You are on page 1of 11

The Value of Active Portfolio Management

Ravi Shukla
Finance Department
Syracuse University
Syracuse, NY 13244-2130
(315) 443-3576
First draft: June 1999
This revision: August 1999

Please do not quote. Comments are welcome. I am grateful for helpful comments from
Fernando Diz.
The Value of Active Portfolio Management
Abstract
We calculate the value of active portfolio management to mutual fund
investors as the dierence between the returns mutual funds provide to
their investors and the returns their portfolios would have earned had
there been no activity in the portfolio. We nd that the mutual fund
investors, on an average, lose 0.13% per month due to active management.
Actively managed mutual funds analyze the markets and revise their port-
folios continuously in response to their analysis. These revisions cause activity
and turnover in the portfolios. Portfolio activity may also arise due to other rea-
sons. Anecdotal evidence suggests that mutual funds may hold some securities
during the most of the time and may alter their portfolios to include securities
their investors may want to see in the portfolio for the quarterly or semiannual
reports. Such window-dressing leads to additional turnover. Another source
of activity in the portfolio is the ow of funds due to net purchase or sale of
mutual fund shares by the investors. Mutual funds charge higher fees for the
analysis and expenses associated with portfolio turnover. The objective of this
paper is to estimate the value of active portfolio management to the mutual
fund investors net of the fees paid for active management.
Majority of the literature on mutual fund performance evaluation concludes
that the fees paid by the investors for active management do not provide any
benet to the investors.
1
Recent evidence suggests that some active manage-
ment may be valuable to the investors. Keim (1989) nds that the 910 Fund
from Dimensional Fund Advisors provided a 2.2% annual premium over the
CRSP 910 Index on which the fund is based. He concludes that investment
strategy and trading rules components of the funds design contributed to this
premium. Chen, Jegadeesh, and Wermers (1999) show that stocks purchased by
mutual funds have signicantly higher returns than the stocks sold by them, and
that funds that have high turnover also have high stock selection skills. Ang,
Chen, and Lin (1998), on the other hand, nd that higher returns are most
reliably achieved by lower expenses and that there is only a weak link between
stock selection and subsequent returns.
Conclusions of performance evaluation literature are primarily based on the
risk-adjusted performance measures calculated using returns reported by mutual
funds. Calculating risk-adjusted performance measures is subject to numerous
1
A summary of the literature on performance evaluation is available in Shukla and Trzcinka
(1992).
1
pitfalls
2
. Cornell (1979) proposed a performance evaluation methodology that
is not subject to many of the criticisms. Cornells methodology assesses selectiv-
ity skills of fund managers by determining the change in portfolio weights with
respect to the ensuing returns on the securities: An informed manager will in-
crease weights on securities that perform well subsequently and decrease weights
on securities that perform poorly. Consequently, a positive covariance between
portfolio weights and security returns is an indicator of superior performance.
Cornells methodology has found only a limited application because of the
lack of availability of portfolio composition data.
3
Grinblatt and Titman, in a
series of papers, have used portfolio based measures of performance. In Grin-
blatt and Titman (1989a), they calculate a time series of hypothetical returns for
mutual funds as the weighted average return on the equity portion of the port-
folio and estimate the transaction costs incurred by the mutual fund investors
as the dierence between the Jensen measures calculated using the hypothetical
and actual returns. In Grinblatt and Titman (1993), they propose a portfolio
change measure for performance evaluation. The portfolio change measure is in
the spirit of Cornells measure and does not require benchmarks for risk adjust-
ments. Daniel, Grinblatt, Titman, and Wermers (1997) measure performance
by comparing fund returns with benchmarks constructed to match the salient
characteristics of the mutual fund portfolios.
We calculate the value of active portfolio management using a portfolio com-
position based approach similar to that of Grinblatt and Titman (1989a). Unlike
Grinblatt and Titman, who calculate a measure of transaction cost incurred by
the investors by taking the dierence between the Jensen alphas based on the
passive (hypothetical) and reported (actual) returns, we calculate the value of
active portfolio management to the mutual fund investors as the dierence be-
tween the reported and passive returns,
Our results show that the value of active management is 0.13% per month.
In other words, mutual fund mutual fund investors pay 0.13% per month in
fees than the performance they receive from active portfolio management. This
nding is consistent with the extant evidence on mutual fund performance.
1 The Value of Active Management
Suppose at time point t, a portfolio consists of n securities (including cash) with
the portfolio weights X = {x
1
, x
2
, . . . , x
n
}. If the returns on these securities
during a period following time point t are R = {r
1
, r
2
, . . . , r
n
}. Then the
passive return on the portfolio, without any revision during the period, would
be:
2
As summarized in Grinblatt and Titman (1989b), the major problems arise due to issues
related to the appropriateness of benchmark and the role of timing ability.
