mean		Median		Std. Dev.
Minimum Purchase Amount	The minimum initial investment amount in dollars for the series/purchase option as reported in the simplified prospectus at the time of purchase.		1209285	78970		500	460110.84			308336	10335		500		130544.94
Fund Expense Variables
Deferred Sales Charge Year 1	The deferred sales charge rate, as reported in the simplified prospectus at the time of purchase, that applied if the units/shares were subsequently redeemed during year 1.	1209285			4.7928	5.5		1.5246	308336			4.2111		5.5		2.0585
Deferred Sales Charge Slope	The average of the percentage change in deferred sales charges from one period to the next.	868685			-0.186	-0.1831		0.0486	299778			-0.1488		-0.2031		0.0939
Sales Commission for Deferred Sales Charge	The sales commission rate paid by the fund company to the fund dealer as reported in the simplified prospectus at the time of purchase.	1209285			1.6704	0		2.1481	308336			0.4838		0		1.3345
Maximum Front End Commission	The maximum front end commission rate as reported in the simplified prospectus at the time of purchase.	1209285			1.7338	0		2.3883	308336			0.9815		0		2.0542
Front End Commissions Paid	The total amount of front end commissions paid each month divided by assets under management at the start of the month.	1102377			0.0196	0		0.0871	294232			0.1065		0.0037		0.176
Maximum Posted Switch Fee	The maximum switch fee rate as reported in the simplified prospectus at the time of purchase.	1209285			1.9964	2		0.083	308336			1.6108		2		0.7918
Performance Fee	The percentage incentive fee charged by the fund manager to the fund.	1209285			0.0845	0		0.561	308336			0		0		0
Negotiated Management Fees Paid	Where the management fees of a particular fund series/purchase option are typically negotiated, the total amount of management fees received each month, divided by assets under management at the start of the month. This amount includes any payments made to the fund and any payments made directly to the fund company (or an affiliate) for fund management.	1102377			0.0244	0		0.0909	294232			0.0861		0		0.1655
Management Expense Ratio	The management expense ratio (MER) after waivers and absorptions as reported in the management report of fund performance (or the financial statements before 2006) at the time of purchase. Please refer to National Instrument 81-106 part 15 for the calculation.	1110152			2.0736	2.1604		0.6552	251669			1.9686		2.0258		0.5988
Trading Expense Ratio	The trading expense ratio (%) for the fund series/purchase option as reported in the management report of fund performance at the time of purchase. For periods before 2006, please calculate (estimates are acceptable) and report the TER as outlined in National Instrument 81-106F1.	1209285			1.3137	0.2467		2.3835	308336			0.6946		0.25		1.6239
Maximum Posted Initial Trailer	For "FE", "DSC", or "NL" as the purchase option type, the maximum trailer fee annualized rate that applied to assets held during the period reported.	1110152			0.5823	0.5		0.3628	251669			0.689		0.75		0.3302
Trailer Slope	The average of the percentage change in the trailer fee from one period to the next.	1074741			0.0664	0		0.1644	251360			-0.049		0		0.3601
Other Payments to Dealer-Brokers	The total amount of payments made each month to dealer-brokers that do not qualify as trailer fees, DSC sales commissions or referral fees and that are tied to fund series/purchase type inflows or assets under management held in each fund series/purchase type, divided by assets under management at the start of each period.	1102377			0	0		0.0001	294232			0		0		0

 

(0.02%), average flows attributable to switches were 0.22% (0.37%), attributable to reinvestments and distributions were -0.02% (0.12%), affiliated dealer flows were 0.18% (0.29%), and affiliated investment fund flows were 0.57% (1.59%).

            Table 1 further indicates that the average 12-month Fama-French 4-factor alpha in the sample is 0.25% for stand-alone funds (and 0.13% for fund-of-funds), with a median of 0.25% (0.001% for fund of funds), and standard deviation of 0.74% (0.57% for fund of funds).[13]  To the best of our ability and fund companies’ ability to provide the data, we have created a survivorship bias free dataset by including live and defunct FundSERV codes.[14]

            A total of 45.8% of the monthly observations in the sample comprise funds purchased under deferred sales charges for stand-alone funds (36.9% for fund-of-funds), while 39.7% (25.0% for fund-of-funds) were purchased under front end, 8.4% (22.9% for fund of funds) purchased under fee based, and 6.3% (15.2% for fund of funds) purchased under no load.

            Table 1 provides definitions and summary statistics for fee variables.  The summary statistics for the fee variables in Table 1 are indicated for the full sample of monthly fund observations across all purchase options.  For stand-alone funds (fund-of-funds), the average maximum posted initial trailer for the entire sample of month-year observations is 0.58% (0.69%), with a median of 0.50% (0.75%) and a standard deviation of 0.36% (0.33%).  Appendix I provides the information across each purchase option.  Appendix I Table I.1 shows that for stand-alone funds (Panel A), the average maximum posted initial trailer fee was 0.45% for deferred sales charge purchase options, 0.72% for front-end load, and 0.65% for no load.  Subsequent trailer fees on average are higher for deferred sales charge purchase options and front-end purchase options, and lower for no load purchase options, as indicated by the trailer slope variable. Among the fund-of-funds (Panel B), the average maximum posted initial trailer fee is 0.53% for deferred sales charge purchase options, 0.81% for front-end load, and 0.85% for no load, and subsequent trailers are lower for these funds. 

            Table 1 shows that for stand-alone funds (fund-of-funds), the average initial deferred sales charge for the entire sample of month-year observations is 4.79% (4.21%), with a median of 5.50% (5.50%) and a standard deviation of 1.52% (2.06%).  The data indicate subsequent years deferred sales charges are lower (as indicated by the deferred sales charge slope variable).  

Table 1 shows that for stand-alone funds (fund-of-funds), the average MER in the -monthly observations in the sample is 2.07% (1.97%), with a median of 2.16% (2.02%), and a standard deviation of 0.66% (0.60%).[15]  A number of other details pertaining to fees are summarized in Table 1 and in Appendix I.

