by John C. Bogle
Founder and Former Chairman, The Vanguard Group
Journal of Portfolio Management
Spring 2002, Volume 28, Number 3
In 1997, I prepared a study1 of the
returns for the mutual funds in each of the nine Morningstar "style
boxes," a matrix with large-, mid-, and small-capitalization
funds on one axis and value, blend, and growth funds on the other.
The study, covering the five-year period 1992 through 1996, presented
powerful evidence that the low-cost quartile of funds in each box
had not only earned higher returns than those in the high-cost quartile,
but returns that significantly exceeded the cost differential.
The results can be summarized as follows: Average return of low-cost
funds, 14.9%; average return of high-cost funds, 12.3%. This difference
of 2.6 percentage points was double the 1.3% expense ratio
differential of the funds (annual expenses ratio of low-cost quartile,
0.7%; expense ratio of high-cost quartile, 2.0%). The differential
increased slightly when risk-adjusted returns were substituted
for total returns. As a result, I concluded: An investor who doesn't
seriously consider limiting selections to funds in the low-expense
group and eschewing funds in the high-expense group is someone who
should take off the blindersperhaps even a bit of a fool.
The Role of Costs
Emboldened by the magnitude and consistency across the nine style
boxes, I then asked, in effect: Since the lowest-cost funds in
the marketplace today are index funds, why not just buy index funds
in each of the style boxes? I then tested that proposition,
and I found the results equally compelling. In seven of the nine
boxes, the comparable-style index produced higher returns,
and in all nine boxes, the index funds assumed lower risks. In terms
of risk-adjusted returns, the index fund's superiority was substantial
in eight boxes, and marginally lower in but one (small-cap growth).
Holding risk constant, the indexes delivered a return surplus of
3.6 percentage points per year (16.5% vs. 12.9%) in the large-cap
group, 4.2 percentage points (18.0% vs. 13.8%) in the mid-cap group,
and 4.4 percentage points (19.5% vs. 15.1%) in the small-cap group.
Armed with this data on the relationship between fund costs and
fund performance, I concluded: "The magnitudes are so large
and so consistent as to devastate the concept of high cost active
management." Prudently, however, I added the obvious caveat:
"We should go only so far with five-year numbers in a strong
equity market . . . But a shorter period would be even less satisfactory,
and a longer (ten-year) period would cut the number of funds we
could observe by half, making for a less reliable sample. (But)
the five-year data deserves testing in other periods and under a
variety of market conditions."2
This paper will do exactly that, using the ten-year period ending
June 30, 2001.
The decade-long period from June 30, 1991, to June 30, 2001, covered
in the new study clearly included a variety of conditionsthe
quiet stock market of 1992-1994, the boom of 1995-1999, and the
subsequent bust in 2000-2001. Interestingly, however, the annual
return of the S&P 500 Stock Index was virtually the same during
the past decade (15.1%) as during the earlier study (15.2%). However,
the variation in actual returns between the best and worst style
boxes was large in the prior study: 3.2 percentage points (15.1%
to 11.9%). In the present study, the variation in average return
between the extremes was only remarkably small: 1.3 percentage points
(14.5% to 13.2%). This table presents the data:
|Annual Rate of Return*
Ten-Years Ended June 30, 2001
* Source: Morningstar. Includes 641 mutual funds
in operation throughout the period.
The hypothesis that the funds in the low-cost quartile would outperform
those listed in the high-cost quartile was again clearly validated
during this period, as this table shows.
|Annual Rate of Return
Ten-Years Ended June 30, 2001
The expense ratio differential during this period was 1.2% (0.6%
for the low-cost funds, 1.8% for the high-cost funds), a bit less
than the 1.3% spread in the prior study (0.7% to 2.0%). But the
performance differential was once again approximately double the
cost differential, 2.2% and 2.6% respectively. Each $1.00 of extra
cost, then, resulted in a loss of $1.83 of return in the ten-year
period, as compared to $2.00 in the five-year period.
Unlike the previous period, in which the risk
exposure of the high-cost funds (standard deviation, 12.2%) was
only slightly higher than for the low-cost funds (11.8%), the risk
exposure differential during the past ten years had increased sharply.
The standard deviation of the low-cost funds averaged 17.4%, vs.
