capture

Document Sample
capture Powered By Docstoc
					              Investor Monitoring and Differences in Mutual Fund Performance
                                                       Christopher James
                                                     University of Florida
                                                Warrington College of Business
                                                        308 Stuzin Hall
                                                 Gainesville, FL 32611-7168
                                                    Office: (352) 392-3486
                                                      Fax: (352) 392-0301
                                                christopher.james@cba.ufl.edu

                                                                    and

                                                       Jason Karceski*
                                                     University of Florida
                                                Warrington College of Business
                                                       308 Stuzin Hall
                                                 Gainesville, FL 32611-7168
                                                   Office: (352) 846-1059
                                                     Fax: (352) 392-0301
                                                 jason.karceski@cba.ufl.edu



Since the early 1990s, a number of mutual funds have emerged that cater exclusively to
institutional investors, i.e. pension funds, for fee financial advisors, trusts and corporate benefit
plans. Although institutional funds might be a natural place to look for “smart money”, agency
costs associated with delegated monitoring may lead to less monitoring of investment
performance and lower overall performance. To test this hypothesis, we split institutional funds
based on proxies for the degree of investor oversight, and we find substantial cross-sectional
variation in performance among institutional funds. Specifically, we find that institutional funds
with low initial investment requirements and funds with retail mates perform worse than other
institutional funds both before and after adjusting for risk and expenses. Finally, cash flows into
institutional funds with high minimum investment requirements are significantly more sensitive
to risk-adjusted measures of performance than are flows into small institutional funds or retail
funds.


JEL classifications: G21, G24




*
  Corresponding author. This is a substantially revised version of an earlier paper entitled, “Captured Money? Differences in the Performance
Characteristics of Retail and Institutional Mutual Funds” presented at the 2003 AFA meetings. We are grateful to two anonymous referees,
Pierluigi Balduzzi, Zoran Ivkovic, Woodrow Johnson, Jay Ritter, and seminar participants at the 2003 AFA meetings, the University of Florida,
and the University of Oregon for helpful comments.
I. Introduction

        Studies of portfolio choice and investor behavior typically distinguish between individual
investors and large institutional investors. For example, studies of mutual fund selection often
assume that individual investors face significant search and information costs. This may explain
why individual investors select funds on the basis of past performance even though historical
results may not accurately predict future fund performance.1 On the other hand, institutional
investors and large private investors are generally assumed to be better informed than smaller
“retail” investors, reflecting economies of scale in information production and their continued
presence in the market place. Lower search costs of institutional investors should lead to
different and presumably more sophisticated investment selection criteria. However, as
Lakonishok, Shleifer and Vishny (1992) point out, investment decisions by some institutional
investors (pension funds, foundations, fiduciaries and corporate investors) are influenced by
several layers of agency conflicts. Specifically, pension fund sponsors, corporate treasurers and
trustees may delegate money management to outside managers in order to avoid responsibility
for poor performance. This can lead to the selection of money managers based on prior
performance—similar to the way retail customers appear to select mutual funds.2
        Since the early 1990s, a number of mutual funds have emerged that cater exclusively to
institutional investors. Morningstar classifies mutual funds as retail or institutional, and defines
institutional funds as those with minimum initial investment requirements of at least $100,000 or
funds that designate themselves as institutional. Given the large initial investment required
(often greater than $500,000 or $1 million), institutional mutual funds compete with other
institutional money managers including limited partnerships, hedge funds and direct money
managers.
        Using Morningstar’s classification, we find that the size of the institutional segment of
the mutual fund market has grown dramatically in recent years, both in terms of the number of
funds and assets under management. For example, in 1986 there were only 22 open-end bond
and equity funds classified as institutional. By the end of 1998, the number of institutional funds

1
  For example, see Sirri and Tufano (1998), Chevalier and Ellison (1997), and Goetzmann and Peles (1997) on the
relationship between fund inflows and past performance. Examples of studies on the persistence in fund returns
include Carhart (1997), Brown and Goetzmann (1995), Grinblatt and Titman (1992), and Gruber (1996).
2
  For a discussion of how principal-agent problems can trigger herding behavior, including chasing “hot” stocks or
mutual funds, see Devenow and Welch (1996).


                                                         1
grew to 873 (a 40-fold increase).3 Assets under management increased from $3.2 billion in 1986
to over $302 billion by the end of 1998.4
         Information on the performance and flows into institutional mutual funds provides a
unique opportunity to examine the factors influencing the investment decisions of institutional
investors. For example, do institutional funds display superior performance relative to retail
funds after controlling for differences in expenses and risk? Does fund performance vary with
the degree of investor oversight? Is the institutional mutual fund industry best thought of as
somewhat monolithic where all funds compete for the same types of customers, or alternatively,
are there important differences among institutional funds where different types of funds target
different clienteles?
         In this paper, we examine these questions by comparing the performance of retail mutual
funds to mutual funds that cater exclusively to institutional investors and by examining cross-
sectional differences in the performance of institutional funds. We find that, despite significantly
lower management expenses, the average return on institutional funds is no better than the
average return on retail funds. Even on a risk-adjusted basis, institutional funds performance is
similar to retail funds.
         We also explore differences in the performance characteristics among institutional funds.
Specifically, if agency conflicts between institutional investors and their clients affect fund
performance, we would expect performance among institutional funds to vary with the degree of
investor oversight or monitoring. While investor oversight is not directly observable, we
consider three potential proxies—the minimum initial investment requirement, whether the fund
is affiliated with a bank, and whether the institutional fund is offered by a fund complex that also
offers the identical fund to retail investors. We argue that investor oversight tends to be lower
for funds with low minimum initial investment requirements and for funds that have an identical
retail fund.
         A review of offering prospectuses for funds in our sample reveals that mutual funds with
large minimum initial investment requirements, for example, over $500,000 (what we call “big
institutional” funds) are typically offered to large pension funds, foundations, corporate accounts

3
  In contrast, the number of retail open-end funds increased by about 650% (from 786 to 5,076).
4
  The assets of open-end funds that are designated as institutional significantly understate the importance of
institutional investors in the mutual fund market. The Investment Company Institute estimates that institutions
(fiduciaries, business organizations, unions and non profit organizations) held almost two trillion dollars in equity,
bond and hybrid open-end mutual funds at year-end 1999 (out of a total of $5.2 trillion in these funds).


                                                           2
and high net worth individuals. These funds may attract a more sophisticated clientele (or a
clientele less subject to agency conflicts) than funds with relatively low initial minimum
investments requirements (what we call “small institutional” funds). Moreover, given the size of
each account in big institutional funds, one would expect greater monitoring on the part of the
beneficiaries of these investments.
         In contrast, institutional funds with low minimum denominations are most often offered
to trust, custodial accounts, insurance agents, or through “for-fee” financial advisors. In essence,
these funds are offered to intermediaries who in turn offer them to their retail clients. For
example, the Neuberger & Berman fund family offers a number of institutional funds to life
insurance companies to serve as investment vehicles under their variable annuity and variable
life insurance contracts which themselves charge management fees.
          We refer to institutional funds associated with a retail fund with the identical target
portfolios as “institutional funds with retail mates”. While the institutional and retail funds hold
the same portfolio, the institutional and retail shares are claims on separate asset pools or trusts.
This structure is used because Section 18f-3 of the Investment Company Act of 1940 (17 Code
of Federal Regulations 270) requires different classes of shares of the same fund to have the same
return before distribution expenses.5 Institutional funds with retail mates may draw investors
attracted to the brand name of the fund or fund complex. This clientele may be either less
sophisticated or more subject to agency conflicts.6 Thus, small institutional funds with retail
mates are likely to be funds where agency conflicts are particularly acute.
          If monitoring affects performance, we would expect big institutional funds and funds
without retail mates to outperform small institutional funds and funds with retail mates. We also
expect flows into and out of small institutional funds and funds with mates to be significantly
less sensitive to performance than flows into and out of big institutional funds and funds without
mates.


5
  While institutional funds and retail funds are frequently referred to as different share classes, technically they are
claims on different trusts. The reason for this structure is because institutional funds require different services from
the fund manager. For example, Fidelity Advisor shares are offered through 403b plans and investment advisors.
Since the plan sponsor provides bookkeeping services and transacts with Fidelity through an omnibus account,
management fees are lower for these plans than Fidelity funds purchased through the direct or retail channel. The
institutional fund and the retail mate file separate prospectuses.
6
  Nanda, Wang and Zheng (2004) find evidence of spillover effects from one “star” fund to other funds in the same
family and suggest that fund families may seek to create this kind of brand name recognition by cultivating star fund
performers.


                                                           3
         Overall, we find that institutional funds with larger minimum investment requirements
and those without retail mates significantly outperform other institutional funds, and this
difference in returns cannot be explained by differences in fees. For example, the average return
on big institutional funds exceeds the return on small institutional funds with mates by more than
200 basis points per year. However, only 17 basis points of this return difference is due to lower
expenses. The differences in returns persist after controlling for differences in investment styles
and risk characteristics. This finding is consistent with the argument that greater investor
monitoring improves performance. Of course, another possibility is that we have failed to
completely account for differences in risk or investment objectives. While we cannot rule out
this possibility, it is unclear why risk differences would be systematically related to the minimum
initial investment funds require.7
          We also find that, although institutional fund flows are significantly less sensitive to
past fund raw returns than retail fund flows, institutional fund flows are more sensitive to risk-
adjusted measures of performance than retail fund flows. Moreover, this heightened risk-
adjusted flow-performance sensitivity for institutional funds results entirely from the higher
sensitivity of big institutional funds. The sensitivity of cash flows to risk-adjusted returns is
more than twice as high for big institutional funds as it is for small institutional or retail funds.
         While the poor performance of small institutional funds coupled with the superior
performance of big institutional funds is consistent with the hypothesis that increased monitoring
improves performance, it is puzzling why small institutional funds with retail mates perform so
poorly. In particular, if managers of these funds are exploiting a complacent clientele of
investors, then one would expect the poorer performance of these funds to be explained solely by
higher management expenses. However, higher reported expenses explain only a fraction of the
lower returns earned by small institutional funds.
         One potential explanation is that institutional funds with retail mates are forced to engage
in more liquidity-based trading which, as Edelen (1999) explains, can adversely affect
performance. Alternatively, the poorer relative performance of small institutional funds with
retail mates may arise from higher undisclosed costs or other activities that adversely affect
performance. For example, less monitoring may result in smaller funds being exposed to more so


7
 Chalmers, Edelen and Kadlec (2001) find that transactions costs not included in the expense ratio vary
considerably across funds. In addition, they find that transactions costs are negatively correlated with performance.


                                                          4
called market timing trades or late trading (see Zitzewitz (2003)). Another possibility is that
small institutional fund performance is adversely affected because of higher brokerage costs (or
poorer trade execution) for these funds relative to big institutional funds. Brokerage costs are
capitalized into the purchase price of the securities and are not reported in fund expense ratios.
Higher brokerage costs arising from so called soft dollar brokerage may be particularly important
for institutional funds with retail mates. Soft dollar brokerage refers to the bundling of research
costs and in some cases the cost of distribution into a single commission for the brokerage
client.8 While commission rates and execution costs are higher than if a discount broker were
used, whether fund investors are adversely effected by this practice depends on whether the
value of the services received by investors through rebates exceed the higher commissions paid.9
         Unfortunately, it is difficult to empirically test these explanations because neither market
timing agreements nor soft money arrangements are reported in the fund’s prospectus. Since the
funds in our sample are all domestic equity funds and performance differences do not vary by
fund objective, market timing seems an unlikely explanation for the differences in performance.
Soft dollar arrangements are also not directly observable. However, the net cost or benefit of
these arrangements can be measured for the sub-sample of institutional and retail funds that have
mates. Specifically, how managers allocate net costs of benefits of these arrangements between
the institutional fund and the retail mate will affect the performance of the institutional fund
relative to its retail mate. Since the two funds hold identical portfolios and have identical


8
  Soft dollar brokerage is not necessarily illegal. Concern that investment managers might be held liable for breach
of fiduciary duty if they failed to obtain the lowest possible commission (regardless of execution quality and the
value of research received) lead to the passage of Section 28(e) of the Securities and Exchange Act. Section 28(e),
which provides a safe harbor against lawsuits when managers pay higher soft dollar commission if the commission
paid is “reasonable in relation to the value of brokerage and research services provided” (15 U.S.C. SS78bb(e)
(1998)). See Johnsen (2000) for a discussion of the law and economics of soft dollar brokerage. The current debate
over soft dollar brokerage focuses on the lack of disclosure concerning brokerage commissions and the use of
brokerage commissions to pay securities firms to sell, distribute or promote funds. For a discussion, see Lauricella
and Oster (2003, 2004). The Investment Company Institute (ICI) recently proposed revising Section 28(e) to exclude
from the safe harbor payments certain types of computer hardware and software, publications available to the
general pocket and third party research services (see ICI (2003)).
   Several funds in our sample discuss soft dollar distributions in their prospectuses. For example, Mass Mutual’s
                               s
prospectus state, “The Trust' Distributor receives distribution fees and may allow all or a portion of them as dealer
discounts and brokerage commissions to dealers, including MML Investors Services, Inc. ("MMLISI"), a wholly
owned subsidiary of Mass Mutual. From time to time, dealers who receive dealer discounts and brokerage
                               s
commissions from the Trust' Distributor may allow all or a portion of them to other dealers or brokers. The service
fees will be paid to Mass Mutual.”
9
  Conrad, Johnson and Wahal (2001) find that the incremental costs of soft brokerage to between 25 and 29 basis
points. While these costs are substantial, they are unable to determine whether the higher costs are offset by rebates
or the value of research provided.


