CANADIAN MUTUAL FUND FLOWS AND PERFORMANCE: NON … · Jonathan Michael LaBerge This thesis investigates the nature of the relationship between mutual fund flows and fund performance
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
CANADIAN MUTUAL FUND FLOWS AND PERFORMANCE: NON-LINEARITY, FRICTIONS AND DIMINISHING RETURNS TO SCALE
Jonathan Michael LaBerge
A Thesis
In
The John Molson School of Business
Presented in Partial Fulfillment of the Requirements
for the Degree of Master of Science in Administration at
NOTICE: The author has granted a nonexclusive license allowing Library and Archives Canada to reproduce, publish, archive, preserve, conserve, communicate to the public by telecommunication or on the Internet, loan, distribute and sell theses worldwide, for commercial or noncommercial purposes, in microform, paper, electronic and/or any other formats.
AVIS: L'auteur a accorde une licence non exclusive permettant a la Bibliotheque et Archives Canada de reproduire, publier, archiver, sauvegarder, conserver, transmettre au public par telecommunication ou par Plntemet, prefer, distribuer et vendre des theses partout dans le monde, a des fins commerciales ou autres, sur support microforme, papier, electronique et/ou autres formats.
The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.
L'auteur conserve la propriete du droit d'auteur et des droits moraux qui protege cette these. Ni la these ni des extraits substantiels de celle-ci ne doivent etre imprimes ou autrement reproduits sans son autorisation.
In compliance with the Canadian Privacy Act some supporting forms may have been removed from this thesis.
Conformement a la loi canadienne sur la protection de la vie privee, quelques formulaires secondaires ont ete enleves de cette these.
While these forms may be included in the document page count, their removal does not represent any loss of content from the thesis.
Canada
Bien que ces formulaires aient inclus dans la pagination, il n'y aura aucun contenu manquant.
Abstract
Canadian Mutual Fund Flows and Performance: Non-Linearity, Frictions and Diminishing Returns to Scale
Jonathan Michael LaBerge
This thesis investigates the nature of the relationship between mutual fund flows
and fund performance in Canada. Specifically, we investigate the non-linearity (or
"asymmetry") of the relationship, and how the relationship differs for equity and fixed-
income funds using both excess and raw returns. We also test whether frictions prevent
investors from punishing poor performers, and whether there are decreasing returns to
scale in mutual fund management.
Using a sample of 119 equity and 44 fixed-income funds managed by public fund
sponsors, we find that when fund performance is measured on a risk-adjusted basis versus
benchmark returns, there is evidence of rational asymmetry in the flow-performance
relationship for Canadian equity funds. We note that non-performance factors, especially
prior flow, are significant predictors of funds flow for both equity and fixed-income
funds. We also find that frictions do affect the flow-performance relationship, and that
returns to scale in mutual fund management appear to be constant.
- in -
Acknowledgements
I want to thank Dr. Lawrence Kryzanowski for his invaluable guidance as my
supervisor throughout the thesis process. His sage advice helped keep me on track when
direction was sorely needed.
I also wish to thank Dr. Abraham Brodt and Dr. Ian Rakita, for taking the time to
review my work.
Lastly, I want to thank my parents for their emotional support, and Nadine for her
understanding and patience.
- IV -
TABLE OF CONTENTS
LIST OF TABLES vi 1. INTRODUCTION 1 2. REVIEW OF THE MUTUAL FUND PERFORMANCE PERSISTENCE ASYMMETRY LITERATURE 2
2.1 Asymmetric Persistence in the Performance of U.S. Funds 3 2.2 Asymmetric Persistence in the Performance of Canadian Funds 5 2.3 Implications of this Literature 5
3. RELATIONSHIP BETWEEN FUND FLOWS AND FUND PERFORMANCE 6 3.1 Review of the Relevant Literature 6
3.1.1 U.S. Studies 6 3.1.2 Canadian Studies 10 3.1.3 Implications of the Funds Flow Literature 11
3.2 Hypothesis 12 3.3 Sample and Data Collection 12
3.3.1 Determining Public Sponsors 13 3.3.2 Data Sources 13 3.3.3 Equity and Fixed-Income Samples 15
3.4 Methodology 16 3.4.1 Flow and Performance Variables 16 3.4.2 Non-Performance Variables 18 3.4.3 Estimation of the Flow-Performance Relationship 18
3 Note that the yearly components of flow as shown in the numerator of equation (2) are calculated by simply summing the monthly component values, as the IFIC reports exist on a monthly basis.
-16-
where Non-Reinvested Distributions i?t — Total Distributions iit — Reinvested Distributions u,
if Reinvested Distributionsu < Total Distributions i>t, and 0 otherwise. While one would
expect that Reinvested Distributions would always be less than Total Distributions by
definition, there are occasional exceptions to this in the IFIC dataset when there are
positive/negative adjustments, even when no declarations occur. When these infrequent
adjustments exist, they are excluded from the flow calculations.
Rearranging (1), fund performance is then endogenously defined as:
PERF,, = ' Assets!, — Assets,. ,_r
•FLOW,, (3) Assets! ,_1
Equation (3) represents the base specification using raw return performance. Alternative
specifications for performance are described in Section 3.4.3 below.
This thesis also differs with the literature in terms of the twelve month period used to
calculate the variables. As the flow-performance relationship is determined by mutual
fund investor reaction, flow and performance are calculated "off-calendar" to capture the
effect of fund reporting. This is based on the assumption that investors are more likely to
base decisions about MERs on the fund's reports once they have received them than to do
so at the beginning of the fund's fiscal period.
Consider the following example as an illustration of the off-calendar calculation of
flow and performance. Assume fund "XYZ" has December 31 as its fiscal year end, and
that the year end in question is December 1999. Allowing 3 months for fund XYZ's
annual report to reach investors, flow and performance for the 12 month period are
calculated using IFIC's April 2000 to March 2001 monthly reports. Using Gross Sales as
an example, fund XYZ's monthly Gross Sales values from April 2000 to March 2001 are
summed, and then added to the other components of flow which are calculated in the
-17-
same manner. The summed value is then divided by the fund's net assets at the end of
March 2000, as described in equation (2) above. Raw performance is then calculated as
the change in fund XYZ's net assets from the end of March 2000 to the end of March
2001 minus the fund's flow during the same period. One-year lagged values for these
variables are calculated in the same fashion, except that the period used is from April
1999 to March 2000.
3.4.2 Non-Performance Variables
Several non-performance variables are incorporated into the models described in
Section 3.4.3 below. Consistent with the findings of the literature, these include the
where the variables are sorted by fund and by time period within each fund's
observations. FUNDDUM and YEARDUM represent the fixed-effects tested by the
model, and they consist of dummy variables for each fund, and for each semi-annual
period. The difference between the two models is simply the number of fund dummy
variables, corresponding to the difference in sample size. As this methodology requires at
least two fund-year observations for each fund (as discussed in Section 3.3.3), funds with
only one fund-year observation are dropped from the sample.
To deal with the issue of multicollinearity, correlated predictor variables are
orthogonalized prior to their inclusion in the model. In the results of the model discussed
in Section 3.5 below, orthogonalized variables are identified by the inclusion of "RESID"
in the variable name, and the nature of the multicollinearity is specified in the relevant
table.
-19-
The first hypothesis of the thesis (see Section 3.2) is accepted or rejected based on the
PERF and DIFFRES variables. Specifically, at least one of the four variables needs to be
statistically significant in order to reject the null hypothesis of no relationship between
flow and performance. If the alternative hypothesis is concluded, the nature of the
relationship and the presence of non-linearity for both the equity and fixed-income
sample are examined for similarity.
As a test of robustness, the model is also estimated with excess returns, calculated
relative to benchmarks and fund type. Specifically:
EXCESSPERF,, = PERF,, -TPERF,,
EXCESSPERF,, = PERF,, - Benchmark Return,
(7)
A list of the benchmarks and the return series used are provided in Appendix 5.
TPERF is removed as an independent variable in the resulting performance models,
and its effect is now accounted for either directly or indirectly within EXCESSPERF. The
resulting models are:
FLOW,, = 0o+fa (EXCESSPERF,,) + fa (EXCESSPERF,,_,)
+fa (DIFFRES.,) + fa (DIFFRES.,^) + fa (FLOW^) + fa (MER,^)
+fa (STDEV,,,) + fa (STDEV,,,_}) + fa (TFLOW,,) + fa0 (TFLOW,,_x) ( 8 ) 119 19
+fa3 (SIZE,^) + £ fa (FUNDDUM) + £ fa (YEARDUM) + e„
FLOW,, =00+$ (EXCESSPERF,,) + fa (EXCESSPERF,,^)
+fa (DIFFRES.,) + fa (DIFFRES,,^) + fa (FLOW,,,,) + fa (MERU_, )
+fa (STDEV,,) + fa (STDEV.,^) + fa (TFLOW,,) + faQ (TFLOW,^) ( 9 ) 44 19
+fa3 (SIZE,,., ) + £ # (FUNDDUM) + £ fa (YEARDUM) + eu 7=14 /c=133
-20-
3.5 Empirical Results
Tables 1 and 2 report the results for equations (5) and (6), respectively. In each case,
two subsets of the model are presented, due to multicollinearity that is persistent after
orthogonalization. Subset A' represents the removal of outliers from each sample, that are
defined as observations whose studentized residuals are more/less than ±3. The outliers
removed from the equity and fixed-income samples represent 2.45% and 1.40% of all
observations in each sample, respectively. In both cases, this subset results in a better fit
of the model, which is fairly high in both cases, and it is these results that are discussed.
The Hausman test for random effects tests the null hypothesis that a random effects
model is consistent and efficient when compared to a fixed effects model. The F-test for
"No Fixed Effects" tests the null hypothesis that none of the cross-sectional or time-series
effects are statistically significant. Therefore, the rejection of both null hypotheses
implies that the model should be structured as a fixed effects rather than a random effects
model, and that at least one of the fixed effects is significant (this justifies their inclusion
in the model). A 10% significance level is used for both tests.
3.5.1 Basic Flow-Performance Relationship
Clear evidence of a relationship between fund flows and performance exists on both a
lagged and contemporaneous basis for the equity sample. On a contemporaneous basis,
the estimates for PERF and DIFFRESRESID are positive and significant (10% and 5%
confidence levels, respectively). On a lagged basis, the estimate for PERF and
DIFFRESRESID are positive and significant (1% and 5% confidence levels,
respectively). These results contrast with those of the fixed-income sample where only
the contemporaneous variables are positive and significant at the 1% confidence level.
-21-
This implies that equity fund investors respond, both on a current and lagged basis, in an
overall positive manner that becomes disproportionate when performance is above
average. However, this is only the case on a current basis for fixed-income investors.
While the null hypothesis detailed in Section 3.2 is rejected, the alternative is only
partially concluded as fixed-income investors do react similarly to equity investors on a
contemporaneous basis.
With regard to the presence of non-linearity in the relationship, the significant
DIFFRES variables, in combination with the significance of the base performance
variables, indicate that Canadian investors do punish below-average performers, but also
reward above-average performers in a disproportionate fashion. This finding is in contrast
to most of the U.S. literature, as well as the findings presented in both Deaves (2004),
who finds non-linearity only among good performers, and Sinha and Jog (2005).
For both the equity and fixed-income sample, it is evident from Tables 1 and 2 that
the prior flows for both the fund and its fund type are strong predictors of the current
period flow (both significant at the 1% level), as reported in Sinha and Jog (2005). They
are strong indicators that future studies in this field should take the autocorrelation of
fund flows into account. The results for the effect of fees are unlike those reported in U.S.
studies but like those reported in Canadian studies. The estimate for MER is
insignificantly negative for both the equity and fixed-income samples. The effect of size
is similar across both samples and consistent with most flow-performance studies in that
the estimate is negative and highly significant.
