SlideShare a Scribd company logo
1 of 19
Download to read offline
The Journal of Financial Research • Vol. XXIX, No. 3 • Pages 349–366 • Fall 2006


MUTUAL FUND PERFORMANCE PERSISTENCE AND COMPETITION:
               A CROSS-SECTOR ANALYSIS


                                          Aneel Keswani
                                       Cass Business School

                                            David Stolin
                                    Toulouse Business School



                                             Abstract

          Existing work on mutual fund performance persistence obtains diverse results,
          depending on the group of funds studied. We examine whether performance per-
          sistence within a peer group of competing mutual funds depends on the group’s
          composition. The U.K. mutual fund industry is ideal for such an examination be-
          cause funds compete within strictly defined sectors. We consider several attributes
          related to the intensity of competition within a sector and use them to explain
          sector-level persistence. We find robust evidence that persistence is higher in
          sectors where concentration of assets under management is higher.

          JEL Classification: G23


                                         I. Introduction

It is well established in the industrial organization literature that the structure of
a sector affects its competitiveness. In more competitive sectors we expect to see
few firms doing persistently well and those performing poorly being forced to exit
the sector. This reasoning is tested by Waring (1996), who finds a strong negative
relation between competitiveness within an industrial sector and the persistence of
profitability for firms in that sector.
         We translate this logic to the mutual fund context. To use the terminology of
industrial organization, mutual funds compete with each other using a combination
of price and nonprice competition strategies. Price competition involves funds

      We thank Vladimir Atanasov, Andrew Clare, Zsuzsanna Fluck, Gordon Gemmill, Brian Kluger,
Tobias Kretschmer, Gordon Midgley, Kenneth Moon, Dennis Stanton, Dylan Thomas, Giovanni Urga, and
especially William T. Moore (former editor) and Jonathan Fletcher (the referee) for insightful comments.
We also acknowledge comments received from participants at the 2003 Financial Management Association
meeting in Denver, and seminars at Cass Business School and the universities of Porto, Reading, Warwick,
and Oxford. We thank Benjamin Kogan, Jan Steinberg, James Sullivan, the Allenbridge Group, and the
Investment Management Association for help with data. Part of the research reported here was conducted
while Stolin was visiting at the Stockholm Institute for Financial Research. All errors and omissions are
ours.

                                                  349
350                    The Journal of Financial Research

varying the fees they charge to obtain a competitive advantage. Nonprice com-
petition involves (among other things) funds competing to produce superior invest-
ment performance. Numerous studies show that higher investment returns have an
important influence on fund market share (e.g., Siggelkow 2003).
         We expect funds from more competitive sectors to compete more aggres-
sively for abnormal returns. This should result in the exit of funds that underperform
and a low probability of remaining funds doing repeatedly well. Competing funds
should be able to close the performance gap on “star” funds by devoting more
resources to researching investment opportunities, by learning to imitate the best
performers, or even by poaching their managers. Thus, in more competitive sectors
we expect to see less persistence in funds’ performance relative to their rivals (i.e.,
less relative persistence).
         To examine the influence of competition on investment performance per-
sistence, we focus on the U.K. unit trust (open-ended mutual fund) industry. This
environment is ideal for our purpose because U.K. mutual funds compete in a
large number of unambiguously defined peer groups (sectors), whose membership
is monitored and enforced by the industry trade body. This is unlike the United
States, where multiple sector definitions coexist and managers are free to game
their sector affiliations (Cooper, Gulen, and Rau 2005).
         Although research into fund performance persistence has a long history,
Brown et al. (1992) show that early studies exaggerate the extent of persistence by re-
lying on survivorship-biased data sets. Carhart (1997) finds that in his survivorship-
free sample of U.S. equity funds, persistence largely disappears after accounting
for momentum in stock returns. However, recent studies argue that after properly
considering fund styles, there is persistence in U.S. equity mutual funds (Ibbotson
and Patel 2002; Teo and Woo 2001; Wermers 2003).
         Outside the United States, there has been debate as well. In the United
Kingdom, it has involved academics (Blake and Timmermann 1998; Allen and Tan
1999; Fletcher and Forbes 2002), practitioners (Quigley and Sinquefield 2000),
the trade association (Giles, Wilsdon, and Worboys 2002), and the regulatory body
(Rhodes 2000; Blake and Timmermann 2003). This literature agrees that perfor-
mance persistence is an important issue but disagrees on whether and to what extent
persistence is present.
         The preceding studies all focus on funds investing in domestic equity secu-
rities. The availability of well-accepted benchmarks for risk adjustment is a major
reason for this focus. Several studies examine persistence for funds investing in
other asset classes (e.g., see Blake, Elton, and Gruber 1993 for bond mutual funds)
and obtain diverse results, depending on the period used and the fund sector stud-
ied. This raises the possibility that levels of persistence may vary depending on the
economic circumstances. In particular, the market structure of a mutual fund sector
may influence funds’ ability to perform consistently.
         Our research examines how mutual fund performance persistence at the
fund sector level is influenced by competition within the sector. A few studies
Mutual Fund Performance Persistence                                           351

consider determinants of persistent performance at the individual fund level (e.g.,
Volkman and Wohar 1995). Several other studies note persistence differences across
sectors or fund objectives (Blake and Timmermann 1998; Kosowski et al. 2003;
Wermers 2003). No study, however, conducts a sector-level statistical analysis of
persistence, and none investigates the effect of competition on persistence. Massa’s
(2003) empirical demonstration that sector-level variables related to competition ex-
plain sector-level performance suggests that such an analysis is potentially fruitful.
         We construct several variables to capture the intensity of competition in
a sector. These include the number of funds in a sector, the proportion of mature
funds, and the Herfindahl index of asset concentration. We find robust evidence that
persistence is higher in sectors where concentration of assets under management
is higher. Our results suggest that the degree of persistence exhibited by a sector’s
investment managers depends on how competitive that sector is.


                                       II. Data and Method

The U.K. Mutual Fund Industry
Unlike in the United States, a survivorship-bias-free electronic database of mutual
funds does not exist in the United Kingdom. To conduct our study, we therefore
manually collected data from 11 consecutive editions of the annual Unit Trust
Yearbook.1 Our data span from 1991 to 2001 and include names of funds and their
management groups, annual returns (including reinvested income and excluding
fees), fund assets under management, launch dates, and of course the name of the
sector to which each fund belongs. We use fund names and an index of name changes
to link fund data across years. We consider mergers between funds as creating a
new fund.
         As our primary focus is at the sector level (rather than at the individual
fund level), we track the evolution and membership of official fund sectors as
defined by the Association of Unit Trust and Investment Funds (AUTIF) and by its
successor, the Investment Management Association (IMA).2 To do this, we use data
on fund movement across sectors, as well as historical announcements by AUTIF
and IMA. Appendix A summarizes the history of U.K. mutual fund sectors. We use


     1
       In the editions corresponding to 2000 and 2001 year-ends, the yearbook had a new publisher, and
several smaller fund families did not supply information on their funds. However, there is no survivorship bias
due to selective reporting of funds. Post-2001 data are unavailable as the yearbook has been discontinued.
     2
       In the United Kingdom, all information providers use the official classification system. The IMA
enforces its sector definitions, and if the asset allocation of a fund contravenes the allocation rules of its
current sector, the IMA will warn the fund to change its allocation if it does not wish to change sectors. If
the fund does not comply, the IMA will move the fund to a new sector reflecting its new asset allocation.
By contrast, in the United States there is a proliferation of methods for assigning funds to a peer group.
This ambiguity allows fund managers to “game” their objectives (Cooper, Gulen, and Rau 2005) and makes
objective-level measures of competition less meaningful.
352                     The Journal of Financial Research

official sector descriptions to group sectors into four broad categories: domestic
equity, global equity, domestic nonequity, and global nonequity. The appendix paints
a picture of substantial innovation at the sector level—with numerous instances
of sectors being opened, discontinued, redefined, or merged—consistent with an
industry seeking to respond to changing conditions. Additionally, there is much
variability in the number of funds within a sector. Our sample period thus captures
an industry in transition, which is helpful for our analysis of the role of market
structure characteristics.


Measurement of Persistence
Measures of performance persistence quantify to what extent performance in one
period (the “ranking” period) continues into the subsequent period (the “evaluation”
period). In this study, we focus on persistence at the one-year frequency (i.e., our
ranking and evaluation periods each equals one year). There are several reasons
for this choice. First, researchers who find evidence of persistence generally find
it for one-year horizons. Second, investors and fund managers tend to evaluate
performance over annual periods. Third, tests of performance persistence require
return availability for both ranking and evaluation periods. This leads to a look-
ahead bias, which can influence how much persistence is detected (Brown et al.
1992; Ter Horst, Nijman, and Verbeek 2001). Lengthening the horizon over which
persistence is measured makes this bias more severe. Over one-year periods, Ter
Horst, Nijman, and Verbeek (2001) find the look-ahead bias to be negligible.
          Performance persistence can be measured using both absolute and relative
performance. We measure persistence using relative performance for two reasons.
First, existing research highlights that in determining mutual fund money flows,
relative performance matters beyond absolute performance. Second, measures of
absolute performance persistence depend on the volatility of securities invested
in by a given sector, making comparisons of absolute persistence across sectors
misleading.
          To measure relative performance persistence, we use raw and not risk-
adjusted returns. In our context, examining persistence on a risk-adjusted basis is
problematic for two reasons. First, we do not have access to monthly returns for
existing and extinct U.K. mutual funds. Second, and more important, the quality of
any risk adjustment would inevitably vary across sectors. For example, domestic
equity returns can be analyzed with well-researched multifactor models, whereas
this is less likely for global or nonequity funds. This means that sector characteristics
related to our ability to risk-adjust would have a spurious effect on a cross-sectional
analysis of persistence in risk-adjusted returns. Persistence measured on the basis of
raw returns, on the other hand, is important in its own right. Numerous information
providers such as Money Management, Unit Trust Yearbook, Standard & Poor’s Web
site, and others rank funds based on raw returns within a sector. Indeed, evidence
Mutual Fund Performance Persistence                          353

on the return-flow relation indicates that investors react to raw returns. Moreover,
implicit in looking at within-sector persistence, as we do, is a peer-group adjustment
of fund returns.
         Commonly used statistics for studying relative persistence within a peer
group include the Spearman rank-correlation coefficient, and quantities based on
2 × 2 winner/loser contingency tables. The latter include the log-odds ratio and
the chi-squared statistic. The chi-squared statistic is disqualified for our purpose
(which is to explain the extent of persistence) because high values correspond
to either persistence or reversal of performance. The advantage of the Spearman
correlation over the log-odds ratio is that the latter uses the performance rank
of each fund rather than just its winner/loser status. This generally means more
powerful tests for persistence (Carpenter and Lynch 1999). The advantage of the
log-odds ratio is that it has a more straightforward economic interpretation, as
we show shortly. We use both the log-odds ratio and the Spearman correlation as
our measures of persistence (equations are given in Appendix B). To avoid our
results being influenced by the small-sample properties of these statistics, we use
only sector-years with at least 20 funds in existence over both years for which
performance is measured.
         Table 1 shows the extent of relative performance persistence across U.K.
mutual fund sectors based on raw returns over consecutive years. In Panel A,
we present the distribution of the Spearman correlation coefficient and of the
log-odds ratio by type of sector. The first group of eight rows pertains to the
log-odds ratio. In column 1, for all sector-years combined (162), the average log-
odds ratio is 0.357. The null hypothesis that the mean log-odds ratio is zero can
be rejected ( p-value < .001) based on applying Student’s t-test to our set of 162
sector-years. The median sector-year has a log-odds ratio of 0.405, and the dis-
tribution ranges from −2.837 to 4.317. The log-odds ratio is positive for 62% of
the sector-years. Furthermore, the table reports the proportion of sector-years for
which the hypothesis of no persistence is rejected in favor of the one-sided alter-
native of positive persistence. For the log-odds ratio, this is the case for 29% of the
sector-years at the .05 confidence level, and for 17% of the sector-years at the .01
confidence level.
         The next eight rows characterize the distribution of the Spearman corre-
lation across sector-years. For all sector-years combined, the average Spearman
correlation is 0.143 and is significantly different from zero. We note that even at
the .01 confidence level, the hypothesis of no persistence is rejected in favor of
the hypothesis of positive persistence for 29% of sector-years. This suggests the
Spearman-based test is more powerful than the log-odds ratio.
         To ensure that performance persistence in our sample is not driven by a
particular subperiod, we separately consider sector-years for which the evaluation
years are 1992 through 1996, and those for which the evaluation years are 1997
through 2001 (results not reported in a table). The average log-odds ratio for the
354                             The Journal of Financial Research

TABLE 1. Relative Performance Persistence Across Sectors.