                                                          5
turnover rates, once portfolio returns have been grossed up for expenses, the only source of
performance differences will be the allocation of “soft money” costs and benefits. We find a
small but statistically significant difference (10 basis points annually) between the expense-
adjusted return for institutional funds compared to their retail mates. More important, the
difference in performance is significantly greater for institutional funds with high initial
minimum investment requirements and low portfolio turnover rates. These results are consistent
with at least a portion of the poorer relative performance of small institutional funds with mates
being due to higher soft money expenses.
         Our results concerning flow performance differences between retail and institutional
funds are consistent with those of Del Guercio and Tkac (2002). They find that investors use
different selection criteria to select managers in the retail mutual fund and pension fund segments
of the market. Specifically, they show that pension fund clients rely more heavily on
sophisticated performance criteria such as Jensen’s alpha than do investors in retail mutual funds.
On the contrary, retail investors focus more on past returns and fund rankings. Our results
support the notion that performance criteria differ across retail and institutional investors. In
addition, our results suggest that differences in agency problems among institutional investors
influence performance and flow performance relationships.10
         The remainder of the paper is organized as follows. Section II describes the data used in
our study and the market for institutional mutual funds. Section III examines the flow
performance characteristics of institutional and retail mutual funds. We also examine differences
in expenses, returns, and performance persistence for retail and institutional funds. We conclude
with a brief discussion of the implications of our findings for the market for institutional money.


II. Data and Methodology

A. Data
         Data on open-end equity funds are collected from the CRSP mutual fund database. To
distinguish between institutional and retail mutual funds, we match the CRSP mutual fund


10
  One explanation for the differences in selection criteria is that pension fund sponsors contract with pension fund
managers for a fixed period of time (generally three years). There is no such contractual constraint for mutual fund
investors. By comparing the selection criteria of institutional and retail investors for open-end mutual funds, we
avoid this potential problem.


                                                          6
database with the November 1996 edition of the Morningstar Principia database. Funds are
categorized as institutional if Morningstar designates them as such or if Morningstar reports that
the minimum initial purchase requirement is $100,000 or more. Our sample includes the four
main Morningstar objective categories for equity mutual funds: aggressive growth, growth,
growth and income, and equity income.
         Some of the institutional funds in our data set have a no-load retail counterpart or “mate”
while others do not. A retail mate is an equity fund with the same name and fund advisor as the
institutional fund, but with a different share class. We verify that these two funds hold the same
equity portfolio and have identical fractional cash balances. In all but a few cases, the
institutional fund and its retail mate also have identical turnover rates. Since these funds hold
identical portfolios, differences in their returns can only be due to differences in expenses and
brokerage commissions paid.
         Lipper Analytical Services identifies bank-sponsored mutual funds. Specifically, Lipper
provided us a list of commercial bank proprietary open-end mutual funds as of year-end 1999.
Lipper defines bank proprietary as “those banks with their own fund families and for which the
majority of sales are attributable to bank distribution.”11 All other data items besides
institutional designation, the retail mate identifier, bank proprietary fund designation, minimum
initial purchase requirement, and investment objective are from CRSP.


B. Description of the Institutional Mutual Fund Market

         Table 1 provides the number of funds and total net assets for retail and institutional funds
in our sample. Using Morningstar’s institutional fund classification and information on
minimum initial investment requirements, we identify 977 institutional mutual funds in 1996
with a full year of returns data available on the CRSP mutual fund database. Of these, 23 percent
by number and 52 percent by net assets are in the four main Morningstar investment categories
for equity funds.12 We follow the funds identified in Morningstar forward in time through 2001
and as a result, the sample from 1996 onward is free of survivorship bias. We also follow


11
   See the Cerulli-Lipper Analytical Report: The State of the Bank Securities Industry, 1996 for a description of the
bank proprietary fund market. Lipper excludes Mellon Bank’s Dreyfus Group from the list of proprietary funds
because the majority of sales are not through commercial banks.
12
   There is a similar concentration of retail funds in these categories of equity funds (21 percent by number and 42
percent by net asset value).


                                                          7
institutional funds back to 1990, but for this time period there is the potential for survivorship
bias.13 There are relatively few institutional funds in our sample before the mid-1990s. Because
of this, we focus on performance and flow performance characteristics of funds from 1995
onward (although our results are similar when we use data beginning in 1990).
        Mutual funds do not publish information on the specific identity of their shareholders. As
a proxy for the client base of institutional funds, we use the fund’s minimum initial investment
requirement. Specifically, we classify funds as big institutional funds if they require an initial
investment of at least $500,000. Institutional funds with initial investment requirements below
$500,000 are classified as small institutional funds. Big institutional funds may cater to a
different clientele than small institutional funds. Given the size of the initial investment,
investors in big institutional funds may have a greater incentive to monitor the performance of
their investment.
        To examine whether big institutional funds target a different type of investor than small
institutional funds, we read the offering prospectuses for all of the big institutional funds and 75
randomly selected small institutional funds in our sample. In particular, mutual fund
prospectuses include a section describing how to purchase shares that often contains a discussion
of investors the fund is targeting as well as restrictions on share purchases. Details provided in
the offering prospectus concerning each fund’s clients vary widely. Some funds simply state that
shares are offered to institutional investors while other prospectuses provide a comprehensive list
of the types of institutional investors that may invest in the fund (e.g. pension plans,
endowments, foundations, trust accounts, etc.). Other institutional funds are offered only to
fiduciary or custodial account holders of a particular financial institution (most frequently
commercial bank trust clients). Some institutional funds are limited to members of particular
professional associations (e.g., the AHA Diversified fund is open only to members of the
American Hospital Association). Appendix A provides examples of the investment restrictions
for some of the funds in our sample.
        Our review of offering prospectuses suggests that big institutional funds cater more
frequently to endowments, foundations and corporate pension and benefit plans than small
institutional funds. The targeted clients of small institutional funds are more frequently fee-


13
  See Elton, Gruber and Blake (2001) for a discussion of survivorship-bias problems with the CRSP mutual fund
database.


                                                       8
based financial planners and trust accounts. Clients of institutional funds also appear to differ by
whether or not the institutional fund has a retail mate (i.e. the institutional fund is simply a
different class shares on a portfolio that is also offered to retail account holders).14 Funds with
retail mates typically state that their clients consist of trusts and for-fee financial advisors and do
not mention corporate benefit plans, endowments or foundations as part of their clientele.15 This
suggests that clients of big institutional funds without retail mates differ most from the clients of
small institutional funds with retail mates.
         Table 2 reports the fraction of institutional funds separated according to minimum initial
investment, whether the fund has a retail mate, and whether the fund is distributed by a
commercial bank (“bank-sponsored” fund). As shown in Table 2, about 25 percent of
institutional funds in our sample are big institutional funds, and 51 percent of institutional funds
have retail mates. Not surprisingly, bank sponsored institutional funds are more likely to have a
retail mate (71 percent) than non-bank sponsored funds (35 percent), reflecting the fact that bank
institutional funds are frequently marketed to trust and custodial accounts.


C. Definition of Flows, Performance and Fees

         To remain consistent with prior research, we follow Sirri and Tufano’s (1998) variable

definitions of flows, relative performance, and fees. Annual net new cash flow into fund i during

year t is calculated as

                                                   TNAi , t − (1 + Ri ,t ) * TNAi ,t −1
                                     FLOWi , t =                                          ,
                                                                 TNAi , t −1

where TNAi,t is fund i’s ending total net assets for year t, and Ri,t is fund i’s annual rate of return

during year t.

         We are also interested in how institutional funds perform relative to their retail
counterparts. Performance differences may be explained by institutional funds following
different, perhaps less risky, investment strategies. To investigate this issue, we measure risk-

14
   For example, Fidelity offers an institutional shares for many of its funds. Each institutional fund files a separate
prospectus.
15
   A notable exception is Vanguard’s Institutional Plus Shares. These funds have minimum initial investment
requirements of $200 million and are offered exclusively to benefit plans.


                                                             9
adjusted performance using a 5-factor model similar to the one used in Carhart (1997) with an
additional international equities factor. The market factor (VWRFt) is proxied by the CRSP
value-weighted return of all NYSE, Amex, and NASDAQ stocks in excess of the one-month T-
bill return. Carhart’s 4-factor model uses the 3-factor model of Fama and French (1993) plus an
additional factor to capture the momentum effect of Jegadeesh and Titman (1993). Our 5-factor
model includes the excess return on a value-weighted market proxy (RMRFt ), the return on the
EAFE index (EAFEt), and factor mimicking portfolios for size (SMBt), book to market (HMLt)
and one-year momentum in stock returns (UMDt).16 Factor realizations are retrieved from Ken
French via his website, and EAFE returns are retrieved from Morgan Stanley Capital
International Inc’s website at www.msci.com.
        To examine the performance of retail and various types of institutional funds, we form
equally-weighted portfolios based on fund type—retail funds, no-load retail funds, institutional
funds, big institutional funds and institutional funds with retail mates. Next, we estimate 5-factor
models of stock returns for each portfolio of funds. The intercept measures risk- or factor-
adjusted performance.
        The definition of fees amortizes a fund’s total load over a seven-year holding period and
is given by

                               FEESi , t = EXPENSESi ,t + 1 * TOTLOADi , t ,
                                                           7
where EXPENSESi,t is the expense ratio of fund i for year t and TOTLOADi,t is the total load
(both front-end and back-end loads) for the fund in year t.
        For institutional funds with retail mates, we compute the annual difference in returns
between the institutional fund and its retail mate and add to this the difference between the
expense ratio of the retail and institutional fund. This provides us with a return difference
grossed up for expenses. When making institutional fund to retail mate comparisons, we exclude
the few funds that do not have identical portfolio turnover rates.17



16
   EAFE (Europe, Australasia, Far East) is a popular benchmark for international equity money managers that does
not include U.S. stocks. We include this index to capture different international equity exposures across mutual
funds.
17
   Institutional fund turnover did not match the retail fund turnover in 44 four out of 880 fund-year matches. The
mismatches do not appear to be systematic. For example, 21 of the mismatches occurred with institutional funds
having higher turnover rates than their retail mates. Moreover, mismatches were not concentrated in a few fund
families. Our results are similar when we include these observations in our sample.


