The Corporate Governance Premium , Returns, and Mutual Funds

ahemhootΒιοτεχνολογία

5 Δεκ 2012 (πριν από 4 χρόνια και 8 μήνες)

445 εμφανίσεις





The
Corporate Governance
Premium
,
Returns,

and Mutual Funds


Julia Chou and William G. Hardin III


Florida International University



Abstract


Mutual funds with a preference for strong corporate governance have performance similar to mutual funds
with a preference for weak corporate governance.
We find a
direct relation between overall mutual fund
corporate governance preference and the corporate

governance premium. Furthermore, the investment
preferences of mutual funds forecast the change in the corporate governance premium.
We

provide
evidence that the investment activities of institutional investors can affect stock performance
,

and that shifts
by institutional investors in corporate governance preference impact the appearance of the corporate
governance premium.






Keywords:

corporate governance,
premium,
mutual funds




JEL Classifications:

G11, G23, G34














We would like to thank the editor (Bonnie Van Ness) as well as the reviewers for helpful comments.
1


1
.

Introduction



Gompers, I
shi
i

and Metrick (
2003, henceforth GIM
)

conten
d

that firms with strong shareholder rights
perform better

than firms with weak shareholder rights
when

measured by

both
operating
and
stock
performance.
1

Agency cost
s

are

postulated as
the
primary
reason
for

the underperformance
of
firms with
weak
shareholder rights.

E
ven with GIM’s theoretical and intuitive appeals,

however,
empirical
assessment
s
that
cluster

governance by industry

and
extend the time period investigated
show that GIM’s

results are
diminished

and
invoke question
s

as to
the
persistence o
f

the
corporate governance premium
(anomaly)
and
our
ability to generalize

existing
results and implications

(Core, Guay, and Rusticus (2006),
Johnson, Moorman, and Sorescu (2009), and Lewellen and Metrick (2010))
.
2


S
tudies

have
noted

concerns

with

the governance index (G
-
index) provided by
GIM

and as applied in
the literature
.
Core, Guay, and Rusticus

(
2006
) demonstrate that strong shareholder rights do not impact
stock price
s

and
that
the outperformance of firms
with
strong

governance can be
associated with

the
I
nternet boom.
After controlling for industry effects,
Johnson, Moorman
,

and Sorescu (
20
09
)

s
how

that
well
-
governed firms do not outperform poorly

governed firms
.
Lewell
e
n and Metrick (2010) argue that
the
use of
broader industry delineations
actually supports the corporate governance premium

while
Bebchuk,
Cohen, and Wang (2010)
posit
that the
corporate governance
premium disappears due to learning

by
analysts

and investors
.

S
ince there are
now
19

years

for which
the

G
-
index

is
available
from the Investor
Responsibility Research Center (IRRC) database
,

issues related to time
frame,
index construction
,

and
changes in the appearance of
t
he corporate governance premium

can
now
be
addressed

in more depth
.





1

Other
studies
, for example, are concerned w
ith
the
affect of corporate governance on firm decisions such as capital
structure (Jiraporn and Gleason (2007),
cash (
Harford, Mansi, and Maxwell (2005)
)
,

leverage (
Agca and
Mansi

(2008),

leasing (Robicheaux, Fu, and Ligon (2008), acquisitions (Masulis,
Wang, and Xie (2007), and dividends
(Borokhovich, Brunarski, Harman
,

and Kehr (2005) and Jiraporn, Kim, and Kim (2011). Also see Fields and Keyes
(2003) for a general survey of corporate governance

as well as Bebchuk, Cohen, and Ferrell (2009).

2

In this p
aper, we refer
to
strong shareholder rights as good corporate governance.

Also, we use corporate governance
premium as the standard term for the corporate governance premium.