                                                         All Sectors
                                                Domestic Other Than       Global     Domestic   Global
                                        All      Equity   Domestic        Equity     Nonequity Nonequity
                                      Sectors    Sectors   Equity         Sectors     Sectors   Sectors
Variable                                (1)        (2)       (3)            (4)         (5)       (6)
Panel A. Sector-Year Statistics
Number of sector-years                  162          36          126          63          30          33
Log-odds ratio by sector-year
  Average                              0.357      0.448        0.331       0.173       0.565        0.421
  p-value for H0 : mean = 0            0.000      0.006        0.002       0.192       0.019        0.057
  Median                               0.405      0.382        0.405       0.365       0.555        0.525
  Minimum                             −2.837     −1.168       −2.837      −2.837      −1.455       −2.485
  Maximum                              4.317      2.711        4.317       2.398       3.008        4.317
  Proportion positive                  0.623      0.639        0.619       0.571       0.667        0.667
  Proportion positive and              0.290      0.306        0.286       0.270       0.367        0.242
    significant at .05 level
  Proportion positive and               0.167      0.306        0.127       0.143       0.133        0.091
    significant at .01 level
Spearman correlation by sector-year
  Average                            0.143        0.147        0.142       0.097       0.180        0.191
  p-value for H0 : mean = 0          0.000        0.004        0.000       0.017       0.005        0.001
  Median                             0.162        0.135        0.166       0.133       0.257        0.176
  Minimum                           −0.642       −0.477       −0.642      −0.642      −0.544       −0.476
  Maximum                            0.817        0.804        0.817       0.672       0.656        0.817
  Proportion positive                0.691        0.583        0.722       0.698       0.733        0.758
  Proportion positive and            0.426        0.472        0.413       0.429       0.433        0.364
    significant at .05 level
  Proportion positive and            0.290         0.333        0.278       0.254       0.267        0.333
    significant at .01 level

Panel B. Aggregate Statistics
Fund-years in winner-winner             2,724       900         1,824       1,240        286         298
  category
Fund-years in loser-loser               2,672       895         1,777       1,219        273         285
  category
Fund-years in winner-loser              2,276       728         1,548       1,090        225         233
  category
Fund-years in loser-winner              2,296       740         1,556       1,096        229         231
  category
Aggregate log-odds ratio                0.331      0.402        0.297       0.235       0.416        0.456
p-value for aggregate log-odds          0.000      0.000        0.000       0.000       0.001        0.000
  ratio
Frequency of repeat performance         0.541      0.550        0.537       0.529       0.552        0.557

Note: This table reports descriptive statistics for measures of relative performance persistence across U.K.
mutual fund sectors, 1991–2001. Sector-years are included if at least 20 funds had returns available in the
ranking and evaluation years. Panel A reports the distribution of the Spearman correlation coefficient and
the log-odds ratio across sector-years. Both the Spearman correlation coefficient and the log-odds ratio are
based on raw annual returns in consecutive calendar years (formulae are given in Appendix B). In Panel B,
sector-years are pooled into an aggregate contingency table.
Mutual Fund Performance Persistence                                         355

earlier (later) period is 0.358 (0.357), and the average Spearman correlation coef-
ficient is 0.134 (0.152). All of these averages are statistically significant at the .01
level. Moreover, the average log-odds ratio and the average Spearman correlation
coefficient are not significantly different between the two periods ( p-values = .99
and .70, respectively).
         Because research on mutual fund performance persistence tends to focus
on domestic equity funds, we separately report results for these sectors in the second
column. The persistence measures are positive and, despite a sample size of only 36
sector-years, highly statistically significant. The average log-odds ratio, at 0.448, is
slightly lower than the 0.516 average log-odds ratio in Fletcher and Forbes (2002),
which is based on raw annual returns for U.K. equity mutual funds from 1982 to
1996. The average Spearman correlation, at 0.147, is slightly lower than the 0.188
reported by Allen and Tan (1999) for raw annual returns of U.K. equity mutual
funds from 1989 to 1995.
         Column 3 presents results for sectors other than domestic equity. The level
of persistence is comparable to that in the preceding column. In fact, unreported
tests show that differences between the two columns are never significant. The
last three columns further disaggregate sectors other than domestic equity into
global equity, domestic nonequity, and global nonequity. In each category, perfor-
mance persistence is positive and significant, at least for the Spearman correlation
coefficient.
         As further evidence on the level of performance persistence in our sample,
in Panel B we pool data from different sector-years to present an aggregate con-
tingency table. In 2,724 (2,672) instances, funds are two-period winners (losers)
in their respective sectors, and in 2,276 (2,296) instances, funds win in the rank-
ing (evaluation) period and lose in the evaluation (ranking) period. The resulting
aggregate log-odds ratio equals ln((2724 × 2672)/(2276 × 2296)) = 0.331 and is
highly significant. The aggregate log-odds ratios for the different sector groups in
columns 2 through 6 are all significant at the .01 level.
         The economic significance of performance persistence in our sample is
perhaps best addressed through the probability that a fund’s winner/loser status
carries over from the ranking period to the evaluation period. This probability of
repeat performance can be estimated as the number of fund-years corresponding
to two-period winners or two-period losers divided by the total number of fund-
years. For all sectors together, this quantity (reported in the last row of the table)
equals (2,724 + 2,672)/(2,724 + 2,672 + 2,276 + 2,296) = 54.1%, as compared
to the 50.0% that one would expect in the absence of performance persistence or
performance reversal.3

      3
        Because we define a winner (loser) as a fund that places in the top (bottom) half in its sector in a
given year, the cell counts in the contingency table are not independent. In fact, if there were no ties and
if the number of funds in a sector were always divisible by four, the winner-winner fund count would be
356                          The Journal of Financial Research

         Overall, there is strong evidence that the U.K. mutual fund industry exhibits
persistence in relative investment performance. If at least some of this persistence is
due to sector-level attributes, in particular to those related to sector competitiveness,
a cross-sector analysis may reveal this. Such analysis is conducted in the next
section.

                   III. Determinants of Sector-Level Persistence

Sector Attributes
Broadly speaking, systematic differences in persistence between sectors can be due
to differences in the composition of sector membership, or to differences in the
types of assets sector members invest in. We construct several variables designed to
quantify how competitive a sector is, and the distribution of these variables is given
in Panel A of Table 2. N is simply the number of funds in a sector at the end of the
ranking year. The largest number of funds in a sector is 302, corresponding to the
UK All Companies sector in 1999 (after sectors dedicated to domestic “growth” and
“growth and income” stocks were merged). Because we drop sectors comprising
fewer than 20 funds with recorded returns, the minimum number of funds in a sector
is 24, the median is 79, and the average is 87.4 It is reasonable to conjecture that
consistent performance is harder to attain in a more crowded sector. For example, in
studying fund performance Siggelkow (2003) regards the number of mutual funds
in a category as “a measure of general competition, for instance, for mis-priced
securities” (p. 133).
         We recognize, however, that in such competition, small funds may have rel-
atively little effect. We therefore also use the Herfindahl index, which is commonly
considered as a measure of intra-industry rivalry. Specifically, HERFINDAHL is
the concentration index of assets under management. Because several funds from
a single family of funds can coexist within a sector, we aggregate assets by fam-
ily to calculate this measure. Thus, the value of HERFINDAHL for each sector is
the sum across families of the square of each family’s assets as a proportion of
a sector’s total assets. Although we use only sectors with at least 20 funds, there
is substantial variation in the value of the Herfindahl index, ranging from 0.027
to 0.629 (by construction, the smallest possible value of the Herfindal index is 0


exactly equal to the loser-loser count, and the winner-loser count would be exactly equal to the loser-winner
count. Using this insight, it is straightforward to show that the probability of repeat performance can be
obtained directly from the log-odds ratio (L) as 1/(1 + e−L/2 ). For example, using the aggregate log-odds
ratio of 0.331, the probability of repeat performance is 1/(1 + e−0.331/2 ) = 0.541. We subsequently use this
conversion to assess the economic significance of our regression results.
      4
        For comparison, Massa (2003) uses several data providers’ fund descriptions to assign U.S. mutual
funds to 1 of 23 categories. The numbers of funds in his categories range from 14 to 1,149, the median is
343, and the average is 411.
Mutual Fund Performance Persistence                                           357

TABLE 2. Descriptive Statistics for Sector-Level Variables.
Panel A. Moments
Variable                Mean         Standard Deviation          Median          Minimum            Maximum

N                         87                  50                   79                 24               302
MATURITY                0.552               0.188                 0.626             0.000             0.846
HERFINDAHL              0.088               0.070                 0.068             0.027             0.629

Panel B. Correlations
Variable                            N               MATURITY               HERFINDAHL                  LATER

DOMESTIC EQUITY                   0.420                0.420                   −0.247                  −0.007
N                                                      0.476                   −0.464                  −0.163
MATURITY                                                                       −0.586                  −0.040
HERFINDAHL                                                                                             −0.078

Note: This table contains descriptive statistics for the set of sector-year explanatory variables. Sector-years
are included if at least 20 funds had returns available in the ranking and evaluation years: 162 sector-years
meet this requirement. The variables are as follows. N is the number of funds within the sector. MATURITY
is the proportion of sector funds that are at least five years old. HERFINDAHL is the Herfindahl index
measuring the concentration of fund assets within the sector, where funds from the same family are
aggregated. DOMESTIC EQUITY is a dummy variable equal to 1 if sector funds are primarily invested in
U.K. equities, and 0 otherwise.



and the largest possible value is 1). We hypothesize that less concentrated (more
competitive) sectors exhibit lower persistence.
         Finally, MATURITY is the proportion of funds that are at least five years
old. In the average sector, most funds are “seasoned,” but the minimum value of 0
for the maturity variable indicates that for some sector-years, all of the funds are
relatively recent entrants. Berk and Green (2004) give a powerful reason why mutual
fund performance persistence should decrease with fund vintage. If investment
management returns to scale are decreasing, managers have differential ability, and
investors channel money to best performers, then superior funds grow to the point
where outperformance is no longer possible. Empirically, Waring (1996) finds that
earnings persistence in an industrial sector tends to decay over time, as competitive
forces have acted over a longer period.
         Panel B of Table 2 shows a correlation matrix for the preceding sector
attributes and for a dummy variable indicating sector membership in the U.K.
equity category (DOMESTIC EQUITY ), as well as a dummy variable that equals
1 for the second half of our sample period (evaluation years from 1997 to 2001),
and 0 otherwise (LATER). DOMESTIC EQUITY is included because most studies
of performance persistence focus on domestic equity sectors. LATER controls for
the possibility that the level of persistence may have changed in more recent years.
We note that pairwise correlations between N, MATURITY, and HERFINDAHL are
high in magnitude: sectors with more funds in them tend to be more mature, and
358                     The Journal of Financial Research

assets invested in these sectors are more dispersed across fund families. Therefore,
in the regressions to follow, we enter these three variables one at a time.

Regression Results
Table 3 presents the results of a pooled regression of sector-level measures of relative
persistence on our set of sector-level explanatory variables. Persistence is measured
over years T (the ranking year) and T+1 (the evaluation year). Explanatory variables
are measured as of the end of year T (with one exception explained below). Thus,
we examine whether sector characteristics observed at the end of year T tell us to
what extent year T performance of the sector’s funds persist into year T+1. We
do not require that funds remain in the same sector until the end of year T+1 (or,
indeed, that the sector itself continue to exist until the end of year T+1) because
doing so would constitute a look-ahead bias.
         In regressions (1) through (3) we use the log-odds ratio as the measure of
persistence and proxy for sector competitiveness with N, MATURITY, or
HERFINDAHL, respectively. Although the number of funds in a sector is not signif-
icantly related to persistence, maturity of funds is significant (t-statistic = −2.71),
as is the concentration of assets under management (t-statistic = 3.01). In other
words, sectors that are less mature and have more concentrated assets—that is,
sectors that may be described as less competitive—are characterized by greater
persistence. The other variables are not statistically significant.
         Regressions (4) through (6) parallel regressions (1) through (3) but include
an additional control variable. CROSSRET is defined as the product of average
sector returns in years T and T+1. Although we do not adjust for differences in
fund exposure to different risk factors, these differences can generate persistence in
raw returns when there is persistence in factor realizations. CROSSRET is intended
to capture spurious persistence due to ex post momentum for the sector as a whole.
The regression results confirm this intuition in that CROSSRET is positive and
highly significant ( p-value < .001). Its only other influence is to enhance slightly
the significance of MATURITY and HERFINDAHL (t-statistics = −2.80 and 3.24,
respectively).
         In regressions (7) through (12), the Spearman correlation coefficient is the
dependent variable. The results are similar to those based on the log-odds ratio.
Once again, MATURITY and HERFINDAHL are statistically significant at the .01
level, and CROSSRET continues to capture persistence due to momentum in all
specifications.

Robustness Checks
The preceding subsection presents evidence that HERFINDAHL and MATURITY
are sector attributes that are systematically related to the persistence exhibited by
the sector. We now report on the robustness of our results to alternative sample-
selection criteria, econometric methods, and other variations.
TABLE 3. Explaining Sector-Level Persistence.