 

3.2. Comparison Tests across All Fund Months

While comparison tests do not enable conclusive statements (they do not control for other things being equal), we present some comparison of means and medians tests in Table 2 to show some patterns in the data.  The comparisons are across all fund-month observations in the dataset, and hence the comparisons are crude and inconclusive. Net flows and alphas for each purchase option are compared to net flows and alphas for the rest of the sample. Similarly, we group the sample into above and below median along each fee variable and then net flows and alphas are compared. The more formal econometric tests are provided in the next subsection.

Amongst the stand-alone funds that cannot be purchased directly from the fund company (Table 2 Panel A), the data indicate deferred sales charges have significantly lower flows, and the average flow is significantly lower at the 10% level, while the median flow is significantly lower at the 5% level of significance.[16]  Funds with higher trading expense ratios, higher maximum posted initial trailers, and higher deferred sales charges have higher median and average flows, and these differences are significant at the 1% level.  Funds with increasing subsequent trailer fees, and funds with higher minimum purchase amounts have lower average and median flows, and these differences are significant at the 1% level.

Stand-alone funds that cannot be purchased directly (Table 2 Panel A) with lower trading expense ratios, and funds with lower minimum initial purchase amounts have significantly higher average and median alphas (significant at the 1% level).  Funds with a no load purchase option have higher average and median alphas (significant at the 1% and 5% levels, respectively).

