Smith
Faculty Opinion Article
Another
Look At S&P 500 Retail Index
Funds
Written by John A. Haslem,
H. Kent Baker and David M. Smith
Monday, 24 March 2008 23:05
A long-standing debate exists
about whether a mutual fund’s
performance is due to the quality of
management, other fund attributes or
just luck. Although some portfolio
managers of actively managed funds
may have stock-picking talent,
actively managed funds, on average,
underperform benchmark portfolios
with equivalent risk.
An implication of the
underperformance of most actively
managed mutual funds is that
investors should be better off in
low-cost index funds. The
often-presumed commoditylike nature
of index funds suggests that price
competition should be stronger than
for actively managed funds. Managers
operate index funds not to beat
their benchmarks or actively managed
funds, but to mimic benchmark
portfolios and performance before
expenses. This “passive management”
requires skill, but of a different
sort than in actively managed funds.
These skills include minimizing
transaction costs in matching the
composition of the benchmark index,
and using futures to compensate for
the effects of “cash drag.”
If strong price competition
exists among index mutual funds,
only nominal size-adjusted
differences in the expense ratios of
S&P 500 Index funds should be
present. But are all of these funds
equally adept at tracking the
“returns” of the S&P 500 Index? The
answer is “no.” S&P 500 Index funds
should not be viewed as financial
commodities. Thus, examining the
size of their expense ratios,
attributes related to them and fund
performance results is important.
Data And Method
We examine 136 retail S&P 500
Index funds identified by
Morningstar at year-end 2006 and use
expense ratios as a percent to
measure mutual fund costs.1 Expenses
are particularly relevant because
they tend to be stable. Because fund
expenses are one of the few
predictable aspects of investing,
investors selecting a fund could
benefit by considering these
expenses.
The expense ratio is total
expenses divided by fund average net
assets.2 Total expenses consist of
three components: (1) management
fees, (2) Rule 12b-1 fees, and (3)
other expenses including transfer
agent fees, securities custodian
fees, shareholder accounting
expenses, legal fees, auditor fees
and independent director fees. The
regulatory definition of total
expenses is not an all-inclusive
measure. It excludes sales charges
and fees directly charged to
shareholder accounts and portfolio
transaction costs (brokerage fees,
bid/ ask spreads, and market impact
and other trading costs) that reduce
portfolio and shareholder returns.
Univariate Analysis
We begin by classifying the 136
index mutual funds into quartiles
based on their expense ratios. Funds
in Quartiles 1 and 4 have the lowest
and highest expense ratios,
respectively. Next, we examine the
relation between expense ratio class
and selected performance measures.
We use three well-known performance
measures: three-year Sharpe ratios,
three-year Jensen’s alphas and
annualized total returns for
multiple periods (one, three and
five years). The Sharpe ratio
assesses risk-adjusted portfolio
returns using standard deviation as
the measure of total risk, and
Jensen’s alpha assesses
risk-adjusted portfolio returns
using systematic risk.
We use the Wilcoxon two-sample
test to determine whether the values
in Quartile 1 and Quartile 4 differ
significantly. Using this univariate
test, we expect that performance, as
measured by each median Sharpe
ratio, Jensen’s alpha and annualized
total return is statistically
greater in the low-expense ratio
class (Quartile 1) than in the
high-expense ratio class (Quartile
4). Thus, we expect a negative
relation between expense ratio class
and each performance measure.
Mutual fund expenses are known to
be an important determinant of
returns. Front loads and deferred
charges reduce investor returns but
are not included in the regulatory
expense ratio. Thus, funds with low
(high) loads are also likely to have
low (high) expense ratios. Unlike
front loads and deferred charges,
12b-1 fees are a component of the
regulatory expense ratio. Marketing
expenses paid for by the fees
allowed under Rule 12b-1 could
increase or decrease a fund’s
expense ratio.
Proponents of the 12b-1 fee argue
that imposing the fee could allow
funds to decrease other loads,
especially front-end loads, which
would in turn attract investors and
reduce the expense ratio due to
increased economies of scale.
We anticipate that median
front-end loads, deferred charges
and 12b-1 fees are statistically
smaller in the lowexpense ratio
class (Quartile 1) than in the
high-expense ratio class (Quartile
4). Thus, we expect a positive
relation between expense ratio class
and each of these three attributes.