                                        Dependent Variable: Log-Odds Ratio                                   Dependent Variable: Spearman Correlation Coefficient
Explanatory Variable           (1)       (2)          (3)        (4)          (5)           (6)        (7)           (8)          (9)         (10)         (11)       (12)

Intercept          0.586                1.097        0.020       0.568       1.040     −0.070          0.214        0.348        0.048        0.208        0.328       0.016
                  (2.93)∗∗∗            (3.90)∗∗∗    (0.12)      (2.90)∗∗    (3.76)∗∗∗ (−0.40)         (3.82)∗∗∗    (4.41)∗∗∗    (0.98)       (3.84)∗∗∗    (4.29)∗∗∗   (0.34)
DOMESTIC EQUITY    0.224                0.302        0.270       0.257       0.306       0.279         0.038        0.057        0.051        0.050        0.058       0.054
                  (0.98)               (1.40)       (1.29)      (1.14)      (1.45)      (1.36)        (0.59)       (0.93)       (0.87)       (0.80)       (0.99)      (0.96)
N                −0.003                                        −0.004                                −0.001                                 −0.001
                (−1.13)                                       (−1.48)                               (−1.21)                                (−1.66)
MATURITY                               −1.297                               −1.315                                 −0.353                                 −0.359
                                      (−2.71)∗∗∗                           (−2.80)∗∗∗                             (−2.63)∗∗∗                             (−2.76)∗∗∗
HERFINDAHL                                            3.780                                 3.975                                 1.104                               1.172
                                                     (3.01)∗∗∗                             (3.24)∗∗∗                             (3.15)∗∗∗                           (3.47)∗∗∗
LATER                       −0.225     −0.226       −0.139     −0.175       −0.168        −0.075      −0.060       −0.058       −0.033      −0.041      −0.037      −0.011
                           (−1.25)    (−1.30)      (−0.80) (−0.99)         (−0.98)       (−0.44)     (−1.16)      (−1.18)      (−0.68)     (−0.84)     (−0.78)     (−0.22)
CROSSRET                                                         4.271        4.062         4.272                                             1.489       1.422       1.485
                                                                (2.80)       (2.73)∗∗∗     (2.89)∗∗∗                                         (3.53)∗∗∗   (3.45)∗∗∗   (3.64)∗∗∗
R2                            0.017     0.053         0.063      0.063        0.090         0.110      0.015        0.047         0.064       0.087       0.114       0.137

Note: This table contains the results of regressing measures of persistence in relative investment performance on sector-level variables. Sector-years are included if at least
20 funds have returns available in the ranking and evaluation years: 162 sector-years meet this requirement. The explanatory variables are as follows. N is the number of funds
within a sector. MATURITY is the proportion of sector funds that are at least five years old. HERFINDAHL is the Herfindahl index measuring the concentration of fund assets
                                                                                                                                                                                  Mutual Fund Performance Persistence




within a sector, where funds from the same family are aggregated. DOMESTIC EQUITY is a dummy variable equal to 1 if sector funds are primarily invested in U.K. equities,
and 0 otherwise. LATER is a dummy variable equal to 1 when the evaluation year is 1997 or later, and 0 otherwise. CROSSRET is the product of average sector returns in the
ranking and evaluation years. The t-statistics are shown in parentheses.
∗∗∗
      Significant at the 1% level.
∗∗
     Significant at the 5% level.
                                                                                                                                                                                  359
360                     The Journal of Financial Research

          Because both the log-odds ratio and the Spearman correlation are estimated
with differing degrees of precision across sectors, the resulting heteroskedasticity in
our regression may lead to inefficient estimation. We therefore use the inverse of the
standard error of the log-odds ratio and of the Spearman correlation coefficient as
weights in a generalized least squares regression using these measures as dependent
variables. The results are not significantly different from those reported earlier.
          We also investigate whether time-series correlation affects our results. First,
we test for serial correlation in a panel but fail to find evidence of this. Second, we
include lagged persistence measures in our regressions. This reduces the number of
observations from 162 to 136. HERFINDAHL and MATURITY remain significant at
the .05 level or better, and N remains insignificant. The lagged persistence measure
itself is never statistically significant.
          To check that our results using Spearman correlation are not influenced
by having a dependent variable limited to the [+1,−1] range, we estimate our
model using the Papke and Wooldridge (1996) generalized linear approach, which
is designed for estimating models with a fractional dependent variable. Our results
are broadly unchanged. All coefficient estimates that are significant using ordinary
least squares at the .05 level and above are also significant using the new approach,
and the signs of all significant coefficient estimates remain the same as before.
          To address the possibility that our results may be influenced by the small-
sample properties of our persistence measures, we exclude sector-years with fewer
than 30 funds. This reduces the number of sector-years to 124. When we do this,
MATURITY becomes insignificant regardless of the econometric method used.
The statistical significance of HERFINDAHL, however, is .05 or better in all
specifications.
          As the coverage of the last two editions of the Unit Trust Yearbook (corre-
sponding to calendar years 2000 and 2001) is reduced because of nonreporting by
several fund families, we repeat our regressions after omitting these years. We also
conduct Fama-MacBeth regressions, drop outlier observations, and use different
ranges to winsorize our persistence measures. Our results remain basically un-
changed: HERFINDAHL is always significant at least at the .10 level and generally
at the .05 level. None of our other variables is consistently significant.


Economic Significance
Our results indicate that the concentration of funds’ assets is statistically signifi-
cantly related to the persistence level in that sector. We now assess the economic
significance of this relation. Consider a fund sector not restricted to U.K. equities
(DOMESTIC EQUITY = 0) in the second half of our sample period (LATER =
1). When HERFINDAHL is set to its full-sample 10th percentile value of 0.038,
using the estimated coefficients in regression (3) of Table 3, the fitted value of
the log-odds ratio equals 0.025. Using the conversion formula in footnote 3, this
Mutual Fund Performance Persistence                                            361

translates into a 50.3% probability that a fund’s winner or loser status is retained
from the ranking period to the evaluation period. This probability exceeds by only
0.3% the corresponding probability that one would expect by mere chance in the
absence of any persistence.
         We now reset HERFINDAHL to its 90th percentile value of 0.162. The
corresponding fitted value of the log-odds ratio is 0.493, which translates into a
56.1% probability of repeat performance, or 6.1% higher than would be expected in
the absence of persistence. In other words, if a sector goes from the 10th to the 90th
percentile of concentration of assets, the excess (relative to the no-persistence case)
probability of remaining in the same half of performance rankings increases from
0.3% to 6.1%. These numbers indicate that the effect of sector-level concentration
on performance persistence is substantial in economic terms.

                                 IV. Longer Term Persistence

Because we find a link between sector characteristics and persistence, we check
whether the results hold when persistence is measured over a longer period. To do
this, rather than examining adjacent ranking and evaluation periods as we did in
the preceding section, we use a lagged ranking period (as in Teo and Woo 2001).
In other words, one year is allowed to pass between the end of the ranking period
and the start of the evaluation period.
         Recall that when the ranking period is not lagged, the average (across
sector-years) log-odds ratio is 0.357 and highly statistically significant. When we
lag the ranking period by one year, the average log-odds ratio becomes −0.013 and
is not significant. Likewise, the average Spearman coefficient drops from 0.143 to
−0.012 and is no longer significant.
         Even though the average level of longer term persistence across sector-
years is close to zero, it is still possible that variation in longer term persistence is
related to sector competitiveness. We therefore repeat our regression analysis when
the dependent variable is the longer term measure of persistence.5 Consistent with
the notion that our sample exhibits little or no persistence at the longer horizons,
HERFINDAHL does not have a significant effect on longer term persistence, and
neither do the other sector-level variables.6


     5
       These results are available from the authors on request.
     6
       Although we find it important to document that our significant results are limited to the one-year
horizon, we note that tests for the existence of persistence, and by extension tests for the association
of our sector-level variables with persistence, are weaker when the horizon is longer. First, survivorship
conditioning becomes more serious when the horizon is longer, and depending on the characteristics of the
fund attrition process, this can either strengthen or weaken persistence. Second, measurement of persistence
is noisier when the horizon is longer (e.g., because fund characteristics change over time). Indeed, few studies
detect persistence beyond the one-year horizon.
362                     The Journal of Financial Research

                                   V. Conclusion

Performance persistence is important to all parties connected with fund manage-
ment. Its existence has been the subject of an intense and ongoing debate. We
contribute to this debate by studying variation in performance persistence across
peer groups. The focus of our study is the U.K. mutual fund industry, where official
sectors unambiguously define such peer groups.
         We study the effect of several sector-level variables on sector-level persis-
tence. Our choice of variables is based on the notion that the more competitive a
sector is, the less likely it is to be characterized by persistence in its funds’ perfor-
mance. The variables used to capture intra-sector rivalry are: the number of funds
in the sector, the concentration of fund family assets under management in the
sector, and the proportion of mature funds in the sector. We additionally control for
the types of assets in which the sector’s funds are invested. Only the concentration
index of fund family assets is consistently significant: the less dispersed the sector’s
assets are, the more persistence is observed. In all, our results indicate that the com-
petitiveness of a fund sector influences the persistence in the relative performance
of its members. The exact channels through which the competitive environment
affects investment managers’ performance are a subject for future research.



                                APPENDIX A
              Evolution of U.K. Unit Trust Sectors, 1991–2001
Mutual Fund Performance Persistence   363
364                     The Journal of Financial Research




                               APPENDIX B
            Calculation of Performance Persistence Statistics

Spearman Rank-Correlation Coefficient
First, funds that existed in years T (the ranking year) and T+1 (the evaluation year)
are identified. Define N to be the size of this sample. For each fund in the sample,
the difference d i in the rank of fund i between years T and T+1 is calculated. The
Spearman rank-correlation statistic is defined as

                                         N
                         rs = 1 − 6           di    (N 3 − N )
                                        i=1

and lies between −1 and +1. For sufficiently large N, it is appropriate to test for
the statistical significance of rs using a t-test where the critical t-statistic is given
by
Mutual Fund Performance Persistence                                         365

                                                       N −2
                                           ts = r s
                                                       1 − rs2
and has N−2 degrees of freedom.

Log-Odds Ratio
Funds within a sector are classified as winners (W ) (losers [L]) if their returns are
in the top (bottom) half of funds for each of the years T and T+1. WW denotes the
number of two-period winners, LW denotes the number of losers in the first year
and winners in the second year, WL reverses this order, and LL denotes the number
of two-period losers. The log-odds ratio is defined as
                                                WW ∗ LL
                                          ln            .
                                                WL ∗ LW
The log-odds ratio is asymptotically normally distributed with mean zero and stan-
dard error given by

                                          1    1   1    1
                               σ =          +    +   +    .
                                         WW   WL LW    LL
The z-statistic of the log odds ratio refers to the log-odds ratio divided by its standard
error.


                                             References

Allen, D. E. and M. L. Tan, 1999, A test of the persistence in the performance of UK managed funds,
           Journal of Business Finance and Accounting 26, 559–93.
Berk, J. B. and R. C. Green, 2004, Mutual fund flows and performance in rational markets, Journal of
           Political Economy 112, 1269–95.
Blake, C. R., E. J. Elton, and M. J. Gruber, 1993, The performance of bond mutual funds, Journal of Business
           66, 371–403.
Blake, D. and A. Timmermann, 1998, Mutual fund performance: Evidence from the UK, European Finance
           Review 2, 57–77.
Blake, D. and A. Timmermann, 2003, Performance persistence in mutual funds: An independent assessment
           of the studies prepared by Charles River Associates for the Investment Management Association,
           Report commissioned by the Financial Services Authority.
Brown, S. J., W. N. Goetzmann, R. G. Ibbotson, and S. A. Ross, 1992, Survivorship bias in performance
           studies, Review of Financial Studies 5, 553–80.
Carhart, M. M., 1997, On persistence in mutual fund performance, Journal of Finance 52, 57–82.
Carpenter, J. N. and A. W. Lynch, 1999, Survivorship bias and attrition effects in measures of performance
           persistence, Journal of Financial Economics 54, 337–74.
Cooper, M. J., H. Gulen, and P. R. Rau, 2005, Changing names with style: Mutual fund name changes and
           their effects on fund flows, Journal of Finance 60, 2825–58.
Fletcher, J. and D. Forbes, 2002, An exploration of the persistence of UK unit trust performance, Journal
           of Empirical Finance 9, 475–93.
366                         The Journal of Financial Research