                                                        10
D. Summary Statistics
         Table 3 contains summary statistics for our sample of mutual funds at year-end 1995 and
2001. Both the number and the average size of institutional funds grew at a faster pace than
retail funds, with the number of institutional funds increasing 57 percent (from 138 to 216 funds)
and the average size of the fund increasing 81 percent (from $330 million to $596 million). The
number of retail funds increased 36 percent and the average size of a retail fund increased 72
percent from $834 million to about $1.4 billion. The differences in growth arise in part from less
shrinkage in the average size of institutional funds from 1999 to 2001.18
         As shown in Table 4, the average institutional fund is significantly smaller than the
average retail fund. Pooling across years 1995 to 2001, the average institutional fund had $547
million under management versus $1.292 billion for the average retail fund. However, the larger
average size for retail funds is the result of a few very large retail funds. For example, the largest
five retail funds had net assets ranging from $54 to $84 billion. In contrast, the largest
institutional fund in our sample had assets of only $22 billion. The median asset size of
institutional funds is slightly larger than the median asset size of retail funds, though the
difference is not statistically significant.
         Not surprisingly, institutional funds have significantly lower expenses than retail funds.
As shown in Table 4, the average (median) annual expense ratio for institutional funds is 0.86
(0.91) percent, while the average (median) annual expense ratio for retail funds is 46 (20) basis
points higher. However, despite lower expenses, the average return for institutional funds (14.44
percent) is about the same as the average return for retail funds (14.22 percent). While the
similarity in returns is suggestive of poorer performance for the underlying portfolios of
institutional funds, performance differences may be explained by differences in risk or
investment objectives of institutional funds. We explore these issues in detail below.
         The results in Table 4 reveal economically significant differences across disparate types
of institutional funds in expenses and performance. For example, small institutional funds and
institutional funds with retail mates have higher expenses and lower average returns than big
institutional funds or institutional funds without mates. Average expenses for big institutional

18
   The Wall Street Journal reported that one reason bank and institutional funds have been able to maintain their size
in the bear market since 2000 is because they are sold through “external platforms” such as broker dealers and
brokerage houses, which are less likely to attract hot money (WSJ, 12/08/02, p. D7). We note that these funds are
the focus of the controversy over using soft dollars to pay for fund distribution (see Lauricella and Oster (2004)).



                                                         11
funds are about 15 basis points lower than for small institutional funds and institutional funds
retail mates. Moreover, the average return differences between big institutional funds and other
institutional funds are larger than the differences in expenses. Performance differences are even
larger between big institutional funds without retail mates and small institutional funds with
mates. Big institutional funds without mates earned on average 238 basis points more per year
than did small institutional funds without mates. The differences are similar when performance
is measured relative to funds with same Morningstar investment objective. In contrast, small
institutional funds with retail mates underperform funds with the same investment objective.
Thus, both the level of the minimum initial purchase requirement and the presence of a retail
mate are useful in explaining cross-sectional patterns in expenses and performance among
institutional funds.
       One potential explanation for the superior performance of big institutional funds is
perhaps they have lower portfolio turnover resulting in lower transactions costs. However, Table
4 shows that turnover is higher for big institutional funds than it is for small institutional funds.
Indeed, small institutional funds have lower average turnover than retail funds. While turnover
may be a noisy measure of transactions costs (see Wermers (2000)), these results do not suggest
that the return differences result from a greater propensity by small institutional funds to trade.


III. Empirical Results

A. Performance Differences
       Although the average return of institutional funds and retail funds are similar despite
substantial differences in expense ratios, this may be explained by institutional funds following
less risky investment strategies. To investigate this issue, we estimate risk-adjusted performance
using a 5-factor model.
       Table 5 provides excess returns and estimates of the 5-factor model of stock returns for
each portfolio of mutual funds. On average, excess returns and risk-adjusted returns are slightly
higher for institutional funds than for retail funds but the differences are close to the differences
in expenses. For example, the average excess return and 5-factor alpha for institutional funds are
5 and 4.1 basis points higher per month than the averages for retail funds, while the expense
difference averages about 4 basis points per month.




                                                  12
          The results in Table 5 reveal differences in performance among different types of
institutional funds as well. Big institutional funds without mates earned an average monthly
excess return of 0.929 percent while institutional funds with mates had the lowest average excess
return of 0.751 percent. The annual difference in returns between these two groups of funds is
213 basis points, which is well in excess of the difference in average annual expenses
(approximately 16 basis points). The difference in average returns is not the result of one year of
particularly good performance by big institutional funds or a single year of unusually poor
performance by institutional funds with mates—the average return for big institutional funds
exceeds the return on affiliated institutional funds by more than the expense ratio each year in
our sample.
         The 5-factor models do not explain the superior performance of big institutional funds
relative to affiliated institutional funds or retail funds. The alphas from these models are larger
for big institutional funds than for retail funds or small institutional funds with retail mates. The
differences in alphas are not trivial. For instance, the difference in the 5-factor alphas between
big institutional funds without retail mates and small institutional funds with retail mates is over
13 basis points per month or about 156 basis points per year (this difference is statistically
significant at the 10 percent level). Because small institutional funds are typically offered to
trusts and through for-fee investment advisors who assess additional fees, the relative
performance of these funds to their ultimate investors is even worse than reported in Table 5.19
         To test whether the performance differences among funds is statistically significant, we
regress the estimated alphas for individual funds on a set of variables representing our various
fund categories and fund characteristics including the minimum initial investment requirement
and whether the fund has a retail (or institutional) mate. Alphas are estimated via the 5-factor
model using all monthly returns available from the CRSP mutual fund database from 1995 to
2001, and we require a fund to have at least 24 months of returns to be included in the sample.
         Panel A of Table 6 provides regression results comparing the performance of institutional
and retail funds. To highlight differences in performance arising from fund manager stock



19
  In this analysis, we have controlled for style differences across fund groupings. According to the 5-factor
coefficient estimates on SMB and HML, institutional funds are slightly tilted towards large company stocks and
growth stocks than are retail funds. Within the institutional fund group, small institutional funds appear to be tilted
more toward value and momentum stocks than big institutional funds.


                                                          13
picking ability, we add back monthly expenses (computed as 1/12 of the most recent annual
expense ratio) to individual fund alphas.20
         As shown in Column 1 of Table 6, the risk-adjusted performance of institutional funds is
statistically indistinguishable from the performance of retail funds. Column 2 reveals a positive
and significant (at the 10 percent level) relation between fund performance and the size of the
initial investment requirement. This finding is consistent with increased investor monitoring of
fund performance for funds with larger initial investment requirements. To estimate economic
significance, the 90th (10th) percentile for minimum initial investment requirement for
institutional funds is $2.25 million ($1,000), so the coefficient of 0.012 on LOG (MINIMUM
INITIAL PURCHASE) in Column 3 means that the performance difference between these two
funds is about 111 basis points per year. From the same regression model in Table 6, mutual
funds with mates underperform funds offered on a stand-alone basis by 154 basis points per year.
This result is consistent with less monitoring of mutual funds with brand name capital.
         Results (unreported) are similar for our two proxies for investor oversight when we
include only retail funds in the regression models. For retail funds only, funds with institutional
mates (which comprises 10 percent of the sample) underperform those without mates by 158
basis points per year. For retail funds, the 90th (10th) percentile for minimum initial investment
requirement is $5,000 ($500), and the coefficient on LOG MINIMUM INITIAL PURCHASE is
0.0157 when only retail funds are included in the model. Consequently, the 10th percentile retail
fund underperforms the 90th percentile retail fund by about 43 basis points per year.
         Panel B of Table 6 compares the performance of various types of institutional funds.
Institutional funds with mates underperform other institutional funds by an average of about 167
basis points per year as shown in Column 2. The average annual risk-adjusted return for big
institutional funds without mates (shown in Column 4) exceeds the return on other institutional
funds by about 163 basis points per year.21 Finally as shown in Column 5, alphas are increasing
in the minimum required investment.


20
   To determine whether certain fund managers show superior ability relative to other fund managers, other papers in
the literature have added back expenses to standard multi-factor alphas. For example, Elton, Gruber and Blake
(2003) utilize expense-adjusted alphas to determine that funds with performance-based incentive fees employ more
skillful managers than funds without an incentive fee structure.
21
   We also explored whether risk differences not accounted for by our 5-factor models explain performance
differences. Specifically, we examined whether performance among institutional funds varies due to differences in
cash holdings. We find no significant difference among institutional funds in terms of the proportion of cash to total


                                                         14
         Overall, the results in Table 6 indicate that big institutional funds outperform small
institutional funds and that institutional funds with retail mates are especially poor performers
relative to retail funds and other institutional funds.22 Moreover, while higher expenses explain
some of the performance differences, the results in Table 6 indicate that the poor performance of
small institutional funds (and the superior performance of big institutional funds without mates)
arises from differences in performance in the stocks these funds invest in. Assuming the 5-factor
model is well-specified, the poor relative performance of small institutional funds suggests that
managers of these funds are either consistently selecting underperforming stocks or engaging in
value destruction through excessive transactions costs.23
         To investigate further whether high transactions costs through soft dollar brokerage
explains why institutional funds with mates perform so poorly, we examine performance of
institutional funds with mates relative to the performance of their retail counterpart. Since the
institutional fund and its retail mate by design hold identical portfolios and almost always have
identical portfolio turnover, the only reason their returns (grossed up for expenses) will be
different is because of differences in brokerage commissions paid net of any benefits from
revenue sharing arrangements. Note that even if fund managers allocate brokerage costs on a
prorated basis between the retail and institutional fund, unless the benefits are also allocated on a
prorated basis, performance differences will arise. For example, if retail distribution costs are
reimbursed through higher brokerage fees, and brokerage costs are allocated on a prorated basis,
then institutional clients will subsidize the distributional costs associated with retail clients. If
monitoring of fund performance is related to the size of the minimum initial investment required,


assets in their portfolios. In addition, when each fund’s ratio of average cash holdings to total assets is included in
Table 6’s cross sectional regressions, the size and significance of coefficient estimates are virtually unchanged.
   In addition to cash holdings, approximately 14 percent of our equity funds report holding some amount of fixed
income securities in their portfolios. To ensure that our results are not driven by bond market exposures, we
replicated Tables 5, 6, and 9 including only mutual funds that reported exactly zero bond holdings. None of our
main results were affected by making this change.
22
   Of course, there is always the possibility that the 5-factor model is misspecified. If so, one concern is that
performance differences are driven by differences in investment objectives across different groups of funds. For
example, perhaps institutional funds with mates are mostly growth and income funds while institutional funds
without mates are growth funds. To address this concern, we ran all of the regressions in Table 6 using only growth
and income funds, and also separately for only growth funds (we also ran them separately for aggressive growth
funds and for equity income funds, but in these cases the number of observations is small). After controlling for
fund objective in this way, the statistical significance of the coefficients on the mate variable and the log of the
minimum initial purchase requirement are very similar to the significance levels for the overall sample reported in
Table 6.
23
   Since turnover is higher for big institutional funds, transactions costs can explain the difference in performance
only if, as Chalmers, Edelen and Kadlec (2001) contend, per trade costs vary substantially across funds.


                                                          15
   we expect more net benefits (fewer net costs) will be allocated to funds with higher initial
   investment requirements.
           We measure the performance differences between institutional funds and their retail
   mates by taking the annual difference in returns and adding the difference in the expense ratio of
   between the retail fund and the institutional mate. This provides a measure of the difference in
   portfolio returns adjusted for difference in reported expenses. Our sample contains 158
   institutional and retail pairs and 880 fund-year observations over the 1995 though 2001 period.
           On average, institutional funds outperformed their retail mates by 11 basis points per
   year. The median difference is 4 basis points. While the mean difference is small, it is
   statistically significant at the one percent level (the t statistic is 5.97). Institutional funds
   outperformed their retail mate in 75 percent of fund years. Performance differences increase with
   the size of the initial purchase requirement. For example, approximately 75 percent of the
   institutional funds with retail mates had minimum purchase requirements in excess of $100,000.
   For these funds, the average performance difference was 13 basis points per year. In contrast, for
   institutional funds with no minimum investment requirement, the performance difference was 4
   basis points per year. The difference in performance is significant at the one percent level (the t
   statistic is 5.25).
           To examine whether the performance difference varies with the size of the initial
   investment and portfolio turnover (a proxy for the importance of soft dollar brokerage), we
   estimate the following regression:


PERF DIFFERENCE = 0.048 + 0.0081 * LOG(MINIMUM PURCHASE) + 0.00634 * TURNOVER
                  (1.36) (1.96)                            (3.79)

                    R2 (between) = 0.18              F = 16.97
   where PERF DIFFERENCE is the percentage difference in the annual return plus expenses

   between the institutional fund and its retail mate, LOG(MINIMUM PURCHASE) is the natural

   log of the minimum initial purchase requirement for the institutional fund, and TURNOVER is

   the annual turnover rate for both funds. T-statistics are reported in parentheses below the

   coefficient estimates. Since performance differences for a particular fund may persist over time,




                                                       16
we estimate the regression using a between effects model which groups all observations for each

fund pair over time.24

           Consistent with the hypothesis that higher minimum purchase requirements result in
more intensive monitoring, the coefficient on LOG(MINIMUM PURCHASE) is positive and
statistically significant at the five percent level. The positive and statistically significant
coefficient on TURNOVER is consistent with argument that the return difference arises from
soft dollar brokerage allocations, and suggests that the higher the minimum initial purchase
requirement, the more soft dollars are allocated to the retail fund, not the institutional fund.