Other attributes may have a
negative relation with mutual fund
expense ratios. Fund size, with
increasing economies of scale,
provides reductions in fund expense
ratios. This reduction in expense
ratios is further encouraged by
funds with stated reductions in
management fees at increasingly
large asset breakpoints. The extent
to which these size-induced savings
are passed along to shareholders
remains an issue of debate.
Fund age and size are closely
related. Older mutual funds tend to
be larger than the younger ones and
tend to benefit more from economies
of scale. As such, fund age is
generally considered to be
negatively associated with expense
ratios.
We expect that median fund size
and age are statistically larger and
older in the low-expense ratio class
(Quartile 1) than in the
high-expense ratio class (Quartile
4).
A relation could also exist
between expense ratio class and
other attributes. For example, cash
holdings and turnover may help to
explain variations in expenses of
actively managed mutual funds. Funds
with larger cash holdings may incur
relatively smaller portfolio
management expenses and transaction
costs in anticipating and providing
liquidity to meet shareholder
redemptions. On the other hand,
large cash holdings are likely to
produce an unacceptable degree of
“cash drag” in a generally rising
market. Index fund portfolios are
managed to match those of their
index benchmarks, which suggests
they have smaller cash holdings with
less variance.
Further, index funds do not
adjust their cash holdings in
efforts to time the market. Thus, in
this context, cash holdings of S&P
500 Index funds may well not differ
significantly by expense ratio
class.
Turnover serves as a proxy for
the level of trading activity and
associated transaction costs. These
costs reduce fund and investor
returns and are not included in the
regulatory expense ratio. Portfolio
turnover may have some small
positive association with management
fees and other expenses, which are
included in the expense ratio.
Unlike actively managed funds, the
level of trading activity of S&P 500
Index funds is unlikely to differ
significantly regardless of the
expense ratio class. We expect that
median cash holdings and turnover do
not differ statistically between the
low-expense ratio class (Quartile 1)
and the high-expense ratio class
(Quartile 4).
|
Figure 1
Source: Morningstar
Principia
|
Multivariate Analysis
We use a multiple regression model to examine whether mutual fund
descriptors, specifically expense ratios and related attributes, are
useful in explaining fund performance.3 Both models of performance
contain a dummy variable indicating the presence or absence of a
12b-1 plan. Our model includes this variable and five others from
the latter study that control for fund size, cash holdings, turnover
and the magnitude of front-end and deferred charges. For this
portion of the analysis, we combine with our index fund sample all
actively managed large-cap blend domestic equity funds (the same
Morningstar equity-style box cell occupied by S&P 500 Index funds)
to accompany the index fund sample. To facilitate pooling of
disparate funds for the regression, our model includes a dummy
variable indicating whether funds are passively or actively managed.
Empirical Results
We present our empirical results in Figures 1 through 4. Figure 1
shows the number of retail S&P 500 Index funds and their expense
ratios for the period 1992–2006. Over the 15-year period, the number
of these index funds increased from nine in 1992 to 136 in 2006. The
median expense ratio also trended upward, from 0.350 percent in 1992
to 0.630 percent in 2006.
|
Figure 2
Source: Morningstar Principia
|
The unweighted mean has increased over time. The mean as weighted
by each fund’s net assets remained between 23 and 26 basis points
for many years, and recently it has fallen to about 20 basis points.
The standard deviation suggests that the range of expense ratios is
tightening. Finally, each year we sort funds by expense ratio and
then identify the fund in which the median dollar is invested. We
find that over time, the median dollar has been invested in a fund
charging between 18 and 20 basis points per year. Clearly, the
proliferation of new funds and fund classes has not resulted in
additional low-cost choices for investors, as evidenced by the
steady upward trend for the median and unweighted mean. On the other
hand, the rightmost two columns suggest that most investors continue
to pour money into the lowest-cost alternatives.
Figure 2 summarizes the median expense ratios for the 136 S&P 500
Index funds in each quartile for 2006. The expense ratios increase
with each higher class from 0.27 percent in the low-expense ratio
class (Quartile 1) to 1.29 percent in the high-expense ratio class
(Quartile 4). As later shown in Figure 3, the median expense ratio
in Quartile 1 is significantly lower than that in Quartile 4.