20.1% for the high cost funds, a 15% greater risk exposure. As a
result, the risk-adjusted returns of the low-cost funds averaged
13.8%, versus 10.8% for the high-cost funds, raising the performance
differential to 3.0% annually during the past decade. That is, each
$1.00 of extra cost resulted in a loss of $2.50 in risk-adjusted
return. It is not possible to understate the significance of these
differences. Costs matter, and they matter even more now
than the earlier study suggested.3
The consistency of the advantage in risk-adjusted
return that low-cost funds have achieved over high-cost funds is
remarkable, as this table shows.
|Risk-Adjusted Returns 4
Ten-Years Ended June 30, 2001
The Sharpe ratio provides another way
of viewing risk-adjusted returns. In the previous study, the average
Sharpe ratio for the low-cost funds was 1.13, or 35% higher than
the 0.84 for the high-cost funds. But even that substantial difference
widened in the ten-year study. The Sharpe ratio of 0.77
for the low-cost funds compared to 0.52 for the high-cost funds,
an improvement of fully 48%.
This differential is even more consistent through the nine style
boxes than was the case in the prior study, when eight of the nine
style boxes fit the pattern. In the ten-year study the low-cost
funds demonstrated substantial superiority in all nine of the style
Ten-Years Ended June 30, 2001
||5-Years Ended Dec. 31, 1996
As a result of the powerful link between cost and return evidenced
in my earlier article, I then asked the obvious question: If costs
matter so muchas they obviously doand if index funds
are the lowest cost fundswhy not just hold index funds
that replicate each of the nine style boxes? That proved to
be a profitable avenue of exploration. Taking all mutual funds as
a group, and comparing them to a mix of comparable index funds,
the earlier results showed the following:
|Five-Years Ended December 31, 1996
* Standard deviation of returns, 1992 to 1996.
As the table shows, the Sharpe ratio
of the index funds (1.23) exceed that of the average managed fund
(0.99) by fully 24%; that of the high-cost funds (0.84) by 0.39,
or 46%; and even that of the low-cost funds (1.13) by 0.10, or 9%.
The consistency of this relationship between index funds and managed
funds throughout the nine style boxes was remarkable. In eight of
the nine boxes, the appropriate index fund Sharpe ratio exceeded
that of the average managed fund by from 0.16 to 0.46. (In the four
fund groups with the largest and therefore more statistically
significant populationsthe range was narrower +0.16 to +0.31).
Only in the small-cap growth fund segment did the small-cap growth
index fund fall short, by 0.06. (More about that group later).
The new study clearly confirms the finding of the earlier study.
During the ten-years ended June 30, 2001, the index fund advantage
was again compelling:
|Ten-Years Ended June 30, 2001
*Standard deviation of returns,
6/30/91 to 6/30/01.
The Index fund advantage over the average fund was slightly smaller
than in the previous study18% above the Sharpe ratio of the
average fund (0.79 vs. 0.67), versus 24%. The advantage increased
from 46% to 52% over that of the high-cost funds, (0.79 versus 0.52),
but declined from 9% to 4% above that of the low-cost funds
(0.79 versus 0.77).
|Sharpe Ratio: Index Funds vs. Managed
Ten-Years Ended June 30, 2001
||5-Years Ended Dec. 31, 1996
Once again, the index funds prevail over active managers, albeit
at somewhat lower margins of advantage. The uniformity of advantage
is striking, with the index funds providing higher risk-adjusted
returns in eight of the nine style boxes, with the sole exception
being the apparent superiority of active managers in the
small-cap growth category, as evidenced in both the present and
Summing up the Studies
It is highly significant that the ten-year study so powerfully
echoes and reinforces the findings of the five-year study. Once
again, low-cost funds outpace high-cost funds. Once again, costs
matter even more than we expect (i.e., a 1% reduction in costs generates
an increase in risk-adjusted return that is much larger than 1%).
Once again, index fundsthe fund category with the lowest costsgive
an excellent account of themselves. The earlier study concluded:
(1) higher returns are directly associated with lower costs;
(2) the notion that indexing works only in large capitalization
markets no longer has the ring of truth. Both conclusions are
reinforced in the current study.
Mutual Fund Returns are Consistently Overstated
The consistency of the data in the two studies is impressive.
But however one regards their validity, it must be recognized that
the average returns of the actively managed mutual fund that
I have presented are significantly overstated. First and foremost,
they are survivor-biased. Only the funds that lived through the
decade to report their performance at the close of the period are
included in the sample. The 634 funds for which Morningstar reported
ten-year records represent the survivors of an estimated 890 funds
which began the decade. The records of the remaining 256 funds are
lost in the dustbin of history. It is reasonable to postulate that
the poorer performers dropped by the wayside, thereby biasing the
study results in favor of the manager.
How much bias? We can't be sure. Independent
studies confirm that survivor bias is substantial. In the Carhart
and Malkiel studies5, survivor bias
ranged from 1.5% to 3.1% per year. If we were to assume a bias of
2% during the ten-year period ended June 30, 2001 (larger for each
of the small-cap groups, smaller for the large-cap groups), the
annual risk-adjusted return of the average managed fund would drop
to from 12.5% to 10.5%, a 3.9 percentage point shortfall to the
14.4% return of the total stock market, more than double the active
fund shortfall of 1.9% suggested in these data. When they fail to
acknowledge the role of survivor bias in the data, studies that
purport to show that indexing doesn't work leave much to be desired.