B. Flow Performance Relationships
           If the clientele of small institutional funds do not monitor fund performance closely, then
we expect the poor relative performance of these funds to be accompanied by cash inflows and
outflows that are less sensitive to fund performance than retail or big institutional funds. We
investigate this issue by examining whether institutional investors select funds using the same
criteria that retail customers use and whether, for a given performance criteria, retail fund cash
flows are more sensitive to performance than flows into institutional funds. We address these
issues in two ways. First, following Sirri and Turfano (1998), we examine the cross-sectional
relationship between fund flows and the fund’s performance rank at the end of the prior year.
Second, we examine the pooled cross-section time series relationship between flows and various
returns measures.
           Each fund’s relative annual performance is calculated by comparing the return of the

fund against all other funds with the same investment objective. A fund’s fractional rank

(RANKi,t) ranges from 0 to 1 and represents its percentile performance relative to other funds

with the same investment objective in year t. To examine the cross-sectional flow performance

relationship using a piecewise linear regression, we include the following five quintile variables

from worst to best performance:




24
     We obtain similar results if we use OLS and assume independence over time.


                                                         17
PERF1i ,t = min[ RANK i ,t ,0.2],
PERF 2i ,t = min[ RANK i ,t − PERF1i ,t ,0.2],
PERF 3i ,t = min[ RANK i , t − PERF 2i , t − PERF1i ,t ,0.2],
PERF 4i ,t = min[ RANK i ,t − PERF 3i , t − PERF 2i , t − PERF1i ,t ,0.2],
PERF 5i, t = min[ RANK i , t − PERF 4i , t − PERF 3i ,t − PERF 2i ,t − PERF1i , t ,0.2].

        Table 7 provides regression results relating fund flows to relative performance, return
volatility, and expenses. Following Sirri and Tufano (1998), flow performance regressions are
estimated for each year, and coefficients and standard errors are computed from the annual
coefficients as described in Fama and MacBeth (1973). As a starting point, we estimate the
regression using data from 1990 onward. Column A of Table 7 presents results for all funds in
the sample, and Columns B through D provide estimates for subsamples consisting of retail
funds and institutional funds.
        Consistent with Sirri and Tufano (1998), Column A reports a significant positive
relationship between fund flows and relative performance for the top performance quintile.
While the flow performance relationship is strongest in the top performance quintile, we find a
positive and statistically significant relationship between flows and performance in the lowest
and middle performance quintiles as well.
        The positive flow performance relationship reported in Column A is due entirely to the
positive flow performance relationship among retail funds. In particular, we do not find any
significant relationship between inflows and relative performance for the institutional funds in
our sample. As shown in Column E, for the highest performance quintile, the sensitivity of the
flow performance relationship for institutional funds is statistically lower than for retail funds.
Although we report the Fama-MacBeth t-statistics in Table 7, there is one cross-sectional
regression model for each year and a t-statistic associated with each of the explanatory variables
in each model. In each year from 1990 to 2001, the cross-sectional t-statistic for the rank of the
highest performance quintile (PERF5) is positive and statistically significant for all retail funds
but is never positive and statistically significant for all institutional funds. Overall, these results
suggest that investors in institutional funds do not chase returns in the same way as their retail
counterparts.
        In contrast to Sirri and Tufano (1998), we find no significant relationship between fund
inflows and fees or the variability in returns (as measured by the standard deviation in monthly


                                                     18
returns over the prior year). One explanation for this difference is that we examine the flow
performance relationship beginning in 1990, while Sirri and Tufano’s sample covers the period
from 1970 though 1990. A second and related explanation is that we have only twelve annual
observations, and with so few observations, the Fama-MacBeth methodology provides very
conservative estimates of the standard errors. We investigate this issue next.
         Given the small number of institutional funds in the early 1990s, we estimate the flow
performance relationship using data from 1995 onward. With only seven annual observations,
the Fama-MacBeth technique lacks power in this case. To account for potential dependence in
the annual observations, we estimate the model using Generalized Least Squares (GLS) to
correct for first order autocorrelations in the individual fund flows.25
          Estimates of the flow performance relationship for no-load retail and several categories
of institutional funds are presented in Table 8. Consistent with Table 7, we find a positive and
statistically significant flow performance relationship only for retail funds. On the contrary, we
find no relationship between fund flows and relative performance for institutional funds.
Moreover, the sensitivity of mutual fund inflows to performance does not appear to vary among
different types of institutional funds. For example, big institutional funds and institutional funds
with retail mates have about the same sensitivity to past historical performance in each of the
performance quintile rankings.26 The lack of a flow performance relationship suggests that either
institutional investors use a different (perhaps more sophisticated) performance measure in
selecting funds or that institutional investors do not chase past performance in the same way
retail investors do.
         In Table 8, flows into both retail and institutional funds are significantly related to fees
charged, with more cash flowing into funds with lower fees. While the point estimates indicate
that institutional flows are more sensitive to fees than retail funds, we cannot reject the
hypothesis that the sensitivity to fees is the same for both retail and institutional funds. The
negative coefficient on the standard deviation of lagged monthly returns suggests that investors


25
   This involves first estimating the first order autocorrelation for each fund separately and then including these
estimates in the variance-covariance matrix. Unfortunately, Stata limits the number of cross sectional observations
to 800. Because of this restriction, we estimate the flow performance relationship for institutional and no-load retail
funds separately. The results using OLS are similar, although significance levels are higher in the GLS case.
26
   We find no significant difference in the flow performance relationships for categories of big institutional funds
(i.e., funds with or without mates) or for categories of small institutional funds. However, given that some
performance quintiles have as few as four funds in a given year, some difference tests almost certainly lack power.


                                                          19
in both retail and institutional funds are averse to volatility, with investors in institutional funds
displaying more risk aversion than retail investors.27
         While the results in Table 8 suggest that investors in institutional funds do not select
funds based on relative historical performance, there are several other potential explanations for
the lack of a significant flow performance relationship among institutional funds. One
possibility is that institutional funds are distributed differently than retail funds across the
performance rank quintiles. For example, perhaps there are very few institutional funds with
adequately poor returns to put them into the lowest performance quintile. However, this is not
the case since we find institutional funds are distributed across all performance rankings. For
example, 14 percent of institutional funds are in the bottom performance quintile, with the
remaining 86 percent distributed about equally across the remaining four performance ranks.
The distribution for big institutional funds is only slightly more skewed with 11 percent of the
observations in the lowest performance quintile and 28 percent of the observations in the top
performance quintile.
         A second possible explanation for differences in the flow performance relationship
between retail and institutional funds is that the investment strategies (within each fund category)
or portfolio composition of institutional funds differ from the strategies employed by retail funds.
If this is the case, historical performance may be less informative about future performance for
institutional funds than for retail funds, causing institutional investors to utilize different fund
selection criteria. One way to address this issue is to compare the flow performance relationship
of institutional funds with retail mates to the flow performance of the retail mates themselves.
Since these funds hold identical portfolios, any difference in the flow performance relation
cannot be due to differences in portfolio holdings or attributes of the portfolio manager.
         Estimates of the flow performance relationships for retail and institutional funds with
mates are reported in Columns E and F of Table 8. While the coefficient estimates on fund size,
fees and risk are similar for retail and institutional funds, inflows to top-performing retail funds
are much more sensitive to performance than inflows for their institutional mates. Indeed, for the
top performance quintile, flows into retail funds are more than ten times more sensitive to


27
  The coefficient estimates for fees and standard deviation in returns variables are quite sensitive to the estimation
technique we employ. Using data from 1990 through 2001, we find negative and significant coefficients for these
variables when we estimate the model using OLS or GLS. In contrast as reported in Table 7, coefficient estimates
for these variables are positive, though not statistically significant, when we use the Fama-MacBeth methodology.


                                                          20
performance than flows into institutional funds. Thus, we find no evidence that differences in
portfolio composition can explain the higher flow performance sensitivities of retail funds.
         A third potential explanation for the lack of a flow performance relationship for
institutional funds is that institutional investors also chase returns, but they base their decisions
on different performance criteria than retail investors.28 For example, Del Guercio and Tkac
(2002) argue that pension fund sponsors are more quantitatively sophisticated than retail mutual
fund investors. As a result, pension funds rely more on risk-adjusted measures than on raw
returns when assessing performance. If this argument applies to investors in institutional funds,
we would expect flows into institutional funds to be more sensitive to risk-adjusted measures of
performance and less sensitive to raw returns than flows into retail funds.
         We test for differences in flow performance relationships between retail and institutional
investors by relating flows to lagged fund returns, 1- and 5-factor alphas, and tracking error
volatility (the standard deviation of the return difference between the fund and the market as
proxied by the CRSP value weighted index). We include tracking error volatility based on Del
Guercio and Tkac’s (2002) discovery of a significant relation between performance and tracking
error for pension funds. Our tracking error variable and the 1- and 5-factor alphas are estimated
using monthly returns over the prior 24 months. To account for potential dependence in the
annual observations, we estimate the regression using GLS to correct for first order
autocorrelation in individual fund flows. However, the results are virtually identical when using
OLS with fixed fund effects and a lagged flow variable.




28
   Another explanation is that institutional investors base their investment decisions on performance over a different
time horizon than retail investors, possibly a shorter or a longer horizon. To investigate the possibility of a shorter
horizon, we examine the time series properties of aggregate monthly fund inflows and examine the relationship
between fund inflows and contemporaneous and one month lagged market returns. In particular, Warther (1995)
finds that monthly fund flows follow an AR(3) process and inflows are correlated with contemporaneous and lagged
market performance. Predictability in monthly flows and the correlations between monthly flows and
contemporaneous and lagged returns is consistent with investors in stock mutual funds in the aggregate chasing
returns. We estimated an AR(3) time series models of net new cash flows to mutual funds grouped by retail and
institutional categories. For retail funds, the first and third monthly lags are positive and statistically significant, and
longer lags are insignificant. On the contrary, there is no significant relationship between current and past inflows
for institutional funds. Consequently, one cannot predict future monthly cash flows into institutional funds using
past inflows, while inflows into retail funds have a large predictable component.
     To test whether institutional fund investors chase returns over a longer time horizon than retail investors, we
used the Fama-MacBeth methodology with relative performance ranks based on the prior two or three years instead
of just one year. Although PERF5 becomes significant at the five percent level for all institutional funds using a
two-year horizon, we still find that PERF5 is significantly lower for institutional funds than for retail funds.


                                                            21
        Table 9 presents estimates of the flow performance relationship for mutual funds grouped
by clientele.29 Consistent with the idea that clients of institutional funds employ more
sophisticated performance criteria, institutional fund flows are more sensitive to risk-adjusted
performance as measured by 5-factor alpha than are retail fund flows (although the difference is
statistically significant at only the ten percent level). In contrast to no-load retail fund flows,
institutional fund flows are negatively and significantly related to return volatility in returns (as
measured by the standard deviation in the prior year) and tracking error volatility.
        The differences in the flow performance relationships among different types of
institutional funds are particularly illuminating. Lagged raw returns are associated with higher
cash flows for small institutional funds, just as they are for no-load retail funds. However, there
is much less of a relationship between lagged raw returns and inflows into big institutional funds,
and lagged returns are less important for institutional funds with retail mates. Moreover, flows
into big institutional funds appear quite sensitive to risk-adjusted performance as measured by 5-
factor alpha. This heightened sensitivity of institutional funds to risk-adjusted returns is almost
entirely driven by big institutional funds. Compared to small institutional fund flows, big
institutional fund flows are significantly less sensitive to lagged returns and more sensitive to
risk-adjusted returns than small institutional funds. We find similar differences when comparing
big institutional funds without retail mates to small institutional funds with retail mates (not
shown).
        Overall, our results suggest that clients of big institutional funds employ more
sophisticated performance criteria than do investors in small institutional funds. Moreover, the
poor relative performance of small institutional funds, particularly for those with retail mates,
coupled with the less sensitive flow performance relationship for these funds is consistent with
the capture hypothesis for these funds.