Giles, T., T. Wilsdon, and T. Worboys, 2002, Performance persistence in UK equity funds—An empirical
            analysis, Report prepared by Charles River Associates for the Investment Management Associ-
            ation, London, United Kingdom.
Ibbotson, R. and A. Patel, 2002, Do winners repeat with style?, Yale ICF Working paper 00-70.
Kosowski, R., A. Timmermann, H. White, and R. Wermers, 2003, Can mutual fund “stars” really pick
            stocks? New evidence from a bootstrap analysis, Working paper, INSEAD.
Massa, M., 2003, How do family strategies affect fund performance? When performance-maximization is
            not the only game in town, Journal of Financial Economics 67, 249–304.
Quigley, G. and R. Sinquefield, 2000, Performance of UK equity unit trusts, Journal of Asset Management
            1, 72–92.
Papke, L. and J. Wooldridge, 1996, Econometric methods for fractional response variables with an applica-
            tion to 401(k) plan participation rates, Journal of Applied Econometrics 11, 619–32.
Rhodes, M., 2000, Past imperfect? The performance of UK equity managed funds, Financial Services
            Authority Occasional Paper No. 9.
Siggelkow, N., 2003, Why focus? A study of intra-industry focus effects, Journal of Industrial Economics
            51, 121–50.
Teo, M. and S.-J. Woo, 2001, Persistence in style-adjusted mutual fund returns, Working paper, Harvard
            University.
Ter Horst, J., T. Nijman, and M. Verbeek, 2001, Eliminating look-ahead bias in evaluating persistence in
            mutual fund performance, Journal of Empirical Finance 8, 345–73.
Volkman, D. A. and M. E. Wohar, 1995, Determinants of persistence in relative performance of mutual
            funds, Journal of Financial Research 18, 415–30.
Waring, G. F., 1996, Industry differences in the persistence of firm-specific returns, American Economic
            Review 86, 1253–65.
Wermers, R., 2003, Is money really smart? New evidence on the relation between mutual fund flows,
            manager behavior, and performance persistence, Working paper, University of Maryland.
Fund performance persistence and competition keswani

More Related Content

What's hot

Stock liquidity 2
Stock liquidity 2Stock liquidity 2
Stock liquidity 2ZNiazi2
 
Event study on the impact of mergers and acquisitions
Event study on the impact of mergers and acquisitionsEvent study on the impact of mergers and acquisitions
Event study on the impact of mergers and acquisitionseleclasson
 
Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...
Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...
Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...inventionjournals
 
Impact of mergers and acquisitions
Impact of mergers and acquisitionsImpact of mergers and acquisitions
Impact of mergers and acquisitionsprjpublications
 
Journal of Business Venturing : Alliance portfolios and shareholder value in ...
Journal of Business Venturing : Alliance portfolios and shareholder value in ...Journal of Business Venturing : Alliance portfolios and shareholder value in ...
Journal of Business Venturing : Alliance portfolios and shareholder value in ...Université Internationale de Rabat
 
M&A
M&AM&A
M&A- -
 
Ownership structure and dividend policy.doc=2
Ownership structure and dividend policy.doc=2Ownership structure and dividend policy.doc=2
Ownership structure and dividend policy.doc=2Liza Khanam
 
Event Study: Market Reactions to New CEO Announcement
Event Study: Market Reactions to New CEO AnnouncementEvent Study: Market Reactions to New CEO Announcement
Event Study: Market Reactions to New CEO AnnouncementZhuting Meng
 
The Failure of Risk Management for Non-Financial Companies in the Context of ...
The Failure of Risk Management for Non-Financial Companies in the Context of ...The Failure of Risk Management for Non-Financial Companies in the Context of ...
The Failure of Risk Management for Non-Financial Companies in the Context of ...Fundação Dom Cabral - FDC
 
9922608_Abi
9922608_Abi9922608_Abi
9922608_AbiAbi Lin
 
Thirty years of_mergers_and_acquisitions_research-2006
Thirty years of_mergers_and_acquisitions_research-2006Thirty years of_mergers_and_acquisitions_research-2006
Thirty years of_mergers_and_acquisitions_research-2006thangngovan
 
Distress risk and stock returns in an emerging market
Distress risk and stock returns in an emerging marketDistress risk and stock returns in an emerging market
Distress risk and stock returns in an emerging marketAlexander Decker
 
Understanding the Dynamics of Business Group Advantages and Affiliate Level A...
Understanding the Dynamics of Business Group Advantages and Affiliate Level A...Understanding the Dynamics of Business Group Advantages and Affiliate Level A...
Understanding the Dynamics of Business Group Advantages and Affiliate Level A...inventionjournals
 

What's hot (18)

Stock liquidity 2
Stock liquidity 2Stock liquidity 2
Stock liquidity 2
 
Event study on the impact of mergers and acquisitions
Event study on the impact of mergers and acquisitionsEvent study on the impact of mergers and acquisitions
Event study on the impact of mergers and acquisitions
 
Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...
Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...
Financial Distress Prediction With Altman Z-Score And Effect On Stock Price: ...
 
Impact of mergers and acquisitions
Impact of mergers and acquisitionsImpact of mergers and acquisitions
Impact of mergers and acquisitions
 
Journal of Business Venturing : Alliance portfolios and shareholder value in ...
Journal of Business Venturing : Alliance portfolios and shareholder value in ...Journal of Business Venturing : Alliance portfolios and shareholder value in ...
Journal of Business Venturing : Alliance portfolios and shareholder value in ...
 
M&A
M&AM&A
M&A
 
10320140503003
1032014050300310320140503003
10320140503003
 
10320140503003
1032014050300310320140503003
10320140503003
 
Ownership structure and dividend policy.doc=2
Ownership structure and dividend policy.doc=2Ownership structure and dividend policy.doc=2
Ownership structure and dividend policy.doc=2
 
Event Study: Market Reactions to New CEO Announcement
Event Study: Market Reactions to New CEO AnnouncementEvent Study: Market Reactions to New CEO Announcement
Event Study: Market Reactions to New CEO Announcement
 
The Failure of Risk Management for Non-Financial Companies in the Context of ...
The Failure of Risk Management for Non-Financial Companies in the Context of ...The Failure of Risk Management for Non-Financial Companies in the Context of ...
The Failure of Risk Management for Non-Financial Companies in the Context of ...
 
9922608_Abi
9922608_Abi9922608_Abi
9922608_Abi
 
Thirty years of_mergers_and_acquisitions_research-2006
Thirty years of_mergers_and_acquisitions_research-2006Thirty years of_mergers_and_acquisitions_research-2006
Thirty years of_mergers_and_acquisitions_research-2006
 
Artikel 7
Artikel 7Artikel 7
Artikel 7
 
Why Do Firms Buyback Their Shares
Why Do Firms Buyback Their SharesWhy Do Firms Buyback Their Shares
Why Do Firms Buyback Their Shares
 
Distress risk and stock returns in an emerging market
Distress risk and stock returns in an emerging marketDistress risk and stock returns in an emerging market
Distress risk and stock returns in an emerging market
 
Understanding the Dynamics of Business Group Advantages and Affiliate Level A...
Understanding the Dynamics of Business Group Advantages and Affiliate Level A...Understanding the Dynamics of Business Group Advantages and Affiliate Level A...
Understanding the Dynamics of Business Group Advantages and Affiliate Level A...
 
10220140503002
1022014050300210220140503002
10220140503002
 

Viewers also liked

Mutual fund performance and manager style by james l. davis(11)
Mutual fund performance and manager style by james l. davis(11)Mutual fund performance and manager style by james l. davis(11)
Mutual fund performance and manager style by james l. davis(11)bfmresearch
 
How poor stock mkt perf affects fund f lows shrider
How poor stock mkt perf affects fund f lows shriderHow poor stock mkt perf affects fund f lows shrider
How poor stock mkt perf affects fund f lows shriderbfmresearch
 
Performance persistence brown_goetzmann
Performance persistence brown_goetzmannPerformance persistence brown_goetzmann
Performance persistence brown_goetzmannbfmresearch
 
Does fund size erode mutual fund performance the role of liquidity and organ...
Does fund size erode mutual fund performance  the role of liquidity and organ...Does fund size erode mutual fund performance  the role of liquidity and organ...
Does fund size erode mutual fund performance the role of liquidity and organ...bfmresearch
 
Nl bfm january 2012
Nl bfm january 2012Nl bfm january 2012
Nl bfm january 2012bfmresearch
 
Should investors avoid active managed funds baks
Should investors avoid active managed funds baksShould investors avoid active managed funds baks
Should investors avoid active managed funds baksbfmresearch
 

Viewers also liked (6)

Mutual fund performance and manager style by james l. davis(11)
Mutual fund performance and manager style by james l. davis(11)Mutual fund performance and manager style by james l. davis(11)
Mutual fund performance and manager style by james l. davis(11)
 
How poor stock mkt perf affects fund f lows shrider
How poor stock mkt perf affects fund f lows shriderHow poor stock mkt perf affects fund f lows shrider
How poor stock mkt perf affects fund f lows shrider
 
Performance persistence brown_goetzmann
Performance persistence brown_goetzmannPerformance persistence brown_goetzmann
Performance persistence brown_goetzmann
 
Does fund size erode mutual fund performance the role of liquidity and organ...
Does fund size erode mutual fund performance  the role of liquidity and organ...Does fund size erode mutual fund performance  the role of liquidity and organ...
Does fund size erode mutual fund performance the role of liquidity and organ...
 
Nl bfm january 2012
Nl bfm january 2012Nl bfm january 2012
Nl bfm january 2012
 
Should investors avoid active managed funds baks
Should investors avoid active managed funds baksShould investors avoid active managed funds baks
Should investors avoid active managed funds baks
 

Similar to Fund performance persistence and competition keswani

Selection termination goyalwahal
Selection termination goyalwahalSelection termination goyalwahal
Selection termination goyalwahalbfmresearch
 
Fund pick activeshare
Fund pick activeshareFund pick activeshare
Fund pick activesharebfmresearch
 
Cap srtucture hotdebt
Cap srtucture hotdebtCap srtucture hotdebt
Cap srtucture hotdebtaditi_ds
 
07. the determinants of capital structure
07. the determinants of capital structure07. the determinants of capital structure
07. the determinants of capital structurenguyenviet30
 
Evaluation of the Development and Performance of Selected GCC and Non-GCC St...
Evaluation of the Development and Performance  of Selected GCC and Non-GCC St...Evaluation of the Development and Performance  of Selected GCC and Non-GCC St...
Evaluation of the Development and Performance of Selected GCC and Non-GCC St...Mace Abdullah
 
Exploring Index Effects - Tamas Toth
Exploring Index Effects - Tamas TothExploring Index Effects - Tamas Toth
Exploring Index Effects - Tamas TothTamas Toth, CFA
 
Short term persistence in mutual fund performance(12)
Short term persistence in mutual fund performance(12)Short term persistence in mutual fund performance(12)
Short term persistence in mutual fund performance(12)bfmresearch
 
Fund performance wermers
Fund performance wermersFund performance wermers
Fund performance wermersbfmresearch
 
邬梦曲论文
邬梦曲论文邬梦曲论文
邬梦曲论文Meng Qu Wu
 
Aerospace and Defense Value Creators Report 2015
Aerospace and Defense Value Creators Report 2015Aerospace and Defense Value Creators Report 2015
Aerospace and Defense Value Creators Report 2015Seda Eskiler
 
Capital structure and eps a study on selected financial institutions listed o...
Capital structure and eps a study on selected financial institutions listed o...Capital structure and eps a study on selected financial institutions listed o...
Capital structure and eps a study on selected financial institutions listed o...Alexander Decker
 
30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH RECENT ADVANCES AND FUTURE OPP...
30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH  RECENT ADVANCES AND FUTURE OPP...30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH  RECENT ADVANCES AND FUTURE OPP...
30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH RECENT ADVANCES AND FUTURE OPP...Rhonda Cetnar
 
Whom You Know Matters in Venture Capital
Whom You Know Matters in Venture CapitalWhom You Know Matters in Venture Capital
Whom You Know Matters in Venture CapitalBreatheBusiness
 
EFFECT OF DIVIDENDS ON STOCK PRICES
EFFECT OF DIVIDENDS ON STOCK PRICES EFFECT OF DIVIDENDS ON STOCK PRICES
EFFECT OF DIVIDENDS ON STOCK PRICES Long Nguyen
 
International Journal of Business and Management Invention (IJBMI)
International Journal of Business and Management Invention (IJBMI)International Journal of Business and Management Invention (IJBMI)
International Journal of Business and Management Invention (IJBMI)inventionjournals
 

Similar to Fund performance persistence and competition keswani (20)

Selection termination goyalwahal
Selection termination goyalwahalSelection termination goyalwahal
Selection termination goyalwahal
 
Fund pick activeshare
Fund pick activeshareFund pick activeshare
Fund pick activeshare
 
Ssrn id1685942
Ssrn id1685942Ssrn id1685942
Ssrn id1685942
 
Ssrn id1685942
Ssrn id1685942Ssrn id1685942
Ssrn id1685942
 
Cap srtucture hotdebt
Cap srtucture hotdebtCap srtucture hotdebt
Cap srtucture hotdebt
 
07. the determinants of capital structure
07. the determinants of capital structure07. the determinants of capital structure
07. the determinants of capital structure
 
Shahbaz
ShahbazShahbaz
Shahbaz
 
Evaluation of the Development and Performance of Selected GCC and Non-GCC St...
Evaluation of the Development and Performance  of Selected GCC and Non-GCC St...Evaluation of the Development and Performance  of Selected GCC and Non-GCC St...
Evaluation of the Development and Performance of Selected GCC and Non-GCC St...
 