Several years ago, in another study, Morningstar estimated the
survivor bias for each of its style boxes over the five-year
period 1992-1996. Even in that relatively short period, the bias
was equal to almost 1% per year. Interestingly, in the light of
my earlier finding that only small-cap growth funds had succeeded
in outpacing their target index, the annual survivor bias in that
style box was 1.7%. If we assume, for the purpose of argument, that
the (necessarily larger) ten-year bias was 3.0% per year, the data
showing a 1.7% percentage point annual advantage over the
index for small-cap managers becomes a 1.3% disadvantage.
Some Fund Returns are Inflated
Second, even the records of those funds that do survive
are to some degree suspect. It is hardly without precedent for small
funds, often those run by large advisers, to inflate their records
by purchasing IPOs, quickly "flipping" them and generating
returns that do not recur when the fund gets large. Two managers
have been fined by the SEC for this practice: One managed a fund
which reported a 62% return for 1996, an excess return largely accounted
for purchasing just 100 to 400 shares of 31 hot IPOs. The other
rose 119% during the 18 months following its initial offering, 83
percentage points of which came from first-day gains realized
on newly-public stocks. In yet another case, a fund advertised (in
bold-face type) a 196.88% return in 1999, acknowledging (in small
print) that a significant portion came from IPOs. Yet these records
are included in the industry data as if they were the holy writ.
Third, actively-managed funds surrender a substantially larger
portion of their pre-tax performance to taxes, in an amount that
could have increased index fund superiority by as much as another
1.5% per year or more during the past decade. The 13.7% pre-tax
annual return reported by the average mutual fund fell to an after-tax
return of 11.1%, a loss of fully 2.6 percentage points to taxes.
Since only one index fund has operated during the entire past decade,
after-tax style-box returns for the indexes are not available. But
the largest S&P 500 Index fund carried a tax burden of just
0.9%far less than the tax burden for the average fund. Ignoring
taxes represents one more overstatement of fund returns by most
studies of manager performance.
Fourth, fund sales charges are ignored in most fund comparisons,
including the data used in this study. Nonetheless, sales charges
represent a hidden reduction in reported returns. If we assume that
a decade ago three-quarters of all funds carried an average initial
sales charge of 6%, the cost, amortized over the ten years, would
reduce returns reported by funds by another 0.5% annually. The high
turnover of fund shares by investors, however, indicates that the
average holding period is no more than five years. Thus, the actual
reduction in annual return engendered by sales charges would be
significantly larger than 0.5%, another substantial reduction in
the return of managed funds.
When we consider all of these factors, it must be clear that,
whatever relationship exists between style-box returns in managed
funds and index funds, the reported returns of managed funds are
significantly overstated. Nonetheless, even accepting the overstated
fund data as presented, mutual funds as a group, style-box by style-box,
with only one exception, fall well short of their index fund benchmarks,
largely as a result of the costs they incur. Index funds win.
The Data vs. the Facts
One might say: So what else is new? For it must be obvious that
if we take all stocks as a group, or any discrete aggregation of
stocks in a particular style, an index that owns all of those stocks
and precisely measures their returns must, and will, outpace the
return of the investors who own that same aggregation of
stocks but incur management fees, administrative costs, trading
costs, taxes, and sales charges. Active managers as a group will
fall short of the index return by the exact amount of the costs
the active managers incur. If the data we have available to us
do not reflect that self-evident truth, well, the data are wrong.
There are infinite ways in which the data can mislead. We count
each mutual fund as a unit in calculating average returns, while
the industry's actual aggregate record is reflected only in an asset-weighted
return. Funds rarely stay rigidly confined to their style boxes;
a growth fund may own some value stocks; a small-cap fund may own
mid-cap and large-cap stocks.
Of course, it is at least theoretically possible that mutual fund
managers as a group may be smarter than other investors, and in
fact consistently outpace the market by an amount sufficient to
overcome their substantial costs. But let's think about that. Is
it realistic to believe that fund managers whoincluding the
pension accounts they managecontrol the investment process
applicable to upwards of 35% of the value of all U.S. equities,
can outpace other managers, advisers, and individuals? For example,
for fund managers to outpace the market by 1% annually after costs
of, say, 2% (excluding taxes) would require an excess return of
3%. In that case, all other investors as a group would then lose
to the market by about 2% per year, or by 4% after costs. In reasonably
efficient markets such as those in the U.S., where prices are set
largely by professional investors, such a gap would seem inconceivable.