IV. Summary and Conclusions

        A large and growing segment of the mutual fund market is targeted towards institutional
clients. Institutional funds cater to a wide variety of customers, from corporate benefit plans to


29
  Because Stata constraints the number variables for GLS estimates, we test for difference between retail and
institutional funds using OLS with fixed firm effects and lagged fund flows as an explanatory variable.


                                                        22
trust accounts. These funds provide a unique opportunity to compare the performance and fund
selection criteria of institutional investors to retail investors. We find evidence that institutional
investors do not chase returns in the same way that retail investors do. There is no significant
relationship between fund inflows and past relative performance in the institutional segment of
the market, and the flow performance relationship of top performing institutional funds is
statistically different from top performing retail funds. However, our results suggest that at least
for big institutional funds, the lack of a significant flow performance relationship is because
investors in these funds employ more sophisticated performance measures than do retail
investors.
       We examine performance differences between retail and institutional funds and between
different types of institutional funds. Institutional funds earn average returns close to those for
retail funds despite having substantially lower expenses. The lower pre-expense returns for
institutional funds can be explained by unusually poor performance of small institutional funds
and institutional funds with retail mates. Not only do these funds have higher expenses than
other institutional funds, but also their average return is significantly lower than the average
return on other types of institutional funds as well as retail funds. One explanation for the lack
of a significant relationship between fund inflows and past performance and the consistently
poor performance of these small institutional funds is the lack of monitoring by investors in these
funds. Consistent with this explanation, we find that small institutional funds are offered
primarily to trust accounts and though for-fee financial advisors.
       The institutional mutual fund industry exhibits important heterogeneity in performance
and flow performance sensitivity along dimensions related to agency conflicts as proxied by the
fund’s minimum initial investment requirement, whether the fund is affiliated with a commercial
bank, and whether the fund has a mate (an identical fund offered to retail customers). These
three variables are also useful in explaining risk-adjusted returns for retail mutual funds. Our
empirical results are consistent with the idea that agency conflicts are more acute in certain types
of institutional mutual funds. In particular, investors in small institutional funds behave as if
they are captured. This capture is reflected in higher fund expenses, lower overall returns and
the significantly lower sensitivity of fund inflows or outflows to overall performance.
       The poor performance of these small institutional funds remains somewhat puzzling. In
particular, if managers of these funds are exploiting investor capture, one would expect that the



                                                  23
poorer performance of these funds would be explained entirely by the higher fees managers of
these funds charge. However, higher management fees explain only a portion of the lower
average return for these funds.
        Hidden fees through soft dollar brokerage payments appear to play a role in this
performance difference. Because soft dollar payments are not explicitly reported, it is difficult to
directly analyze their effect on mutual fund performance. We use the performance difference
between the institutional fund and its retail mate as a proxy for how soft dollar payments made
by the fund are allocated to the no-load retail fund versus the institutional class of the fund. We
find that as turnover and the minimum initial purchase requirement of the institutional fund
increase, the performance difference between the two funds also increases. This suggests that
more intensive monitoring by institutional investors, as indicated by higher minimum initial
purchase requirements, induces funds to allocate more soft dollar expenses to the retail fund and
less to the institutional fund.




                                                 24
References
Brown, Stephen, and William Goetzmann, 1995, Attrition and Mutual Fund Performance,
   Journal of Finance 50, 679-698.
Carhart, Mark, 1997, On the Persistence of Mutual Fund Performance, Journal of Finance 52,
   57-82.
Cerulli-Lipper Analytical Report: The State of the Bank Securities Industry 1996.
Chalmers, John M.R., Roger M. Edelen and Gregory B. Kadlec, 2001, Mutual Fund Returns and
   Trading Expenses: Evidence on the Value of Active Fund Management, working paper
   Lundquist College of Business, University of Oregon.
Chevalier, Judith and Glenn Ellison, 1997, Risk Taking by Mutual Funds as a Response to
   Incentives, Journal of Political Economy 105,1167-1200.
Conrad, Jennifer S., Kevin M. Johnson and Sunil Wahal, 2001, Instutional Trading and Soft
   Dollars, Journal of Finance 56, 397-416.
Devenow, Andrea, and Ivo Welch, 1996, Rational Herding in Financial Economics, European
   Economic Review 40, 603-615.
Del Guercio, Diane and Paula Tkac, 2002, The Determinants of the Flow of Funds on Managed
   Portfolios: Mutual Funds vs. Pension Funds, Journal of Financial and Quantitative Analysis
   37, 523-557.
Edelen, Roger M., 1999, Investor Flows and the Assessed Performance of Open-End Mutual
   Funds, Journal of Financial Economics 53, 439-466.
Elton, Edwin J., Martin J. Gruber, and Christopher R. Blake, 2001, A First Look at the Accuracy
   of the CRSP Mutual Fund Database and a Comparison of the CRSP and Morningstar Mutual
   Fund Databases, Journal of Finance 56, 2415-2430.
Elton, Edwin J., Martin J. Gruber, and Christopher R. Blake, 2003, Incentive Fees and Mutual
   Funds, Journal of Finance 58, 779-804.
Fama, Eugene F., and Kenneth R. French, 1993, Common Risk Factors in the Returns on Bonds
   and Stocks, Journal of Financial Economics 33, 3-53.
Fama, Eugene F., and James D. MacBeth, 1973, Risk, Return, and Equilibrium: Empirical Tests,
   Journal of Political Economy 91, 607-636.
Goetzmann, William, and Nadav Peles, 1997, Cognitive Dissonance and Mutual Fund Investors,
   Journal of Financial Research 20, 145-158.



                                               25
Grinblatt, Mark, and Sheridan Titman, 1992, The Persistence of Mutual Fund Performance,
   Journal of Finance 47, 1977-1984.
Gruber, Martin J., 1996, Another Puzzle: The Growth in Actively Managed Mutual Funds,
   Journal of Finance 51, 783-810.
ICI, 2003, ICI Urges SEC Rulemaking on soft Dollars, Directed Brokerage, December 2003,
   www.ici.org/issues/timing/03f_sec_soft_com.
Jegadeesh, Narasimhan, and Sheridan Titman, 1993, Returns to Buying Winners and Selling
   Losers: Implications for Stock Market Efficiency, Journal of Finance 48, 65-92.
Johnsen, Bruce D., 2000, The Welfare Effects of Soft Dollar Brokerage: Law and Economics
   The Research Foundation of the Association for Investment Management and Research.
Kothari S.P. and Jerold B. Warner, 2001, Evaluating Mutual Fund Performance, Journal of
   Finance 56, 1985-2010.
Lakonishok, Josef, Andrei Shleifer, and Robert W. Vishny, 1992, The Structure and Performance
   of the Money Management Industry, Brookings Papers on Economic Activity:
   Microeconomics, 339-390.
Lauricella, Tom, and Christopher Oster, 2004, Why Your Mutual Funds Trading Matters, The
   The Wall Street Journal, January 16.
Lauricella, Tom, and Christopher Oster, 2003, Mutual Fund Soft Fees Getting a Harder Look,
   The Wall Street Journal, December 26.
Nanda, Vikram, Z. Jay Wang, and Lu Zheng, 2004, Family Values and the Star Phenomenon:
   Strategies of Mutual Fund Families, Review of Financial Studies 17, 667-698.
Sirri, Erik R., and Peter Tufano, 1998, Costly Search and Mutual Fund Flows, Journal of
   Finance 53, 1589-1621.
Warther, Vincent A., 1995, Aggregate Mutual Fund Flows and Security Returns, Journal of
   Financial Economics 39, 209-235.
Wermers, Russ, 2000, Mutual Fund Performance: An Empirical Decomposition into Stock-
   Picking Talent, Style, Transactions Costs and Expenses, Journal of Finance 55, 1655-1695.
Zitzewitz, Eric, 2003, Who Cares About Shareholders? Arbitrage-Proofing Mutual Funds,
   Journal of Law, Economics and Organization 19, 245-280.




                                              26
                                            Table 1
        Number and Average Size of Retail and Institutional Equity Mutual Funds 1990-2001

The sample includes open-end mutual funds with investment objectives of aggressive growth, growth and income, growth, or
equity income, as classified by Morningstar. Funds are required to have the following valid data from the CRSP Mutual Fund
Database: total net assets at the beginning and end of the year, monthly returns for the current and prior years, and expense ratios
and total loads for the current and prior years. Funds are also required to have valid minimum initial purchase requirements from
the November 1996 Morningstar Principia database. Funds are designated as institutional if they have a minimum initial purchase
requirement of at least $100,000 or are categorized as institutional by Morningstar (using Principia data item “Purchase
Constraints”). “Big institutional” funds have a minimum purchase requirement of at least $500,000.

                                                                                                              Big Institutional
                   Retail Funds              No Load Retail Funds             Institutional Funds
                                                                                                                   Funds

                          Year-End                       Year-End                       Year-End                        Year-End
                           Total                          Total                          Total                           Total
                         Net Assets                     Net Assets                     Net Assets                      Net Assets
 Year         Number     ($millions)         Number     ($millions)         Number     ($millions)         Number      ($millions)

 1990             368       157,269              179         64,170              10          2,503                0               0
 1991             368       223,981              175         96,856              12          4,628                0               0
 1992             386       280,115              145       102,721               22          9,107                3          1,851
 1993             524       395,306              219       156,703               64         20,301               17          7,722
 1994             648       455,907              267       189,990               88         24,859               23          8,939
 1995             828       690,743              323       286,398              138         45,230               39        18,370
 1996           1,032       931,390              384       397,913              183         69,313               51        27,576
 1997           1,189     1,267,699              411       548,070              229        107,100               58        37,469
 1998           1,254     1,617,381              418       720,345              241        141,159               55        49,648
 1999           1,225     2,076,731              411       939,778              234        160,746               54        57,967
 2000           1,168     1,904,046              390       840,861              223        148,978               54        54,568
 2001           1,128     1,617,804              376       707,146              216        128,762               53        45,950




                                                                27
                                             Table 2
Distribution of Institutional Funds by Minimum Initial Investment, Affiliation with a Retail Fund
                                       and Bank Affiliation

The sample includes all institutional and retail funds from Table 1. “Big institutional” funds have a minimum purchase
requirement of at least $500,000. An institutional fund has a retail mate if the November 1996 Morningstar Principia database
reports a retail fund with the same fund name, advisor and portfolio holdings. Identified by Lipper Analytical Services, bank
sponsored funds have bank fund families and more than half of their sales come from bank distribution. Fund characteristics are
taken from the last year of available data for each fund.

                                                                  All            Bank           Non-Bank
                                                             Institutional     Sponsored        Sponsored
                                                                Funds            Funds            Funds
                  Fund type                                    (N=268)          (N=92)           (N=176)

                  Big Institutional                             24.6%            22.8%            25.6%
                  Institutional funds with Mate                 51.1%            70.7%            35.2%
                  Big Institutional without Mate                13.8%            7.6%             19.3%
                  Small Institutional with Mate                 40.3%            55.4%            29.0%
                  Bank Sponsored                                34.3%            100%              0%




                                                             28
                                              Table 3
       Summary Statistics of Retail and Institutional Equity Mutual Funds in 1995 and 2001

The sample includes all retail and institutional funds from Table 1. Minimum initial purchase requirement is from the 1996
Morningstar Principia database, and all other data is from the CRSP mutual fund database. Total fees are calculated as the
expense ratio plus amortized total load, where the load is amortized over seven years. Monthly standard deviation is the standard
deviation of the twelve monthly returns during the year. Funds are designated as institutional if they have a minimum initial
purchase requirement of at least $100,000 or are categorized as institutional by Morningstar (using Principia data item “Purchase
Constraints”). All equity funds include 966 funds in 1995 and 1,344 funds in 2001. Retail funds include 828 funds in 1995 and
1,128 funds in 2001. Institutional funds include 138 funds in 1995 and 216 funds in 2001.