Exploring Index Effects - Tamas Toth
Exploring Index Effects - Tamas TothExploring Index Effects - Tamas Toth
Exploring Index Effects - Tamas Toth
 
The Shift from Active to Passive Investment
The Shift from Active to Passive InvestmentThe Shift from Active to Passive Investment
The Shift from Active to Passive Investment
 
Short term persistence in mutual fund performance(12)
Short term persistence in mutual fund performance(12)Short term persistence in mutual fund performance(12)
Short term persistence in mutual fund performance(12)
 
Fund performance wermers
Fund performance wermersFund performance wermers
Fund performance wermers
 
邬梦曲论文
邬梦曲论文邬梦曲论文
邬梦曲论文
 
Aerospace and Defense Value Creators Report 2015
Aerospace and Defense Value Creators Report 2015Aerospace and Defense Value Creators Report 2015
Aerospace and Defense Value Creators Report 2015
 
Capital structure and eps a study on selected financial institutions listed o...
Capital structure and eps a study on selected financial institutions listed o...Capital structure and eps a study on selected financial institutions listed o...
Capital structure and eps a study on selected financial institutions listed o...
 
30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH RECENT ADVANCES AND FUTURE OPP...
30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH  RECENT ADVANCES AND FUTURE OPP...30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH  RECENT ADVANCES AND FUTURE OPP...
30 YEARS OF MERGERS AND ACQUISITIONS RESEARCH RECENT ADVANCES AND FUTURE OPP...
 
Whom You Know Matters in Venture Capital
Whom You Know Matters in Venture CapitalWhom You Know Matters in Venture Capital
Whom You Know Matters in Venture Capital
 
EFFECT OF DIVIDENDS ON STOCK PRICES
EFFECT OF DIVIDENDS ON STOCK PRICES EFFECT OF DIVIDENDS ON STOCK PRICES
EFFECT OF DIVIDENDS ON STOCK PRICES
 
E02101035043
E02101035043E02101035043
E02101035043
 
International Journal of Business and Management Invention (IJBMI)
International Journal of Business and Management Invention (IJBMI)International Journal of Business and Management Invention (IJBMI)
International Journal of Business and Management Invention (IJBMI)
 

More from bfmresearch

Impact ofmutualfundclosuresonfundmanagers
Impact ofmutualfundclosuresonfundmanagersImpact ofmutualfundclosuresonfundmanagers
Impact ofmutualfundclosuresonfundmanagersbfmresearch
 
Standard & poor's 16768282 fund-factors-2009 jan1
Standard & poor's 16768282 fund-factors-2009 jan1Standard & poor's 16768282 fund-factors-2009 jan1
Standard & poor's 16768282 fund-factors-2009 jan1bfmresearch
 
Performance emergingfixedincomemanagers joi_is age just a number
Performance emergingfixedincomemanagers joi_is age just a numberPerformance emergingfixedincomemanagers joi_is age just a number
Performance emergingfixedincomemanagers joi_is age just a numberbfmresearch
 
Portfolio turnover white paper
Portfolio turnover white paperPortfolio turnover white paper
Portfolio turnover white paperbfmresearch
 
Mauboussin skill manager
Mauboussin skill managerMauboussin skill manager
Mauboussin skill managerbfmresearch
 
Is alphadead researchnote
Is alphadead researchnoteIs alphadead researchnote
Is alphadead researchnotebfmresearch
 
Fis group study on emerging managers performance drivers 2007
Fis group   study on  emerging managers performance drivers 2007Fis group   study on  emerging managers performance drivers 2007
Fis group study on emerging managers performance drivers 2007bfmresearch
 
Barclays manager selection0312
Barclays   manager selection0312Barclays   manager selection0312
Barclays manager selection0312bfmresearch
 
Active managementmostlyefficientmarkets faj
Active managementmostlyefficientmarkets fajActive managementmostlyefficientmarkets faj
Active managementmostlyefficientmarkets fajbfmresearch
 
2012 0224 active share
2012 0224 active share2012 0224 active share
2012 0224 active sharebfmresearch
 
12 182-china webcast
12 182-china webcast12 182-china webcast
12 182-china webcastbfmresearch
 
Scoring For Returns-Stuart Investment
Scoring For Returns-Stuart InvestmentScoring For Returns-Stuart Investment
Scoring For Returns-Stuart Investmentbfmresearch
 
Persistence inmutualfundperformance carhart
Persistence inmutualfundperformance carhartPersistence inmutualfundperformance carhart
Persistence inmutualfundperformance carhartbfmresearch
 
Ownership and fund performance evans
Ownership and fund performance evansOwnership and fund performance evans
Ownership and fund performance evansbfmresearch
 
Information ratio mgrevaluation_bossert
Information ratio mgrevaluation_bossertInformation ratio mgrevaluation_bossert
Information ratio mgrevaluation_bossertbfmresearch
 
Performance teammgmtvsindividual bliss
Performance teammgmtvsindividual blissPerformance teammgmtvsindividual bliss
Performance teammgmtvsindividual blissbfmresearch
 
Returns basedstyleanalysisinexcel mcdermott
Returns basedstyleanalysisinexcel mcdermottReturns basedstyleanalysisinexcel mcdermott
Returns basedstyleanalysisinexcel mcdermottbfmresearch
 

More from bfmresearch (20)

Impact ofmutualfundclosuresonfundmanagers
Impact ofmutualfundclosuresonfundmanagersImpact ofmutualfundclosuresonfundmanagers
Impact ofmutualfundclosuresonfundmanagers
 
Standard & poor's 16768282 fund-factors-2009 jan1
Standard & poor's 16768282 fund-factors-2009 jan1Standard & poor's 16768282 fund-factors-2009 jan1
Standard & poor's 16768282 fund-factors-2009 jan1
 
Performance emergingfixedincomemanagers joi_is age just a number
Performance emergingfixedincomemanagers joi_is age just a numberPerformance emergingfixedincomemanagers joi_is age just a number
Performance emergingfixedincomemanagers joi_is age just a number
 
Spiva mid2011
Spiva mid2011Spiva mid2011
Spiva mid2011
 
Portfolio turnover white paper
Portfolio turnover white paperPortfolio turnover white paper
Portfolio turnover white paper
 
Mauboussin skill manager
Mauboussin skill managerMauboussin skill manager
Mauboussin skill manager
 
Jp littlebook
Jp littlebookJp littlebook
Jp littlebook
 
Is alphadead researchnote
Is alphadead researchnoteIs alphadead researchnote
Is alphadead researchnote
 
Fis group study on emerging managers performance drivers 2007
Fis group   study on  emerging managers performance drivers 2007Fis group   study on  emerging managers performance drivers 2007
Fis group study on emerging managers performance drivers 2007
 
Barclays manager selection0312
Barclays   manager selection0312Barclays   manager selection0312
Barclays manager selection0312
 
Active managementmostlyefficientmarkets faj
Active managementmostlyefficientmarkets fajActive managementmostlyefficientmarkets faj
Active managementmostlyefficientmarkets faj
 
2012 0224 active share
2012 0224 active share2012 0224 active share
2012 0224 active share
 
12 182-china webcast
12 182-china webcast12 182-china webcast
12 182-china webcast
 
Vanguard dc
Vanguard dcVanguard dc
Vanguard dc
 
Scoring For Returns-Stuart Investment
Scoring For Returns-Stuart InvestmentScoring For Returns-Stuart Investment
Scoring For Returns-Stuart Investment
 
Persistence inmutualfundperformance carhart
Persistence inmutualfundperformance carhartPersistence inmutualfundperformance carhart
Persistence inmutualfundperformance carhart
 
Ownership and fund performance evans
Ownership and fund performance evansOwnership and fund performance evans
Ownership and fund performance evans
 
Information ratio mgrevaluation_bossert
Information ratio mgrevaluation_bossertInformation ratio mgrevaluation_bossert
Information ratio mgrevaluation_bossert
 
Performance teammgmtvsindividual bliss
Performance teammgmtvsindividual blissPerformance teammgmtvsindividual bliss
Performance teammgmtvsindividual bliss
 
Returns basedstyleanalysisinexcel mcdermott
Returns basedstyleanalysisinexcel mcdermottReturns basedstyleanalysisinexcel mcdermott
Returns basedstyleanalysisinexcel mcdermott
 

Recently uploaded

VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...
VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...
VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...Suhani Kapoor
 
Q3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast SlidesQ3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast SlidesMarketing847413
 
Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...
Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...
Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...makika9823
 
House of Commons ; CDC schemes overview document
House of Commons ; CDC schemes overview documentHouse of Commons ; CDC schemes overview document
House of Commons ; CDC schemes overview documentHenry Tapper
 
(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办
(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办
(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办fqiuho152
 
AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...
AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...
AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...yordanosyohannes2
 
(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
 
VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130
VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130
VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130Suhani Kapoor
 
Log your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaignLog your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaignHenry Tapper
 
fca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdffca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdfHenry Tapper
 
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一S SDS
 
Malad Call Girl in Services 9892124323 | ₹,4500 With Room Free Delivery
Malad Call Girl in Services  9892124323 | ₹,4500 With Room Free DeliveryMalad Call Girl in Services  9892124323 | ₹,4500 With Room Free Delivery
Malad Call Girl in Services 9892124323 | ₹,4500 With Room Free DeliveryPooja Nehwal
 
How Automation is Driving Efficiency Through the Last Mile of Reporting
How Automation is Driving Efficiency Through the Last Mile of ReportingHow Automation is Driving Efficiency Through the Last Mile of Reporting
How Automation is Driving Efficiency Through the Last Mile of ReportingAggregage
 
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...shivangimorya083
 
The Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh KumarThe Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh KumarHarsh Kumar
 
call girls in Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️
call girls in  Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️call girls in  Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️
call girls in Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️9953056974 Low Rate Call Girls In Saket, Delhi NCR
 
Attachment Of Assets......................
Attachment Of Assets......................Attachment Of Assets......................
Attachment Of Assets......................AmanBajaj36
 

Recently uploaded (20)

Monthly Economic Monitoring of Ukraine No 231, April 2024
Monthly Economic Monitoring of Ukraine No 231, April 2024Monthly Economic Monitoring of Ukraine No 231, April 2024
Monthly Economic Monitoring of Ukraine No 231, April 2024
 
VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...
VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...
VIP High Class Call Girls Saharanpur Anushka 8250192130 Independent Escort Se...
 
🔝+919953056974 🔝young Delhi Escort service Pusa Road
🔝+919953056974 🔝young Delhi Escort service Pusa Road🔝+919953056974 🔝young Delhi Escort service Pusa Road
🔝+919953056974 🔝young Delhi Escort service Pusa Road
 
Q3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast SlidesQ3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast Slides
 
Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...
Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...
Independent Lucknow Call Girls 8923113531WhatsApp Lucknow Call Girls make you...
 
House of Commons ; CDC schemes overview document
House of Commons ; CDC schemes overview documentHouse of Commons ; CDC schemes overview document
House of Commons ; CDC schemes overview document
 
(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办
(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办
(办理原版一样)QUT毕业证昆士兰科技大学毕业证学位证留信学历认证成绩单补办
 
🔝9953056974 🔝Call Girls In Dwarka Escort Service Delhi NCR
🔝9953056974 🔝Call Girls In Dwarka Escort Service Delhi NCR🔝9953056974 🔝Call Girls In Dwarka Escort Service Delhi NCR
🔝9953056974 🔝Call Girls In Dwarka Escort Service Delhi NCR
 
AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...
AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...
AfRESFullPaper22018EmpiricalPerformanceofRealEstateInvestmentTrustsandShareho...
 
(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(DIYA) Bhumkar Chowk Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
 
VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130
VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130
VIP Call Girls Service Begumpet Hyderabad Call +91-8250192130
 
Log your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaignLog your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaign
 
fca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdffca-bsps-decision-letter-redacted (1).pdf
fca-bsps-decision-letter-redacted (1).pdf
 
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
(办理学位证)加拿大萨省大学毕业证成绩单原版一比一
 
Malad Call Girl in Services 9892124323 | ₹,4500 With Room Free Delivery
Malad Call Girl in Services  9892124323 | ₹,4500 With Room Free DeliveryMalad Call Girl in Services  9892124323 | ₹,4500 With Room Free Delivery
Malad Call Girl in Services 9892124323 | ₹,4500 With Room Free Delivery
 
How Automation is Driving Efficiency Through the Last Mile of Reporting
How Automation is Driving Efficiency Through the Last Mile of ReportingHow Automation is Driving Efficiency Through the Last Mile of Reporting
How Automation is Driving Efficiency Through the Last Mile of Reporting
 
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
 
The Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh KumarThe Triple Threat | Article on Global Resession | Harsh Kumar
The Triple Threat | Article on Global Resession | Harsh Kumar
 
call girls in Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️
call girls in  Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️call girls in  Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️
call girls in Nand Nagri (DELHI) 🔝 >༒9953330565🔝 genuine Escort Service 🔝✔️✔️
 
Attachment Of Assets......................
Attachment Of Assets......................Attachment Of Assets......................
Attachment Of Assets......................
 