2


The focal points of the present study are the
persistence of the corporate governance premium

and the
factors associated with
its

appearance and disappearance
. Schwert (2003)
indicates that there

are two
primary
reasons
associated with
an anomaly
’s
disappe
ar
ance
.
The first is
that the anomaly
initially was
discovered by sample selection bias
,

and the second is that

investors have
discovered

th
e

anomaly and
implement
ed

advantageous
investment strategies
eliminating

the anomaly
.

As an initial assessment, we
i
nvestigate the persistence of the corporate governance premium
.
Subsequent

analysis looks at the actions
of institutional investors, proxied by mutual funds, to investigate the impact of
changes in
investment
focus
and strategy

on the appearance of the corporate governance premium.


To
investigate
whether firms
with
strong

governance
generate higher returns
, we follow GIM’s
methodology
by forming

Democracy and Dictatorship portfolios and
calculating
monthly value
-
weighted
retur
ns for each portfolio.
The

Democracy portfolio
outperforms the Dictatorship portfolio
only
for the
period

ending in year

1999
. After and including year 2000,
the
Dictatorship
portfolio
has

better
performance than

the
Democracy
portfolio
.
T
hese initial
findings
are confirmed using
Carhart’s (1997)

four
-
factor model.
W
ell
-
governed firms generate
statistically
significant higher abnormal returns
only
before 2000.

When
the sample period is extended to 2008, there is no
statistically
significant difference
i
n

returns between Democracy and Dictatorship portfolios.

The
appearance of
the
corporate governance
premium is
temporal or
time frame dependent.

Also, c
onsistent with
Core, Guay, and Rusticus (2006)
, we

find that the corporate governance
premium

in the period 1990
to

1999 is driven by the year
s

1998 and 1999
,

which are
associated with
a

boom
in

technology and
healthcare

stocks
.
If these two years are excluded

from
the period

assessed
, no corporate governance premium
is evidenced
even for
the earli
er

time period
.


More significantly,
examining whether mutual funds capture any prospective corporate governance
premium or contribute to the disappearance of the corporate governance premium further extends the
analysis
.
Specifically
,
we examine
the
association between mutual fund corporate governance p
reference
and fund performance
segmented
by
fund

investment style. To
analyze
fund investment styles, an
3


additional factor

termed the
Corporate Governance Factor

(
CGF)

is introduced. The CGF is
calculat
ed as
the monthly return of the investment
portfolio

which

buy
s

the Democracy portfolio and
sells

the
Dictatorship portfolio
. It
p
rox
ies
the corporate governance
premiu
m

and is used
along with

the Carhart
fou
r
-
fac
tor model to classify mutual funds into different categories
based on

investment strateg
y
.

All
equity
mutual funds are sorted into groups according to their loadings on th
e factors of SMB, HML and CGF for

the five
-
factor model

(Houge and Loughran

(
2006)
)
.

Higher loadings on the CGF indicate funds

that

invest
more capital in well
-
governed firms.


Evidence shows that
mutual funds do not
systematically
earn statistically significant abnormal returns
by investing in firms with strong shareholder rights. B
efore
2000
,

m
utual funds
with a preference for

strong

corporate governance

generate
d

statistically
significant
higher

annual returns than their
various
counterparts
(
between
1.8%
and
6.0%
)
. However,
after 2000 and through 2008,
mutual funds with the same
prefere
nce
for
good
corporate governance
underperform
ed

their counterparts
.

If all years are pooled,
mutual funds
preferring

weakly

governed firms
actually
generate higher return
s
. The results
are robust
even
after controlling for fund characteristics that impact fund performance.
T
he
initial
excess returns to mutual
funds with strong preferences for corporate governance
do

not persist over the entire
sample period
.

Further results allow us to address the

issue of whether the disappearance of the corporate governance
premium is related to sample selection bias or a change in investor behavior.
The

average loadings on the
CGF variable for mutual funds are statistically significant
,
positive
,

and

higher in the period of 1990 to
1999
when compared to

the period of 2002 to 2008.
Mutual

funds
prefer

good corporate governance in the
earlier

period
and
weak corporate governance in the latter period.