Table 2. Comparison of Means and Medians Tests for New Flows and Alpha   This table presents comparison of means and medians for Alphas and inflows Net of PAC, SWP, Switches, Reinvestments, Distributions, and Affiliated Dealer and Affiliated Investment Funds. Alpha is calculated based on monthly gross return and Fama-French North America 4 factors. Monthly gross returns are winsorized at 1% level; Fama-French 4 factors (market, SMB, HML and WML) and risk-free rate come from Kenneth R. French - Data Library. Average/median net flows and alpha are first calculated at individual fund level then averaged out across the whole data-set. Fund of funds are excluded in this table..  *, **, *** Significant at the 10%, 5%, and 1% levels, respectively. N/A is used when at least one of the comparison pairs has insufficient observations.               Panel A. Stand-Alone Funds where series cannot be purchased directly from the fund company.       Average Net Flows With This Purchase Option, or Above Median for Fee Levels Average Net Flows Without This Purchase Option, or Below Median for Fee Levels Comparison of Means Median Net Flows With This Purchase Option, or Above Median for Fee Levels Median Net Flows Without This Purchase Option, or Below Median for Fee Levels Comparison of Medians Average Alpha With This Purchase Option, or Above Median for Fee Levels Average Alpha Without This Purchase Option, or Below Median for Fee Levels Comparison of Means Median Alpha With This Purchase Option, or Above Median for Fee Levels Median Alpha Without This Purchase Option, or Below Median for Fee Levels Comparison of Medians   Purchase Option Deferred Sales Charge -0.034 -0.015 -1.870* -0.034 -0.014 -1.990** 0.235 0.225 0.320 0.220 0.230 -0.350 Purchase Option Front End -0.013 -0.029 1.470 -0.012 -0.029 1.550 0.204 0.243 -1.280 0.222 0.228 -0.200 Purchase Option Fee Based -0.029 -0.023 -0.360 -0.029 -0.022 -0.390 0.235 0.229 0.130 0.221 0.226 -0.110 Purchase Option No Load -0.012 -0.024 0.440 -0.011 -0.023 0.470 0.429 0.222 2.730*** 0.410 0.219 2.550** Management Expense Ratio -0.030 -0.033 0.490 -0.027 -0.029 0.290 0.181 0.194 -0.980 0.182 0.207 -1.960** Trading Expense Ratio -0.011 -0.047 6.330*** -0.006 -0.047 7.350*** 0.157 0.188 -2.610*** 0.145 0.206 -4.950*** Maximum Posted Initial Trailer -0.003 -0.055 3.420*** -0.003 -0.056 3.460*** 0.237 0.266 -0.790 0.238 0.257 -0.510 Trailer Slope -0.016 0.009 -3.080*** -0.016 0.006 -2.900*** 0.229 0.235 -0.190 0.229 0.245 -0.520 Deferred Sales Charge Year 1 -0.005 -0.046 3.840*** -0.004 -0.046 4.030*** 0.241 0.205 1.270 0.244 0.194 1.780* Deferred Sales Charge Slope -0.038 -0.012 -1.520 -0.037 -0.008 -1.760* 0.223 0.236 -0.280 0.212 0.239 -0.600 Sales Commission for Deferred Sales Charge -0.034 0.000 N/A -0.033 0.000 N/A 0.229 0.000 N/A 0.218 0.000 N/A Maximum Front End Commission -0.014 0.000 N/A -0.013 0.000 N/A 0.194 0.000 N/A 0.211 0.000 N/A Maximum Posted Switch Fee 0.000 -0.007 N/A 0.000 -0.006 N/A 0.000 0.009 N/A 0.000 -0.011 N/A Other Payments to Dealer-Brokers -0.007 0.000 N/A -0.007 0.000 N/A 0.249 0.000 N/A 0.192 0.000 N/A Front End Commissions Paid -0.018 0.000 N/A -0.013 0.000 N/A 0.267 0.000 N/A 0.323 0.000 N/A Performance Fee 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A Negotiated Management Fees Paid -0.035 0.000 N/A -0.035 0.000 N/A 0.306 0.000 N/A 0.219 0.000 N/A Minimum Purchase Amount -0.046 0.007 -5.550*** -0.045 0.008 -5.580*** 0.180 0.304 -4.440*** 0.174 0.295 -4.440***                                   Table 2 (Continued) Panel B. Stand-Alone Funds where series can be purchased directly from the fund company.         Average Net Flows With This Purchase Option, or Above Median for Fee Levels Average Net Flows Without This Purchase Option, or Below Median for Fee Levels Comparison of Means Median Net Flows With This Purchase Option, or Above Median for Fee Levels Median Net Flows Without This Purchase Option, or Below Median for Fee Levels Comparison of Medians Average Alpha With This Purchase Option, or Above Median for Fee Levels Average Alpha Without This Purchase Option, or Below Median for Fee Levels Comparison of Means Median Alpha With This Purchase Option, or Above Median for Fee Levels Median Alpha Without This Purchase Option, or Below Median for Fee Levels Comparison of Medians   Purchase Option Deferred Sales Charge 0.000 -0.001 N/A 0.000 -0.004 N/A 0.000 0.118 N/A 0.000 0.109 N/A Purchase Option Front End 0.005 -0.002 0.650 -0.001 -0.005 -0.170 0.142 0.113 0.280 0.103 0.111 -0.070 Purchase Option Fee Based 0.000 -0.001 N/A 0.000 -0.004 N/A 0.000 0.118 N/A 0.000 0.109 N/A Purchase Option No Load -0.008 0.006 -0.710 -0.008 -0.001 -0.490 0.095 0.140 -0.600 0.099 0.120 -0.290 Management Expense Ratio 0.009 -0.002 1.120 0.000 -0.002 0.530 0.150 0.138 0.270 0.140 0.161 -0.460 Trading Expense Ratio 0.006 0.005 0.100 0.004 -0.002 0.980 0.214 0.178 0.610 0.241 0.130 2.050** Maximum Posted Initial Trailer -0.006 -0.007 0.030 -0.002 -0.009 0.210 0.034 0.061 -0.320 0.062 0.058 0.040 Trailer Slope -0.001 0.023 -0.420 -0.004 0.034 -0.520 0.133 0.357 -0.920 0.130 0.390 -1.040 Deferred Sales Charge Year 1 0.000 0.005 N/A 0.000 0.000 N/A 0.000 0.160 N/A 0.000 0.155 N/A Deferred Sales Charge Slope 0.004 -0.013 0.690 0.000 -0.012 0.380 0.112 0.010 1.480 0.087 0.027 0.830 Sales Commission for Deferred Sales Charge 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A Maximum Front End Commission 0.005 0.000 N/A -0.001 0.000 N/A 0.142 0.000 N/A 0.103 0.000 N/A Maximum Posted Switch Fee 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A Other Payments to Dealer-Brokers 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A Front End Commissions Paid 0.004 0.000 N/A -0.001 0.000 N/A 0.245 0.000 N/A 0.193 0.000 N/A Performance Fee 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A 0.000 0.000 N/A Negotiated Management Fees Paid -0.007 -0.008 0.040 -0.023 -0.005 -0.530 0.537 0.039 3.820*** 0.701 0.047 4.560*** Minimum Purchase Amount -0.001 0.000 N/A -0.002 0.000 N/A 0.134 0.000 N/A 0.219 0.000 N/A                                                 Table 2 (Continued) Panel C. Fund of funds where series cannot be purchased directly from the fund company.       Average Net Flows With This Purchase Option, or Above Median for Fee Levels Average Net Flows Without This Purchase Option, or Below Median for Fee Levels Comparison of Means Median Net Flows With This Purchase Option, or Above Median for Fee Levels Median Net Flows Without This Purchase Option, or Below Median for Fee Levels Comparison of Medians Average Alpha With This Purchase Option, or Above Median for Fee Levels Average Alpha Without This Purchase Option, or Below Median for Fee Levels Comparison of Means Median Alpha With This Purchase Option, or Above Median for Fee Levels Median Alpha Without This Purchase Option, or Below Median for Fee Levels Comparison of Medians   Purchase Option Deferred Sales Charge -0.0040 -0.0056 4.51*** 0.0000 0.0000 -0.0200 0.2949 0.2993 -1.86* 0.2548 0.2539 0.3900 Purchase Option Front End -0.0025 -0.0063 10.17*** 0.0000 0.0000 -1.0800 0.3126 0.2900 9.28*** 0.2816 0.2537 11.88*** Purchase Option Fee Based -0.0139 -0.0035 -20.48*** 0.0000 0.0000 -0.1600 0.2948 0.2981 -1.00 0.2336 0.2545 -6.65*** Purchase Option No Load -0.0058 -0.0049 - 1.65* 0.0000 0.0000 0.6200 0.2906 0.2985 -2.16** 0.2539 0.2543 0.1300 Management Expense Ratio -0.0062 -0.0033 - 8.19*** 0.0000 0.0000 -0.9200 0.2866 0.3127 -11.17*** 0.2524 0.2551 -1.2100 Trading Expense Ratio -0.0048 -0.0058 2.87*** 0.0000 0.0000 -1.4900 0.3137 0.2870 11.39*** 0.2558 0.2510 2.16** Maximum Posted Initial Trailer -0.0072 0.0031 -28.85*** 0.0000 0.0000 -0.8400 0.2902 0.3062 -4.92*** 0.2523 0.2561  -1.20 Trailer Slope -0.0052 -0.0005  -4.62 *** 0.0000 -0.0001 1.14 0.3099 0.0244  43.25*** 0.2561 -0.0025  41.47*** Deferred Sales Charge Year 1 -0.0190 -0.0121 -6.71*** 0.0000 -0.0001 1.38 0.2265 0.2856  -15.17*** 0.1144 0.2539  -38.29*** Deferred Sales Charge Slope -0.0074 -0.0020 -13.21*** -0.0001 0.0009 -9.80*** 0.3186 0.3298  -4.34 *** 0.2576 0.2598  -0.88 Sales Commission for Deferred Sales Charge -0.0030 0.0000 N/A 0.0000 0.0000 N/A 0.2982 0.0000 N/A 0.2545 0.0000 N/A Maximum Front End Commission -0.0046 0.0000 N/A 0.0000 0.0000 N/A 0.2897 0.0000 N/A 0.2467 0.0000 N/A Maximum Posted Switch Fee 0.0000 0.0500 N/A 0.0000 0.0355 N/A 0.0000 0.0020 N/A 0.0000 0.0025 N/A Other Payments to Dealer-Brokers -0.0197 0.0000 N/A 0.0000 0.0000 N/A 0.3490 0.0000 N/A 0.3519 0.0000 N/A Front End Commissions Paid -0.0191 0.0000 N/A 0.0000 0.0000 N/A 0.2500 0.0000 N/A 0.1433 0.0000 N/A Performance Fee -0.0123 0.0000 N/A -0.0085 0.0000 N/A 0.3204 0.0000 N/A 0.2996 0.0000 N/A Negotiated Management Fees Paid -0.0192 0.0000 N/A 0.0000 0.0000 N/A 0.3534 0.0000 N/A 0.3758 0.0000 N/A Minimum Purchase Amount -0.0061 0.0011 -22.27*** 0.0000 0.0001 -3.34*** 0.1936 0.4155  -85.05*** 0.1788 0.4073  -88.67***             Table 2 (Continued) Panel D. Fund of funds where series can be purchased directly from the fund company.