The effects of risk differ between the equity and fixed-income samples. For equity
funds, the estimates are negative but statistically insignificant. Like the fee results, these
-22-
findings are similar to Deaves (2004) and Sinha and Jog (2005), but contrast with those
reported by U.S. studies. For the fixed-income sample, the effect of risk is also negative
but marginally significant at the 10% confidence level for the contemporaneous effect
and at the 5% confidence level for the lagged effect. This suggests that the flow of funds
is related (inversely) to the risk level of fixed-income but not equity funds. This is
consistent with investment theory, as we would expect fixed-income investors to be more
risk-averse than equity investors.
3.5.2 Alternative Flow-Performance Relationship
Tables 3 and 4 report the results of equations (8) and (9), respectively. For the equity
sample, regardless of whether benchmark returns or fund-type performance are used to
calculate excess returns, both contemporaneous and lagged performance estimates are
similarly positive and significant at the 1% confidence level. However, when testing for
differential returns, the results depend on the basis of comparison. Comparing fund
performance to benchmark returns, which are more commonly used in the literature, we
observe negative and significant differential response on a lagged basis. When fund-type
returns are used instead, we observe positive and significant differential response on a
contemporaneous basis. The fits of the models are almost the same, so in both cases
about the same percentage of variation is explained.
For the fixed-income sample, we observe similar reactions to absolute flow on a
contemporaneous basis. Lagged excess returns are only marginally significant when
fund-type returns are used. The model fit is slightly better for the BM model by only
about 3%.
-23-
In summary, the results from the alternative performance specifications suggest that
the basis of comparison matters to equity but not fixed-income investors. Of particular
importance from the equity results is the negative and significant estimate for the
differential response when calculating excess returns using benchmarks. As this model is
the most similar to the methodology employed in academic studies, the parameter
estimate provides evidence of a muted response to risk-adjusted outperformance. Taken
together with a linear response to risk-adjusted underperformance, this relationship
matches the theoretical behavior that is expected given the prior literature on performance
persistence asymmetry. It supports the principle of investor rationality. As the results
differ for the fixed-income sample, the same inference can not be drawn for fixed-income
investors. Given these results and academic convention, excess returns compared to
benchmarks are used as the performance specification in the remaining sections of the
thesis.
4. RESPONSE OF FUND FLOW COMPONENTS TO PERFORMANCE
As mentioned earlier, the prevalent findings of both the performance persistence and
fund flows literatures suggest that investors may not be acting rationally. While evidence
to the contrary is presented in the previous section for Canada, this evidence contrasts
with the results of the U.S. studies. Several authors in the U.S. have attempted to provide
an explanation for the observed behavior of investors, mainly focusing on transactional
arguments. As both a further test of rationality, as well as an attempt to document how
the structural effects of the Canadian fund industry affect investor decision making, this
section tests the role of frictions in the flow-performance relationship.
-24-
4.1. Relevant Literature
As discussed in Section 3.1.1.2, Ippolito (1992) finds evidence of flow asymmetry.
The author rationalizes this finding within the context of rational learning by stating that
transaction costs act as a barrier for investors to reallocate their funds to top performers.
Specifically, he hypothesizes that the flow reaction to performance should be stronger for
fund purchases than for redemptions. While the author interprets his findings of
disproportionately higher flow for good performers as evidence of differences between
purchases and redemptions, he does not explicitly model or test the variables as such.
Fant (1999) examines the effect of frictions on stock market returns instead of on
fund performance. He tests the behavior of investors by dividing fund flows into its four
major components; namely, new sales, redemptions, exchanges-in and exchanges-out.
The relationship between aggregate fund flows and stock market returns is then examined
by testing the hypothesis that differences in the relationship will be observed among the
components due to frictions. Specifically, Fant argues that a stronger flow-performance
relationship should be observed among exchanges than among sales and redemptions
since investors can more easily move money between funds in the same fund complex via
exchanges. His findings confirm his conjecture as the relationship between fund flows
and market performance occurs entirely between market returns and exchanges-in and
out. Fant's findings provide support for investor rationality as frictions appear to play a
significant role in investors' decisions to react to performance.
4 Exchanges-in and out are defined in a similar fashion to the transfers-in and out defined in Appendix 3.
-25-
Sigurdsson (2004) reports evidence to the contrary in that the flow-performance
relationship observed in the literature is mainly driven by the net inflow of new money to
the industry. The rationale behind his hypothesis stems from the idea that new investors,
or old investors allocating new money, will rationally take performance into account
when selecting which funds to invest in. Once the decision to invest new money is made,
frictions are a certainty and are unlikely to act as barriers to chasing performance. In
contrast, existing investors are not likely to exercise their transfer option due to
uncertainty about future returns, and transaction costs.
According to IFIC documentation, it is common for Canadian fund sponsors to allow
investors to switch among funds in the same fund family without facing a redemption fee.
While investors may be required to pay a transfer fee, mutual fund dealers may only levy
a charge of up to 2% of assets.5 By contrast, if investors redeem an existing position in
order to move to funds that they believe will provide better risk-adjusted returns in the
future, they face fees significantly in excess of 2%. In the case of rear-load funds,
deferred sales charge (DSC) fees regularly exceed 6% in the first year after purchase, and
often do not fall below 2% until the fund has been held for a total of 6 years. Ruckman
(2003) cites anecdotal evidence that Canadian investors "buy rear-loaded funds four
times as often as front-loaded funds". While front-end fees are generally lower than their
DSC counterparts, they are still generally higher than 2%. Moreover, if an investor sells a
position in a front-end load fund and purchases a rear-loaded fund to avoid redemption
and sales fees, she would only be able to do this once before she would again face
friction.
5 From "Mutual Fund Fees", in Series 1 of IFIC's Investor Frequently Asked Questions. Last updated October 24th, 2005. http://ific.ca/eng/frames.asp?ll=Investor_Education
where funds are defined as winners (WINNER =1) if their excess returns relative to the
benchmarks are positive, and losers otherwise (WINNER = 0). The SIZE variable is the
same as employed previously.
The third hypothesis of the thesis is accepted or rejected based on the significance of
the coefficient of the SIZE variable. In the event of a significantly negative parameter
estimate, the null hypothesis is rejected with confidence.
5.4 Empirical Results
The results for equation (21) are presented in Tables 9 and 10. While SIZE has a
negative parameter estimates in both cases, it is not a significant factor for either the
equity or fixed-income sample. This evidence does not lead to a rejection of the null
hypothesis, implying that returns to scale in Canadian mutual fund management are
constant.
There is evidence of performance persistence in the equity sample since the parameter
estimate of the lagged WINNER variable is highly significant, and the odds ratio
estimates imply that it is roughly one and a half to three times as likely that a prior
winner/loser stands to be a future winner/loser. Consistent with most of the findings in
the literature, there is evidence that a majority of this effect is due to the persistence of
poor performers rather than good performers.6
6 Of the 559 observations in the equity sample, 340 observations have identical lagged and contemporaneous WINNER variables (either 1/1 or 0/0). However, out of these 340 cases, 220 are due to poor performer persistence (0/0), leaving only 120 cases due to persistence of good performers.
-34-
6. CONCLUSIONS
In this thesis, the nature of the relationship between mutual fund flows and
performances for Canadian mutual funds was examined. Specifically, we investigated the
non-linearity of the relationship, and how the relationship differs for equity and fixed-
income funds using excess and raw returns. We also test whether frictions prevent
investors from punishing poor performers, and whether there are decreasing returns to
scale in mutual fund management in Canada.
There are several relevant findings presented in this thesis. First, Canadian equity
investors do punish poor performers by avoiding funds that underperform on a risk-
adjusted basis. This finding contrasts with most of the U.S. evidence. Funds that
outperform on a risk-adjusted basis do experience increases in funds flow, but at a lower
rate than the loss in funds incurred by underperformers. On the other hand, fixed-income
investors avoid poor performers and disproportionately embrace good performers.
Second, funds flows in Canada are heavily autocorrelated. There is substantial
evidence of a momentum effect in the movement of money among funds at both the
fund-specific and fund-objective level for both equity and fixed-income funds. Fixed-
income (equity) investors respond negatively (are indifferent) to risk even when
performance is measured on a risk-adjusted basis. Larger funds grow more slowly than
their smaller counterparts.
Third, frictions affect investor decision making, as a stronger relationship exists
between flows and performances for sales and redemptions than for inter-fund transfers
for equity funds. The reverse is true for fixed-income funds. However, the specifics of
how frictions affect investor decision making are as yet undetermined.
-35-
Fourth, no evidence exists in either sample that fond size affects the likelihood of
future outperformance. This implies that investors should be as willing to (re)allocate
money to larger fonds as they would be to growing funds. While evidence of
performance persistence is present in the equity sample, the majority of this effect is due
to the persistence of poor rather than good performers. This finding is consistent with the
performance persistence literature. No evidence of persistence is found among the fixed-
income funds, which implies that fixed-income mutual fonds in Canada add little value.
There are a number of possibilities for future research in this area. It would be
interesting to assess how flow-performance non-linearity has evolved over time, and
whether there is any evidence of learning on the part of investors. The rationality of
investors can be further tested by controlling for the status of the fonds flow, namely,
whether it is due to a new allocation or to a reallocation.
-36-
Tab
le 1
- A
ggre
gate
d E
quit
y F
low
Mod
el w
ith
Abs
olut
e P
erfo
rman
ce
Res
ults
of
a fi
xed-
effe
cts
pane
l re
gres
sion
of
the
equi
ty s
ampl
e w
ith c
urre
nt y
ear
Fund
Flo
w (
FLO
W ,)
as
the
depe
nden
t va
riab
le. M
ER
rep
rese
nts
the
fund
's pr
ior
year
's M
anag
emen
t E
xpen
se R
atio
, FL
OW
t-i r
epre
sent
s th
e fu
nd's
prio
r ye
ar F
und
Flow
, PE
RF
t and
PE
RF
t-i r
epre
sent
the
fun
d's
curr
ent
and
prio
r ye
ar a
bsol
ute
(raw
) re
turn
, ST
DE
V
t an
d ST
DE
V ,
.[ re
pres
ent
the
fund
's cu
rren
t an
d pr
ior
12-m
onth
sta
ndar
d de
viat
ion
of r
etur
ns,
DIF
FRE
S t.
and
DIF
FRE
S ,_i
rep
rese
nt
curr
ent
and
prio
r ye
ar i
ndic
ator
var
iabl
es t
akin
g th
e va
lue
of (
PE
RF
- T
PER
F) i
f PE
RF
> T
PE
RF
, or
zer
o ot
herw
ise,
TFL
OW
t a
nd T
FL
OW
n r
epre
sent
cur
rent
an
d pr
ior
year
flo
ws
to t
he f
und'
s ob
ject
ive
clas
s (o
r "f
und-
type
"),
TPE
RF
t and
TPE
RF
,.j r
epre
sent
cur
rent
and
pri
or y
ear
abso
lute
(ra
w)
retu
rns
of t
he f
und'
s ob
ject
ive
clas
s (o
r "f
und-
type
"),
and
SIZ
E i
s th
e na
tura
l lo
gari
thm
of
the
fund
's pr
ior
year
Net
Ass
ets.
DIF
FRE
SRE
SID
, a
nd D
IFFR
ESR
ESI
D t_i
rep
rese
nt t
he
resi
dual
s of
DIF
FRE
S t a
nd D
IFFR
ES
M r
egre
ssed
aga
inst
PE
RF
t and
PE
RF
t_i,
resp
ectiv
ely.
TFL
OW
RE
SID
tA r
epre
sent
s th
e re
sidu
als
of T
FLO
W M
reg
ress
ed
agai
nst
TFL
OW
t. T
PER
FRE
SID
, re
pres
ents
the
res
idua
ls o
f T
PE
RF
, reg
ress
ed a
gain
st P
ER
F t
All
fund
-typ
e va
riab
les
are
calc
ulat
ed u
sing
all
avai
labl
e eq
uity
fu
nds
in th
e IF
IC d
atas
et,
not j
ust
thos
e of
pub
lic s
pons
ors.