T
here is no shift toward greater
preference
for

well
-
governed firms
.
M
utual funds actually
shift

their
investment
preference
to

weak
corporate governance after
2002
. Hence
,

the disappearance of the corporate governance

premium is not
reflective of mutual funds investing in firms with
strong

corporate governa
nce

in an attempt to arbitrage
the
4


corporate governance anomaly.

This

contrasts with

GIM

s
argument

that agency cost
s

w
ere

underestimated
during
the
1990s and
that

agency problems

caused the

corporate governance premium.

Additional

results

indicate

that mutual funds investing in well
-
governed firms during
the
1990s

were

not
doing so due to
c
orporate governance

concerns
.
M
utual funds with
an initial
preference
for

strong

corporate governance
actually
reallocated

their portfolio
s

to
include

fewer firms with
strong

corporate
governance
. In the period 1990 to 1999, only
about
40% of funds
with a preference for

good corporate
governance
retained

the

same

degree of
preference after
three

years.
Furthermore
, the shift of preference
to
firms with

weak corporate governance
by

mutual funds after 2002 is driven by
funds

with

initial
preference
s

for

good corporate governance in
the
1990s.
This implies that f
und managers do not
believe

the
excess returns for the period
were

solely
due
to good governance

practices
, but
may
instead
be related to
other market forces such as
the boom of some
industries

or
the ability of analysts to project fi
rm
performance (
Core,
Guay, and Rusticus (2006)
, and
Lewelle
n and Metrick
(
2010)
)
.


W
e
also
provide

evidence

that
mutual fund
investment preference ca
n be used to predict the
corporate
governance
premium
.
There is a positive relation between

the
corporate governance
premium

and
prior
period
mutual fund
preference
for
good corporate governance.
T
he

corporate governance

preference

of

equity funds

forecast
s

the future

corporate governance
premium
.
Conversely
, the
prior
year
stock
performance of well
-
governed firms does not forecast
subsequent mutual fund
corporate governance
preference
s
.
Furthermore
, w
hen mutual fund holdings are used to test
for
corporate governance preference,
results
indicate
that industry
preference
s

dominate governance

preference
s
.
Mutual fund preference
for
good corporate governance is
in large measure
subsumed
by
an

industry effect.

The

results are consistent
with the view that
the
trading behavior of institutional investors affects future stock returns documented in
Gompers and Metrick (2001)
, Froot and Teo (2008),

and Yan
g

and
Zhang

(2009).

The findings suggest tha
t
the

corporate governance
premium

primarily
refle
cts institutional investors


demand
for

s
tocks
in a specific
industry
i
nstead of the quality of corporate governance
of a specific firm
.


5



Our paper makes several contributions to existing research.
First, we provide
additional

evidence
show
ing

that corporate governance
premium

is
not economically significant
.
E
xisting
studies

mainly revisit
the relation between the quality of corporate governance and stock performance
by examining
characteristics

at
the firm level

(see,

Core, Guay, and Rusticus (
2006
),

and Johnson, Moorman and Sorescu

(
2009)
)
. We examine this
association
from

the view
of

both firm
s

and investor
s
.

At the firm level, we extend
the sample period to 2008 and provide
a
clearer
picture of

h
ow corporate governance quality
is related to
stock performance

and how
the
corporate governance
premium
varies yearly

and between sub
-
periods
.
O
ur
analys
i
s at the investor level

demonstrate
s

that
sophisticated
practitioners
do not

generate excess returns
by
investing

based on
a preference for
strong

corporate governance
.


Second, we
show
t
hat
changes in the investment preference
s

of

institutional investors may be
one
reason
return
anomalies

disappear
. Anomalies may be artifacts that mask changes in investor preference
while investor preference

may actually decrease or eliminate the appearance of an anomaly.
Previous
studies

ha
ve

shown that institutional investor behavior
affects

the appearance
or disa
ppearance
of
a

performance
anomaly
.