Fund performance persistence and competition keswani

  • 1. The Journal of Financial Research • Vol. XXIX, No. 3 • Pages 349–366 • Fall 2006 MUTUAL FUND PERFORMANCE PERSISTENCE AND COMPETITION: A CROSS-SECTOR ANALYSIS Aneel Keswani Cass Business School David Stolin Toulouse Business School Abstract Existing work on mutual fund performance persistence obtains diverse results, depending on the group of funds studied. We examine whether performance per- sistence within a peer group of competing mutual funds depends on the group’s composition. The U.K. mutual fund industry is ideal for such an examination be- cause funds compete within strictly defined sectors. We consider several attributes related to the intensity of competition within a sector and use them to explain sector-level persistence. We find robust evidence that persistence is higher in sectors where concentration of assets under management is higher. JEL Classification: G23 I. Introduction It is well established in the industrial organization literature that the structure of a sector affects its competitiveness. In more competitive sectors we expect to see few firms doing persistently well and those performing poorly being forced to exit the sector. This reasoning is tested by Waring (1996), who finds a strong negative relation between competitiveness within an industrial sector and the persistence of profitability for firms in that sector. We translate this logic to the mutual fund context. To use the terminology of industrial organization, mutual funds compete with each other using a combination of price and nonprice competition strategies. Price competition involves funds We thank Vladimir Atanasov, Andrew Clare, Zsuzsanna Fluck, Gordon Gemmill, Brian Kluger, Tobias Kretschmer, Gordon Midgley, Kenneth Moon, Dennis Stanton, Dylan Thomas, Giovanni Urga, and especially William T. Moore (former editor) and Jonathan Fletcher (the referee) for insightful comments. We also acknowledge comments received from participants at the 2003 Financial Management Association meeting in Denver, and seminars at Cass Business School and the universities of Porto, Reading, Warwick, and Oxford. We thank Benjamin Kogan, Jan Steinberg, James Sullivan, the Allenbridge Group, and the Investment Management Association for help with data. Part of the research reported here was conducted while Stolin was visiting at the Stockholm Institute for Financial Research. All errors and omissions are ours. 349
  • 2. 350 The Journal of Financial Research varying the fees they charge to obtain a competitive advantage. Nonprice com- petition involves (among other things) funds competing to produce superior invest- ment performance. Numerous studies show that higher investment returns have an important influence on fund market share (e.g., Siggelkow 2003). We expect funds from more competitive sectors to compete more aggres- sively for abnormal returns. This should result in the exit of funds that underperform and a low probability of remaining funds doing repeatedly well. Competing funds should be able to close the performance gap on “star” funds by devoting more resources to researching investment opportunities, by learning to imitate the best performers, or even by poaching their managers. Thus, in more competitive sectors we expect to see less persistence in funds’ performance relative to their rivals (i.e., less relative persistence). To examine the influence of competition on investment performance per- sistence, we focus on the U.K. unit trust (open-ended mutual fund) industry. This environment is ideal for our purpose because U.K. mutual funds compete in a large number of unambiguously defined peer groups (sectors), whose membership is monitored and enforced by the industry trade body. This is unlike the United States, where multiple sector definitions coexist and managers are free to game their sector affiliations (Cooper, Gulen, and Rau 2005). Although research into fund performance persistence has a long history, Brown et al. (1992) show that early studies exaggerate the extent of persistence by re- lying on survivorship-biased data sets. Carhart (1997) finds that in his survivorship- free sample of U.S. equity funds, persistence largely disappears after accounting for momentum in stock returns. However, recent studies argue that after properly considering fund styles, there is persistence in U.S. equity mutual funds (Ibbotson and Patel 2002; Teo and Woo 2001; Wermers 2003). Outside the United States, there has been debate as well. In the United Kingdom, it has involved academics (Blake and Timmermann 1998; Allen and Tan 1999; Fletcher and Forbes 2002), practitioners (Quigley and Sinquefield 2000), the trade association (Giles, Wilsdon, and Worboys 2002), and the regulatory body (Rhodes 2000; Blake and Timmermann 2003). This literature agrees that perfor- mance persistence is an important issue but disagrees on whether and to what extent persistence is present. The preceding studies all focus on funds investing in domestic equity secu- rities. The availability of well-accepted benchmarks for risk adjustment is a major reason for this focus. Several studies examine persistence for funds investing in other asset classes (e.g., see Blake, Elton, and Gruber 1993 for bond mutual funds) and obtain diverse results, depending on the period used and the fund sector stud- ied. This raises the possibility that levels of persistence may vary depending on the economic circumstances. In particular, the market structure of a mutual fund sector may influence funds’ ability to perform consistently. Our research examines how mutual fund performance persistence at the fund sector level is influenced by competition within the sector. A few studies
  • 3. Mutual Fund Performance Persistence 351 consider determinants of persistent performance at the individual fund level (e.g., Volkman and Wohar 1995). Several other studies note persistence differences across sectors or fund objectives (Blake and Timmermann 1998; Kosowski et al. 2003; Wermers 2003). No study, however, conducts a sector-level statistical analysis of persistence, and none investigates the effect of competition on persistence. Massa’s (2003) empirical demonstration that sector-level variables related to competition ex- plain sector-level performance suggests that such an analysis is potentially fruitful. We construct several variables to capture the intensity of competition in a sector. These include the number of funds in a sector, the proportion of mature funds, and the Herfindahl index of asset concentration. We find robust evidence that persistence is higher in sectors where concentration of assets under management is higher. Our results suggest that the degree of persistence exhibited by a sector’s investment managers depends on how competitive that sector is. II. Data and Method The U.K. Mutual Fund Industry Unlike in the United States, a survivorship-bias-free electronic database of mutual funds does not exist in the United Kingdom. To conduct our study, we therefore manually collected data from 11 consecutive editions of the annual Unit Trust Yearbook.1 Our data span from 1991 to 2001 and include names of funds and their management groups, annual returns (including reinvested income and excluding fees), fund assets under management, launch dates, and of course the name of the sector to which each fund belongs. We use fund names and an index of name changes to link fund data across years. We consider mergers between funds as creating a new fund. As our primary focus is at the sector level (rather than at the individual fund level), we track the evolution and membership of official fund sectors as defined by the Association of Unit Trust and Investment Funds (AUTIF) and by its successor, the Investment Management Association (IMA).2 To do this, we use data on fund movement across sectors, as well as historical announcements by AUTIF and IMA. Appendix A summarizes the history of U.K. mutual fund sectors. We use 1 In the editions corresponding to 2000 and 2001 year-ends, the yearbook had a new publisher, and several smaller fund families did not supply information on their funds. However, there is no survivorship bias due to selective reporting of funds. Post-2001 data are unavailable as the yearbook has been discontinued. 2 In the United Kingdom, all information providers use the official classification system. The IMA enforces its sector definitions, and if the asset allocation of a fund contravenes the allocation rules of its current sector, the IMA will warn the fund to change its allocation if it does not wish to change sectors. If the fund does not comply, the IMA will move the fund to a new sector reflecting its new asset allocation. By contrast, in the United States there is a proliferation of methods for assigning funds to a peer group. This ambiguity allows fund managers to “game” their objectives (Cooper, Gulen, and Rau 2005) and makes objective-level measures of competition less meaningful.
  • 4. 352 The Journal of Financial Research official sector descriptions to group sectors into four broad categories: domestic equity, global equity, domestic nonequity, and global nonequity. The appendix paints a picture of substantial innovation at the sector level—with numerous instances of sectors being opened, discontinued, redefined, or merged—consistent with an industry seeking to respond to changing conditions. Additionally, there is much variability in the number of funds within a sector. Our sample period thus captures an industry in transition, which is helpful for our analysis of the role of market structure characteristics. Measurement of Persistence Measures of performance persistence quantify to what extent performance in one period (the “ranking” period) continues into the subsequent period (the “evaluation” period). In this study, we focus on persistence at the one-year frequency (i.e., our ranking and evaluation periods each equals one year). There are several reasons for this choice. First, researchers who find evidence of persistence generally find it for one-year horizons. Second, investors and fund managers tend to evaluate performance over annual periods. Third, tests of performance persistence require return availability for both ranking and evaluation periods. This leads to a look- ahead bias, which can influence how much persistence is detected (Brown et al. 1992; Ter Horst, Nijman, and Verbeek 2001). Lengthening the horizon over which persistence is measured makes this bias more severe. Over one-year periods, Ter Horst, Nijman, and Verbeek (2001) find the look-ahead bias to be negligible. Performance persistence can be measured using both absolute and relative performance. We measure persistence using relative performance for two reasons. First, existing research highlights that in determining mutual fund money flows, relative performance matters beyond absolute performance. Second, measures of absolute performance persistence depend on the volatility of securities invested in by a given sector, making comparisons of absolute persistence across sectors misleading. To measure relative performance persistence, we use raw and not risk- adjusted returns. In our context, examining persistence on a risk-adjusted basis is problematic for two reasons. First, we do not have access to monthly returns for existing and extinct U.K. mutual funds. Second, and more important, the quality of any risk adjustment would inevitably vary across sectors. For example, domestic equity returns can be analyzed with well-researched multifactor models, whereas this is less likely for global or nonequity funds. This means that sector characteristics related to our ability to risk-adjust would have a spurious effect on a cross-sectional analysis of persistence in risk-adjusted returns. Persistence measured on the basis of raw returns, on the other hand, is important in its own right. Numerous information providers such as Money Management, Unit Trust Yearbook, Standard & Poor’s Web site, and others rank funds based on raw returns within a sector. Indeed, evidence
  • 5. Mutual Fund Performance Persistence 353 on the return-flow relation indicates that investors react to raw returns. Moreover, implicit in looking at within-sector persistence, as we do, is a peer-group adjustment of fund returns. Commonly used statistics for studying relative persistence within a peer group include the Spearman rank-correlation coefficient, and quantities based on 2 × 2 winner/loser contingency tables. The latter include the log-odds ratio and the chi-squared statistic. The chi-squared statistic is disqualified for our purpose (which is to explain the extent of persistence) because high values correspond to either persistence or reversal of performance. The advantage of the Spearman correlation over the log-odds ratio is that the latter uses the performance rank of each fund rather than just its winner/loser status. This generally means more powerful tests for persistence (Carpenter and Lynch 1999). The advantage of the log-odds ratio is that it has a more straightforward economic interpretation, as we show shortly. We use both the log-odds ratio and the Spearman correlation as our measures of persistence (equations are given in Appendix B). To avoid our results being influenced by the small-sample properties of these statistics, we use only sector-years with at least 20 funds in existence over both years for which performance is measured. Table 1 shows the extent of relative performance persistence across U.K. mutual fund sectors based on raw returns over consecutive years. In Panel A, we present the distribution of the Spearman correlation coefficient and of the log-odds ratio by type of sector. The first group of eight rows pertains to the log-odds ratio. In column 1, for all sector-years combined (162), the average log- odds ratio is 0.357. The null hypothesis that the mean log-odds ratio is zero can be rejected ( p-value < .001) based on applying Student’s t-test to our set of 162 sector-years. The median sector-year has a log-odds ratio of 0.405, and the dis- tribution ranges from −2.837 to 4.317. The log-odds ratio is positive for 62% of the sector-years. Furthermore, the table reports the proportion of sector-years for which the hypothesis of no persistence is rejected in favor of the one-sided alter- native of positive persistence. For the log-odds ratio, this is the case for 29% of the sector-years at the .05 confidence level, and for 17% of the sector-years at the .01 confidence level. The next eight rows characterize the distribution of the Spearman corre- lation across sector-years. For all sector-years combined, the average Spearman correlation is 0.143 and is significantly different from zero. We note that even at the .01 confidence level, the hypothesis of no persistence is rejected in favor of the hypothesis of positive persistence for 29% of sector-years. This suggests the Spearman-based test is more powerful than the log-odds ratio. To ensure that performance persistence in our sample is not driven by a particular subperiod, we separately consider sector-years for which the evaluation years are 1992 through 1996, and those for which the evaluation years are 1997 through 2001 (results not reported in a table). The average log-odds ratio for the