                                                           1995                                         2001
                                                                      Standard                                     Standard
                                            Mean          Median      Deviation            Mean        Median      Deviation


                                                    Panel A: All Equity Funds
          Assets Managed ($ millions)        763            139          2,606             1,300         179         4,668


                                                       Panel B: Retail Funds
          Assets Managed ($ millions)        834            135          2,796             1,434         188         5,004
         Minimum Initial Purchase ($)       1,745          1,000         3,338             1,916        1,000        3,780
                Portfolio Turnover (%)       81.8           62.0         76.9               93.1        75.0         81.7
                    Expense Ratio (%)        1.37           1.25         0.62               1.11        1.00         0.56
                        Total Fees (%)       1.76           1.75         0.74               1.52        1.63         0.68
      Monthly Standard Deviation (%)         3.15           3.03         0.73               6.39        5.33         3.04


                                                    Panel C: Institutional Funds
          Assets Managed ($ millions)        328            147          681                596          134         2,058
         Minimum Initial Purchase ($)      568,341         5,000      1,402,137           656,073      17,500      2,654,272
                Portfolio Turnover (%)       70.5           51.0         65.7               83.4        73.0         67.2
                    Expense Ratio (%)        0.83           0.90         0.31               0.85        0.93         0.31
                        Total Fees (%)       0.85           0.90         0.34               0.87        0.93         0.34
      Monthly Standard Deviation (%)         3.07           3.03         0.48               5.79        5.01         2.31




                                                                29
                                           Table 4
  Differences in Performance and Expenses of Retail and Institutional Mutual Funds 1995-2001
The sample includes open-end mutual funds with investment objectives of aggressive growth, growth and income, growth, or
equity income, as classified by Morningstar. Statistics are reported for the pooled time-series cross-section of all fund-years
available for each fund type. Funds are designated as institutional if they have a minimum initial purchase requirement of at least
$100,000 or are categorized as institutional by Morningstar. “Big institutional” funds have a minimum purchase requirement of
at least $500,000, and all other institutional funds are classified as “small institutional” funds. A retail mate is a fund with the
same name and fund advisor as the institutional fund, but with a different share class. The CRSP value-weighted return includes
all NYSE, Amex, and Nasdaq stocks on the CRSP database. EW Objective Return is the equally weighted annual return for all
funds within the investment objective.
                                                                             Standard                                     Standard
                                                    Mean        Median       Deviation          Mean       Median         Deviation
                                                             All Retail Funds                          All Institutional Funds
                                                            (7,824 fund-years)                           (1,464 fund-years)
                 Assets Managed ($ millions)        1,292          156         4,807             547         166           1,796
                           Annual Return (%)        14.22         17.20        20.37            14.44       17.81          18.03
       Annual Return – CRSP VW return (%)           -1.16         -2.17        15.24            -0.62       -0.94          12.57
  Annual Return – EW Objective Return (%)           -0.26         -0.28        13.62             0.25        0.60          11.25
             Monthly Standard Deviation (%)         4.62          4.17          2.26             4.40        4.14           1.95
                           Expense Ratio (%)        1.32          1.11          0.74             0.86        0.91           0.29
                      Portfolio Turnover (%)        87.0          67.1         106.2             77.5        65.7           64.1

                                                        Big Institutional Funds                      Small Institutional Funds
                                                           (364 fund-years)                            (1,100 fund-years)
               Assets Managed ($ millions)          801         181          2,866              463         165           1,246
                        Annual Return (%)          15.26       20.03         18.44             14.16      17.22           17.89
     Annual Return – CRSP VW return (%)            -0.14       -0.18         11.14             -0.78       -1.27          13.01
  Annual Return – EW Objective Return (%)          0.65         1.64         10.02              0.13       0.30           11.63
           Monthly Standard Deviation (%)          4.39         4.23          1.90              4.40       4.11            1.97
                        Expense Ratio (%)          0.75         0.82          0.32              0.90       0.93            0.27
                    Portfolio Turnover (%)         80.9         71.0          71.1              76.4       64.0            61.6

                                                    Institutional Funds with Mates              Institutional Funds without Mates
                                                            (742 fund-years)                              (722 fund-years)
               Assets Managed ($ millions)          422          224          603               676          116           2,478
                        Annual Return (%)          13.75        16.75        17.49             15.14        18.78          18.55
     Annual Return – CRSP VW return (%)            -1.30        -1.38        12.40             0.07         -0.24          12.71
  Annual Return – EW Objective Return (%)          -0.46        0.28         11.03             0.99         1.01           11.43
           Monthly Standard Deviation (%)          4.37         4.07          1.93             4.43         4.18           1.97
                        Expense Ratio (%)          0.91         0.94          0.26             0.82         0.85           0.31
                    Portfolio Turnover (%)         74.8         64.0          60.4             80.3         66.3           67.6

                                                   Big Institutional Funds with Mates            Big Institutional Funds without
                                                             (150 fund-years)                        Mates (214 fund-years)
               Assets Managed ($ millions)          464           248          565             1,037        137          3,694
                        Annual Return (%)          14.18         18.84        17.73            16.02       20.28         18.92
     Annual Return – CRSP VW return (%)            -1.75         -1.02        10.40            0.97        0.97          11.52
  Annual Return – EW Objective Return (%)          -0.92         0.89         10.24            1.75        1.89           9.73
           Monthly Standard Deviation (%)           4.20         3.94          1.73            4.53        4.35           2.01
                        Expense Ratio (%)           0.85         0.95          0.29             0.68        0.66          0.33
                    Portfolio Turnover (%)          69.7          58.1         64.4             88.7        77.0          74.5

                                                 Small Institutional Funds with Mates           Small Institutional Funds without
                                                           (592 fund-years)                          Mates (508 fund-years)
               Assets Managed ($ millions)         412           220          612               524        101           1,710
                        Annual Return (%)         13.64         16.43        17.44             14.77      17.95          18.40
     Annual Return – CRSP VW return (%)           -1.18         -1.47        12.87             -0.31      -0.71          13.18
  Annual Return – EW Objective Return (%)         -0.34          0.10        11.23              0.67       0.43          12.06
           Monthly Standard Deviation (%)         4.41           4.10         1.97              4.39       4.13           1.96
                        Expense Ratio (%)         0.92           0.94         0.25              0.87       0.90           0.29
                    Portfolio Turnover (%)        76.1           66.7         59.3              76.7       59.5           64.2



                                                                30
                                                                                   Table 5
                               Risk-Adjusted Performance for Retail and Institutional Mutual Funds from 1995-2001
Average monthly returns in excess of the riskfree rate and five-factor alphas are computed for equity fund types from 1995 to 2001. The sample includes open-end mutual funds
with investment objectives of aggressive growth, growth and income, growth, and equity income as classified by Morningstar. Funds are designated as institutional if they have a
minimum initial purchase requirement of at least $100,000 or are categorized as institutional by Morningstar. “Big institutional” funds have a minimum purchase requirement of at
least $500,000. An institutional fund has a mate if the November 1996 Morningstar Principia database reports a no-load retail fund with the same fund name, advisor and portfolio
holdings. Portfolio returns are equally weighted each month. RMRF is the excess return on the CRSP value-weighted market index. SMB, HML and UMD are factor-mimicking
portfolios for size, book to market, and momentum obtained from Ken French’s website. EAFE is the return on the EAFE index obtained through Morgan Stanley Capital
International Inc.’s website at www.msci.com. T-statistics are in parentheses.

                                                                                                                      5-Factor Model
                                                                Monthly
                                               Monthly          Standard                                                                                    Adjusted
           Fund type                         Excess Return      Deviation           Apha       RMRF         SMB           HML          UMD       EAFE         R2
                                                 (%)              (%)               (%)

           Retail                                0.748             4.48            -0.120       0.917       0.079        0.043         0.030     0.044       0.982
                                                                                   (-1.58)      (37.1)      (3.94)       (1.58)        (1.52)    (1.76)

           No-load Retail                        0.759             4.49            -0.082       0.895       0.070        0.039         0.022     0.045       0.986
                                                                                   (-1.23)      (41.6)      (4.04)       (1.64)        (1.32)    (2.07)
           Institutional                         0.798             4.33            -0.079       0.923       0.016        0.112         0.038     0.045       0.979
                                                                                   (-0.99)      (35.5)      (0.75)       (3.88)        (1.86)    (1.70)
           Big Institutional                     0.848             4.38            -0.032       0.951       -0.003       0.078         0.019     0.035       0.983
                                                                                   (-0.42)      (39.0)      (-0.17)      (2.86)        (0.99)    (1.41)
           Institutional with Mate               0.751             4.27            -0.115       0.920       0.015        0.109         0.030     0.043       0.977
                                                                                   (-1.37)      (33.8)      (0.67)       (3.59)        (1.40)    (1.54)
           Big Institutional without Mate        0.929             4.42             0.018       0.964       0.044        0.081         0.034     0.040       0.980
                                                                                    (0.22)      (36.6)      (2.05)       (2.75)        (1.61)    (1.51)
           Small Institutional with Mate         0.761             4.28            -0.113       0.918       0.038        0.117         0.038     0.046       0.975
                                                                                   (-1.29)      (32.1)      (1.64)       (3.69)        (1.67)    (1.60)




                                                                                      31
                                                             Table 6
                Performance Differences Among Retail and Institutional Mutual Funds
The dependent variable is the estimated intercept (alpha) of monthly percentage returns plus one-twelfth of the annual expense ratio
from a five-factor pricing model that includes the Fama and French (1993) three factors, a factor-mimicking portfolio for one-year
momentum, and the return on the EAFE index as a proxy for international equities. Alpha is estimated for individual mutual funds
over a 84-month period, 1995 through 2001. The sample includes only funds with at least 24 months of returns over the seven-year
period. Each fund is also required to have a valid expense ratio from the CRSP Mutual Fund database and a valid minimum initial
purchase requirement from the Novermber 1996 Morningstar Principia database. LOG(MINIMUM INITIAL PURCHASE) is the
natural log of one plus the minimum initial puchase requirement measured in dollars. INST equals one if the fund is classified as an
institutional fund, and is zero otherwise. MATE equals one for institutional (retail) funds with a no-load retail (institutional) fund with
the same fund name, advisor and portfolio holdings from 1995 to 2001, and is zero otherwise. INST WITH MATE equals one if the
fund is classified as an institutional fund and has a no-load retail mate, and is zero otherwise. BANK equals one if the fund is a bank
sponsored fund as classified by Lipper, and is zero otherwise. Standard errors are computed using White’s heteroskedasticity
consistent covariance estimator.
                                                       (1)           (2)             (3)           (4)               (5)
  Constant                                          0.269           0.176         0.284           0.207            0.283
                                                    (14.6)          (3.85)        (15.3)          (4.30)           (14.6)

  INST                                              -0.043
                                                    (-1.11)

  LOG (MINIMUM INITIAL PURCHASE)                                    0.0120                        0.0106
                                                                    (2.02)                        (1.76)
  MATE                                                                           -0.128           -0.123
                                                                                 (-3.53)          (-3.33)
  BANK                                                                                                             -0.102
                                                                                                                   (-3.10)

  N                                                 1,601           1,601        1,601            1,601            1,601
  Adjusted R2                                       0.0006          0.0016       0.0055           0.0072           0.0049




                                                               32
                                   (1)      (2)       (3)      (4)      (5)       (6)      (7)
Constant                         0.300     0.230    0.264     0.207    0.113    0.194     0.264
                                 (5.47)    (5.85)   (5.58)    (5.64)   (2.10)   (2.41)    (5.48)

INST WITH MATE                   -0.139                                         -0.099
                                 (-2.00)                                        (-1.80)
BIG INST                                   0.016
                                           (0.19)
SMALL INST WITH MATE                                -0.097
                                                    (-1.43)
BIG INST WITHOUT MATE                                         0.136
                                                              (1.27)
LOG (MINIMUM INITIAL PURCHASE)                                         0.0136   0.0102
                                                                       (2.22)   (1.81)
BANK                                                                                      -0.110
                                                                                          (-1.75)

N                                 272       272      272       272      272      272       272
Adjusted R2                      0.0149    0.0001   0.0069    0.0063   0.0163   0.0228    0.0084