As
the
PER
F ul
I T
PER
F t-i
, DIF
FRE
SRE
SID
t /
TPE
RFR
ESI
D t a
nd D
IFFR
ESR
ESI
D M
/ T
PER
F ,.,
va
riab
les
still
had
hig
h po
sitiv
e or
neg
ativ
e co
rrel
atio
ns a
fter
ort
hogo
naliz
atio
n, S
ubse
ts A
and
B r
efle
ct s
ubst
itutio
n of
the
se v
aria
bles
. Su
bset
A' r
epre
sent
s th
e re
mov
al o
f 14
obs
erva
tions
who
se s
tude
ntiz
ed r
esid
ual
valu
es f
rom
the
Sub
set A
mod
el w
ere
grea
ter/
less
tha
n 3/
-3. B
olde
d va
lues
rep
rese
nt e
stim
ates
sig
nifi
cant
at
the
1%, 5
%, o
r 10
% le
vel.
Par
amet
er E
stim
ates
Inte
rcep
t P
ER
Fi
PE
RF
..i
DIF
FR
ES
RE
SID
t D
IFF
RE
SR
ES
ID t-
i FL
OW
t-i
ME
R
STD
EV
t S
TD
EV
t-i
TF
LO
W.
TF
LO
WR
ES
ID <
-i T
PE
RF
RE
SID
t T
PE
RF
i-i
SIZ
E t-
i
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
Subs
et A
3.16
17
0.05
80
0.20
92
0.60
91
0.10
01
0.08
36
-3.0
821
0.14
87
0.33
86
0.31
55
-0.0
645
-0.2
840
49.1
9 2.
64
P-V
alue
s
<.0
001
0.56
61
0.00
03
0.00
70
0.56
66
<.0
001
0.37
66
0.86
89
0.57
25
0.04
33
0.67
93
<.0
001
<.0
001
<.0
001
Subs
et B
3.45
26
0.11
69
0.12
46
-2.0
260
1.64
52
0.15
53
0.44
93
0.07
11
-0.1
946
-0.1
285
-0.2
852
50.0
9 2.
46
P-V
alue
s
<.0
001
0.34
07
<.0
001
0.56
99
0.05
39
0.79
69
0.00
8 0.
6735
0.
3716
0.
5016
<
.000
1
<.0
001
<.0
001
Subs
et A
*
1.56
38
0.13
17
0.26
58
0.36
66
0.32
71
0.10
56
-0.0
585
-0.1
677
-0.2
777
0.35
44
-0.1
158
-0.1
671
41.6
0 2.
75
P-V
alue
s
0.00
01
0.08
63
<.0
001
0.03
26
0.02
98
<.0
001
0.98
19
0.80
36
0.53
39
0.00
25
0.31
76
<.0
001
<.0
001
<.0
001
Num
bero
fFun
ds
119
119
117
Num
ber
of S
emi-A
nnua
l Tim
e Pe
riod
s 19
19
19
N
umbe
r of
Fun
d-Y
ears
(O
bser
vatio
ns)
572
572
558
Adj
uste
d R
-Squ
are
52.0
6%
49.6
4%
57.8
7%
-37-
Tab
le 2
- A
ggre
gate
d F
ixed
-Inc
ome
Flo
w M
odel
with
Abs
olut
e P
erfo
rman
ce
Res
ults
of
a fix
ed-e
ffec
ts
pane
l re
gres
sion
of
the
fixe
d-in
com
e sa
mpl
e w
ith c
urre
nt y
ear
Fund
Flo
w (
FLO
W t
) as
the
dep
ende
nt v
aria
ble.
ME
R r
epre
sent
s th
e fu
nd's
prio
r ye
ar M
anag
emen
t E
xpen
se R
atio
, FL
OW
t_i r
epre
sent
s th
e fu
nd's
prio
r ye
ar F
und
Flow
, PE
RF
t and
PE
RF
,.] r
epre
sent
the
fun
d's
curr
ent
and
prio
r ye
ar a
bsol
ute
(raw
) re
turn
, ST
DE
V t a
nd S
TD
EV
tA r
epre
sent
the
fun
d's
curr
ent
and
prio
r 12
-mon
th s
tand
ard
devi
atio
n of
ret
urns
, D
IFFR
ES
t and
DIF
FRE
S t.j
re
pres
ent
curr
ent
and
prio
r ye
ar i
ndic
ator
var
iabl
es t
akin
g th
e va
lue
of (
PER
F -
TPE
RF)
if P
ER
F >
TP
ER
F, o
r ze
ro o
ther
wis
e, T
FLO
W t a
nd T
FL
OW
,_i r
epre
sent
cu
rren
t an
d pr
ior
year
flo
w t
o th
e fu
nd's
obje
ctiv
e cl
ass
(or
"fun
d-ty
pe")
, T
PER
F t a
nd T
PER
F ,_i
rep
rese
nt c
urre
nt a
nd p
rior
yea
r ab
solu
te (
raw
) re
turn
of
the
fund
's ob
ject
ive
clas
s (o
r "f
und-
type
"),
and
SIZ
E i
s th
e na
tura
l lo
gari
thm
of
the
fund
's pr
ior
year
Net
Ass
ets.
DIF
FRE
SRE
SID
t a
nd D
IFFR
ESR
ESI
D ,_
i rep
rese
nt
the
resi
dual
s of
DIF
FRE
S , a
nd D
IFF
RE
S t_i
reg
ress
ed a
gain
st P
ER
F t a
nd P
ER
F t-\
, res
pect
ivel
y. A
ll fu
nd-t
ype
vari
able
s ar
e ca
lcul
ated
usi
ng a
ll av
aila
ble
fixe
d-in
com
e fu
nds
in th
e IF
IC d
atas
et,
not j
ust
thos
e of
pub
lic s
pons
ors.
As
the
DIF
FRE
SRE
SID
t /
TP
ER
F t a
nd D
IFFR
ESR
ESI
D M
/ T
PE
RF
t_i v
aria
bles
stil
l ha
d hi
gh n
egat
ive
corr
elat
ions
aft
er o
rtho
gona
lizat
ion,
Sub
sets
A a
nd B
ref
lect
sub
stitu
tion
of t
hese
var
iabl
es.
Subs
et A
' rep
rese
nts
the
rem
oval
of
3 ob
serv
atio
ns
who
se s
tude
ntiz
ed r
esid
ual
valu
es f
rom
the
Sub
set
A m
odel
wer
e gr
eate
r/le
ss t
han
3/-3
. Bol
ded
valu
es r
epre
sent
est
imat
es s
igni
fica
nt a
t th
e 1%
, 5%
, or
10%
le
vel.
Par
amet
er E
stim
ates
Inte
rcep
t P
ER
Ft
PE
RF
t-i
DIF
FR
ES
RE
SID
t D
IFF
RE
SR
ES
ID t-
i F
LO
W ,-
i M
ER
S
TD
EV
, S
TD
EV
t-i
TF
LO
W,
TF
LO
W t-
i T
PE
RF
t T
PE
RF
,.i
SIZ
E ..
i
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
Subs
et A
3.15
40
0.72
92
0.27
12
1.10
32
0.64
32
0.17
05
11.2
035
-2.2
196
-4.3
424
0.36
63
-0.0
213
-0.3
244
78.5
2 2.
88
P-V
alu
es
<.0
001
0.00
64
0.26
73
0.02
29
0.15
11
<.0
001
0.16
36
0.19
31
0.00
76
0.00
39
0.84
83
<.0
001
<.0
001
<.0
001
Subs
et B
3.22
99
0.75
74
0.32
23
0.17
83
11.6
183
-0.5
762
-3.4
726
0.38
36
-0.0
026
-0.1
887
-0.2
704
-0.3
204
62.1
1 2.
61
P-V
alue
s
<.00
01
0.00
92
0.23
41
<.0
001
0.15
58
0.70
70
0.01
72
0.00
52
0.98
20
0.50
52
0.33
98
<.0
001
<.0
001
<.0
001
Su
bse
t A
'
2.62
58
0.76
18
0.24
13
1.12
65
0.56
39
0.18
89
9.74
80
-2.5
441
-3.0
735
0.40
80
-0.0
801
-0.2
740
77.2
4 3.
17
P-V
alue
s
<.0(
)01
0.00
08
0.24
41
0.00
61
0.13
56
<.0
001
0.15
12
0.07
78
0.02
60
0.00
02
0.39
59
<.0
001
<.0
001
<00
01
Num
ber
of F
unds
44
44
44
N
umbe
r of
Sem
i-Ann
ual
Tim
e Pe
riod
s 19
19
19
N
umbe
r of
Fun
d-Y
ears
(O
bser
vatio
ns)
215
215
212
Adj
uste
d R
-Squ
are
66.9
1%
65.5
5%
73.7
9%
-38-
Tab
le 3
- A
ggre
gate
d E
quity
Flo
w M
odel
wit
h R
elat
ive
Per
form
ance
Res
ults
of
a fi
xed-
effe
cts
pane
l reg
ress
ion
of th
e eq
uity
sam
ple
with
cur
rent
yea
r Fu
nd F
low
(FL
OW
,) a
s th
e de
pend
ent
vari
able
. ME
R r
epre
sent
s th
e fu
nd's
prio
r ye
ar M
anag
emen
t E
xpen
se R
atio
, FL
OW
t_]
repr
esen
ts t
he f
und'
s pr
ior
year
Fun
d Fl
ow,
EX
CE
SSPE
RF
t and
EX
CE
SSPE
RF
t_i r
epre
sent
the
fun
d's
curr
ent
and
prio
r ye
ar e
xces
s re
turn
, ST
DE
V ,
and
ST
DE
V t_,
rep
rese
nt t
he f
und'
s cu
rren
t an
d pr
ior
12-m
onth
sta
ndar
d de
viat
ion
of r
etur
ns,
DIF
FRE
S t a
nd D
IFF
RE
S ,_i
re
pres
ent
curr
ent
and
prio
r ye
ar i
ndic
ator
var
iabl
es t
akin
g th
e va
lue
of E
XC
ESS
PER
F if
EX
CE
SSPE
RF
> 0
, or
zer
o ot
herw
ise,
TFL
OW
, a
nd T
FL
OW
n
repr
esen
t cu
rren
t an
d pr
ior
year
flo
w t
o th
e fu
nd's
obje
ctiv
e cl
ass
(or
"fun
d-ty
pe")
, T
PER
F t a
nd T
PER
F ,_i
repr
esen
t cu
rren
t an
d pr
ior
year
abs
olut
e (r
aw)
retu
rn
of t
he f
und'
s ob
ject
ive
clas
s (o
r "f
und-
type
"),
and
SIZ
E i
s th
e na
tura
l lo
gari
thm
of
the
fund
's pr
ior
year
Net
Ass
ets.
DIF
FRE
SRE
SID
t a
nd D
IFFR
ESR
ESI
D t_
i re
pres
ent
the
resi
dual
s of
DIF
FRE
S , a
nd D
IFFR
ES
t., r
egre
ssed
aga
inst
EX
CE
SSPE
RF
, and
EX
CE
SSPE
RF
M,
resp
ecti
vely
. T
FLO
WR
ESI
D M
rep
rese
nts
the
resi
dual
s of
TFL
OW
M r
egre
ssed
aga
inst
TFL
OW
t.