	Average Net Flows With This Purchase Option, or Above Median for Fee Levels	Average Net Flows Without This Purchase Option, or Below Median for Fee Levels	Comparison of Means	Median Net Flows With This Purchase Option, or Above Median for Fee Levels	Median Net Flows Without This Purchase Option, or Below Median for Fee Levels	Comparison of Medians	Average Alpha With This Purchase Option, or Above Median for Fee Levels	Average Alpha Without This Purchase Option, or Below Median for Fee Levels	Comparison of Means	Median Alpha With This Purchase Option, or Above Median for Fee Levels	Median Alpha Without This Purchase Option, or Below Median for Fee Levels	Comparison of Medians
Purchase Option Deferred Sales Charge	0.0142	-0.0053	35.25***	0.0031	-0.0003	14.69 ***	0.1632	0.2728	-43.71 ***	0.0046	0.2849	-126.58***
Purchase Option Front End	0.0062	0.0021	11.10 ***	0.0004	0.0002	2.00**	0.1848	0.2430	-21.60 ***	0.0533	0.1472	-45.11 ***
Purchase Option Fee Based	-0.0140	0.0072	-21.19 ***	-0.0034	0.0014	-10.01***	0.4431	0.1739	84.04 ***	0.6166	0.0089	242.45 ***
Purchase Option No Load	-0.0428	0.0061	-17.58***	-0.0064	0.0005	-6.43***	0.1071	0.2314	-22.50***	0.0006	0.1312	-29.91***
Management Expense Ratio	0.0000	0.0142	-29.96***	0.0000	0.0040	-15.82 ***	0.2883	0.1178	64.35 ***	0.3350	0.0001	130.30***
Trading Expense Ratio	0.0054	0.0326	-19.02***	0.0009	0.0121	-12.97 ***	0.3659	0.0817	85.16 ***	0.4028	-0.0015	141.39***
Maximum Posted Initial Trailer	-0.0057	0.0212	-38.27***	-0.0001	0.0041	-15.12 ***	0.2755	0.1581	44.04 ***	0.2878	0.0027	121.68***
Deferred Sales Charge Year 1	0.0216	0.0289	-3.84***	0.0028	0.0097	-6.47***	0.3403	0.0650	76.70***	0.3702	-0.0019	136.35 ***
Deferred Sales Charge Slope	0.0279	-0.0223	53.32 ***	0.0072	-0.0062	27.33 ***	0.1031	0.3353	-93.47***	-0.0001	0.4552	-180.90***
Sales Commission for Deferred Sales Charge	0.0139	0000	N/A	0.0040	0000	N/A	0.1415	0000	N/A	0.0009	0000	N/A
Maximum Front End Commission	0.0062	0000	N/A	0.0004	0000	N/A	0.1848	0000	N/A	0.0533	0000	N/A
Maximum Posted Switch Fee	0000	-0.0152	N/A	0000	-0.0155	N/A	0000	0.0592	N/A	0000	0.0172	N/A
Other Payments to Dealer-Brokers	0000	0000	N/A	0000	0000	N/A	0000	0000	N/A	0000	0000	N/A
Front End Commissions Paid	-0.0077	0.0313	N/A	-0.0007	0.0098	N/A	0.2330	0.1032	N/A	0.2468	-0.0004	N/A
Negotiated Management Fees Paid	0.0408	0000	N/A	0.0190	0000	N/A	0.0620	0000	N/A	-0.0018	0000	N/A
Minimum Purchase Amount	0.0179	0.0236	-5.62***	0.0040	0.0060	-4.00***	0.3293	0.1327	65.57***	0.3456	0.0016	127.97***

 

            Amongst the fund series that can be purchased directly from the fund company (Table 2 Panel B), funds with a higher trading expense ratio have higher median alphas (significant at the 5% level) and funds with higher negotiated management fees paid have higher alphas.  The other differences in Panel B are not statistically significant.

Table 2 Panel C shows higher average monthly fund flows for deferred sales charges, front end purchase options, and funds with higher trading expense ratios, and lower average monthly flows for fee based and no load purchase options, as well as for funds with higher management expense ratios, higher initial trailers, higher subsequent trailers, funds with higher initial deferred sales charges, funds with increasing subsequent deferred sales charges, and funds with lower minimum purchase amounts; however, these differences are only significant in medians for increasing deferred sales charges and for funds with higher minimum purchase amounts.

Table 2 Panel C shows fund-of-funds that cannot be purchased directly have higher alphas among funds with front end purchase options, lower management expense ratios, higher trading expense ratios, lower maximum posted initial trailer fees, increasing subsequent trailers, lower initial deferred sales charges, lower subsequent deferred sales charges, and lower initial purchase amounts.  Average alphas are lower among fund-of-funds with no load purchase options and deferred sales charge options  Median alphas are higher amongst fund-of-funds with front end purchase options.  Median alphas are lower among fund-of-funds with fee based purchase options, and funds with higher minimum purchase amounts.

Table 2 Panel D shows fund-of-funds that can be purchased directly have higher average and median monthly flows when they have deferred sales purchase options, front end purchase options, higher subsequent deferred sales charges, and lower minimum purchase amounts.  Fund-of-funds with fee-based and no load purchase options, higher management expense ratios, trading expense ratios, higher maximum posted initial trailers, and higher initial deferred sales charges have lower average and median monthly flows.  Table 2 Panel D further shows fund-of-funds that can be purchased directly have higher average and median alphas when they have fee based purchase options, higher management expense ratios, higher trading expense ratios, higher maximum posted initial trailers, higher initial deferred sales charges, lower subsequent deferred sales charges, and higher minimum purchase amounts.  Fund-of-funds that can be purchased directly with no-load, deferred sales charge, and front end purchase options have lower average and median alphas.