3
Mutual funds are required to report their portfolios semiannually. Many mutual funds
report their portfolio compositions on a quarterly basis. Only recently have these data been
available in easy, machine readable form.
2
r
P
p
= X

R. (1)
Suppose the reported return on the portfolio during this period, net of man-
agement fees and brokerage expenses charged, is r
R
p
, and the fees associated with
portfolio revision activity during the period is e, then the gross actual return
earned by the fund is:
r
G
p
= r
R
p
+ e. (2)
We caution that e is not the same as the expense ratio reported by the mutual
funds. The expenses in e consist of:
(a) A fraction of the reported expense ratio associated with the day-to-day
analysis and turnover. The reported expenses usually does not include
brokerage fees.
(b) The brokerage fees and other expenses directly associated with the resulting
turnover. Livingston and ONeal (1996) and Fortin and Michelson (1998)
nd that mutual fund brokerage costs average 0.25% to 0.30% of assets.
The total value generated as a result of active management is simply the
dierence between the gross return and the passive return:
V = r
G
p
r
P
p
= r
R
p
+ e r
P
p
. (3)
The total value generated by active management, V , is unobservable since the
associated expenses (e) charged by the mutual fund management are not ob-
servable.
The value of active management to the mutual fund investors is the value
generated by active management less the fees paid by the investors:
V
I
= V e = r
R
p
r
P
p
. (4)
Equation (4) simply states that the value of the day-to-day active portfolio
management to the investors of the actively managed portfolio is the dierence
between the reported returnthe return realized by the investorsand the
passive returnthe return investors would have realized if the managers had
left the portfolio alone.
2 Data
We calculate the value of active portfolio management to the mutual fund in-
vestors by applying equation (4) in the month following the date of a known
portfolio composition.
To compute the passive return (r
P
p
), we need the portfolio composition and
the realized returns on the securities in the portfolio for the month following the
portfolio date. The portfolio compositions are obtained from quarterly updated
Morningstar Principia CD ROMs. The quarterly updates are released in April,
3
July, October and January, and contain information as of the end of the March,
June, September and December, respectively. Our collection of Morningstar
Principia databases begins in April 1995 (the rst Principia released) and ends
in January 1999. Our primary source for security returns is the Center for
Research in Security Prices (CRSP) stock returns le which contains returns
for stocks listed and traded in the United States. Therefore, our mutual fund
portfolios are restricted to include U.S. stocks only.
We use the various quarterly updated Morningstar Principia databases to
create lists of mutual fund portfolios composed of cash and 100% U.S. equity
positions
4
at the end of various quarters.
To determine the compositions of these portfolios, we utilize a Morningstar
Principia eld that contains the date of the portfolio composition for mutual
funds on that disk.
5
For example, the 1999 Quarter 1 disk contains a value of
1997-09-30 in the Portfolio Date eld for AIM Mid Cap Equity A (Ticker
GTAGX). Therefore, the 1999 Quarter 1 disk contains the portfolio composition
as of September 30, 1997 for AIM Mid Cap Equity A mutual fund. This lag
between the portfolio date and the database date occurs because the mutual
funds usually release the portfolio compositions with a lag of a few months.
By comparing the two lists created above we identify the compositions for
481 portfolios. The stocks in the mutual fund portfolios are carefully matched
with the stocks in CRSP database.
6
Some mutual fund portfolios include stocks
not in the CRSP database (most of these are listed on the NASDAQ bulletin
board). For many of these stocks, we are able to obtain returns from the Yahoo
Finance web site (http://finance.yahoo.com).
The cash position in the portfolio is assumed to earn a rate of return on the
90-day U.S. Treasury bill. The returns on 90-day Treasury bills are obtained
from Morningstar.
We are able to obtain the returns for all the stocks in the portfolios for 342
cases. These 342 cases include some clear outliers in terms of fund size, portfolio
turnover, and expense ratio. We remove these outliers from the sample to arrive
at a more homogenous nal sample of 323 cases. In this paper, we report the
results for these 323 cases.
7
The reported returns (r
R
p
) for the mutual funds in the month after the date
of known portfolio composition are obtained from the various Morningstar Prin-
cipia databases.
4
We do not require that the fund objective should be to invest in U.S. equity.
5
This eld is available starting with 1997 Quarter 1.
6
This is a very time consuming task since stock names in Morningstar rarely match those
in CRSP. Most stocks were matched using ticker symbol, but matching by ticker symbol is
susceptible to errors since tickers are reassigned. All matches were carefully scrutinized to
ensure accuracy. Any doubt about a match resulted in the elimination of the mutual fund
portfolio from the sample. Also, many portfolios were dropped because they included bonds
and non-U.S. stocks despite the indication of 100% U.S. equity.