All
fund
-typ
e va
riab
les
are
calc
ulat
ed u
sing
all
avai
labl
e eq
uity
fun
ds i
n th
e IF
IC d
atas
et,
not j
ust
thos
e of
pu
blic
spo
nsor
s. B
M i
dent
ifie
s th
e m
odel
whe
re b
ench
mar
k re
turn
s w
ere
used
to
calc
ulat
e E
XC
ESS
PER
F. F
T i
dent
ifie
s th
e m
odel
whe
re f
und-
type
re
turn
s (T
PER
F) w
ere
used
to
calc
ulat
e E
XC
ESS
PER
F. B
M' a
nd F
T' r
epre
sent
the
rem
oval
of
13 o
bser
vatio
ns w
hose
stu
dent
ized
res
idua
l va
lues
fro
m t
he B
M a
nd F
T
mod
els
wer
e gr
eate
r/le
ss th
an 3
/-3.
Bol
ded
valu
es r
epre
sent
est
imat
es s
igni
fica
nt a
t the
1%
, 5%
, or
10%
leve
l.
Par
amet
er E
stim
ates
Inte
rcep
t E
XC
ES
SP
ER
F t
EX
CE
SS
PE
RF
t-i
DIF
FR
ES
RE
SID
t
DIF
FR
ES
RE
SID
t-i
FL
OW
t-i
ME
R
ST
DE
V t
ST
DE
V t-
i T
FL
OW
t T
FL
OW
RE
SID
t-i
SIZ
E
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
BM
3.34
87
0.23
98
0.19
50
-0.1
071
-0.5
277
0.09
51
-3.2
312
0.95
74
0.32
56
0.34
37
-0.0
717
-0.2
836
49.9
8 2.
65
P-V
alue
s
<.00
01
0.00
26
0.00
05
0.64
45
0.00
58
<.0
001
0.35
31
0.27
41
0.58
65
0.02
46
0.64
13
<.0
001
<.0
001
<.0
001
FT
3.17
62
0.24
48
0.23
93
0.80
56
-0.2
721
0.08
72
-3.6
047
-0.0
600
0.61
97
0.32
62
0.03
01
-0.2
846
50.9
4 2.
62
P-V
alue
s
<.00
01
0.00
39
<.0
001
0.00
49
0.22
42
<.0
001
0.29
81
0.94
65
0.30
62
0.03
03
0.84
02
<.0
001
<.0
001
<.0
001
BM
'
1.90
70
0.25
10
0.17
76
-0.1
277
-0.4
896
0.09
82
0.13
87
0.24
91
-0.0
439
0.36
46
-0.1
362
-0.1
672
45.0
1 2.
72
P-V
alu
cs
<.0
001
<.0
001
0.00
08
0.46
86
0.00
24
<.0
001
0.95
78
0.70
67
0.92
24
0.00
17
0.24
24
<.0
001
<.0
001
<.0
001
FT"
1.76
70
0.23
01
0.26
79
0.51
61
-0.0
929
0.09
40
-0.2
325
-0.4
474
0.10
67
0.37
52
-0.0
541
-0.1
746
45.4
6 2.
64
P-V
alue
s
<.0
001
0.00
05
<.0
001
0.01
96
0.62
76
<.0
001
0.92
96
0.51
46
0.81
67
0.00
12
0.63
52
<.0
001
<.0
001
<.0
001
Num
ber
of F
unds
11
9 11
9 11
7 11
7 N
umbe
r of
Sem
i-A
nnua
l Tim
e Pe
riods
19
19
19
19
N
umbe
r of
Fun
d-Y
ears
(O
bser
vatio
ns)
572
572
559
559
Adj
uste
d R
-Squ
are
52.0
5%
52.7
8%
56.3
4%
56.1
9%
-39-
Tab
le 4
- A
ggre
gate
d F
ixed
-Inc
ome
Flo
w M
odel
wit
h R
elat
ive
Per
form
ance
Res
ults
of
a fi
xed-
effe
cts
pane
l re
gres
sion
of
the
fixe
d-in
com
e sa
mpl
e w
ith c
urre
nt y
ear
Fund
Flo
w (
FLO
W t)
as t
he d
epen
dent
var
iabl
e. M
ER
rep
rese
nts
the
fund
's pr
ior
year
Man
agem
ent
Exp
ense
Rat
io, F
LO
W t-\
repr
esen
ts t
he f
und'
s pr
ior
year
Fun
d F
low
, E
XC
ESS
PER
F t a
nd E
XC
ES
SP
ER
F M
rep
rese
nt t
he f
und'
s cu
rren
t an
d pr
ior
year
exc
ess
retu
rn,
STD
EV
t a
nd S
TD
EV
,_i
rep
rese
nt t
he f
und'
s cu
rren
t an
d pr
ior
12-m
onth
sta
ndar
d de
viat
ion
of r
etur
ns,
DIF
FRE
S t
and
DIF
FRE
S t-i
rep
rese
nt c
urre
nt a
nd p
rior
yea
r in
dica
tor
vari
able
s ta
king
the
val
ue o
f E
XC
ESS
PER
F if
EX
CE
SSPE
RF
> 0
, or
zer
o ot
herw
ise,
TFL
OW
t a
nd
TFL
OW
t_! r
epre
sent
cur
rent
and
pri
or y
ear
flow
to
the
fund
's ob
ject
ive
clas
s (o
r "f
und-
type
"),
TP
ER
F t a
nd T
PER
F t_i
rep
rese
nt c
urre
nt a
nd p
rior
yea
r ab
solu
te
(raw
) re
turn
of
the
fund
's ob
ject
ive
clas
s (o
r "f
und-
type
"),
and
LO
GFU
ND
NA
is
the
natu
ral
loga
rith
m o
f th
e fu
nd's
prio
r ye
ar N
et A
sset
s. D
IFF
RE
SR
ES
ID,
and
DIF
FRE
SRE
SID
t.i
re
pres
ent
the
resi
dual
s of
D
IFFR
ES
, an
d D
IFFR
ES
tA
regr
esse
d ag
ains
t E
XC
ESS
PER
F t
and
EX
CE
SS
PE
RF
M,
resp
ectiv
ely.
T
FLO
WR
ESI
D M
rep
rese
nts
the
resi
dual
s of
TFL
OW
M r
egre
ssed
aga
inst
TFL
OW
t-
All
fund
-typ
e va
riab
les
are
calc
ulat
ed u
sing
all
avai
labl
e fi
xed-
inco
me
fund
s in
the
IFI
C d
atas
et,
not
just
tho
se O
f pu
blic
spo
nsor
s. B
M i
dent
ifie
s th
e m
odel
whe
re b
ench
mar
k re
turn
s w
ere
used
to
calc
ulat
e E
XC
ESS
PER
F. F
T
iden
tifie
s th
e m
odel
whe
re f
und-
type
ret
urns
(T
PER
F) w
ere
used
to
calc
ulat
e E
XC
ESS
PER
F. F
T' r
epre
sent
s th
e re
mov
al o
f 2
obse
rvat
ions
who
se s
tude
ntiz
ed
resi
dual
val
ues
from
the
FT
mod
el w
ere
grea
ter/
less
tha
n 3/
-3. B
olde
d va
lues
rep
rese
nt e
stim
ates
sig
nifi
cant
at t
he 1
%, 5
%, o
r 10
% le
vel.
Par
amet
er E
stim
ates
Inte
rcep
t E
XC
ES
SP
ER
F t
EX
CE
SS
PE
RF
M
DIF
FR
ES
RE
SID
t
DIF
FR
ES
RE
SID
t-i
FL
OW
t-i
ME
R
ST
DE
Vt
ST
DE
V ,-
i T
FL
OW
t T
FL
OW
t-i
SIZ
E
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
BM
3.30
85
0.54
35
0.19
35
1.21
65
0.08
15
0.18
81
13.5
746
-0.9
024
-3.0
738
0.44
81
-0.1
446
-0.3
176
37.7
3 2.
69
P-V
auic
s
<.0
001
0.01
36
0.32
57
0.00
87
0.85
93
<.0
001
0.09
20
0.53
86
0.02
86
0.00
03
0.22
60
<.0
001
<.0
001
<.0
001
FT
3.32
67
0.57
72
0.35
35
2.03
29
0.61
68
0.18
11
10.7
277
-2.6
329
-4.1
754
0.36
90
-0.0
491
-0.3
377
90.1
2 2.
83
P-V
alu
cs
<00
01
0.01
42
0.13
29
0.00
38
0.35
02
<0
00
1 0.
1809
0.
1301
0.
0126
0.
0026
0.
6406
<
00
01
<.0
001
<.0
001
BM
f
2.75
74
0.64
40
0.14
33
1.48
88
-0.0
565
0.20
45
12.0
495
-1.5
983
-2.2
896
0.47
03
-0.1
971
-0.2
685
31.2
8 3.
30
P-V
alue
s
<.00
01
0.00
04
0.37
11
0.00
01
0.88
11
<.0
001
0.06
60
0.18
25
0.05
23
<.0
001
0.04
49
<.0
001
0.00
10
<.0
001
FT
'
2.77
88
0.63
47
0.34
02
1.89
96
0.35
23
0.20
00
10.1
004
-3.0
029
-2.8
207
0.43
80
-0.0
877
-0.2
842
45.8
3 2.
94
P-V
'alu
es
<.00
01
0.00
23
0.09
98
0.00
21
0.54
32
<.00
01
0.15
16
0.04
97
0.05
63
<.0
001
0.34
34
<.0
001
<.0
001
<.0
001
Num
ber
of F
unds
44
44
43
44
N
umbe
r of
Sem
i-Ann
ual T
ime
Peri
ods
19
19
19
19
Num
ber
of F
und-
Yea
rs (
Obs
erva
tions
) 21
5 21
5 21
0 21
3 A
djus
ted
R-S
quar
e 66
.88%
67
.04%
75
.66%
72
.12%
-40-
Tab
le 5
- C
ompo
nent
Equ
ity E
xter
nal
Flo
w M
odel
s w
ith R
elat
ive
Per
form
ance
Res
ults
of
a fi
xed-
effe
cts
pane
l re
gres
sion
of
the
equi
ty s
ampl
e w
ith c
urre
nt y
ear
Sale
s (S
AL
ES
t) an
d R
edem
ptio
ns (
RE
DE
M t)
as t
he d
epen
dent
var
iabl
es.
EX
CE
SSPE
RF
t and
EX
CE
SSPE
RF
M r
epre
sent
the
fun
d's
curr
ent
and
prio
r ye
ar e
xces
s re
turn
rel
ativ
e to
ben
chm
arks
, D
IFFR
ES
, and
DIF
FR
ES
t-i r
epre
sent
cu
rren
t an
d pr
ior
year
ind
icat
or v
aria
bles
tak
ing
the
valu
e of
EX
CE
SSPE
RF
if E
XC
ESS
PER
F >
0,
or z
ero
othe
rwis
e, S
AL
ES
,_i a
nd R
ED
EM
ul
repr
esen
t th
e fu
nd's
prio
r ye
ar S
ales
and
Red
empt
ions
, ME
R r
epre
sent
s th
e fu
nd's
prio
r ye
ar M
anag
emen
t E
xpen
se R
atio
, ST
DE
V t a
nd S
TD
EV
M r
epre
sent
the
fun
d's
curr
ent
and
prio
r 12
-mon
th s
tand
ard
devi
atio
n of
abs
olut
e re
turn
s, T
SAL
ES
, / T
RE
DE
M t
and
TSA
LE
S t-i
/ T
RE
DE
M t_i
rep
rese
nt c
urre
nt a
nd p
rior
yea
r Sa
les
and
Red
empt
ions
of
the
fund
's ob
ject
ive
clas
s (o
r "f
und-
type
"),
and
SIZ
E i
s th
e na
tura
l lo
gari
thm
of
the
fund
's pr
ior
year
Net
Ass
ets.