As mentioned, it is important to keep in mind that these tests in Table 2 are not presented to be conclusive, as they do not control for other things being equal.  The Table 2 tests simply compare averages and medians of all monthly-fund observations in the data, and show some general patterns.  In section 4, we provide regression evidence to make a more accurate assessment of the patterns in the data.  But first, in subsection 3 we examine comparison tests for a subset of the funds that made permanent changes to fees over time.

3.3. Comparison Tests for the Subset of Funds that Permanently Changed Trailer Fees

More precise comparison tests are provided in Figure 1.  Comparison tests in Figure 1 are more precise than those in Table 2 because Figure 1 focuses on the narrow same set of funds that changed their trailer fees over time (and hence fund characteristics are kept constant because we examine the same funds at different points in time), while Table 2 did not keep other fund characteristics constant.  The data in Figure 1 indicate that 2.5% of funds permanently increased

Figure 1. Abnormal return associated with permanent first year trailer fee jumps (drops)

We analyzed the funds that changed trailer fee in Year 1 over the sample horizon. The total sample contains 22077 different funds; among which, 559 or 2.5% of funds have increased their trailer fee in year 1; 123 or around 0.6% of funds have reduced their trailer fee in year 1. We tracked the fund performance before and after trailer fee changes. In the following two graphs, the horizontal axes show the event month, where the trailer fee change happens in month 0. The vertical axes show the average (median) alpha by each event month for the funds that changed their trailer fees in Year 1. We restrict the sample to funds that have both at least 6 month observations before the event month and at least 24 month observations after the event month.

                      

 

 

 

 

their trailer fees in the sample period, and fees increased from an average of 0.39% to 0.78%. Comparison tests of the alphas for these funds from the 6-month prior fee change period to the 24 month post fee change period show alpha dropped by 32.4% on average and this difference is statistically significant at the 1% level. The data in Figure 1 further indicate that 0.6% of funds permanently decreased their trailer fees in the sample period from an average of 0.43% to 0.27%. Comparison tests of the alphas for these funds from the 6-month prior fee change period to the 24 month post fee change period show alpha increased by 88.4% on average, and this difference is statistically significant at the 1% level.

4. Regression Analyses

            This section first presents flow-performance regressions of monthly flows in subsection 4.1.  Thereafter subsection 4.2 presents regression analyses of yearly alphas.

4.1. Flow-Performance Regression Analyses

Table 3 presents regression analyses of fund flows relative to past performance for all of the funds in the data.  The basic model specification is as follows:

Flow_t+1 = Constant + β₁ * Alpha_t + β₂ * Alpha²_t + β₃ * Purchase Option Dummy + β₄ * Purchase Option Dummy * Alpha_t + β₅ * controls + residuals

The regression is estimated as a panel model with random effects across each fund series/purchase option (FundSERV code) and month.[19]    Random effects are used in Table 3 and not fixed effects because some of the right-hand-side variables are time invariant.  Standard errors are clustered by FundSERV code (Petersen, 2009).[20]  The results are robust to not clustering or clustering on different variables. The dependent variable in Table 3 is the total monthly inflows minus outflows, net of total monthly pre-authorized contribution (PAC) inflows, total monthly systematic withdrawal plan (SWP) outflows, switches in and switches out, reinvested distributions and distributions to unit holders, and affiliated dealer and affiliated investment fund inflows and outflows, and divided by AUM at start of month.  Models 1-5 in Table 3 are presented for the subset of funds that cannot be purchased directly from the fund company, and Models 6-7 are presented for the subset of funds that can be purchased directly from the fund company.  The different models include different right-hand-side variables to show robustness to different specifications and possible collinearity effects.  We do not present all purchase options together in the same model since that would create perfect collinearity problems.  The excluded purchase option variable(s) in each model means that the coefficients are estimating the impact of the included variables relative to the excluded purchase option variables.  Models 6-7 do not include the more complete set of all variables as in Model 5 because there are fewer observations for the purchase option directly from the fund company less variation amongst the sample to enable including all of the other variables. 

 

 

 

            The data in Table 3 indicate the following.  First, prior performance in terms of alpha is positively related to future flow: higher alpha increases future flow, while lower alpha reduces future flow.  A 1-standard deviation increase in prior alpha causes a 10.0% increase in future flow (based on Model 5, and this effect is most conservatively estimated at 4.2% in Model 2 and least conservatively estimated as 16.7% in Model 6), and this effect is statistically significant at the 1% level in all models.  Each of the models includes a squared alpha term to account for possible non-linearity in flow.  Prior work on mutual fund flows (Del Guerci and Tkac, 2002) shows that flow is convex (retail investors are quick to rush to invest into funds that have had recent prior success, but slow to withdraw capital from funds that have had poor recent performance).  The regressions in Table 3 indicate that flow is not convex when the series cannot be purchased directly from the fund company, but is convex when the series can be purchased directly from the fund company.

For funds that cannot be purchased directly from the fund company, the data indicate that deferred sales charges lower the level of flow regardless of prior performance.  This effect is significant at the 5% level in Model 5, and at the 1% level in Models 1 and 7.  The other intercept effects in Table 3 Panel A are not statistically robust.  Model 5 also indicates that funds with deferred sales charges, front-end purchase options, and no load purchase options also flatten the flow-performance relationship.  These effects are all statistically significant at the 1% level in Model 5.  Model 2 indicates that front end purchase options have a higher flow-performance slope, but that is measured relative to the average non-front end fund in Model 2; likewise, Figure 2 shows that the negative impact on the slope of flow-performance for front end purchase options is less pronounced than for deferred sales charge, and no load purchase options.  By contrast, the impact of fee based purchase options (without deferred sales charges, commissions and trailer fees) has a steeper flow-performance line and a higher flow regardless of performance, and these effects are both statistically significant at the 1% level in Model 3.  The economic significance of these effects is shown in Figure 2.  A fund that moves from the top quartile to the bottom performance quartile, for example, experiences a drop in flow relative to AUM by 0.32% under fee based purchase options, but only 0.26% under front end purchase options, 0.19% under no load purchase options, and 0.13% under deferred sales charge purchase options, controlling for other things being equal. 