  • 6. 354 The Journal of Financial Research TABLE 1. Relative Performance Persistence Across Sectors. All Sectors Domestic Other Than Global Domestic Global All Equity Domestic Equity Nonequity Nonequity Sectors Sectors Equity Sectors Sectors Sectors Variable (1) (2) (3) (4) (5) (6) Panel A. Sector-Year Statistics Number of sector-years 162 36 126 63 30 33 Log-odds ratio by sector-year Average 0.357 0.448 0.331 0.173 0.565 0.421 p-value for H0 : mean = 0 0.000 0.006 0.002 0.192 0.019 0.057 Median 0.405 0.382 0.405 0.365 0.555 0.525 Minimum −2.837 −1.168 −2.837 −2.837 −1.455 −2.485 Maximum 4.317 2.711 4.317 2.398 3.008 4.317 Proportion positive 0.623 0.639 0.619 0.571 0.667 0.667 Proportion positive and 0.290 0.306 0.286 0.270 0.367 0.242 significant at .05 level Proportion positive and 0.167 0.306 0.127 0.143 0.133 0.091 significant at .01 level Spearman correlation by sector-year Average 0.143 0.147 0.142 0.097 0.180 0.191 p-value for H0 : mean = 0 0.000 0.004 0.000 0.017 0.005 0.001 Median 0.162 0.135 0.166 0.133 0.257 0.176 Minimum −0.642 −0.477 −0.642 −0.642 −0.544 −0.476 Maximum 0.817 0.804 0.817 0.672 0.656 0.817 Proportion positive 0.691 0.583 0.722 0.698 0.733 0.758 Proportion positive and 0.426 0.472 0.413 0.429 0.433 0.364 significant at .05 level Proportion positive and 0.290 0.333 0.278 0.254 0.267 0.333 significant at .01 level Panel B. Aggregate Statistics Fund-years in winner-winner 2,724 900 1,824 1,240 286 298 category Fund-years in loser-loser 2,672 895 1,777 1,219 273 285 category Fund-years in winner-loser 2,276 728 1,548 1,090 225 233 category Fund-years in loser-winner 2,296 740 1,556 1,096 229 231 category Aggregate log-odds ratio 0.331 0.402 0.297 0.235 0.416 0.456 p-value for aggregate log-odds 0.000 0.000 0.000 0.000 0.001 0.000 ratio Frequency of repeat performance 0.541 0.550 0.537 0.529 0.552 0.557 Note: This table reports descriptive statistics for measures of relative performance persistence across U.K. mutual fund sectors, 1991–2001. Sector-years are included if at least 20 funds had returns available in the ranking and evaluation years. Panel A reports the distribution of the Spearman correlation coefficient and the log-odds ratio across sector-years. Both the Spearman correlation coefficient and the log-odds ratio are based on raw annual returns in consecutive calendar years (formulae are given in Appendix B). In Panel B, sector-years are pooled into an aggregate contingency table.
  • 7. Mutual Fund Performance Persistence 355 earlier (later) period is 0.358 (0.357), and the average Spearman correlation coef- ficient is 0.134 (0.152). All of these averages are statistically significant at the .01 level. Moreover, the average log-odds ratio and the average Spearman correlation coefficient are not significantly different between the two periods ( p-values = .99 and .70, respectively). Because research on mutual fund performance persistence tends to focus on domestic equity funds, we separately report results for these sectors in the second column. The persistence measures are positive and, despite a sample size of only 36 sector-years, highly statistically significant. The average log-odds ratio, at 0.448, is slightly lower than the 0.516 average log-odds ratio in Fletcher and Forbes (2002), which is based on raw annual returns for U.K. equity mutual funds from 1982 to 1996. The average Spearman correlation, at 0.147, is slightly lower than the 0.188 reported by Allen and Tan (1999) for raw annual returns of U.K. equity mutual funds from 1989 to 1995. Column 3 presents results for sectors other than domestic equity. The level of persistence is comparable to that in the preceding column. In fact, unreported tests show that differences between the two columns are never significant. The last three columns further disaggregate sectors other than domestic equity into global equity, domestic nonequity, and global nonequity. In each category, perfor- mance persistence is positive and significant, at least for the Spearman correlation coefficient. As further evidence on the level of performance persistence in our sample, in Panel B we pool data from different sector-years to present an aggregate con- tingency table. In 2,724 (2,672) instances, funds are two-period winners (losers) in their respective sectors, and in 2,276 (2,296) instances, funds win in the rank- ing (evaluation) period and lose in the evaluation (ranking) period. The resulting aggregate log-odds ratio equals ln((2724 × 2672)/(2276 × 2296)) = 0.331 and is highly significant. The aggregate log-odds ratios for the different sector groups in columns 2 through 6 are all significant at the .01 level. The economic significance of performance persistence in our sample is perhaps best addressed through the probability that a fund’s winner/loser status carries over from the ranking period to the evaluation period. This probability of repeat performance can be estimated as the number of fund-years corresponding to two-period winners or two-period losers divided by the total number of fund- years. For all sectors together, this quantity (reported in the last row of the table) equals (2,724 + 2,672)/(2,724 + 2,672 + 2,276 + 2,296) = 54.1%, as compared to the 50.0% that one would expect in the absence of performance persistence or performance reversal.3 3 Because we define a winner (loser) as a fund that places in the top (bottom) half in its sector in a given year, the cell counts in the contingency table are not independent. In fact, if there were no ties and if the number of funds in a sector were always divisible by four, the winner-winner fund count would be
  • 8. 356 The Journal of Financial Research Overall, there is strong evidence that the U.K. mutual fund industry exhibits persistence in relative investment performance. If at least some of this persistence is due to sector-level attributes, in particular to those related to sector competitiveness, a cross-sector analysis may reveal this. Such analysis is conducted in the next section. III. Determinants of Sector-Level Persistence Sector Attributes Broadly speaking, systematic differences in persistence between sectors can be due to differences in the composition of sector membership, or to differences in the types of assets sector members invest in. We construct several variables designed to quantify how competitive a sector is, and the distribution of these variables is given in Panel A of Table 2. N is simply the number of funds in a sector at the end of the ranking year. The largest number of funds in a sector is 302, corresponding to the UK All Companies sector in 1999 (after sectors dedicated to domestic “growth” and “growth and income” stocks were merged). Because we drop sectors comprising fewer than 20 funds with recorded returns, the minimum number of funds in a sector is 24, the median is 79, and the average is 87.4 It is reasonable to conjecture that consistent performance is harder to attain in a more crowded sector. For example, in studying fund performance Siggelkow (2003) regards the number of mutual funds in a category as “a measure of general competition, for instance, for mis-priced securities” (p. 133). We recognize, however, that in such competition, small funds may have rel- atively little effect. We therefore also use the Herfindahl index, which is commonly considered as a measure of intra-industry rivalry. Specifically, HERFINDAHL is the concentration index of assets under management. Because several funds from a single family of funds can coexist within a sector, we aggregate assets by fam- ily to calculate this measure. Thus, the value of HERFINDAHL for each sector is the sum across families of the square of each family’s assets as a proportion of a sector’s total assets. Although we use only sectors with at least 20 funds, there is substantial variation in the value of the Herfindahl index, ranging from 0.027 to 0.629 (by construction, the smallest possible value of the Herfindal index is 0 exactly equal to the loser-loser count, and the winner-loser count would be exactly equal to the loser-winner count. Using this insight, it is straightforward to show that the probability of repeat performance can be obtained directly from the log-odds ratio (L) as 1/(1 + e−L/2 ). For example, using the aggregate log-odds ratio of 0.331, the probability of repeat performance is 1/(1 + e−0.331/2 ) = 0.541. We subsequently use this conversion to assess the economic significance of our regression results. 4 For comparison, Massa (2003) uses several data providers’ fund descriptions to assign U.S. mutual funds to 1 of 23 categories. The numbers of funds in his categories range from 14 to 1,149, the median is 343, and the average is 411.
  • 9. Mutual Fund Performance Persistence 357 TABLE 2. Descriptive Statistics for Sector-Level Variables. Panel A. Moments Variable Mean Standard Deviation Median Minimum Maximum N 87 50 79 24 302 MATURITY 0.552 0.188 0.626 0.000 0.846 HERFINDAHL 0.088 0.070 0.068 0.027 0.629 Panel B. Correlations Variable N MATURITY HERFINDAHL LATER DOMESTIC EQUITY 0.420 0.420 −0.247 −0.007 N 0.476 −0.464 −0.163 MATURITY −0.586 −0.040 HERFINDAHL −0.078 Note: This table contains descriptive statistics for the set of sector-year explanatory variables. Sector-years are included if at least 20 funds had returns available in the ranking and evaluation years: 162 sector-years meet this requirement. The variables are as follows. N is the number of funds within the sector. MATURITY is the proportion of sector funds that are at least five years old. HERFINDAHL is the Herfindahl index measuring the concentration of fund assets within the sector, where funds from the same family are aggregated. DOMESTIC EQUITY is a dummy variable equal to 1 if sector funds are primarily invested in U.K. equities, and 0 otherwise. and the largest possible value is 1). We hypothesize that less concentrated (more competitive) sectors exhibit lower persistence. Finally, MATURITY is the proportion of funds that are at least five years old. In the average sector, most funds are “seasoned,” but the minimum value of 0 for the maturity variable indicates that for some sector-years, all of the funds are relatively recent entrants. Berk and Green (2004) give a powerful reason why mutual fund performance persistence should decrease with fund vintage. If investment management returns to scale are decreasing, managers have differential ability, and investors channel money to best performers, then superior funds grow to the point where outperformance is no longer possible. Empirically, Waring (1996) finds that earnings persistence in an industrial sector tends to decay over time, as competitive forces have acted over a longer period. Panel B of Table 2 shows a correlation matrix for the preceding sector attributes and for a dummy variable indicating sector membership in the U.K. equity category (DOMESTIC EQUITY ), as well as a dummy variable that equals 1 for the second half of our sample period (evaluation years from 1997 to 2001), and 0 otherwise (LATER). DOMESTIC EQUITY is included because most studies of performance persistence focus on domestic equity sectors. LATER controls for the possibility that the level of persistence may have changed in more recent years. We note that pairwise correlations between N, MATURITY, and HERFINDAHL are high in magnitude: sectors with more funds in them tend to be more mature, and
  • 10. 358 The Journal of Financial Research assets invested in these sectors are more dispersed across fund families. Therefore, in the regressions to follow, we enter these three variables one at a time. Regression Results Table 3 presents the results of a pooled regression of sector-level measures of relative persistence on our set of sector-level explanatory variables. Persistence is measured over years T (the ranking year) and T+1 (the evaluation year). Explanatory variables are measured as of the end of year T (with one exception explained below). Thus, we examine whether sector characteristics observed at the end of year T tell us to what extent year T performance of the sector’s funds persist into year T+1. We do not require that funds remain in the same sector until the end of year T+1 (or, indeed, that the sector itself continue to exist until the end of year T+1) because doing so would constitute a look-ahead bias. In regressions (1) through (3) we use the log-odds ratio as the measure of persistence and proxy for sector competitiveness with N, MATURITY, or HERFINDAHL, respectively. Although the number of funds in a sector is not signif- icantly related to persistence, maturity of funds is significant (t-statistic = −2.71), as is the concentration of assets under management (t-statistic = 3.01). In other words, sectors that are less mature and have more concentrated assets—that is, sectors that may be described as less competitive—are characterized by greater persistence. The other variables are not statistically significant. Regressions (4) through (6) parallel regressions (1) through (3) but include an additional control variable. CROSSRET is defined as the product of average sector returns in years T and T+1. Although we do not adjust for differences in fund exposure to different risk factors, these differences can generate persistence in raw returns when there is persistence in factor realizations. CROSSRET is intended to capture spurious persistence due to ex post momentum for the sector as a whole. The regression results confirm this intuition in that CROSSRET is positive and highly significant ( p-value < .001). Its only other influence is to enhance slightly the significance of MATURITY and HERFINDAHL (t-statistics = −2.80 and 3.24, respectively). In regressions (7) through (12), the Spearman correlation coefficient is the dependent variable. The results are similar to those based on the log-odds ratio. Once again, MATURITY and HERFINDAHL are statistically significant at the .01 level, and CROSSRET continues to capture persistence due to momentum in all specifications. Robustness Checks The preceding subsection presents evidence that HERFINDAHL and MATURITY are sector attributes that are systematically related to the persistence exhibited by the sector. We now report on the robustness of our results to alternative sample- selection criteria, econometric methods, and other variations.