DIF
FR
ES
RE
SID
, a
nd
DIF
FRE
SRE
SID
M r
epre
sent
the
res
idua
ls o
f D
IFFR
ES
, and
DIF
FRE
S ,.]
reg
ress
ed a
gain
st E
XC
ESS
PER
F t a
nd E
XC
ESS
PER
F t.i
, res
pect
ivel
y. A
ll fu
nd-t
ype
vari
able
s ar
e ca
lcul
ated
usi
ng a
ll av
aila
ble
equi
ty f
unds
in
the
IFIC
dat
aset
, no
t jus
t th
ose
of p
ublic
spo
nsor
s. M
odel
s w
ith a
n ap
ostr
ophe
rep
rese
nt t
he r
emov
al o
f 15
(S
AL
ES
') &
7 (
RE
DE
M')
obse
rvat
ions
who
se s
tude
ntiz
ed r
esid
ual
valu
es f
rom
the
572
-obs
erva
tion
mod
el w
ere
grea
ter/
less
tha
n 3/
-3.
Bol
ded
valu
es
repr
esen
t es
timat
es s
igni
fica
nt a
t the
1%
, 5%
, or
10%
leve
l.
Par
amet
er E
stim
ates
Inte
rcep
t E
XC
ES
SP
ER
F t
EX
CE
SS
PE
RF
t-i
DIF
FR
ES
RE
SID
t
DIF
FR
ES
RE
SID
t-i
SA
LE
S t-i
/ R
ED
EM
M
ME
R
ST
DE
V t
ST
DE
V t-
i T
SA
LE
S t /
TR
ED
EM
t T
SA
LE
S t-i
/ T
RE
DE
M t-
i S
IZE
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
SAL
ES
1.61
65
0.19
68
0.15
27
-0.0
701
-0.2
042
0.11
17
-1.8
790
0.12
76
0.33
31
0.52
76
0.25
67
-0.1
466
31.3
6 2.
31
P-V
alu
es
<.0
001
0.00
01
<.0
001
0.64
39
0.09
73
<.0
001
0.40
97
0.82
71
0.39
59
0.00
07
0.15
07
<.0
001
0.00
10
<.0
001
RE
DE
M
0.05
27
0.04
76
-0.0
120
0.08
50
0.02
72
0.18
05
-1.2
289
0.13
16
-0.1
470
0.77
17
-0.0
372
-0.0
012
49.3
1 2.
57
P-V
alue
s
0.55
24
0.00
01
0.15
56
0.02
07
0.36
20
<.0
001
0.02
54
0.33
64
0.12
05
<.0
001
0.29
93
0.81
08
<.0
001
<.0
001
SA
LE
S'
0.83
63
0.20
21
0.06
55
0.05
41
-0.3
405
0.20
15
-1.5
329
-0.0
029
0.19
16
0.44
75
0.12
89
-0.0
758
27.1
0 1.
75
P-V
alue
s
0.00
03
<.0
001
0.04
35
0.59
79
0.00
05
<.0
001
0.31
14
0.99
41
0.46
57
<.0
001
0.27
68
<.0
001
0.00
44
<.0
001
RE
DE
M'
0.01
11
0.05
69
-0.0
115
0.08
38
0.01
85
0.18
87
-1.1
603
0.16
18
-0.1
032
0.77
93
-0.0
423
-0.0
001
55.1
3 3.
43
P-V
alue
s
0.88
36
<.0
001
0.10
71
0.00
77
0.46
45
<.0
001
0.01
29
0.16
21
0.19
78
<.0
001
0.16
40
0.97
68
<.0
001
<.0
001
Num
ber
of F
unds
11
9 11
9 11
5 11
9 N
umbe
r of
Sem
i-Ann
ual T
ime
Peri
ods
19
19
19
19
Num
ber
of F
und-
Yea
rs (
Obs
erva
tions
) 57
2 57
2 55
7 56
5 A
djus
ted
R-S
quar
e 56
.46%
58
.35%
64
.80%
64
.95%
-41-
Tab
le 6
- C
ompo
nent
Equ
ity
Inte
rnal
Flo
w M
odel
s w
ith
Rel
ativ
e P
erfo
rman
ce
Res
ults
of
a fix
ed-e
ffec
ts p
anel
reg
ress
ion
of th
e eq
uity
sam
ple
wit
h cu
rren
t ye
ar T
rans
fers
In
(TR
AN
SIN
t)
and
Tra
nsfe
rs O
ut (
TR
AN
SOU
T t)
as t
he d
epen
dent
va
riab
les.
EX
CE
SSPE
RF
t and
EX
CE
SSPE
RF
t_x
repr
esen
t th
e fu
nd's
curr
ent
and
prio
r ye
ar e
xces
s re
turn
rel
ativ
e to
ben
chm
arks
, D
IFF
RE
S , a
nd D
IFFR
ES
M
repr
esen
t cu
rren
t an
d pr
ior
year
ind
icat
or v
aria
bles
taki
ng th
e va
lue
of E
XC
ESS
PER
F if
EX
CE
SSPE
RF
> 0
, or
zero
oth
erw
ise,
TR
AN
SIN
t.j
and
TR
AN
SOU
T t_
] re
pres
ent
the
fund
's pr
ior
year
Tra
nsfe
rs I
n an
d O
ut,
ME
R r
epre
sent
s th
e fu
nd's
prio
r ye
ar M
anag
emen
t E
xpen
se R
atio
, ST
DE
V t a
nd S
TD
EV
,.[
repr
esen
t th
e fu
nd's
curr
ent
and
prio
r 12
-mon
th s
tand
ard
devi
atio
n of
abs
olut
e re
turn
s, T
TR
AN
SIN
t /
TR
AN
SOU
T a
nd T
TR
AN
SIN
M /
TT
RA
NSO
UT
,.,
repr
esen
t cu
rren
t an
d pr
ior
year
Sal
es a
nd R
edem
ptio
ns o
f th
e fu
nd's
obje
ctiv
e cl
ass
(or
"fun
d-ty
pe")
, an
d SI
ZE
is
the
natu
ral
loga
rith
m o
f th
e fu
nd's
prio
r ye
ar N
et A
sset
s.
DIF
FRE
SRE
SID
t
and
DIF
FRE
SRE
SID
,.,
repr
esen
t th
e re
sidu
als
of D
IFFR
ES
, and
DIF
FRE
S tA
reg
ress
ed a
gain
st E
XC
ESS
PER
F ,
and
EX
CE
SSPE
RF
t4,
resp
ectiv
ely.
TT
RA
NSO
UT
RE
SID
M r
epre
sent
s th
e re
sidu
als
of T
TR
AN
SOU
T t_
i reg
ress
ed a
gain
st T
TR
AN
SOU
T t. A
ll fu
nd-t
ype
vari
able
s ar
e ca
lcul
ated
usi
ng
all
avai
labl
e eq
uity
fun
ds i
n th
e IF
IC d
atas
et,
not
just
tho
se o
f pu
blic
spo
nsor
s. M
odel
s w
ith
an a
post
roph
e re
pres
ent
the
rem
oval
of
11 (
TR
AN
SIN
') &
8
(TR
AN
SO
UT
') ob
serv
atio
ns w
hose
stu
dent
ized
res
idua
l va
lues
fro
m t
he 5
72-o
bser
vatio
n m
odel
wer
e gr
eate
r/le
ss t
han
3/-3
. Bol
ded
valu
es r
epre
sent
est
imat
es
sign
ific
ant
at th
e 1%
, 5%
, or
10%
leve
l.
Par
amet
er E
stim
ates
Inte
rcep
t E
XC
ES
SP
ER
F t
EX
CE
SS
PE
RF
M
DIF
FR
ES
RE
SID
t
DIF
FR
ES
RE
SID
t-i
TR
AN
SIN
t-i /
TR
AN
SO
UT
t-i
ME
R
ST
DE
V t
ST
DE
V t-
i T
TR
AN
SIN
t /
TT
RA
NS
OU
T t
TT
RA
NS
IN t
-i /
TT
RA
NS
OU
TR
ES
ID M
S
IZE
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
TR
AN
SIN
1.66
54
0.03
09
-0.0
568
0.62
05
-0.0
551
0.10
58
-6.8
228
2.65
37
0.37
53
1.03
61
0.84
47
-0.1
309
35.8
3 2.
19
P-V
alue
s
0.04
65
0.80
13
0.48
31
0.09
21
0.85
61
0.00
92
0.21
64
0.05
53
0.69
82
0.00
13
0.63
93
0.01
16
0.00
02
<.0
001
TR
AN
SOU
T
-0.1
182
-0.0
278
-0.0
731
0.49
90
0.24
97
0.19
09
-4.1
023
1.58
95
0.51
06
0.78
09
1.52
58
0.01
10
39.1
4 1.
96
P-V
alue
s
0.86
89
0.79
79
0.27
89
0.13
08
0.35
24
<.0
001
0.40
77
0.20
20
0.55
46
0.00
80
0.00
04
0.81
21
<.0
001
<.0
001
TR
AN
SIN
'
1.34
99
0.19
18
0.02
99
-0.0
157
-0.1
669
0.09
23
-2.0
460
-0.0
904
-0.3
203
0.71
39
-0.9
336
-0.0
637
73.3
1 4.
87
P-V
alue
s
0.00
18
0.00
26
0.48
16
0.93
48
0.28
36
<.0
001
0.46
79
0.90
22
0.51
88
<.0
001
0.31
49
0.01
88
<.0
001
<.0
001
TR
AN
SO
UT
0.30
59
0.05
42
-0.0
100
0.25
38
0.15
38
0.12
19
-0.8
507
0.53
39
0.06
33
0.60
22
0.64
32
-0.0
106
147.
71
5.48
P-V
alue
s
0.38
18
0.30
84
0.76
22
0.11
94
0.24
26
<.00
01
0.72
48
0.38
20
0.88
08
<.0
001
0.00
28
0.63
95
<00
01
<.0
001
Num
ber
of F
unds
11
9 11
9 11
9 11
9 N
umbe
r of
Sem
i-Ann
ual T
ime
Peri
ods
19
19
19
19
Num
ber
of F
und-
Yea
rs (
Obs
erva
tions
) 57
2 57
2 56
1 56
4 A
djus
ted
R-S
quar
e 42
.56%
47
.51%
61
.66%
68
.18%
-42-
Tab
le 7
- C
ompo
nent
Fix
ed-I
ncom
e E
xter
nal
Flo
w M
odel
s w
ith
Rel
ativ
e P
erfo
rman
ce
Res
ults
of
a fi
xed-
effe
cts
pane
l re
gres
sion
of
the
fixe
d-in
com
e sa
mpl
e w
ith
curr
ent
year
Sal
es (
SAL
ES
t) an
d R
edem
ptio
ns (
RE
DE
M ,
) as
the
dep
ende
nt
vari
able
s. E
XC
ESS
PER
F , a
nd E
XC
ESS
PER
F tA
rep
rese
nt t
he f
und'
s cu
rren
t an
d pr
ior
year
exc
ess
retu
rn r
elat
ive
to b
ench
mar
ks,
DIF
FRE
S t a
nd D
1FFR
ES
t.i
repr
esen
t cu
rren
t an
d pr
ior
year
ind
icat
or v
aria
bles
tak
ing
the
valu
e of
EX
CE
SSPE
RF
if E
XC
ESS
PER
F >
0,
or z
ero
othe
rwis
e, S
AL
ES
M a
nd R
ED
EM
t.j
repr
esen
t th
e fu
nd's
prio
r ye
ar S
ales
and
Red
empt
ions
, M
ER
repr
esen
ts t
he f
und'
s pr
ior
year
Man
agem
ent
Exp
ense
Rat
io,
STD
EV
t a
nd S
TD
EV
t_]
repr
esen
t the
fu
nd's
curr
ent
and
prio
r 12
-mon
th s
tand
ard
devi
atio
n of
abs
olut
e re
turn
s, T
SA
LE
S t /
TR
ED
EM
t a
nd T
SA
LE
SH
/ T
RE
DE
Mj.j
rep
rese
nt c
urre
nt a
nd p
rior
yea
r Sa
les
and
Red
empt
ions
of t
he f
und'
s ob
ject
ive
clas
s (o
r "f
und-
type
"),
and
SIZ
E i
s th
e na
tura
l lo
gari
thm
of
the
fund
's pr
ior
year
Net
Ass
ets.