7
The removal of outliers did not inuence the nal inferences. The results for the full
sample are available upon request.
4
Table 1: Details of the Mutual Fund Portfolio Sample
This table shows the descriptive details of our sample of mutual fund portfolios. Panel
A shows the distribution of portfolios by the portfolio dates, Panel B shows the number
of repeated occurrences mutual funds have in our sample, and Panel C shows the
distribution of the fund portfolios by their stated objectives.
Panel A
Portfolio dates
Date Freq
30-Sep-95 2
31-Dec-95 1
31-Mar-96 9
30-Jun-96 26
30-Sep-96 42
31-Dec-96 58
31-Mar-97 68
30-Jun-97 56
30-Sep-97 61
323
Panel B
Cases by funds
Funds Cases
79 1
46 2
24 3
10 4
8 5
167
Panel C
Stated objectives
Objective Freq
Aggressive Growth 13
Asset Allocation 5
Equity-Income 8
Growth 167
Growth and Income 46
Small Company 52
Specialty 6
SpecialtyFinancial 4
SpecialtyReal Estate 18
SpecialtyUtilities 4
323
3 Results
Table 1 provides details of the mutual fund portfolio sample. Panel A of the
Table shows that the sample consists of 323 cases between 30-Sep-95 and 30-
Sep-97, with most of the cases coming after 30-Jun-96. Each case represents
a mutual fund portfolio on a particular date. Panel B shows the distribution
of cases by funds. The sample consists of a total of 167 unique funds. 79 of
these funds appear only once in the sample, 46 funds have two cases for two
distinct portfolio dates, 24 funds have 3, etc. Finally, Panel C shows that the
fund objectives include a wide range. The majority of the cases, however, are
concentrated in Growth, Growth and Income, and Small Company.
Table 2 shows selected attributes of the mutual fund sample. Our sample
represents a wide range of mutual funds from as small as $0.10 million in assets
to almost $6 billion. The median turnover in our sample is 52% though the
sample includes funds with signicantly higher levels of turnover. The median
expense ratio (Exp) of 1.25% per year is representative of the mutual fund
industry. The median cash position of our portfolios is 4.86%. The median
number of securities is 47 and ranges from 15 to 172. The total weight in the
top 10 securities (Top10Wt), which is an indicator of diversication, is about
30%.
Table 3 shows the results of our analysis. The average passive return during
5
Table 2: Attributes of the Mutual Fund Portfolio Sample
This table shows selected attributes of the 323 mutual fund portfolios in our
sample. Assets are measured in millions of dollars. Turnover is expressed in %
per year. Exp is the expense ratio in % per year. Cash is % of portfolio held as
cash. NSec is the number of securities in the portfolio. Top10Wt is the weight of
the top 10 securities in the portfolio.
Assets Turnover Exp Cash NSec Top10Wt
Maximum 5931.20 706.00 4.23 37.46 172.00 91.44
Minimum 1.20 3.00 0.38 0.00 15.00 8.17
Median 56.60 52.00 1.25 4.86 47.00 30.70
Average 172.05 70.23 1.35 6.83 55.56 32.68
Std Dev 455.93 78.38 0.45 6.87 28.84 12.62
the month following the reported portfolio composition is 1.63%. The average
reported return is 1.50%. The value of active management to the investors is
0.13% with a t-statistic is 4.96 which is statistically signicant at 5% level.
A simple interpretation of this number is that the investors paid 0.13% per
month more for active portfolio management than what they received. These
measure of the value of active management is consistent with Grinblatt and
Titman (1989a) who estimate the transaction cost incurred by the mutual fund
investors to be between 0.09% per month and 0.20% per month, depending on
the benchmark(s) used for risk adjustment.
As the following regression equation shows, the reported returns track the
passive returns very closely:
r
R
p
= 0.00120 +0.99393 r
P
p
R
2
= 0.9887 F = 28122.794
(4.30) (167.70)
Figure 1 shows that the value of active management has a symmetric, uni-
modal distribution with most of the observations lying between 1.00% and
0.75%. The highest and the lowest values are 1.91% and 3.41%, respectively.
Of the 323 portfolios, 117 (36%) provided a positive value their investors while
206 (64%) provided a negative value. It is often argued that mutual funds that
charge higher fees have better managers and processes, and provide better per-
formance to their investors. To investigate the relationship between the value
of active management to investors and the fees paid for active management, we
estimate the following simple regression:
V
I
= 0.00089 0.00031 Exp R
2
= 0.0008 F = 0.277
(1.08) (0.53)
Clearly, there is no meaningful relationship between expenses and the value of
active management to the investors.