Univariate Results
Figure 3 summarizes the median attributes of the retail S&P 500
Index funds by expense ratio class. We generally report the results
using medians instead of means because the underlying variables tend
to be non-normally distributed. Figure 3 also reports the results
for a univariate analysis between expense ratio classes (Quartiles 1
and 4) and various fund attributes.
Panel A of Figure 3 shows the results of our univariate tests
involving the implied impact of expense ratios on returns for the
136 retail S&P 500 Index funds. As expected, funds in the
low-expense ratio class (Quartile 1) significantly outperform those
in the high-expense ratio class (Quartile 4) based on the Sharpe
ratio, Jensen’s alpha and annualized total returns over one-, three-
and five-year periods. Each performance measure decreases
monotonically when moving across expense ratio classes from Quartile
1 to Quartile 4. For example, three-year annual total returns
decrease as follows: 10.16 percent (Quartile 1), 9.85 percent
(Quartile 2), 9.71 percent (Quartile 3) and 9.03 percent (Quartile
4). These findings are consistently strong. Panel B of Figure 3
shows the relation of expense ratio class to other mutual fund
attributes. As expected, funds in the low-expense ratio class
(Quartile 1) have significantly smaller deferred charges and 12b-1
fees than those in the high-expense ratio class (Quartile 4). None
of the 33 funds in Quartile 1 has deferred charges, but 32 of the 35
funds (91.4 percent) in Quartile 4 do. Only six of the 33 funds
(18.2 percent) in Quartile 1 have 12b-1 fees, but all 35 funds in
Quartile 4 do.
Front-end loads do not differ significantly between funds in the
low- versus high-expense ratio classes. Only 24 of the 136 funds
(17.6 percent) have front-end loads, possibly because informed
investors recognize them as avoidable costs. The fact that only two
of 33 funds (6.1 percent) in Quartile 1 and zero of 35 funds in
Quartile 4 charge front-end loads helps to explain the lack of a
significant difference between Quartiles 1 and 4.
The findings are also consistent with our expectations involving
the relation between expense ratio class and other attributes. The
median net assets in the low-expense ratio class ($459.50 million)
are significantly larger than those in the high-expense ratio class
($17.30 million). Funds are significantly older in the low-expense
ratio class (10.67 years) than those in the high-expense ratio class
(6.96 years). Thus, a negative relation exists between expense ratio
class and both fund size and age. Although median cash holdings are
larger in the high- versus low-expense ratio class (1.90 percent and
1.50 percent, respectively), no significant difference exists in
cash holdings between Quartiles 1 and 4. Findings of no significant
difference also apply to turnover. As expected, the median turnover
is only 6 percent in both Quartiles 1 and 4.
Sorting our sample of 136 retail S&P 500 Index mutual funds by
expense ratio class and aggregating net fund assets, we find that
some 15 percent of assets are in funds in expense ratio classes
above the group median (Quartiles 3 and 4). Thus, retail investors
have placed about 85 percent of their assets in low-expense S&P 500
Index funds (Quartiles 1 and 2). These findings are consistent with
a degree of price competition among the higher-performing funds.
|
Figure 3
Source: Morningstar Principia
|
Note: Sharpe ratios
and Jensen’s alphas are three-year. Returns are annualized for the
period ending December 31, 2006. Loads are one-year front-end loads.
The rightmost column indicates whether the values in Quartile 1 and
Quartile 4 differ significantly, according to Wilcoxon two-sample
tests. The number of funds is 136 unless otherwise indicated.
* and ** indicate significance at the 0.05 and 0.01 levels,
respectively.
What may explain the sizable expense dispersion among retail S&P
500 Index mutual funds despite the relative homogeneity of their
portfolios? One explanation is that for retail S&P 500 Index funds,
expense dispersions are not explained by portfolio performance
alone, but also by nonportfolio differentiation and
information-search frictions.4 As for nonportfolio differentiation,
our evidence may help to explain why high-priced index funds exist.
High-expense funds, on average, have lower minimum initial purchase
amounts. As Panel B of Figure 3 shows, the median minimum initial
purchase for funds is $1,000 in Quartiles 2, 3 and 4 but $3,000 in
Quartile 1. Thus, some retail investors may be unable to buy funds
in the low-expense category because they cannot meet the minimum
initial purchase requirement.