DIF
FRE
SRE
SID
t a
nd
DIF
FRE
SRE
SID
M
re
pres
ent
the
resi
dual
s of
DIF
FRE
S t
and
DIF
FRE
S ,.,
re
gres
sed
agai
nst
EX
CE
SS
PE
RF
, an
d E
XC
ES
SP
ER
F^,
re
spec
tivel
y,
and
RE
DE
MR
ESI
D ,
., an
d T
SAL
ESR
ESI
D t
rep
rese
nt t
he r
esid
uals
of
RE
DE
M ,_
i and
TSA
LE
S t r
egre
ssed
aga
inst
ME
R, r
espe
ctiv
ely.
All
fun
d-ty
pe v
aria
bles
are
ca
lcul
ated
usi
ng a
ll a
vaila
ble
fixe
d-in
com
e fu
nds
in t
he I
FIC
dat
aset
, no
t jus
t th
ose
of p
ublic
spo
nsor
s. M
odel
s w
ith
an a
post
roph
e re
pres
ent
the
rem
oval
of
6 (S
AL
ES
') &
7 (
RE
DE
M')
obse
rvat
ions
who
se s
tude
ntiz
ed r
esid
ual
valu
es f
rom
the
215
-obs
erva
tion
mod
el w
ere
grea
ter/
less
tha
n 3/
-3. B
olde
d va
lues
rep
rese
nt
estim
ates
sig
nifi
cant
at t
he 1
%, 5
%, o
r 10
% le
vel.
Par
amet
er E
stim
ates
Inte
rcep
t E
XC
ES
SP
ER
F t
EX
CE
SS
PE
RF
t-i
DIF
FR
ES
RE
SID
t
DIF
FR
ES
RE
SID
t-i
SA
LE
S n
/ R
ED
EM
RE
SID
t-i
ME
R
ST
DE
V t
ST
DE
V t-
i T
SA
LE
SR
ES
ID t
/ T
RE
DE
M t
TR
ED
EM
t-i
SIZ
E
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
SAL
ES
2.20
86
0.15
17
-0.1
224
1.52
08
0.11
41
0.28
51
-8.5
322
4.97
79
0.60
59
0.50
03
-0.1
689
18.2
4 3.
08
P-V
alue
s
0.00
09
0.47
54
0.53
40
0.00
23
0.81
64
<.0
001
0.40
20
0.00
17
0.68
43
<.0
001
0.00
11
0.05
10
<.0
001
RE
DE
M
0.70
36
-0.0
514
-0.0
753
0.66
37
0.28
14
0.55
64
-26.
3120
3.
6997
-0
.539
7 0.
7548
0.
0650
-0
.010
4
15.4
9 1.
26
P-V
alue
s
0.08
39
0.68
77
0.54
25
0.02
60
0.34
37
<.0
001
<.0
001
0.00
01
0.56
19
0.00
03
0.46
36
0.72
99
0.16
11
0.13
37
SA
LE
S'
2.20
33
0.25
69
-0.0
963
0.58
89
-0.0
329
0.25
09
-2.7
063
0.86
68
1.36
45
0.29
10
-0.1
779
22.9
8 5.
73
P-V
alue
s
<.0
001
0.10
03
0.48
38
0.11
24
0.92
33
<.0
001
0.70
26
0.65
38
0.18
94
0.00
12
<.0
001
0.01
08
<.0
001
RE
DE
M'
1.07
19
-0.0
334
-0.0
205
-0.0
841
0.24
31
0.59
41
-29.
7428
1.
1780
0.
3745
0.
5035
0.
0748
-0
.029
7
18.0
4 1.
97
P-V
alue
s
0.00
01
0.71
69
0.80
46
0.69
17
0.23
17
<0
00
1 <
.000
1 0.
2982
0.
5505
0.
0007
0.
2189
0.
1489
0.08
06
0.00
06
Num
ber
of F
unds
44
44
43
44
N
umbe
r of
Sem
i-Ann
ual T
ime
Peri
ods
19
19
19
19
Num
ber
of F
und-
Yea
rs (
Obs
erva
tions
) 21
5 21
5 20
9 20
8 A
djus
ted
R-S
quar
e 83
.05%
90
.39%
88
.68%
94
.77%
-43-
Tab
le 8
- C
ompo
nent
Fix
ed-I
ncom
e In
tern
al F
low
Mod
els
wit
h R
elat
ive
Per
form
ance
Res
ults
of
a fi
xed-
effe
cts
pane
l re
gres
sion
of
the
fixe
d-in
com
e sa
mpl
e w
ith c
urre
nt y
ear
Tra
nsfe
rs I
n (T
RA
NSI
N t
) an
d T
rans
fers
Out
(T
RA
NSO
UT
t)
as t
he
depe
nden
t va
riab
les.
EX
CE
SSPE
RF
t and
EX
CE
SSPE
RF
t_! r
epre
sent
the
fun
d's
curr
ent
and
prio
r ye
ar e
xces
s re
turn
rel
ativ
e to
ben
chm
arks
, D
IFFR
ES
t and
D
IFFR
ES
t-i r
epre
sent
cur
rent
and
pri
or y
ear
indi
cato
r va
riab
les
taki
ng t
he v
alue
of
EX
CE
SSPE
RF
if E
XC
ESS
PER
F >
0,
or z
ero
othe
rwis
e, S
AL
ES
,.] a
nd
RE
DE
M n
rep
rese
nt t
he f
und'
s pr
ior
year
Sal
es a
nd R
edem
ptio
ns,
ME
R re
pres
ents
the
fun
d's
prio
r ye
ar M
anag
emen
t E
xpen
se R
atio
, ST
DE
V t a
nd S
TD
EV
,_!
repr
esen
t th
e fu
nd's
curr
ent
and
prio
r 12
-mon
th s
tand
ard
devi
atio
n of
abs
olut
e re
turn
s, T
SAL
ES
t / T
RE
DE
M t a
nd T
SAL
ES
t-i /
TR
ED
EM
,_] r
epre
sent
cur
rent
an
d pr
ior
year
Sal
es a
nd R
edem
ptio
ns o
f th
e fu
nd's
obje
ctiv
e cl
ass
(or
"fun
d-ty
pe")
, an
d SI
ZE
is
the
natu
ral
loga
rith
m o
f th
e fu
nd's
prio
r ye
ar N
et A
sset
s.
DIF
FRE
SRE
SID
, a
nd D
IFFR
ESR
ESI
D ,
., re
pres
ent
the
resi
dual
s of
DIF
FRE
S t
and
DIF
FRE
S ,.,
reg
ress
ed a
gain
st E
XC
ES
SP
ER
F ,
and
EX
CE
SSPE
RF
,.,,
resp
ectiv
ely,
and
TT
RA
NSO
UT
RE
SID
t
repr
esen
ts t
he r
esid
uals
of
TT
RA
NSO
UT
t
agai
nst
ME
R.
Bot
h la
gged
fun
d-ty
pe
flow
var
iabl
es a
re e
xclu
ded
due
mul
ticol
linea
rity
tha
t is
per
sist
ent
afte
r or
thog
onal
izat
ion.
All
fun
d-ty
pe v
aria
bles
are
cal
cula
ted
usin
g al
l av
aila
ble
fixe
d-in
com
e fu
nds
in t
he I
FIC
dat
aset
, no
t ju
st t
hose
of
publ
ic s
pons
ors.
Mod
els
with
an
apos
trop
he r
epre
sent
the
rem
oval
of
6 (T
RA
NS
IN')
& 7
(T
RA
NSO
UT
') ob
serv
atio
ns w
hose
stu
dent
ized
res
idua
l va
lues
fro
m t
he 2
15-o
bser
vatio
n m
odel
wer
e gr
eate
r/le
ss t
han
3/-3
. Bol
ded
valu
es r
epre
sent
est
imat
es s
igni
fica
nt a
t the
1%
, 5%
, or
10%
leve
l.
Par
amet
er E
stim
ates
Inte
rcep
t E
XC
ES
SP
ER
F t
EX
CE
SS
PE
RF
t-i
DIF
FR
ES
RE
SID
, D
IFF
RE
SR
ES
ID t-
i T
RA
NS
IN t-
i / T
RA
NS
OU
T t-
i M
ER
ST
DE
V t
STD
EV
t-i
TT
RA
NS
INR
ES
ID t
/ T
TR
AN
SO
UT
RE
SID
t
SIZ
E
Hau
sman
Tes
t fo
r R
ando
m E
ffec
ts
F-T
est
for
No
Fix
ed E
ffec
ts
TR
AN
SIN
2.34
92
1.32
69
1.45
25
5.36
06
-0.8
230
0.20
90
-109
.334
0 -7
.089
2 -5
.667
6 4.
1390
0.07
36
19.6
6 1.
22
P-V
alue
s
0.46
81
0.19
83
0.12
49
0.02
59
0.73
83
0.00
78
0.01
19
0.35
73
0.43
48
<.0
001
0.77
05
0.03
26
0.17
30
TR
AN
SOU
T
-0.3
529
2.76
57
1.49
61
4.54
33
-5.7
730
0.34
79
-36.
5985
-1
4.31
76
-0.8
528
1.29
92
0.21
99
21.9
2 1.
04
P-V
aluc
s
0.91
49
0.00
71
0.13
26
0.06
82
0.01
60
0.00
02
0.38
92
0.06
57
0.90
84
0.17
76
0.39
58
0.01
55
0.41
86
TR
AN
SIN
'
2.70
94
-0.2
925
0.61
82
1.76
03
-0.0
056
0.38
60
-64.
8865
2.
2748
-2
.224
5 2.
6655
-0.0
946
25.3
7 1.
47
P-V
all
ies
0.18
97
0.65
81
0.30
19
0.25
16
0.99
71
<.0
001
0.01
82
0.64
27
0.62
47
<.0
001
0.55
74
0.00
47
0.03
38
TR
AN
SO
UT
1.60
54
0.74
52
1.07
64
0.74
19
-3.9
161
0.50
29
-23.
5473
-2
.501
7 1.
6624
0.
7599
-0.0
407
34.6
6 1.
29
P-V
alue
s
0.44
86
0.26
50
0.09
55
0.64
58
0.01
10
<.0
001
0.38
52
0.61
87
0.72
51
0.23
31
0.80
69
0.00
01
0.11
41
Num
ber
of F
unds
44
44
44
44
N
umbe
r of
Sem
i-A
nnua
l Tim
e Pe
riod
s 19
19
19
19
N
umbe
r of
Fun
d-Y
ears
(O
bser
vatio
ns)
215
215
211
211
Adj
uste
d R
-Squ
are
58.3
5%
53.3
0%
65.2
7%
60.0
8%
-44-
Tab
le 9
-Ret
urns
to
Scal
e M
odel
: E
quit
y Sa
mpl
e
Res
ults
of
a si
mpl
e lo
gist
ic r
egre
ssio
n w
ith t
he d
epen
dent
var
iabl
e, W
INN
ER
t,
taki
ng t
he v
alue
of
1 if
the
fun
d's
exce
ss r
etur
n, r
elat
ive
to i
ts b
ench
mar
k, i
s po
siti
ve,
and
0 ot
herw
ise.
SIZ
E i
s th
e na
tura
l lo
gari
thm
of
the
fund
's pr
ior
year
Net
Ass
ets.
The
obs
erva
tions
em
ploy
ed i
n th
e m
odel
cor
resp
ond
to t
he e
quity
B
M' m
odel
pre
sent
ed i
n T
able
3 a
bove
.
Res
pons
e P
rofil
e:
Win
ner
•
Val
ue
Freq
uenc
y 1
209
0 35
0 T
otal
55
9 Max
imum
Lik
elih
ood
Est
imat
es
Est
imat
e SE
W
ald
Chi
-Squ
are
P-V
alue
0.