6
Table 3: The Value of Active Portfolio Management
This table shows the statistics for passive (r
P
p
) and
reported (r
R
p
) returns, and the value of active man-
agement to the mutual fund investors (VI) on of the
323 mutual fund portfolios in our sample.
r
P
p
r
R
p
VI
Maximum 15.49 15.63 1.91
Minimum 11.61 11.05 3.41
Median 1.92 1.76 0.09
Average 1.63 1.50 0.13
Std Dev 4.44 4.44 0.47
t 6.61 6.08 4.96
Figure 1: Frequency Distribution of the Value of Active Portfolio Management
0
5
10
15
20
25
30
35
-
2
.
0
0
%
-
1
.
5
0
%
-
1
.
0
0
%
-
0
.
5
0
%
0
.
0
0
%
0
.
5
0
%
1
.
0
0
%
1
.
5
0
%
2
.
0
0
%
Net Benefit
F
u
n
d
s
7
Table 4: The Value of Active Portfolio Management for Vari-
ous Stated Objective Groups
This table shows the value of active management for the various
stated objective groups in the sample.
Stated Objective N Avg Std t
Aggressive Growth 13 0.06 0.40 0.56
Asset Allocation 5 0.04 0.27 0.37
Equity-Income 8 0.04 0.17 0.63
Growth 167 0.15 0.48 4.13
Growth and Income 46 0.16 0.39 2.78
Small Company 52 0.18 0.52 2.52
Specialty 6 0.03 0.40 0.16
SpecialtyFinancial 4 0.15 0.22 1.33
SpecialtyReal Estate 18 0.15 0.48 1.31
SpecialtyUtilities 4 0.19 0.06 6.85
Figure 4 shows the average, standard deviation and t-statistics for the value
of active management for various stated objective groups. Only Equity-Income
and SpecialtyReal Estate funds have positive values, though these are not
statistically signicant. Growth funds, which constitute a large component of
the mutual fund universe as well as our sample, generate a statistically signi-
cant loss of 0.15% due to active management.
4 Conclusion
By comparing the returns realized by the mutual fund investors and the po-
tential passive returns on the mutual fund portfolios we nd that the mutual
fund investors, on an average, lose 0.13% per month due to active portfolio
management.
References
[1] Ang, James S., Chen, Carl R., and Lin, James Wuh, 1998, Mutual fund
managers eorts and performance, Journal of Investing, Volume 7, Num-
ber 4 (Winter), Pages 6875.
[2] Chen, Hsiu-Lang, Jegadeesh, Narasimhan, and Wermers, Russ, 1999. The
value of active mutual fund management: An examination of the stock-
holdings and trades of fund managers. Unpublished Manuscript.
8
[3] Cornell, Bradford, 1979, Asymmetric information and portfolio perfor-
mance measurement, Journal of Financial Economics, Volume 7, Number 4
(July), Pages 381390.
[4] Daniel, Kent, Grinblatt, Mark, Titman, Sheridan, and Wermers, Russ,
1997, Measuring mutual fund performance with characteristic-based bench-
marks, Journal of Finance, Volume 52, Number 3 (July), Pages 10351058.
[5] Fortin, Rich and Michelson, Stuart, 1998, Mutual fund trading costs, Jour-
nal of Investing, Volume 7, Number 1 (Spring), Pages 6670.
[6] Grinblatt, Mark and Titman, Sheridan, 1989a, Mutual fund performance:
An analysis of quarterly portfolio holdings, Journal of Business, Volume 62,
Number 3, Pages 393416.
[7] Grinblatt, Mark and Titman, Sheridan, 1989b, Portfolio performance eval-
uation: Old issues and new insights, Review of Financial Studies, Volume 2,
Number 3 (Fall), Pages 393421.
[8] Grinblatt, Mark and Titman, Sheridan, 1993, Performance measurement
without benchmarks: An examination of mutual fund returns, Journal of
Business, Volume 66, Number 1 (January), Pages 4768.
[9] Keim, Donald B., 1989, An analysis of mutual fund design: The case of
investing in small-cap stocks, Journal of Financial Economics, Volume 51,
Number 2 (1 February), Pages 173194.
[10] Livingston, Miles and ONeal, Edward S., 1996, Mutual fund brokerage
commissions, Journal of Financial Research, Volume 19, Number 2 (Sum-
mer), Pages 273292.
[11] Shukla, Ravi K. and Trzcinka, Charles A., 1992, Performance measurement
of managed portfolios, Financial Markets, Institutions & Instruments, Vol-
ume 1, Number 4, Pages 158.
9

You might also like