For example,
Gompers an
d Metrick (2001) provide a

possible reason

for the
disappearance of trading anomalies and
demonstrate

that institutional investor

trading behavior
may

be

responsible for the disappearance

of trading anomalies
.

Ng
and Wang
(2004)
show the relation between
institutional trading
and

anomal
ies

related to
the
January
effect and other
seasonal

anomalies
. Phalippou
(2005) demonstrate
s

that
the
value premium decreases
with increase
d

institutional
ownership. Cremers and
Pareek (2010)
attribute
many anomalies

to
sho
rt
-
term institutional investors and show that t
he effect of
institutional
trading

on anomalies is especially strong when short
-
term investors have higher
share
holdings.
Our e
mpirical
evidence support
s

Gompers and Metrick’s view that institutional investors’ trading
preference
s

and investment
s

can
cause
premiums
or
anomalies to
appear,
disappear or
dissipate
.



Third, we show that institutional investors may
give limited
real weight

to
the quality of corporate
governance
, even if they say they do
,

and
that
the disappearance of
the
corporate governance
premium
is
6


not due to investors
shifting investment to

well
-
governed firms
(Starks (2009)
)
. One
important

argument
from GIM is that
invest
ors emphasizing the importance of corporate governance drove the increase in firm
value for well
-
governed firms at the end of 1990s
.

O
ur results

contrast with

this argument and show that
institutional investors
,

proxied by
mutual funds
,

do not
actually
focus on the

quality of corporate
governance.

Although mutual funds d
id

prefer good corporate governance
for

the period
of
1990

to 1
999,
mutual funds change
d

their preference
s

to

firms with
weak

corporate

governance
structures
in the latter
period.



2
.

Data

and Sample Selection



The

data are

mainly
from t
hree

sources
.
Mutual fund
returns and
fund c
haracteristic
s
are

from the
CRSP Survivor Bias
-
Free U
.
S
.

Mutual Fund Database
.
S
tock returns
are
from
the
CRSP
stock file and

t
he
corporate

governance index (
G
-
index
)

is

from

the Investor Responsibility Research Center (IRRC). The
sam
ple period is from 1990 to 2008
.
T
he fund returns
are reported
a
s net returns.


2.1

Corporate
g
overnance
i
ndex
and
c
orporate
g
overnance
p
remium



We use
GIM

s methodology
with

the IRRC
governance index

(G
-
index)

as a proxy for the quality of
corporate governance. The G
-
index takes into account 24 different provisions in
five

categories


tactics for
delaying hostile bidders, proxy voting rights, director/officer protection
s
, oth
er takeover defenses, and state
laws.

T
he G
-
index

is constructed by adding one point
for
every specific provision in place and zero
,

otherwise.
A
lower G
-
index score

indicates
higher corporate governance

quality
.


GIM’s classification
is used
to form stock

portfolios. Firms are separated into Democracy, Fair, and
Dictatorship portfolios
. If a

firm

s G
-
index is equal to or greater than 14
,

it is classified
a
s a

Dictatorship.
7


Firms having

a
G
-
index between 1
4

and
5

are in the Fair category.

F
irms

where the

G
-
index is equal to or
less

than 5

are in the Democracy category
.

If the G
-
index is not available in a y
ear, the latest G
-
index is used
.
The corporate governance
premium
is calculated monthly as the difference
between

value
-
weighted returns
of the
Democra
cy and Dictatorship portfolios.

The portfolios are
re
form
ed

each time

IRRC publishes a new
version of
its
governance data.


2.2

Mutual
f
und
i
nvestment
p
reference



The

mutual fund
sample includes all funds

with

CRSP data
having

positive total net assets and
with
investment
s

of
at least 75% of fund assets
primarily
in
domestic
common stocks.
Mutual funds are excluded
from the sample if
they have
Wiesenberger fund c
odes
of
INT and C&I,

Strategic Insight fund objective
codes
of
ECH, ECN, EGG, EGS, EGT, EGX, EID, EIG, EIS, EIT, EJP, ELT, EPC, EPX, ERP, ESC, FLG,
and GLE
,

or

Lipper Objective Codes
of
EM, EMD, EU, GL, GLI, GS, G
X, IF, IS, JA, LT, PC, SWM, XJ.