For funds that can be purchased directly from the fund company (Models 6 and 7 in Table 3), the data do not enable an accurate assessment of the interaction effects with prior performance and purchase option types due to the smaller degree of variation in the sample.  Nevertheless, for the interaction terms that can be included, they are positive and significant, meaning that the effect is more pronounced relative to the excluded interactions, and generally consistent with the evidence in Models 1-5.  Likewise, similar evidence is observed for fund-of-funds in Table 3 Panel B, with the exception that no-load purchase options do not appear to flatten the flow-performance line for fund-of-funds, unlike that for stand-alone funds.

 

 

Figure 2. Purchase Options and Flow-Performance Intercept and Slope

This figure shows the impact of different purchase options on the flow-performance diagram.  Estimates are based on Table 3 coefficient estimates.  The first figure shows the change in slope, and the second figure shows the intercept change.

 

Table 4 presents regression analyses of fund flows relative to past performance for the subset of funds that do not allow for fee based purchase options (i.e., only including funds that  allow for deferred sales charges, commissions and trailer fees).  The basic model specification is as follows:

Flow_t+1 = Constant + β₁ * Alpha_t + β₂ * Alpha²_t + β₃ * MER_t + β₄ * MER_t * Alpha_t + β₅ * Alpha²_t * MER_t + β₆ * Trailer Fee_t + β₇ * Trailer Fee_t * Alpha_t + β₈ * Alpha²_t * Trailer Fee_t + β₉ * Other Types of Fee_t + β₁₀ * Other Types of Fee_t * Alpha_t + β₁₁ * controls + residuals

Table 4 presents six different specifications to show robustness.  Fund fixed effects are used in Table 4 because the right-hand-side variables are time variant, and because the Hausmann (1978) confirmed the validity of the random effects specification.  There is a change in fees in 8.52% of the sample months (some of which are permanent, and others transitory or temporary and subsequently reversed), and hence there is sufficient variation to not warrant any adjustments to the fixed effects estimation procedure suggested by Plumper and Troeger (2007), for example. Standard errors are clustered by FundSERV Code (Petersen, 2009).  The results are robust to not clustering or clustering on different variables.  Table 4 examines the subset of funds that do not include the fee-based purchase option because those funds do not allow for trailer fees.  In Appendix II we present a very large number of additional specifications to assess robustness, including different sets of variables, different variables, and different subsets of the data such as those that exclude various purchase option types.  The main findings from the data are quite robust across different specifications, with some differences that are noted therein.

 

Table 4. Flow Sensitivity Analysis for Subsample Excluding Fee-Based

 

This table presents monthly FundSERV code fixed effects panel regressions of the determinants of the percentage flow as a function of prior month’s alpha, fee variables, interaction terms between fees and alphas, and control variables. The dependent variable is Flows Net of PAC, SWP, Switches, Reinvestments, Distributions, and Affiliated Dealer and Affiliated Investment Funds. Variables are as defined in Table 1.  Standard errors are clustered by FundSERV code.  T-statistics are in parentheses. *, **, *** Significant at the 10%, 5%, and 1% levels, respectively.

 

Panel A. Stand-Alone Funds,  cannot be directly purchased from fund manager

 

 

 

 

 

 

Table 4. Panel A. (Continued)

 

 

 

 

 

 

 

 

 

 

Table 4 (Continued)

Panel B. Stand-Alone Funds, can be directly purchased from fund manager

 

 

 

	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Alpha Lagged	0.00368***	0.00181**	0.00266***	0.00266***	0.00266***	0.00254***
	(11.85)	(2.04)	(2.93)	(2.93)	(2.85)	(2.72)
Alpha Lagged^2	0.000819***	0.000900***	0.000882***	0.000898***	0.000908	-0.00352***
	(4.51)	(4.61)	(4.52)	(4.60)	(1.57)	(-3.64)
Trading Expense Ratio (TER) %	-0.000505***	-0.000511***	-0.000513***	-0.000513***	-0.000513***	-0.000480***
	(-3.85)	(-3.74)	(-3.76)	(-3.76)	(-3.76)	(-3.52)
Alpha Lagged * Trading Expense Ratio	0.000248*	0.000348**	0.000379**	0.000374**	0.000374**	0.000381**
	(1.72)	(2.29)	(2.50)	(2.46)	(2.46)	(2.51)
Alpha Lagged^2 * Trading Expense Ratio	-0.000250***	-0.000284***	-0.000286***	-0.000287***	-0.000287***	-0.000363***
	(-2.72)	(-2.92)	(-2.94)	(-2.95)	(-2.94)	(-3.70)
Maximum Posted Initial Trailer Fee (%)		-0.0544***	-0.0548***	-0.0539***	-0.0539***	-0.0540***
		(-9.08)	(-9.15)	(-8.88)	(-8.88)	(-8.90)
Alpha Lagged *Maximum Posted Initial Trailer Fee		0.00199*	-0.000471	-0.000419	-0.000414	-0.000262
		(1.80)	(-0.38)	(-0.34)	(-0.33)	(-0.21)
Maximum Front End Commission (%)			-282.1	-228.5	-228.7	-261.0
			(-0.01)	(-0.01)	(-0.01)	(-0.01)
Alpha Lagged * Maximum Front End Commission			0.000601***	0.000591***	0.000591***	0.000594***
			(4.66)	(4.54)	(4.54)	(4.43)
Minimum Purchase Amount ($)				-0.000000126***	-0.000000126***	-0.000000125***
				(-8.70)	(-8.70)	(-8.61)
Alpha Lagged * Minimum Purchase Amount				-4.37e-09	-4.37e-09	6.96e-10
				(-1.17)	(-1.17)	(0.18)
Management Expense Ratio (MER) %				0.00131	0.00131	0.00102
				(1.09)	(1.09)	(0.84)
Alpha Lagged^2 *Maximum Posted Initial Trailer Fee					-0.0000120	0.00113
					(-0.02)	(1.41)
Alpha Lagged^2 * Management Expense Ratio						0.00223***
						(5.85)
Alpha Lagged^2 * Maximum Front End Commission						0.0000426
						(0.55)
Constant	-0.0426***	-0.0172***	231.8	187.8	188.0	214.5
	(-258.08)	(-4.28)	(0.01)	(0.01)	(0.01)	(0.01)
Number of Observations	164078	134505	134505	134505	134505	134505
Number of Groups	2407	2010	2010	2010	2010	2010
R2 within	0.0017	0.0023	0.0025	0.0031	0.0031	0.0033
R2 between	0.0098	0.004	0.0655	0.0655	0.0655	0.0655
R2 overall	0.0036	0.002	0.0372	0.0372	0.0372	0.0372
F	54.14***	43.71***	36.41***	33.78***	31.18***	29.32***