6565
1.
2790
0.
2633
0.
6079
0.
8051
0.
1792
20
.190
0 <.
0001
-0
.115
8 0.
8907
1.
4998
0.
2207
Odd
s R
atio
Est
imat
es
^__
Est
imat
e 95
% W
ald
C.L
. W
inne
r t-i
2.
237
1.57
4 3.
178
Size
0.
891
0.74
1.
072
Mod
el I
'll S
Uiti
stic
s C
rite
rion
Im
. Onl
v In
t. v\
.'Con
cilia
tes
* A
IC
740.
986
721.
701
SC
745.
312
734.
68
-2L
og
L
738.
986
715.
701
-45-
Inte
rcep
t W
inne
r t-i
Size
Tab
le 1
0 -
Ret
urns
to
Scal
e M
odel
: F
ixed
-Inc
ome
Sam
ple
Res
ults
of
a si
mpl
e lo
gist
ic r
egre
ssio
n w
ith t
he d
epen
dent
var
iabl
e, W
INN
ER
t,
taki
ng t
he v
alue
of
1 if
the
fun
d's
exce
ss r
etur
n, r
elat
ive
to i
ts b
ench
mar
k, i
s po
siti
ve,
and
0 ot
herw
ise.
SIZ
E i
s th
e na
tura
l lo
gari
thm
of
the
fund
's pr
ior
year
Net
Ass
ets.
The
obs
erva
tions
em
ploy
ed i
n th
e m
odel
cor
resp
ond
to t
he f
ixed
-in
com
e B
M' m
odel
pre
sent
ed i
n T
able
4 a
bove
.
Res
pons
e P
rofi
le:
Win
ner
t
Val
ue
1 0
Freq
uenc
y 89
12
1 T
otal
21
0 Max
imum
Lik
elih
ood
Est
imat
es
Inte
rcep
t W
inne
r t-i
Size
Est
imat
e -0
.272
2 0.
0689
-0
.005
0
SE
1.47
86
0.28
24
0.11
35
Wal
d C
hi-S
quar
e 0.
0339
0.
0595
0.
0019
P-V
alue
0.
8540
0.
8073
0.
9651
Odd
s R
atio
Est
imat
es
Est
imat
e 95
% W
ald
C.L
. W
inne
r .-i
1.
071
0.61
6 1.
863
Size
0.
995
0.79
7 0.
1243
Mod
el F
it S
tati
stic
s C
rite
rion
In
t. O
nly
Int.
w/
Cov
aria
tes
AIC
SC
-2 Log L
288.227
291.574
286.227
292.166
302.208
286.166
-46-
REFERENCES
Barber, B. M., Odean, T., & Zheng, L. (2005). Out of sight, out of mind: The effects of expenses on mutual fund flows. Journal of Business, 78(6), 2095-2119.
Berk, J. B., & Green, R. C. (2004). Mutual fund flows and performance in rational markets. Journal of Political Economy, 112(6), 1269-1295.
Berkowitz, M. K., & Qiu, J. (2003). Ownership, risk and performance of mutual fund management companies. Journal of Economics and Business, 55(2), 109-134.
Brown, S. J., & Goetzmann, W. N. (1995). Performance persistence. The Journal of Finance, 50(2), 679-698.
Carhart, M. M. (1997). On persistence in mutual fund performance. The Journal of Finance, 52(1), 57-82.
Capon, N., Fitzsimons, G.J. & Prince, R.A. (1996). An individual level analysis of the mutual fund investment decision. Journal of Financial Services Research, 10(1), 59-82.
Chevalier, J., & Ellison, G. (1997). Risk taking by mutual funds as a response to incentives. Journal of Political Economy, 105(6), 1167-1200.
Deaves, R. (2004). Data-conditioning biases, performance, persistence and flows: The case of Canadian equity funds. Journal of Banking & Finance, 28(3), 673-694.
Del Guercio, D., & Tkac, P. A. (2002). The determinants of the flow of funds of managed portfolios: Mutual funds vs. pension funds. Journal of Financial and Quantitative Analysis, 37(A), 523-557.
Droms, W. G., & Walker, D. A. (2001). Persistence of mutual fund operating characteristics: Returns, turnover rates, and expense ratios. Applied Financial Economics, 11(4), 457-466.
Edelen, R. M. (1999). Investor flows and the assessed performance of open-end mutual funds. Journal of Financial Economics, 53(3), 439-466.
Fant, L. F. (1999). Investment behavior of mutual fund shareholders: The evidence from aggregate fund flows. Journal of Financial Markets, 2(4), 391-402.
Goetzmann, W. N., & Peles, N. (1997). Cognitive dissonance and mutual fund investors. Journal of Financial Research, 20(2), 145-158.
Goetzmann, W. N., & Ibbotson, R. G. (1994). Do winners repeat? Journal of Portfolio Management, 20(2), 9-18.
-47-
Grinblatt, M., & Titman, S. (1992). The persistence of mutual fund performance. The Journal of Finance, 47(5), 1977-1984.
Gruber, M. J. (1996). Another puzzle: The growth in actively managed mutual funds. The Journal of Finance, 5/(3), 783-810.
Harless, D. W., & Peterson, S. P. (1998). Investor behavior and the persistence of poorly-performing mutual funds. Journal of Economic Behavior & Organization, 37(3), 257-276.
Hendricks, D., Patel, J., & Zeckhauser, R. (1993). Hot hands in mutual funds: Short-run persistence of relative performance, 1974-1988. The Journal of Finance, 48(1), 93-130.
Ippolito, R. A. (1992). Consumer reaction to measures of poor quality: Evidence from the mutual fund industry. Journal of Law and Economics, 35(1), 45-70.
Lynch, A. W., & Musto, D. K. (2003). How investors interpret past fund returns. The Journal of Finance, 58(5), 2033.
Patel, J., Zeckhauser, R. J., & Hendricks, D. (1994). Investment flows and performance: Evidence from mutual funds, cross-border investments, and new issues. In R. Sato, R. M. Levich & R. V. Ramachandran (Eds.), Japan, Europe, and International Financial Markets (pp. 51-72). New York: Cambridge University Press.
Ruckman, K. (2003). Expense ratios of North American mutual funds. Canadian Journal of Economics, 36(1), 192-223.
Sinha, R., & Jog, V. (2005). Fund flows and performance: A study of Canadian equity funds. (Working Paper). Available at SSRN: http://ssrn.com/abstract=696621
Sigurdsson, K. (2004, March) The Effect of New Money Inflows on the Flow-Performance Relationship in the U.S. Mutual Fund Industry (Working Paper). Available at SSRN: http://ssrn.com/abstract=537823
Sirri, E. R., & Tufano, P. (1998). Costly search and mutual fund flows. Journal of Finance, 53(5), 1589-1622.
Smith, K. (1978). Is fund growth related to fund performance? Journal of Portfolio Management, 4(3), 49-54.
Spitz, A. E. (1970). Mutual fund performance and cash inflows. Applied Economics, 2(2), 141-145.
*As reported in the "Stock Header Information" section of the CFMRC database
** Parent company of Dynamic Mutual Funds
*** Parent company of Investors Group, and as of 2003, Mackenzie Investment Management
-50-
Appendix 3 - Variable Definitions
Excerpt from page 2 and 3 of IFIC's Statistical Reporting Guidelines 2002:
"Definitions of Data Fields (within the mutual fund category)
Eleven fields of information are reported for each individual mutual fund including; total net assets, cash and short-term liabilities, sales, reinvested dividends, redemptions, transfers in, transfers out, net sales, number of unitholders, distribution of capital gains and distribution of income.
Further clarification of these terms is as follows:
1. TOTAL NET ASSETS - are defined as the sum of all securities within the fund, at market, at close of the last business day of the current month. Included in this category should be cash, accrued income (i.e. dividend and interest), accrued liabilities (for the current period), commission payable, tax liabilities, broker receivable/payable, any outstanding sales or redemption orders (e.g. wire transfers) and any other outstanding payables or receivables. All assets should be expressed in Canadian dollars.
2. CASH AND SHORT-TERM INVESTMENTS - includes the value of short-term bonds and notes, up to and including a term of 365 days (which is in accordance with standard GAAP convention for short term). Liabilities accrued in net assets must be accrued in cash and short-term also. This is also in accordance with standard GAAP convention for short term. Liabilities accrued in net assets must be accrued in cash and short-term also. This number can be negative (if the fund is over invested), but can never be greater than the asset values. Reported cash must be net of unrealized derivative exposure where applicable. If the derivative is closed out, it would be treated as cash and the value calculated at the date of close out.
3. GROSS SALES - include all sales for the accounting period, exclusive of Reinvested Distributions and Transfers In and Transfers Out. Gross Sales should also include all outstanding wire orders, where applicable and must be net of interfund gross sales, and net of sales charges/loads.
4. REINVESTED DISTRIBUTIONS - must include all capital gains, dividends, interest income and return of capital for the accounting period. To the extent that distributions by a Fund in any year exceed the net income and net realized taxable capital gains of the Fund for that year, those distributions will be a return of capital.
5. REDEMPTIONS - must include all redemptions for the accounting period and must include all outstanding wire orders, where applicable. Redemptions must be net of interfund redemptions.
6. TRANSFERS IN/OUT must include all outstanding wire orders, where applicable and must be net of interfund transfers. Transfers in and out must be equal for the month for each member firm. Where a manager has products other than mutual funds and a customer transfers out of mutual funds into another product (i.e. term deposit), the transfer should be reported as a redemption rather than a transfer out. Where a customer transfers out of a non-mutual fund product (i.e. term deposit) into a mutual fund it should be reported as a sale rather than a transfer in.
7. DISTRIBUTIONS - must include all capital gains, dividends, interest income and return of capital for the accounting period. The total distributions should always be greater than the reinvested distributions. The exception to this would be when there are positive/negative adjustments, even when no declarations have been made."
-51-
Appendix 4 - List of Funds in Equity and Fixed-Income Samples by Sponsor
Equity Sample Sponsor Code Type Most Recent Fund Name
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
Bissett & Associates Investment Mgmt
Bissett & Associates Investment Mgmt
BPI Capital Management Corp.
BPI Capital Management Corp.
BPI Capital Management Corp.
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Guardian Group of Funds Ltd.
Guardian Group of Funds Ltd.
Guardian Group of Funds Ltd.
Guardian Group of Funds Ltd.
Guardian Group of Funds Ltd.
AGFAAA
AGFAEQ
AGFAGB
AGFAGG
AGFASB
AGFCAA
AGFCEB
AGFEGB
AGFGEB
AGFGIB
AGFGRO
AGFIEQ
AGFISS
AGFRIV
AGFSMA
AGFUSG
B1SCDN
BISSC
BPICAF
BPICEI
BPIEQU
DYNCGF
DYNEQU
DYNEUR
DYNFOC
DYNFOF
DYNGRO
DYNIBA
DYNINF
DYNITF
DYNNBF
DYNPAM
DYNPAR
DYNPCA
DYNPOB
DYNPRE
GUGAME
GUGBAL
GUGENT
GUGGLO
GUGGRO
BA
CS
US
us CF
BA
CS
CF
CS
BA
CS
CF
CF
CF
CS
CF
CS
CS
CS
CS
CF
CS
CS
CF
CS
BA
CS
US
BA
CF
BA
US
BA
CS
BA
CS
us BA
CS
CF
CS
AGF World Balanced Fund
AGF Canadian Aggressive Equity Fund
AGF American Growth Class
AGF Aggressive Growth Fund
AGF Asian Growth Class
AGF Canadian Real Value Balanced Fund
AGF Canadian Equity Fund Limited
AGF European Equity Class
AGF Canadian Growth Equity Fund
AGF Canadian Balanced Fund
AGF Canadian Stock Fund
AGF RSP International Equity Allocation Fund
AGF International Stock Class
AGF RSP International Value Fund
AGF Canadian Small Cap Fund
AGF International Value Fund
Bissett Canadian Equity Fund
Bissett Small Cap Fund
BPI Canadian Small Companies Fund
BPI Canadian Equity Value Fund
BPI Global Equity Fund
Dynamic Canadian Growth Fund
Dynamic Focus+ Equity Fund
Dynamic European Value Fund
Dynamic Value Fund of Canada
Dynamic Strategic Growth Portfolio
Dynamic Focus+ Wealth Management Fund
Dynamic Focus+ American Fund
Dynamic Focus+ Balanced Fund
Dynamic International Value Fund
Dynamic Value Balanced Fund
Dynamic Power American Growth Fund
Commonwealth Canadian Balanced Fund
Dynamic Power Canadian Growth Fund
Dynamic Power Balanced Fund
Dynamic Precious Metals Fund
GGOF American Value Fund Ltd.