  • 11. TABLE 3. Explaining Sector-Level Persistence. Dependent Variable: Log-Odds Ratio Dependent Variable: Spearman Correlation Coefficient Explanatory Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Intercept 0.586 1.097 0.020 0.568 1.040 −0.070 0.214 0.348 0.048 0.208 0.328 0.016 (2.93)∗∗∗ (3.90)∗∗∗ (0.12) (2.90)∗∗ (3.76)∗∗∗ (−0.40) (3.82)∗∗∗ (4.41)∗∗∗ (0.98) (3.84)∗∗∗ (4.29)∗∗∗ (0.34) DOMESTIC EQUITY 0.224 0.302 0.270 0.257 0.306 0.279 0.038 0.057 0.051 0.050 0.058 0.054 (0.98) (1.40) (1.29) (1.14) (1.45) (1.36) (0.59) (0.93) (0.87) (0.80) (0.99) (0.96) N −0.003 −0.004 −0.001 −0.001 (−1.13) (−1.48) (−1.21) (−1.66) MATURITY −1.297 −1.315 −0.353 −0.359 (−2.71)∗∗∗ (−2.80)∗∗∗ (−2.63)∗∗∗ (−2.76)∗∗∗ HERFINDAHL 3.780 3.975 1.104 1.172 (3.01)∗∗∗ (3.24)∗∗∗ (3.15)∗∗∗ (3.47)∗∗∗ LATER −0.225 −0.226 −0.139 −0.175 −0.168 −0.075 −0.060 −0.058 −0.033 −0.041 −0.037 −0.011 (−1.25) (−1.30) (−0.80) (−0.99) (−0.98) (−0.44) (−1.16) (−1.18) (−0.68) (−0.84) (−0.78) (−0.22) CROSSRET 4.271 4.062 4.272 1.489 1.422 1.485 (2.80) (2.73)∗∗∗ (2.89)∗∗∗ (3.53)∗∗∗ (3.45)∗∗∗ (3.64)∗∗∗ R2 0.017 0.053 0.063 0.063 0.090 0.110 0.015 0.047 0.064 0.087 0.114 0.137 Note: This table contains the results of regressing measures of persistence in relative investment performance on sector-level variables. Sector-years are included if at least 20 funds have returns available in the ranking and evaluation years: 162 sector-years meet this requirement. The explanatory variables are as follows. N is the number of funds within a sector. MATURITY is the proportion of sector funds that are at least five years old. HERFINDAHL is the Herfindahl index measuring the concentration of fund assets Mutual Fund Performance Persistence within a sector, where funds from the same family are aggregated. DOMESTIC EQUITY is a dummy variable equal to 1 if sector funds are primarily invested in U.K. equities, and 0 otherwise. LATER is a dummy variable equal to 1 when the evaluation year is 1997 or later, and 0 otherwise. CROSSRET is the product of average sector returns in the ranking and evaluation years. The t-statistics are shown in parentheses. ∗∗∗ Significant at the 1% level. ∗∗ Significant at the 5% level. 359
  • 12. 360 The Journal of Financial Research Because both the log-odds ratio and the Spearman correlation are estimated with differing degrees of precision across sectors, the resulting heteroskedasticity in our regression may lead to inefficient estimation. We therefore use the inverse of the standard error of the log-odds ratio and of the Spearman correlation coefficient as weights in a generalized least squares regression using these measures as dependent variables. The results are not significantly different from those reported earlier. We also investigate whether time-series correlation affects our results. First, we test for serial correlation in a panel but fail to find evidence of this. Second, we include lagged persistence measures in our regressions. This reduces the number of observations from 162 to 136. HERFINDAHL and MATURITY remain significant at the .05 level or better, and N remains insignificant. The lagged persistence measure itself is never statistically significant. To check that our results using Spearman correlation are not influenced by having a dependent variable limited to the [+1,−1] range, we estimate our model using the Papke and Wooldridge (1996) generalized linear approach, which is designed for estimating models with a fractional dependent variable. Our results are broadly unchanged. All coefficient estimates that are significant using ordinary least squares at the .05 level and above are also significant using the new approach, and the signs of all significant coefficient estimates remain the same as before. To address the possibility that our results may be influenced by the small- sample properties of our persistence measures, we exclude sector-years with fewer than 30 funds. This reduces the number of sector-years to 124. When we do this, MATURITY becomes insignificant regardless of the econometric method used. The statistical significance of HERFINDAHL, however, is .05 or better in all specifications. As the coverage of the last two editions of the Unit Trust Yearbook (corre- sponding to calendar years 2000 and 2001) is reduced because of nonreporting by several fund families, we repeat our regressions after omitting these years. We also conduct Fama-MacBeth regressions, drop outlier observations, and use different ranges to winsorize our persistence measures. Our results remain basically un- changed: HERFINDAHL is always significant at least at the .10 level and generally at the .05 level. None of our other variables is consistently significant. Economic Significance Our results indicate that the concentration of funds’ assets is statistically signifi- cantly related to the persistence level in that sector. We now assess the economic significance of this relation. Consider a fund sector not restricted to U.K. equities (DOMESTIC EQUITY = 0) in the second half of our sample period (LATER = 1). When HERFINDAHL is set to its full-sample 10th percentile value of 0.038, using the estimated coefficients in regression (3) of Table 3, the fitted value of the log-odds ratio equals 0.025. Using the conversion formula in footnote 3, this
  • 13. Mutual Fund Performance Persistence 361 translates into a 50.3% probability that a fund’s winner or loser status is retained from the ranking period to the evaluation period. This probability exceeds by only 0.3% the corresponding probability that one would expect by mere chance in the absence of any persistence. We now reset HERFINDAHL to its 90th percentile value of 0.162. The corresponding fitted value of the log-odds ratio is 0.493, which translates into a 56.1% probability of repeat performance, or 6.1% higher than would be expected in the absence of persistence. In other words, if a sector goes from the 10th to the 90th percentile of concentration of assets, the excess (relative to the no-persistence case) probability of remaining in the same half of performance rankings increases from 0.3% to 6.1%. These numbers indicate that the effect of sector-level concentration on performance persistence is substantial in economic terms. IV. Longer Term Persistence Because we find a link between sector characteristics and persistence, we check whether the results hold when persistence is measured over a longer period. To do this, rather than examining adjacent ranking and evaluation periods as we did in the preceding section, we use a lagged ranking period (as in Teo and Woo 2001). In other words, one year is allowed to pass between the end of the ranking period and the start of the evaluation period. Recall that when the ranking period is not lagged, the average (across sector-years) log-odds ratio is 0.357 and highly statistically significant. When we lag the ranking period by one year, the average log-odds ratio becomes −0.013 and is not significant. Likewise, the average Spearman coefficient drops from 0.143 to −0.012 and is no longer significant. Even though the average level of longer term persistence across sector- years is close to zero, it is still possible that variation in longer term persistence is related to sector competitiveness. We therefore repeat our regression analysis when the dependent variable is the longer term measure of persistence.5 Consistent with the notion that our sample exhibits little or no persistence at the longer horizons, HERFINDAHL does not have a significant effect on longer term persistence, and neither do the other sector-level variables.6 5 These results are available from the authors on request. 6 Although we find it important to document that our significant results are limited to the one-year horizon, we note that tests for the existence of persistence, and by extension tests for the association of our sector-level variables with persistence, are weaker when the horizon is longer. First, survivorship conditioning becomes more serious when the horizon is longer, and depending on the characteristics of the fund attrition process, this can either strengthen or weaken persistence. Second, measurement of persistence is noisier when the horizon is longer (e.g., because fund characteristics change over time). Indeed, few studies detect persistence beyond the one-year horizon.
  • 14. 362 The Journal of Financial Research V. Conclusion Performance persistence is important to all parties connected with fund manage- ment. Its existence has been the subject of an intense and ongoing debate. We contribute to this debate by studying variation in performance persistence across peer groups. The focus of our study is the U.K. mutual fund industry, where official sectors unambiguously define such peer groups. We study the effect of several sector-level variables on sector-level persis- tence. Our choice of variables is based on the notion that the more competitive a sector is, the less likely it is to be characterized by persistence in its funds’ perfor- mance. The variables used to capture intra-sector rivalry are: the number of funds in the sector, the concentration of fund family assets under management in the sector, and the proportion of mature funds in the sector. We additionally control for the types of assets in which the sector’s funds are invested. Only the concentration index of fund family assets is consistently significant: the less dispersed the sector’s assets are, the more persistence is observed. In all, our results indicate that the com- petitiveness of a fund sector influences the persistence in the relative performance of its members. The exact channels through which the competitive environment affects investment managers’ performance are a subject for future research. APPENDIX A Evolution of U.K. Unit Trust Sectors, 1991–2001
  • 15. Mutual Fund Performance Persistence 363
  • 16. 364 The Journal of Financial Research APPENDIX B Calculation of Performance Persistence Statistics Spearman Rank-Correlation Coefficient First, funds that existed in years T (the ranking year) and T+1 (the evaluation year) are identified. Define N to be the size of this sample. For each fund in the sample, the difference d i in the rank of fund i between years T and T+1 is calculated. The Spearman rank-correlation statistic is defined as N rs = 1 − 6 di (N 3 − N ) i=1 and lies between −1 and +1. For sufficiently large N, it is appropriate to test for the statistical significance of rs using a t-test where the critical t-statistic is given by
  • 17. Mutual Fund Performance Persistence 365 N −2 ts = r s 1 − rs2 and has N−2 degrees of freedom. Log-Odds Ratio Funds within a sector are classified as winners (W ) (losers [L]) if their returns are in the top (bottom) half of funds for each of the years T and T+1. WW denotes the number of two-period winners, LW denotes the number of losers in the first year and winners in the second year, WL reverses this order, and LL denotes the number of two-period losers. The log-odds ratio is defined as WW ∗ LL ln . WL ∗ LW The log-odds ratio is asymptotically normally distributed with mean zero and stan- dard error given by 1 1 1 1 σ = + + + . WW WL LW LL The z-statistic of the log odds ratio refers to the log-odds ratio divided by its standard error. References Allen, D. E. and M. L. Tan, 1999, A test of the persistence in the performance of UK managed funds, Journal of Business Finance and Accounting 26, 559–93. Berk, J. B. and R. C. Green, 2004, Mutual fund flows and performance in rational markets, Journal of Political Economy 112, 1269–95. Blake, C. R., E. J. Elton, and M. J. Gruber, 1993, The performance of bond mutual funds, Journal of Business 66, 371–403. Blake, D. and A. Timmermann, 1998, Mutual fund performance: Evidence from the UK, European Finance Review 2, 57–77. Blake, D. and A. Timmermann, 2003, Performance persistence in mutual funds: An independent assessment of the studies prepared by Charles River Associates for the Investment Management Association, Report commissioned by the Financial Services Authority. Brown, S. J., W. N. Goetzmann, R. G. Ibbotson, and S. A. Ross, 1992, Survivorship bias in performance studies, Review of Financial Studies 5, 553–80. Carhart, M. M., 1997, On persistence in mutual fund performance, Journal of Finance 52, 57–82. Carpenter, J. N. and A. W. Lynch, 1999, Survivorship bias and attrition effects in measures of performance persistence, Journal of Financial Economics 54, 337–74. Cooper, M. J., H. Gulen, and P. R. Rau, 2005, Changing names with style: Mutual fund name changes and their effects on fund flows, Journal of Finance 60, 2825–58. Fletcher, J. and D. Forbes, 2002, An exploration of the persistence of UK unit trust performance, Journal of Empirical Finance 9, 475–93.
  • 18. 366 The Journal of Financial Research Giles, T., T. Wilsdon, and T. Worboys, 2002, Performance persistence in UK equity funds—An empirical analysis, Report prepared by Charles River Associates for the Investment Management Associ- ation, London, United Kingdom. Ibbotson, R. and A. Patel, 2002, Do winners repeat with style?, Yale ICF Working paper 00-70. Kosowski, R., A. Timmermann, H. White, and R. Wermers, 2003, Can mutual fund “stars” really pick stocks? New evidence from a bootstrap analysis, Working paper, INSEAD. Massa, M., 2003, How do family strategies affect fund performance? When performance-maximization is not the only game in town, Journal of Financial Economics 67, 249–304. Quigley, G. and R. Sinquefield, 2000, Performance of UK equity unit trusts, Journal of Asset Management 1, 72–92. Papke, L. and J. Wooldridge, 1996, Econometric methods for fractional response variables with an applica- tion to 401(k) plan participation rates, Journal of Applied Econometrics 11, 619–32. Rhodes, M., 2000, Past imperfect? The performance of UK equity managed funds, Financial Services Authority Occasional Paper No. 9. Siggelkow, N., 2003, Why focus? A study of intra-industry focus effects, Journal of Industrial Economics 51, 121–50. Teo, M. and S.-J. Woo, 2001, Persistence in style-adjusted mutual fund returns, Working paper, Harvard University. Ter Horst, J., T. Nijman, and M. Verbeek, 2001, Eliminating look-ahead bias in evaluating persistence in mutual fund performance, Journal of Empirical Finance 8, 345–73. Volkman, D. A. and M. E. Wohar, 1995, Determinants of persistence in relative performance of mutual funds, Journal of Financial Research 18, 415–30. Waring, G. F., 1996, Industry differences in the persistence of firm-specific returns, American Economic Review 86, 1253–65. Wermers, R., 2003, Is money really smart? New evidence on the relation between mutual fund flows, manager behavior, and performance persistence, Working paper, University of Maryland.