 

 

 

 

 

 

Table 4 (Continued)

Panel C. Fund of  Funds, cannot be directly purchased from fund manager

 

 

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Alpha Lagged	0.00309**	0.00297**	0.00580**	0.0119***	2.550	2.552
	(2.44)	(2.31)	(2.53)	(3.00)	(0.24)	(0.24)
Management Expense Ratio(MER)%	0.00265***	0.00264***	0.00328***	0.00315***	0.00296***	0.00284***
	(3.86)	(3.85)	(4.54)	(4.35)	(4.08)	(3.90)
Alpha Lagged * Management Expense Ratio	-0.000365	-0.000327	-0.00146**	-0.00260***	-0.00194**	-0.00231***
	(-0.68)	(-0.60)	(-2.00)	(-3.26)	(-2.40)	(-2.79)
Trading Expense Ratio (TER) %	0.0000999	0.0000901	0.000219**	0.000160	0.000144	0.000144
	(1.16)	(1.03)	(2.00)	(1.47)	(1.32)	(1.32)
Alpha Lagged * Trading Expense Ratio	-0.0000589	-0.0000701	-0.00000491	0.000111	0.000184	0.000185
	(-0.46)	(-0.53)	(-0.03)	(0.70)	(1.15)	(1.16)
Maximum Posted Initial Trailer Fee (%)	0.000912	0.000920	0.000197	-0.0000921	-0.00120	-0.00136
	(0.36)	(0.36)	(0.05)	(-0.02)	(-0.29)	(-0.33)
Alpha Lagged *Maximum Posted Initial Trailer Fee	-0.00168*	-0.00169*	0.000398	0.000262	0.00436***	0.00210
	(-1.74)	(-1.75)	(0.33)	(0.21)	(3.03)	(1.25)
Trailer Slope	-0.00308	-0.00308	-0.000820	-0.00263	-0.00213	-0.00200
	(-1.34)	(-1.34)	(-0.28)	(-0.90)	(-0.72)	(-0.68)
Alpha Lagged *Trailer Slope	-0.00553***	-0.00552***	-0.00302	-0.00195	-0.00361	-0.00401
	(-3.39)	(-3.38)	(-1.25)	(-0.81)	(-1.48)	(-1.64)
Alpha Lagged*Performance Fee	0.00247***	0.00247***	0.00219***	0.00195***	0.00232***	0.00232***
	(6.68)	(6.63)	(5.26)	(4.11)	(4.85)	(4.84)
Negotiated Management Fee (%)	-0.0246***	-0.0247***	-0.0210***	-0.0209***	-0.0194***	-0.0193***
	(-7.18)	(-7.18)	(-5.90)	(-5.90)	(-5.46)	(-5.44)
Alpha Lagged * Negotiated Management Fee	-0.00141	-0.00138	-0.00559	-0.00501	-0.00936**	-0.00972**
	(-0.37)	(-0.36)	(-1.30)	(-1.17)	(-2.15)	(-2.23)
Alpha Lagged^2		0.0000660	0.000415	-0.00137	-0.00147	-0.00484***
		(0.24)	(1.17)	(-1.39)	(-1.49)	(-3.08)
Alpha Lagged^2 * Trading Expense Ratio		0.0000502	-0.0000982	-0.000189*	-0.000238**	-0.000235**
		(0.60)	(-1.05)	(-1.89)	(-2.38)	(-2.33)
Deferred Sales Charge Slope			-0.142**	-0.138**	-0.135**	-0.136**
			(-2.50)	(-2.44)	(-2.32)	(-2.33)

 

 

 

Table 4. Panel C.  (Continued)

		Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
Alpha Lagged * DSC Slope					0.0102		0.00378		0.00332		0.00356
					(1.31)		(0.48)		(0.42)		(0.44)
Front End Commission paid for DSC purchases (%)					-0.000446		0.000911		0.00179		0.00179
					(-0.20)		(0.39)		(0.76)		(0.76)
Alpha Lagged *Sales Commission paid for DSC					-0.000278		-0.000145		-0.000941***		-0.000885***
					(-1.26)		(-0.65)		(-3.52)		(-3.30)
Alpha Lagged^2* Management Expense Ratio							0.00104**		0.00114**		0.00179***
							(2.02)		(2.21)		(3.16)
Deferred Sales Charge Year 1							-0.0218**		-0.0214*		-0.0212*
							(-2.25)		(-1.79)		(-1.78)
Alpha Lagged *DSC Amount Year 1 (%)							-0.00111*		-0.000423		-0.000563
							(-1.70)		(-0.64)		(-0.85)
Front End Commission Paid (%)							-0.0873***		-0.0876***		-0.0877***
							(-28.62)		(-28.71)		(-28.75)
Alpha Lagged * Front End Commission paid							-0.00641*		-0.00561		-0.00542
							(-1.68)		(-1.47)		(-1.42)
Maximum Front End Commission for FE purchases (%)									-0.000823		-0.000592
									(-0.08)		(-0.06)
Alpha Lagged * Maximum Front End Commission (%)									-0.00142***		-0.00139***
									(-5.40)		(-5.29)
Alpha Lagged * Maximum Posted Switch Fee									-1.270		-1.270
									(-0.23)		(-0.23)
Alpha Lagged^2 *Maximum Posted Initial Trailer Fee											0.00336***
											(2.78)
Minimum Purchase Amount($)											-9.97e-10
											(-0.68)
Alpha Lagged * Minimum Purchase Amount											9.40e-10
											(0.51)
Constant	-0.00958***		-0.00958***		-0.0383***		0.0831		0.0825		0.0815
	(-4.25)		(-4.25)		(-3.04)		(1.55)		(1.49)		(1.47)
Number of Observations	105089		105089		73611		73611		73611		73611
Number of Groups	2027		2027		1468		1468		1468		1468
R2 within	0.0015		0.0015		0.0021		0.0147		0.0151		0.0152
R2 between	0.0001		0.0001		0.0006		0.0315		0.02		0.0215
R2 overall	0.0013		0.0012		0.0012		0.0466		0.035		0.0363
F	12.72***		10.96***		8.471***		46.66***		42.42***		38.32***