Multivariate Results
In our final step, we test a multiple regression model that
relates annualized returns to expense ratios, while holding constant
for other fund characteristics such as fund size, cash holdings,
turnover rate, load and whether the fund has a 12b-1 fee. By the
normal measures of cross-sectional analysis such as adjusted R2 and
the F-statistic, our model performed well in explaining fund
returns.
|
Figure 4
* and ** indicate significance at the 0.05 and 0.01
levels, respectively.
|
As Figure 4 shows, we find a
strong negative relation between the
annualized return and the expense
ratio. The magnitudes of our
regression coefficients show that
after controlling for other major
mutual fund attributes, an increment
of 1 percentage point in the expense
ratio is associated with about a 1
percentage point lower annual
return. This result holds for both
three-year and five-year annualized
returns. This apparent economic and
statistical significance supports
investor use of the expense ratio as
an indicator of relative investment
merit.
Conclusions
We analyze the expense ratio and
attributes affecting the performance
of 136 retail S&P 500 Index funds.
These funds are among the simplest
of financial vehicles. Nonetheless,
the expense ratios and performance
of these benchmark trackers differ
significantly.
Our univariate analysis shows
that mutual fund performance, as
measured by the median Sharpe
ratios, Jensen’s alphas and
annualized total returns, increases
as expense ratios decrease.
In addition, a positive relation exists between expense ratios,
deferred charges and 12b-1 fees. By contrast, fund size and age are
negatively related to expense ratios. Thus, our analysis of mutual fund
expenses suggests expense-conscious retail investors should look
carefully at the deferred charges, 12b-1 fees, size and age of the S&P
500 Index fund before investing. Our multivariate model provides
evidence that the expense ratios help to explain performance; that is,
lower expense ratios are strongly related to higher returns.
What are the implications of our findings? Investors often view index
mutual funds as financial commodities because of the often-held
assumption that funds tracking the same benchmark index should not
differ in any meaningful way. Our evidence of large differences in
expense ratios across retail S&P 500 Index funds casts serious doubt on
this term as a descriptor of these funds. One explanation for the
presence and enduring popularity of high-priced index funds is that some
uninformed investors are paying high costs without receiving
commensurate benefits. Another explanation is that such funds tend to
require significantly lower minimum initial purchase amounts. The
evidence suggests that despite less-than-perfect price competition among
the whole of S&P 500 Index funds, the S&P 500 Index funds with the
lowest expense ratios (Quartile 1) attract the majority of retail
investors’ funds.
Endnotes
1 Morningstar Principia Pro for
Mutual Funds Advanced Module, January 2007 (CD).
2 U.S. Securities and Exchange
Commission, “Report on Mutual Fund Fees and Expenses.” Washington, D.C.:
Division of Investment Management, December 2000.
3 Our model follows those
proposed by D. K. Malhotra and Robert W. McLeod, 1997, “An Empirical
Analysis of Mutual Fund Expenses,” Journal of Financial Research 20 (2),
175–190; and Wilfred L. Dellva and Gerard T. Olson, 1998, “The
Relationship between Mutual Fund Fees and Expenses and Their Effects on
Performance,” Financial Review 33 (1), 85–104.
4 See Ali Hortaçsu and Chad
Syverson, 2004, “Product Differentiation, Search Costs, and Competition
in the Mutual Fund Industry: A Case Study of S&P 500 Index Funds,”
Quarterly Journal of Economics 119 (2), 403–456.
This article is
reprinted with the permission of the Journal of
Indexes, which is the
publication of record for the index
industry. Any rebroadcast or
distribution of this content
requires the expressed permission of
the Journal of Indexes. All
Journal of Indexes content,
including past columns by Professor
Haslem can be accessed on
www.indexuniverse.com/JOI.
John
A. Haslem,
Professor Emeritus of Finance,
University of Maryland.
H.
Kent Baker,
University Professor of Finance,
American University, Washington, DC.
David
M. Smith,
Associate Professor of Finance,
University at Albany, SUNY.
Professor Baker is a doctoral
graduate of the Smith School and
Professor Haslem was his
dissertation advisor.