Houge and Loughran’s (2006) methodology
is used to classify
fund

investme
nt style.

At

the beginning
of
each
year
, we

estimate the following model for each fund by using the
prior

three
-
year monthly returns
.

The regression is:


𝑥 𝑒
𝑡
=

+

1
×
 
𝑡
+

2
×

𝑡
+

3
×

𝑡
+

4
×
 
𝑡
+

5
×

𝑡
+
𝜀


(1)


where
𝑥
𝑒
𝑡

is the excess return in month
t
, and

 
𝑡
,


𝑡
,

𝑡

and
 
𝑡

are the four factors
in Carhart’
s

(1997)

model.


𝑡

is the corporate governance factor.
To calculate
the
CGF,
stocks
are
classified
into three
types

(Democracy,
Fair
, and

Dictatorship
) based on
their
governance index
(G
-
index)
scores.
Then,
a zero
-
investment

portfolio
is constructed
by
buying

the Democracy portfolio
and selling

the
Dictatorship portfolio.
T
he portfolio performance

for

firms

in the Dem
ocracy portfolio and
firms
in the
8


Dictatorship portfolio

are

evaluated using
value
-
weighted monthly returns. We denote “CGF” as th
e

monthly return
s

for this investment portfolio.

Funds with top half loadings

on HML (SMB) are classified as Value (Small cap) funds,
and funds with
bottom half
loadings
are
classified as
Growth (Large cap) funds.
We also
independently
sort
mutual
funds
into corporate governance g
roups. Mutual funds
with
the
top
third

of coefficient

value
s

on
the
CGF are
segmented

into
the

good corporate governance preference
classification
(
strong

CG preference)
and mutual
funds
with the

bottom
third

of coefficient value
s

are
segmented

into
the
weak

corporate governance
preference

classification
(
weak

CG preference)
.

The remaining funds are classified as neutral.

The
portfolios are rebalanced every year.

T
his classification method
better
accounts for
mutual fund st
yle
than the objective codes provided by
data providers

and addresses known misiden
tification
of

mutual fund
investment
objectives.
Chan, Chen
and Lakonishok (2002) document that

fund managers
usually switch

their strategy to
one that

follows a

specific

index, such as S&P

500
, to maintain career and reputational prospects
.
Using
prior

three
-
year
performance to determine style provides a
timely
measure
of
a mutual fund’s

current investment
preferences. If
a
mutual fund leans to
firms with good corporate governance, the
CGF
coefficient w
ill be
positive
.
This

method has

been used
to gauge
mutual fund
corporate governance preferences

(
Kim, Shukla,
and Tomas

(
2000
),

Davis
(
2001
),

Chan, Chen, and Lakonishok

(
2002
),

and Houge and Loughran

(
2006
)
).



3
.

Corporate Governance
Premium



T
he corporate governance
premium
’s

actual

exist
ence

remains

a controversial issue in financ
ial
research
. Cremers and Nair (2005) and Cremers, Nair and John (2009) show that strong shareholder rights
complement internal monitoring and demonstrate that the corporate governance
premium

can be explained
by the
probability

of takeover.
Core, Guay, and Rusticus (2006)
and
Johnson, Moorman
,

and Sorescu (2009)

9


provide evidence
showing

that the corporate governance
premium

does not
persist

and that
its appearance
likel
y

is
due to sample selection bias.

Conversely,
Lewellen and Metrick (2010)

argue that with better
delineation of industries, the corporate governance premium persists.

Given uncertainty as to the existence and persistence of the corporate governance premium
,
additional
assessment

of the
corporate

gover
nance

quality

to
stock performance

relation
is provided
.
A

longer,
more
complete

sample period from 1990 to 2008
is used
. If the corporate governance
premium

exists, it should be
observed
over the entire
period
and not just the period

1990

to
1999
as in
GIM.