GGOF Canadian Balanced Fund
GGOF Enterprise Fund
GGOF Global Growth Fund
GGOF Canadian Equity Fund
-52-
Sponsor
Mackenzie Financial Corporation
Code Type Most Recent Fund Name
INDBAL BA Mackenzie Balanced Fund
Mackenzie Financial Corporation INDCBA BA Mackenzie Universal Canadian Balanced Fund
Mackenzie Financial Corporation INDCDN CS Mackenzie Ivy Canadian Fund
Mackenzie Financial Corporation INDCEF CS Mackenzie Universal Canadian Growth Fund
Mackenzie Financial Corporation INDDIV CS Mackenzie Maxxum Dividend Growth Fund
Mackenzie Financial Corporation INDEUR CF Mackenzie Universal European Opportunities Fund
Mackenzie Financial Corporation INDFUT CS Mackenzie Universal Future Fund
Mackenzie Financial Corporation INDGRO CS Mackenzie Growth Fund
Mackenzie Financial Corporation INDHOR CS Mackenzie Maxxum Canadian Value Fund Mackenzie Financial Corporation INDIAM CF Mackenzie Ivy Foreign Equity Fund
Mackenzie Financial Corporation INDIBO BA Mackenzie Ivy Growth & Income Fund
Mackenzie Financial Corporation INDINC BA Mackenzie Sentinel Income Fund
Mackenzie Financial Corporation INDSMF CF Mackenzie Select Managers Fund
Mackenzie Financial Corporation INDWST CF Universal World Science & Technology Fund
Investors Group Inc. INVAA BA Investors Asset Allocation Fund
Investors Group Inc. INVBAL BA Investors Canadian Balanced Fund
Investors Group Inc. INVCSC CS Investors Canadian Small Cap Fund
Investors Group Inc. 1NVEQU CS Investors Canadian Equity Fund
Investors Group Inc. INVEUR CF Investors European Growth Fund
Investors Group Inc. INVGLO CF Investors Global Fund
Investors Group Inc. INVGP CF Investors Growth Portfolio
Investors Group Inc. INVGPP BA Investors Growth Plus Portfolio
Investors Group Inc. INVGRO US Investors North American Growth Fund
Investors Group Inc. INVGST CF Investors Global Science & Technology Fund
Investors Group Inc. INVIPP BA Investors Income Plus Portfolio
Investors Group Inc. INVIUG US Investors U.S. Large Cap Value RSP Fund
Investors Group Inc. INVMLC CS IG AGF Canadian Growth Fund Investors Group Inc. FNVMUT BA Investors Mutual of Canada
Investors Group Inc. INVPAC CF Investors Pacific International Fund
Investors Group Inc. INVRAL BA IG AGF Canadian Balanced Fund
Investors Group Inc. INVRET CS Investors Canadian Large Cap Value Fund
Investors Group Inc. INVRGP CS Investors Retirement Growth Portfolio
Investors Group Inc. INVRPP BA Investors Retirement Plus Portfolio
Investors Group Inc. INVSUM CS Investors Summa Fund
Investors Group Inc. INVUSG US Investors U.S. Large Cap Value Fund Investors Group Inc. INVWG CF Investors World Growth Portfolio O'Donnell Investment Management OIMCEG CS O'Donnell Canadian Emerging Growth Fund O'Donnell Investment Management OIMGRO CS O'Donnell Growth Fund
-53-
Sponsor Code Type Most Recent Fund Name CT Investment Management Group Inc.
CT Investment Management Group Inc.
CT Investment Management Group Inc.
CT Investment Management Group Inc.
Sceptre Investment Counsel Limited
Sceptre Investment Counsel Limited
Sceptre Investment Counsel Limited
Trimark Investment Management Inc.
Trimark Investment Management Inc.
Trimark Investment Management Inc.
Trimark Investment Management Inc.
Trimark Investment Management Inc.
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.l. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.L Investments
C.L Investments
C.I. Investments
C.L Investments
C.L Investments
C.L Investments
C.L Investments
C.L Investments
C.L Investments
PERBAL
PEREAG
PERGRO PERSPE
SCEBAL
SCEEQU
SCEITL
TRIBAL
TRIFUN
TRIGRO
TRISCG
TRISEL
UNVAEQ
UNVCAF
UNVCBA
UNVCIC
UNVDEF
UNVEME
UNVEQU
UNVGBR
UNVGBS
UNVGEE UNVGEQ
UNVGLO
UNVGOF
UNVHFF
UNVHGI
UNVIBA
UNVIBR
UNVIGF
UNVINC UNVLAM
UNVMSS
UNVPAC
UNVSGB
UNVSGH
UNVSGL
UNVSGT
UNVSGU
BA
US
CS
cs BA
cs CF
BA
CF
CF
CS
cs us cs BA
CS
cs CF
CF
BA
CF
CF
CF
CF
CF
CS
BA
BA
BA
BA
BA
CF
CF
CF
BA
CF
CF
CF
CF
Canada Trust Balanced Fund-Inv
Canada Trust Amerigrowth Fund-Inv
Canada Trust Stock Fund-Inv
Canada Trust Special Equity Fund-Inv
Sceptre Balanced Growth Fund
Sceptre Equity Growth Fund
Sceptre Global Equity Fund
Trimark Select Balanced Fund
Trimark Fund
Trimark Select Growth Fund
Trimark Select Canadian Growth Fund
Trimark RSP Equity Fund
C.L American Equity Fund
C.L Explorer Fund
C.L Canadian Balanced Fund
C.L Canadian Growth Fund
C.I. Signature Select Canadian Fund
C.L Emerging Markets Fund
BPI Global Equity Fund
C.L Global Boomernomics RSP
CT. Global Biotechnology Corporate Class
BPI Global Equity RSP Fund
C.L Global RSP Fund
C.L Global Fund
BPI Global Opportunities I Fund
C.L Harbour Fund
C.L Harbour Growth & Income Fund
C.L International Balanced Fund
C.L International Balanced RSP Fund
C.L Signature Canadian Balanced
C.L Signature Canadian Income Fund
C.L Latin American Fund
C.L Global Managers Corporate Class
C.L Pacific Fund
C.L Global Boomernomics Corporate Class
C.L Global Health Sciences Corporate Class
C.L Global Corporate Class
C.L Global Telecommunications Sector Shares
C.L Global Science & Technology Corporate Class
-54-
Fixed-Income Sample Sponsor Code Type Most Recent Fund Name
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
AGF Funds Inc.
Bissett & Associates Investment Mgmt
Bissett & Associates Investment Mgmt
BPI Capital Management Corp.
BPI Capital Management Corp.
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Dynamic Mutual Funds
Guardian Group of Funds Ltd.
Guardian Group of Funds Ltd.
Guardian Group of Funds Ltd.
Mackenzie Financial Corporation
Mackenzie Financial Corporation
Mackenzie Financial Corporation
Mackenzie Financial Corporation
Mackenzie Financial Corporation
Mackenzie Financial Corporation
Investors Group Inc.
Investors Group Inc.
Investors Group Inc.
Investors Group Inc.
Investors Group Inc.
Investors Group Inc.
Investors Group Inc.
O'Donnell Investment Management
CT Investment Management Group Inc.
CT Investment Management Group Inc.
CT Investment Management Group Inc.
Sceptre Investment Counsel Limited
Sceptre Investment Counsel Limited
Trimark Investment Management Inc.
Trimark Investment Management Inc.
Trimark Investment Management Inc.
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
C.I. Investments
AGFCBB
AGFDIV
AGFHIB
AGFMMB
BISBON
BISMMF
BPIBIL
BPIIF
DYNCDI
DYNDGF
DYNDIV
DYNINC
DYNMMF
GUGHIF
GUGITL
GUGPFD
INDBON
INDCAS
INDCBF
INDDIV
INDMMO
INDSHT
INVBON
INVDIV
INVIP
INVIPP
INVMDF
INVMMF
INVMOR
OIMHIF
PERBON
PERRMO
PERSHO
SCEBON
SCEMMF
TRIABO TRIGIF
TRIINT
UNVCBO UNVGBO
UNVHIF
UNVIF
UNVMMF
UNVWBF
BO
DI
BO
MM
BO
MM
MM
DI
DI
DI
DI
BO
MM
DI
BF
DI
BO
MM
BF
DI
DM
MM
BO
DI
BO
BO
DI
MM
DM
BO
BO
DM
MM
BO
MM
BO
BO
MM
BO BF
DI
DI
MM
Ss-
AGF Canadian Bond Fund
AGF Canadian Large Cap Dividend Fund
AGF Canadian Conservative Income Fund
AGF Canadian Money Market Fund
Bissett Bond Fund
Bissett Money Market Fund
BPI T-Bill Fund
BPI Income Fund
Dynamic Canadian Dividend Fund Ltd.
Dynamic Dividend Value Fund
Dynamic Dividend Fund
Dynamic Income Fund
Dynamic Money Market Fund
GGOF Monthly High Income Fund
GGOF RSP Global Bond Fund
GGOF Monthly Dividend Fund Ltd.
Mackenzie Sentinel Bond Fund
Mackenzie Sentinel Cash Management Fund
Mackenzie Sentinel RRSP Global Bond Fund
Mackenzie Maxxum Dividend Growth Fund
Industrial Mortgage Securities Fund
Mackenzie Sentinel Money Market Fund
Investors Government Bond Fund
Investors Dividend Fund
Investors Income Portfolio Fund
Investors Income Plus Portfolio
IG Mackenzie Maxxum Dividend Growth Fund
Investors Canadian Money Market Fund
Investors Mortgage Fund
O'Donnell High Income Fund
Canada Trust Bond Fund-Inv
Canada Trust Mortgage Fund-Inv
Canada Trust Money Market Fund-Inv
Sceptre Bond Fund
Sceptre Money Market Fund
Trimark Advantage Bond Fund
Trimark Government Income Fund
Trimark Interest Fund
C.I. Canadian Bond Fund
C.I. Global Bond RSP Fund
C.I. Signature High Income Fund
C.I. Signature Dividend Income Fund
C.I. Money Market Fund
C.I. Global Bond Fund
Appendix 5 - Relevant Benchmarks by Fund-Type
Fund-Type
TOTAL
Equity Sample
Code Funds % of Sample Benchmark Return Index Source Corral Canadian Common Foreign Common Balanced 'US Common
Fund-Type Code Funds % of Sample Benchmark Return Index Source Correl* Bond & Income Dividend & Income Canadian Money Market Foreign Bond & Income Mortgage
BO Dl MM BF DM
14 12 11 4 3
31.82% 27.27% 25.00% 9.09% 6.82%
Long-Term GOC Bonds S&P TSX Scotia Capital 30 Day T-Bills Citigroup World Government Bond All Maturity Scotia Capital MBS Overall
CFMRC CFMRC Datastream Datastream Datastream
36.08% 59.91% 25.73% 11.21% 17.94%
TOTAL 44 100.00%
* Correlation of monthly benchmark returns with monthly fund-type returns