                                              33
                                           Table 7
          Flow-Performance Relationships for Retail and Institutional Funds 1990-2001
The sample includes open-end mutual funds with investment objectives of aggressive growth, growth and income, growth,
and equity income as classified by Morningstar. Cross-sectional regressions are run year by year, and t statistics are
calculated from the annual coefficients as in Fama and MacBeth (1973). All regressions use the growth rates of net new cash
flow as the dependent variable. The independent variables are the natural log of the fund’s total net assets at the end of the
prior year (LNTNA), the growth rate of net new cash flow to all funds with the same investment objective in the same year
(FLOWCAT), the standard deviation of monthly returns in the prior year (MSDT), total fees in the prior year (FEES)
calculated as expense ratio plus amortized load, where the load is amortized over seven years and the fractional performance
rank relative to other funds in the same investment category in the prior year (PERF). INST equals one if the fund is
classified as an institutional fund, and is zero otherwise. Funds are designated as institutional if they have a minimum initial
investment requirement of at least $100,000 or are categorized as institutional by Morningstar. T-statistics are reported in
parentheses.
                                         (A)               (B)               (C)                (D)               (E)
                                      All Funds         All Retail        No-Load          Institutional       All Funds
                                                          Funds         Retail Funds          Funds
     Intercept                          0.054             0.046            -0.048              0.928             0.078
                                        (0.61)            (0.50)           (-0.28)            (1.54)             (0.89)

     LNTNA                           -0.090             -0.089            -0.058             -0.185            -0.092
                                     (-7.94)            (-8.92)           (-4.47)            (-3.51)           (-8.27)

     FLOWCAT                          1.170             1.233              1.015             0.906              1.194
                                      (5.47)            (4.82)             (2.60)            (2.00)             (5.30)

     MSDT                             2.339             2.469              3.133             0.321              2.301
                                      (1.25)            (1.28)             (1.20)            (0.11)             (1.25)

     FEES                             1.492             0.679             -0.329             -3.636             0.525
                                      (1.26)            (0.72)            (-0.07)            (-0.75)            (0.54)

     PERF1                            1.138             1.200              1.246             -1.295             1.158
                                      (5.44)            (5.16)             (2.19)            (-0.61)            (5.23)

     PERF2                            0.313             0.279             -0.026             1.561              0.294
                                      (1.98)            (1.57)            (-0.05)            (1.52)             (1.66)

     PERF3                            0.736             0.746              0.551             0.676              0.741
                                      (3.76)            (3.40)             (2.12)            (1.54)             (3.40)

     PERF4                            0.446             0.444              0.334             0.536              0.447
                                      (1.98)            (1.73)             (1.47)            (0.76)             (1.73)

     PERF5                            2.841             3.106              3.184             0.495              3.106
                                      (7.76)            (8.17)             (5.82)            (0.50)             (8.16)

     INST*PERF1                                                                                                -0.004
                                                                                                               (-0.02)

     INST*PERF2                                                                                                -0.528
                                                                                                               (-0.65)

     INST*PERF3                                                                                                 0.651
                                                                                                                (0.81)

     INST*PERF4                                                                                                 0.330
                                                                                                                (0.46)
     INST*PERF5                                                                                                -2.770
                                                                                                               (-3.22)
     Number of observations          11,778             10,118             3,698             1,650             11,778
     Adjusted R2                      0.17               0.18              0.16              0.11                0.18



                                                              34
                                              Table 8
             Flow-Performance Relationships for Retail and Institutional Funds 1995-2001
   The sample includes open-end mutual funds with investment objectives of aggressive growth, growth and income, growth,
   and equity income as classified by Morningstar. The table reports pooled time-series cross-sectional GLS coefficient
   estimates using the growth rates of net new cash flow as the dependent variable. The GLS estimation is based on an AR(1)
   error structure over time. The independent variables are the log of the fund’s total net assets at the end of the prior year
   (LNTNA), the growth rate of net new cash flow to all funds with the same investment objective in the same year
   (FLOWCAT), the standard deviation of monthly returns in the prior year (MSDT), total fees in the prior year (FEES)
   calculated as expense ratio plus amortized load, where the load is amortized over seven years and the fractional performance
   rank relative to other funds in the same investment category in the prior year (PERF). INST equals one if the fund is
   classified as an Institutional fund, and is zero otherwise. Funds are designated as institutional if they have a minimum initial
   purchase requirement of at least $100,000 or are categorized as institutional by Morningstar. “Big institutional” funds have
   a minimum purchase requirement of at least $500,000, and all other institutional funds are classified as “small institutional”
   funds. An institutional (retail) fund has a mate if the November 1996 Morningstar Principia database reports a retail
   (institutional) fund with the same fund name, advisor and portfolio holdings. Z-statistics are reported in parentheses, and the
   chi-square test for goodness of fit has nine degrees of freedom.

                                 (A)                (B)               (C)                (D)               (E)               (F)
                              No-Load          Institutional          Big              Small          Institutional        Retail
                             Retail Funds         Funds          Institutional      Institutional      Funds with        Funds with
                                                                    Funds              Funds             Mates             Mates
Intercept                       0.546             0.989             1.181              1.507             1.278             1.269
                                (5.78)            (6.01)            (3.80)             (7.01)            (5.26)            (5.03)

LNTNA                           -0.102            -0.172            -0.130            -0.284             -0.191            -0.170
                                (-10.5)           (-10.9)           (-4.45)           (-12.1)            (-7.90)           (-8.17)

FLOWCAT                         1.062             1.027               0.402            1.007             0.926             0.581
                                (5.28)            (3.86)              (0.77)           (3.14)            (2.50)            (1.34)

MSDT                            -0.935            -3.632            -3.747            -3.608              -4.01             -2.59
                                (-0.45)           (-3.51)           (-1.86)           (-3.00)            (-2.66)           (-1.47)

FEES                             -8.96            -20.77            -22.69            -26.48             -24.85            -11.56
                                (-3.78)           (-2.52)           (-1.52)           (-2.20)            (-1.99)           (-1.24)

PERF1                           0.206             0.812             -1.774             1.505             0.271             -0.768
                                (0.52)            (1.32)            (-1.36)            (2.29)            (0.30)            (-0.78)

PERF2                           0.380             -0.086              1.151           -0.535             -0.290            0.117
                                (1.13)            (-0.20)             (1.26)          (-1.19)            (-0.48)           (0.17)

PERF3                           0.459             0.617               0.487            0.734             1.471             0.990
                                (1.43)            (1.51)              (0.60)           (1.67)            (2.54)            (1.55)

PERF4                           0.233             0.780             -0.134             0.758             0.134             -0.122
                                (0.72)            (1.95)            (-0.18)            (1.72)            (0.24)            (-0.18)

PERF5                           2.352             0.715               1.532            0.838             0.900             3.782
                                (6.40)            (1.42)              (1.63)           (1.50)            (1.16)            (3.95)

Number of Observations          2,663              1,450               358             1,092              735               687

Chi-square (9 df)               357.0              239.8              44.4             245.8             128.2             133.1




                                                                 35
                                              Table 9
             Flow-Performance Relationships for Retail and Institutional Funds 1995-2001
   The sample includes open-end mutual funds with investment objectives of aggressive growth, growth and income, growth,
   and equity income as classified by Morningstar. The table reports pooled time-series cross-sectional GLS coefficient
   estimates using the growth rates of net new cash flow as the dependent variable. The GLS estimation is based on an AR(1)
   error structure over time. The independent variables are the log of the fund’s total net assets at the end of the prior year
   (LNTNA), the growth rate of net new cash flow to all funds with the same investment objective in the same year
   (FLOWCAT), the standard deviation of monthly returns in the prior year (MSDT), total fees in the prior year (FEES)
   calculated as expense ratio plus amortized load, where the load is amortized over seven years, the realized return for the
   fund in the prior year (LAGGED RETURN), and the five-factor alpha estimated over the prior 24 months (ALPHA), and the
   standard deviation of the difference between fund returns and the CRSP value weighted index over the prior 24 months
   (TRACKING ERROR). Funds are designated as institutional if they have a minimum initial purchase requirement of at
   least $100,000 or are categorized as institutional by Morningstar. “Big institutional” funds have a minimum purchase
   requirement of at least $500,000, and all other institutional funds are classified as “small institutional” funds. An
   institutional (retail) fund has a mate if the November 1996 Morningstar Principia database reports a retail (institutional) fund
   with the same fund name, advisor and portfolio holdings. Z-statistics are reported in parentheses, and the chi-square test for
   goodness of fit has nine degrees of freedom.

                                 (A)                (B)               (C)                (D)               (E)               (F)
                              No-Load          Institutional          Big              Small          Institutional        Retail
                             Retail Funds         Funds          Institutional      Institutional      Funds with        Funds with
                                                                    Funds              Funds             Mates             Mates
Intercept                       0.553             0.800             0.609              0.876             1.143             1.230
                                (7.14)            (6.29)            (2.76)             (5.64)            (5.77)            (6.23)

LNTNA                           -0.086            -0.106            -0.068            -0.120             -0.146            -0.160
                                (-8.70)           (-7.42)           (-2.75)           (-6.91)            (-6.60)           (-8.33)

FLOWCAT                         0.907             0.736               0.363            0.841             0.601             0.076
                                (4.58)            (2.79)              (0.74)           (2.70)            (1.62)            (0.17)

MSDT                            -0.125            -2.461            -4.135            -2.269             -3.059            -0.181
                                (-0.15)           (-2.19)           (-1.99)           (-1.70)            (-1.87)           (-0.09)

FEES                             -6.06            -11.28             -8.71            -12.04             -13.71            -17.97
                                (-2.59)           (-1.48)           (-0.70)           (-1.24)            (-1.14)           (-1.98)

LAGGED RETURNS                  0.337             0.332               0.249            0.361             0.262             0.859
                                (5.05)            (2.99)              (1.23)           (2.76)            (1.52)            (4.12)

ALPHA                           16.96             28.62               29.04            27.73             28.14             28.03
                                (6.25)            (6.28)              (3.41)           (5.07)            (4.22)            (3.35)

TRACKING ERROR                  0.559             -3.019            -1.183            -2.573             -2.640            -2.398
                                (0.40)            (-1.64)           (-2.81)           (-1.07)            (-0.90)           (-0.68)



Number of Observations          2,490              1,308               325              983               669               643

Chi-square (9 df)               184.8              162.3              29.7             112.8              84.4             115.0




                                                                 36
                                         Appendix A
                   Examples of Purchase Requirements for Institutional Funds30

I. Big Institutional Funds without Retail Mates

    1. AHA Diversified Fund
        AHA Investment Funds, Inc. (the "Fund"), is an open-end, diversified management
investment company (commonly known as a "mutual fund"). Shares of the Fund are
available only to participants in the American Hospital Association Investment Program (the
"Program"), and to the American Hospital Association (and its affiliated companies). The
Fund is designed to provide Participants in the Program with a cost-effective method of
pursuing a professionally managed, diversified program for investment of their pension funds
and corporate assets, and implementing asset allocation decisions.
        Shares of each Portfolio are available only to Participants which have entered into
Program Agreements and may be purchased on a continuous basis directly from the Fund at
the net asset value per share of the Portfolio next calculated after receipt of a purchase order
and federal funds. Shares are not available for purchase by individuals. Shares may be
redeemed (see "How to Redeem Shares"), but are non-transferable.
        The minimum initial investment in the Fund is $1 million. The initial minimum
investment and subsequent investments in any Portfolio must be at least $100,000.

     2. American AAdvantage Growth and Income
         Institutional Class shares are offered without a sales charge to investors who make an
initial investment of at least $2 million, including: (1) agents or fiduciaries acting on behalf
of their clients (such as employee benefit plans, personal trusts and other accounts for which
a trust company or financial advisor acts as agent or fiduciary); (2) endowment funds and
charitable foundations; (3) employee welfare plans which are tax-exempt under Section
501(c)(9) of the Internal Revenue Code of 1986, as amended ("Code"); (4) qualified pension
and profit sharing plans; (5) cash and deferred arrangements under Section 401(k) of the
Code; (6) corporations; and (7) other investors who make an initial investment of at least $2
million.