 

 

 

 

 

Table 4 (Continued)

Panel D. Fund-of-Funds, can be directly purchased from fund manager

 

 

 

	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Alpha Lagged	0.0516	0.0499	0.385	0.0512	0.0493	0.579
	(1.36)	(1.32)	(0.57)	(1.35)	(1.30)	(0.85)
Management Expense Ratio (MER) %	0.00384**	0.00443***	0.00444***	0.00389***	0.00444***	0.00443***
	(2.55)	(2.94)	(2.95)	(2.58)	(2.95)	(2.94)
Alpha Lagged * Management Expense Ratio	0.00267*	0.00357**	0.00353**	0.00284**	0.00384***	0.00365**
	(1.89)	(2.51)	(2.48)	(1.99)	(2.67)	(2.51)
Trading Expense Ratio (TER) %	-0.000528**	-0.000667***	-0.000680***	-0.000523**	-0.000633***	-0.000641***
	(-2.24)	(-2.83)	(-2.87)	(-2.22)	(-2.68)	(-2.70)
Alpha Lagged * Trading Expense Ratio	0.000106	0.000370	0.000388	0.0000895	0.000647*	0.000713**
	(0.36)	(1.24)	(1.27)	(0.30)	(1.88)	(1.99)
Maximum Posted Initial Trailer Fee (%)	0.0446	0.0453	0.0507	0.0445	0.0454	0.0507
	(1.29)	(1.31)	(1.47)	(1.29)	(1.31)	(1.47)
Alpha Lagged *Maximum Posted Initial Trailer Fee	0.00768***	0.00659***	0.00699***	0.00775***	0.00679***	0.00677***
	(3.41)	(2.92)	(3.10)	(3.43)	(3.01)	(2.90)
Trailer Slope	0.0129***	0.0134***	0.0134***	0.0130***	0.0135***	0.0135***
	(11.29)	(11.76)	(11.76)	(11.35)	(11.85)	(11.85)
Alpha Lagged * Trailer Slope	-0.00127	-0.000986	-0.00117	-0.00143	-0.00113	-0.00135
	(-1.11)	(-0.86)	(-1.01)	(-1.25)	(-0.99)	(-1.16)
Alpha Lagged *Deferred Sales Charge Year 1	0.00109***	0.00138***	0.00154***	0.00109***	0.00131***	0.00153***
	(2.60)	(3.23)	(2.94)	(2.59)	(3.05)	(2.92)
Front End Commission Paid for DSC purchases (%)	-0.00441	-0.00437	-0.00281	-0.00411	-0.00398	-0.00236
	(-0.33)	(-0.33)	(-0.21)	(-0.31)	(-0.30)	(-0.18)
Alpha Lagged *Sales Commission Paid for DSC	-0.000135	-0.000386	-0.000385	-0.000158	-0.000403	-0.000429
	(-0.48)	(-1.37)	(-1.33)	(-0.56)	(-1.43)	(-1.48)
Alpha Lagged * Maximum Posted Switch Fee	-0.0321*	-0.0314*	-0.199	-0.0322*	-0.0314*	-0.296
	(-1.70)	(-1.66)	(-0.59)	(-1.70)	(-1.66)	(-0.87)
Front End Commission Paid (%)	-0.0835***	-0.0854***	-0.0848***	-0.0835***	-0.0854***	-0.0847***
	(-41.04)	(-42.00)	(-41.60)	(-41.02)	(-41.99)	(-41.58)
Alpha Lagged * Front End Commission Paid	0.00762***	0.00665***	0.00650***	0.00766***	0.00686***	0.00664***
	(4.50)	(3.92)	(3.80)	(4.52)	(4.04)	(3.88)
Negotiated Management Fee (%)		0.0681***	0.0677***		0.0681***	0.0677***
		(17.38)	(17.15)		(17.39)	(17.14)
Alpha Lagged * Negotiated Management Fee		-0.0318***	-0.0302***		-0.0323***	-0.0301***
		(-6.25)	(-5.26)		(-6.33)	(-5.24)

 

 

 

 

 

 

 

 

 

 

 

Table 4. Panel D. (Continued)

 

 

 

	Model 7			Model 8		Model 9		Model 10		Model 11		Model 12
Minimum Purchase Amount ($)					-0.00270***						-0.00270***
					(-5.98)						(-5.98)
Alpha Lagged * Minimum Purchase Amount					-0.00000223						-0.00000353
					(-0.50)						(-0.78)
Alpha Lagged^2							0.00198*		0.00137		-0.000578
							(1.67)		(1.14)		(-0.21)
Alpha Lagged^2 * Management Expense Ratio							-0.00103		-0.000481		0.000275
							(-1.24)		(-0.56)		(0.22)
Alpha Lagged^2 * Trading Expense Ratio									-0.000313*		-0.000370**
									(-1.79)		(-2.02)
Alpha Lagged^2 *Maximum Posted Initial Trailer Fee											0.00125
											(0.78)
Constant		-0.0167	-0.0272		0.627***		-0.0171		-0.0279		0.625***
		(-0.46)	(-0.75)		(5.44)		(-0.47)		(-0.77)		(5.43)
Number of Observations		72081	72081		72081		72081		72081		72081
Number of Groups		1304	1304		1304		1304		1304		1304
R2 within		0.0618	0.0923		0.1164		0.1172		0.1341		0.1457
R2 between		0.0525	0.0622		0.0786		0.0978		0.1134		0.1244
R2 overall		0.052	0.0535		0.0641		0.0843		0.1032		0.1233
F		127.4***	131.1***		119.2***		112.8***		111.9***		98.97***