Further, the available data showing returns earned by individual
investors give every indication that, like institutions, individuals
match the market before costs and lose to the market after costs,
a conclusion that would surprise no one who has ever examined performance
data with care.
Even someone who has never plied the fund performance seas must
understand this central fact of investing: Investment success
is defined by the allocation of financial market returnsstocks,
bonds, and money market instruments alike between investors
and financial intermediaries. Despite the elementary, self-evident,
and eternal nature of this capital market equationgross return
minus cost equals net returnthe dialogue between advocates
of indexing and advocates of active management continues unabated,
for there is a lot of money at stakecertainly well over $100
billion per year. Mutual fund direct costs alone (excluding
sales charges and transaction fees) account for some $70 billion;
fund trading costs likely account for an additional $50 billion
The reality is that the horses ridden by the mutual fund jockeys
are handicapped with so much weight that the entire fund industry
cannot possibly win the race for investment success. Given the limitations
on the data available that I have noted above, of course, if one
searches long enough and hard enough, one can possibly identify
interim periods in which the equation will appear to be disproven.
But the reality is what it is. While there can be debate
over the figures, there can be no debate over the facts: For invesors
in the aggregate, the capital market equation is unyielding. Yes,
some managed funds can, and some do, outpace the indexes, but there
is no sure way to identify them in advance.
Indexing and Market Efficiency
There is one more misconception that needs to be put to rest.
It comes in this form: "If (Bogle) is right (about the role
of cost and the superiority of indexing), he will be wrong; and
if he is wrong, he will be right. The more people become convinced
they can beat the market (i.e, Bogle is wrong) the more efficient
the markets become as more intelligent and capable professionals
enter the market. Ironically, it then becomes less likely they will
outperform it. (But) if managers and investors come to believe active
management is a waste of money (i.e., Bogle is right), money managers
will be replaced by index funds. This will reduce the number of
market participants, and hence worsen market efficiency. The
remaining minority of active managers will then have a better chance
of outperforming their respective markets."6
Alas, that allegation does not meet the test of simple logic.
Whether the markets are efficient or not, as long as the index reflects
the performance of the market (or any given segment of the market),
it follows that the remaining participants (largely active managers)
will also earn the market return (or market segment return) before
their intermediation costs are deducted. Syllogism: (1) All
investors as a group earn the market return. (2) Index funds earn
the market return. Therefore: (3) All non-index investors earn the
market returnbut only before their costs are deducted. Result:
The substantial costs of financial intermediaries doom active investors
as a group to inferior returns.
Admittedly, if our markets turn inefficientsomething
that is hard to imagine in these days of infinite informationthe
"good" managers may be able to enlarge their edge over
"bad" managers. But it must be self-evident, that in effect,
each manger who succeeds in outpacing the stock market by, say,
4% per year before costs over a decade, must be balanced by another
who falls short by 4%, again before costs. Efficient markets or
inefficient, active managersgood and bad togetherlose.
Such is the nature of financial markets.
(requires Adobe Acrobat
1. "The Implications of Style Analysis
for Mutual Fund Performance Evaluation," Journal of Portfolio
Management, [Summer, 1998]. Back
2. Bogle . Recently, Dylan Minor,
financial advisor and vice president at Morgan Stanley in Santa
Barbara (CA, 93101) responded to my challenge by presenting data
for the 1992-1996 period that seemed to contradict my conclusions,
"Beware of Index Fund Fundamentalists" [Summer, 2001].
3. One explanation for this leverage effect,
in which the performance shortfall bears a 2½:1 ratio to
cost may be higher portfolio turnover. The annual turnover of the
high-cost funds averaged 98%, more than 50% higher than the 63%
turnover of the low-cost funds. Back
4. Calculation method described in Modigliani,
Franco and Leah Modigliani, "Risk-Adjusted Performance,"
Journal of Portfolio Management, [Winter, 1997]. For this study,
author used style-specific benchmarks to calculate risk-adjusted
5. Malkiel, Burt, 1995, "Returns from
Investing in Equity Mutual Funds 1971 to 1991," Journal of
Finance, Vol. L, No. 2.
Carhart, Mark, Jennifer Carpenter, Anthony Lynch and David Musto,
2001, "Mutual Fund Survivorship," working paper, NYU Stern
School of Business. Back
6. Minor Journal of Portfolio Management, [Summer,
2001]. Emphasis is author's. Back
Note: The opinions expressed in this article do not necessarily represent the views of Vanguard's present management.
to Speeches in the Bogle Research Center
©2006 Bogle Financial Center. All Rights Reserved.