W
hether mutual funds

actually
can capture this
premium

is subsequently evaluated
.
T
wo purposes

are served
: to t
est whether
firms
with good corporate governance
outperform firms

with poor corporate governance

and
to
evaluat
e

the
corporate governance
premium
’s
economic implications for investors.
Our

tests of mutual fund
performance
also
provide

evidence show
ing

that the
appearance

of
the
corporate governance
premium

may
be caused by
reasons

other than the quality of corporate governance itself
.

Instit
utional investors, proxied
by mutual funds, are shown to be unable to benefit from using an investment strategy based on corporate
governance as a standalone factor.


3.1

The
p
ersistence
of
the c
orporate
g
overnance
p
remium



It is
i
mportan
t

to re
-
examine GIM’s findings
over a
longer period
sinc
e

pr
ior

studies
related to
shareholder rights
cover only
a
short
time
period (1990

to
200
2

or
1990

to
2003
).

A

number of

abnormal
financial activities
occurred
during
the
1990s
,

especially the high values placed on m
any

I
nternet
-

and

technolog
y
-
centered firms in the

computer, communication
s

and biotechnology

fields
. These firms had low
fixed assets compared with firms in other manufacturing industries and
often

were
startups with

few
takeover defenses

as might be expected of firms with venture finance backing interested in preserving exit
strategies
.
After the
technology
lead
stock
crash
around 2000, investors began to evaluate these firms
with
10


better metrics
and the returns
were
adjusted accordingly
.
The implication is that

the higher abnormal returns
for firms with strong shareholder rights
may be

highly
related

to technology

fo
c
used

industries during this
specific period.


Table 1 shows the
yearly
corporate governance
premium

from 1990 to 2008.

From Table

1
, there are
six

positive return
years and
four

negative return
years during the period 1990

to
1999
. The highest positive
returns are
in 1998 and 1999.
Although firms

with good corporate governance

generally have better
perf
ormance
,
these
firms
’ performance is indistinguishable from
firms

with poor corporate governance

from 1990 to 1997
. When the later period (2000

to
2008) is examined, firms

with poor corporate
governance

actually
outperform firms

with good corporate governa
nce
in
seven

out

of

nine

years. In 2008,
firms with
good
corporate governance
outperform again
, although f
irms with strong shareholder right
s
appear to
suffer less than other firms
during

the financial crisis.

Figure 1

show
s

that
well
-
governed
firms

have
better performance in the pr
ior

period than the later period

when

year 2000 is used to
separate

the
sample period
.
The
corporate

governance premiums are
extremely

high in 1998 and
1999, which

suggests
that the
corporate governance
premium

may
be associated

with

special events
,

changes in
investor

behavior
,

or
other unobserved
factors.


We use a

four
-
factor model to test Democracy and Dictatorship portfolios
over

various periods
with

results in Table 2.
Our
evidence confirms GIM’s findings
for the

period
1990

to
1999
where
well
-
governed
firms have
statistically
significant

and
positive abnormal returns
,

while

poorly

governed firms earn
negative abnormal returns.

T
he performance trend
, however,

reverses after 2000
when

Democracy
portfolio
s

actually earn neg
ative abnormal returns

and
Dictatorship
portfolio
s

generate

positive

abnormal
returns
. When the period is extended from 1990 to 2008, the firms in
the
Democracy portfolio do not earn
statistically
significant abnormal returns. Furthermore, if the year
s

199
7

to
1999 are excluded,
well
-
governed firms
no longer

earn
statistically
significant abnormal returns

for the
period

1990

to
1999
.

11



The
se
findings
imply
that corporate governance
quality
may
ha
ve

little

direct

effect on stock
performance
. Since
the corporate governance
premium

only appears between 199
7

and 1999
, the results
indicate that

investors
do

not
direct
ly

benefit simply
by
investing in firms

with good corporate governance
.