    3. Delaware Pooled Trust Aggressive Growth and Defensive Equity Funds
        The Fund is designed to meet the investment needs of discerning institutional
investors who desire experienced investment management and place a premium on personal
service.
        Shares of each Portfolio may be purchased without a sales commission, at net asset
value per share next determined after (i) the Fund has been notified by telephone of your
                                                                        s
purchase order and (ii) Federal Funds have been delivered to the Fund' bank account
maintained with The Morgan Guaranty Trust Company of New York ("Custodian Bank").
At such time as the Fund receives appropriate regulatory approval to do so in the future,
under certain circumstances, the Fund may, at its sole discretion, allow institutional investors
who have an investment counseling relationship with Delaware Investment Advisers or
Delaware International to make investments in the Portfolios by a contribution of securities
in-kind to such Portfolios.
30
     Descriptions are from the funds offering prospectus


                                                           37
       The minimum initial investment for a Portfolio is $1,000,000.

    4. DFA U.S Large Cap and U.S Large Company
         DFA INVESTMENT DIMENSIONS GROUP INC. is an open-end management
investment company whose shares are offered, without a sales charge, generally to
institutional investors and clients of registered investment advisers.
         PURCHASE OF SHARES Minimum initial investment $1 million…
         If accepted by the Fund, shares of the Portfolios may be purchased in exchange for
securities which are eligible for acquisition by the Portfolios (or their corresponding Series)
or otherwise represented in their portfolios as described in this prospectus or in exchange for
local currencies in which such securities of the International Equity Portfolios, the
International Value Series, Enhanced U.S. Large Company Series, DFA Two-Year Global
Fixed Income Series and DFA Global Fixed Income Portfolio are denominated.

    5. MAS Equity, MAS Midcap Growth, MAS Mid Cap Value and MAS Value
       Funds
       Institutional Class Shares are available to clients of the Adviser with combined
investments of $5,000,000 and Shareholder Organizations who have a contractual
                                        s
arrangement with the Fund or the Fund' Distributor, including institutions such as trusts,
foundations or broker-dealers purchasing for the accounts of others.

    6. Masterworks Growth Stock and S&P Stock Funds
        Only the following types of investors are eligible to invest in the Funds: (1)
Participants in Benefit Plans ("Plan Participants"), including retirement plans, that have
                                 s
appointed one of the Company' Shareholder Servicing Agents as plan trustee, plan
administrator or other agent, or whose plan trustee, plan administrator or other agent has a
servicing arrangement with a Shareholder Servicing Agent that permits investments in the
Funds, and Plan Participants who invest pursuant to an agreement between such a Benefit
Plan and a Shareholder Servicing Agent; (2) Foundations, corporations and other business
                                                                       s
entities that have a servicing arrangement with one of the Company' Shareholder Servicing
Agents that permits investments in the Funds and persons who invest pursuant to an
agreement between such an entity and a Shareholder Servicing Agent; and (3) Individuals,
other than those described above, who invest at least $1 million in a Fund pursuant to an
account arrangement with a Shareholder Servicing Agent ("Qualified Buyers").

    7. PIMCO Institutional Funds
        Each Fund offers two classes of shares: the "Institutional Class" and the
"Administrative Class." Shares of the Institutional Class are offered primarily for direct
investment by investors such as pension and profit sharing plans, employee benefit trusts,
endowments, foundations, corporations, other institutions, and high net worth individuals.
They also are offered through certain financial intermediaries that charge their customers
transaction or other fees with respect to the customers' investment in the Funds. Shares of the
Administrative Class are offered primarily through brokers, retirement plan administrators,
and other financial intermediaries. Administrative Class shares pay service fees to such
entities for services they provide to shareholders of that class. Shares of each class of the




                                              38
Funds are offered for sale at the relevant next determined net asset value for that class with
no sales charge.


II. Big Institutional Funds with Retail Mates

    1. Bear Sterns Insiders Select Y class, Large Cap Value Y class, S&P Stars Y class
       The minimum initial investment is $1,000,000; there is no minimum for subsequent
investments. You may buy Class Y shares of the Portfolio through your account
representative at a broker-dealer with whom the Distributor has entered into a sales
agreement (an "Authorized Dealer") or the Transfer Agent by wire only.

   2. Corefund
       The minimum initial investment is $500,000. Institutional investors may acquire
Class Y Shares of a Fund for their own account or as a record owner on behalf of fiduciary,
agency or custody accounts by placing orders with the Distributor.

    3. Crestfunds
         Trust Class shares are offered continuously to tax-advantaged and other employee
benefit or retirement accounts with which Crestar Bank, or an affiliate thereof, has entered
into certain management, administration, and/or servicing agreements. For further
information on opening an account, contact your plan sponsor or financial intermediary for
the services and procedures which pertain to your account. Sales personnel of financial
institutions distributing the Portfolios'shares and other persons entitled to receive
compensation for selling such shares may receive differing compensation.
         Minimum initial investment: $500,000

    4. Fidelity Advisors
        Institutional Class shares are offered to:
    1. Broker-dealer managed account programs that (i) charge an asset-based fee and (ii)
will have at least $1 million invested in the Institutional Class of the Advisor funds. In
addition, employee benefit plans (as defined in the Employee Retirement Income Security
Act), 403(b) programs and plans covering sole-proprietors (formerly Keogh/H.R. 10 plans)
must have at least $50 million in plan assets;

    2. Registered investment advisor managed account programs, provided the registered
investment advisor is not part of an organization primarily engaged in the brokerage
business, and the program (i) charges an asset-based fee and (ii) will have at least $1 million
invested in the Institutional Class of the Advisor funds. In addition, accounts other than an
employee benefit plan, 403(b) program or plan covering a sole-proprietor (formerly a
Keogh/H.R. 10 plan) in the program must be managed on a discretionary basis;

    3. Trust institution and bank trust department managed account programs that (i) charge
an asset-based fee and (ii) will have at least $1 million invested in the Institutional Class of
the Advisor funds. Accounts managed by third parties are not eligible to purchase
Institutional Class shares;



                                               39
    4. Insurance company separate accounts that will have at least $1 million invested in the
Institutional Class of the Advisor funds;

   5. Fidelity Trustees and employees; and

    6. Insurance company programs for employee benefit plans, 403(b) programs or plans
covering sole-proprietors (formerly Keogh/H.R. 10plans) that (i) charge an asset-based fee
and (ii) will have at least $1 million invested in the Institutional Class of the Advisor funds.
Insurance company programs for employee benefit plans, 403(b) programs and plans
covering sole-proprietors (formerly Keogh/H.R. 10 plans) include such programs offered by
a broker-dealer affiliate of an insurance company, provided that the affiliate is not part of an
organization primarily engaged in the brokerage business.

     5. Mass Mutual Institutional Value Equity 2 and Value Equity 3 Funds
     Mass Mutual Institutional Funds (the "Trust") is a professionally managed investment
company… The Trust is designed to offer investors both the opportunity to pursue long-term
investment goals and the flexibility to respond to changes in their investment objectives and
economic and market conditions. Each Fund has a distinct investment objective.
                 s
      The Trust' Distributor receives distribution fees and may allow all or a portion of them
as dealer discounts and brokerage commissions to dealers, including MML Investors
Services, Inc. ("MMLISI"), a wholly owned subsidiary of Mass Mutual. From time to time,
                                                                                  s
dealers who receive dealer discounts and brokerage commissions from the Trust' Distributor
may allow all or a portion of them to other dealers or brokers. The service fees will be paid to
Mass Mutual. Investors in Class 2 and Class 3 shares must meet certain eligibility
requirements. An initial Investor may purchase Class 2 shares of a Fund only if the net asset
value of such shares equals at least $2 million. An initial Investor may purchase Class 3
shares of a Fund only if the net asset value of such shares equals at least $10 million. For
purposes of determining whether a Plan Investor has satisfied the initial investment minimum
with respect to Class 2 or Class 3 shares, all assets invested with Mass Mutual and/or in the
Trust for that Plan will be recognized, except for life insurance assets. For purposes of this
determination, a Plan also shall include those plans established by entities of the Plan
Sponsor that satisfy Section 414(b), (c), (m), or (o) of the Code, provided that all such Plans
have common trustees. Additionally, the Plan Investor will be deemed to have satisfied the
$2 million and $10 million requirements if the Plan Investor certifies that at least such
amounts are available for transfer and will be transferred from the Plan to Mass Mutual
and/or the Trust no later than six months from the date of initial purchase. Alternatively, an
                                                                         s
initial Investor may purchase Class 2 or Class 3 shares if the Investor' employee retirement
benefit plan with Mass Mutual has 400 or more Participants (with respect to Class 2) or 750
or more Participants (with respect to Class 3). All subsequent investments by an Investor will
be of the same Class of shares already held until shares are converted from one Class to
another (see "Converting From Class 1 to Class 2 Shares or Class 2 to Class 3 Shares".)

    6. Vista Large Cap Equity Institutional
       Investors must buy a minimum $1,000,000 worth of Institutional Shares in the Fund
to open an account. There are no minimum levels for subsequent purchases. An investor can



                                              40
combine purchases of Institutional Shares of other Chase Vista Funds (except for money
market funds) in order to meet the minimum. The Fund may waive this minimum at its
discretion.


III. Small Institutional Funds

    1. Biltmore Equity, Equity Index and Quantitative Equity Funds
         Institutional Shares are offered only to accounts held by the Wachovia Banks in a
fiduciary, agency, custodial, or similar capacity and are subject to a minimum initial
investment as provided in the Wachovia Banks'                s
                                                   customer' relevant account agreement.
Institutional Shares are sold at net asset value and are distributed without a Rule 12b-1 Plan.

    2. Benchmark Diversified Growth, Equity Index and Focused Growth Funds
         Units of the Portfolio are offered to Northern Trust, its affiliates and other institutions
and organizations (the "Institutions") acting on behalf of their customers, clients, employees
and others (the "Customers") and for their own account. Institutions may purchase units of
the Portfolio through procedures established in connection with the requirements of their
qualified accounts or through procedures set forth herein with respect to Institutions that
invest directly. Institutions should contact Northern or an affiliate for further information
regarding purchases through qualified accounts. There is no minimum initial investment for
Institutions that maintain qualified accounts with Northern or its affiliates.

    3. Galaxy Growth and Income
         Trust Shares are offered to investors maintaining qualified accounts at banks and trust
institutions including institutions affiliated with Fleet Financial Group, Inc., and to
participants in employer-sponsored defined contribution plans

    4. Neuberger & Berman Focus, Guardian, Manhattan and Part Funds
       These portfolios are offered to life insurance companies to serve as investment
vehicles under their variable annuity and variable life insurance contracts.

    5. One Bank Group Funds Class I shares
         One Group offers the following classes of shares: Class I shares are available to
institutional investors and any organization authorized to act in a fiduciary, advisory,
custodial or agency capacity. We refer to these entities as intermediaries.

    6. SEI Funds Institutional shares
        The Achievement Funds Trust (the "Trust") is a mutual fund that offers separate
classes of shares of beneficial interest in the seven portfolios listed above (the "Portfolios").
This Prospectus relates solely to the Institutional class of shares of the Portfolios (the
"shares") which are designed to offer financial institutions ("shareholders") a convenient
means of investing their own funds or funds for which they act in a fiduciary, agency or
custodial capacity in one or more professionally managed portfolios of securities. Each
Portfolio also offers Retail Class A shares that differ from the Institutional shares with
respect to distribution costs, sales charges and dividends.



                                                41
         Financial institutions may acquire shares of the Portfolios for their own account or as
record owner on behalf of fiduciary, agency or custody accounts by placing orders with the
Distributor. Institutions that use certain SEI proprietary systems may place orders
electronically through those systems. State securities laws may require banks and financial
institutions purchasing shares for their customers to register as dealers pursuant to state laws

     7. UAM Funds
         The minimum initial investment required is $2,500. The minimum initial investment
for IRA accounts is $500. The minimum initial investment for spousal IRA accounts is $250
         Shares of the Portfolio may be purchased by customers of broker-dealers or other
financial intermediaries ("Service Agents") who have established a shareholder servicing
relationship with the Fund on behalf of their customers. Service Agents may impose
additional or different conditions on purchases or redemptions of Portfolio shares and may
charge transaction or other account fees. Each Service Agent is responsible for transmitting
to its customers a schedule of any such fees and information regarding additional or different
purchase or redemption conditions.




                                               42

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:112
posted:2/20/2010
language:English
pages:43