While t
he corporate governance
premium

may be related to the boom
of technology industries
at

the end of
1990s
, it is, in any case, temporal and does not persist over the entire period studied
.
3

The

results support
Johnson, Moorman and Sorescu’
s

(2009)

conclusion that
the
corporate governance premium
may be

related
to
an

industry effect
during

a specific time frame.


3.2

Mutual
f
und
i
nvestment
p
reference
and
p
erformance



To
exam
ine

the

corporate governance
premium
’s
practical relevance
,

it is essential to show
that

investors can make an economic profit by trading well
-
governed

firms (Jensen, 1978; and Sch
wert, 2003).
If
the
profit is
insufficient

to cover transaction cost
s

and all related expense
s
,
one

can
argue

that
corporate
governance
quality
does not have
a meaningful
economic impact
.
Concurrent a
nalysis of mutual f
und
performance
and
preference for
corporate

governance
provides
a
robustness check

as mutual funds
represent a class of investors with scale and knowledge
. If the quality of corporate governance is important
and impact
s

firm performance,
one would

expect
to see mutual funds with
a
preference
for
good

(strong)

corporate governance outperform mutual funds with
a
preference
for
poor

(weak)

corporate governance.


At the beginning of each year,
prior
three
-
year monthly returns
are used
with
a
five
-
factor model to
independently sort mutual funds into groups
by
fund style
based on
the
factor loadings
. Fund investment
style
s

include

a

fund
’s

preference for firm size

(large cap or small cap)
, value or growth stocks and the
quality of corporate gove
rnance. As
noted
, we add
an additional

factor (CGF) to Carhart’s (1997)



3

We provide a summary
table of

the top 30 stocks with the best performance in t
he Democracy portfolio from 1998 to
1999 in the Appendix. From the table, 26 of 30 firms are related to technology or healthcare sectors.

12


four
-
factor model to measure the
corporate governance (
CG
)

preference for each fund.

T
he equal
-
weighted
yearly return for the following year within each
fund
portfolio

is then calculat
ed
.


Table 3 reports the fund performance for each portfolio formed
by

investment style. Considering that
the discovery of
the
corporate governance
premium

may change mutual fund investment decisions
(and
reduce or eliminate the corporate governance premiu
m)
and
that
the economy
adjusted to

the
I
nternet
bubble during 2000 and 2001, we separate the
entire
period into three sub
-
periods.
The f
irst period is from
1990 to 1999,
the
second period is from 2000 to 2001 and
the
third period is from 2002 to 2008
. Data from
the three periods
are shown in the Panel
s

A, B
,

and C, respectively. The first period matches the sample
period used in GIM’s paper
to allow
compar
ison
. The
second period controls for the
I
nternet bubble
’s
potentially
large impact on the corpor
ate governance
premium
. The third period is the sample period after
the date of GIM’s publication to reflect
the possibility
that the
premium

disappear
s

due to
substantial
dissemination of
corporate governance
research.


R
esults
segmenting the data
by time

period
provide conflict
ing

evidence
on

the existence of
the
corporate governance premium and
investors
’ ability

to
capture
the

corporate governance
premium
. In
Panel A, mutual funds
with a

prefer
ence for

good corporate governance outperform mutual funds with
a
preference
for
poor
corporate governance. The return differential between funds with preference
for

good
versus

poor
corporate governance varies from
slightly less than
2% to 6%. The evidence is con
sistent across
fund
s

with different investment styles. However, in Panel
s

B and C, the results are reverse
d
. In the year
s

2000 and 2001, mutual funds with good governance preference underperform almost all other
fund groups
.
The differences are all
statist
ically
significant at conventional level
s
.
T
he underperformance for funds with
a
preference
for

good
corporate

governance is not
surprising

given the link with

technology
-
related stocks

which performed very poorly during the technology firm focused market correction of 2000 and 2001
.
In
Panel C,
which covers the 2002

to
2008 period,
mutual funds with
a
preference

for

good corporate
13


governance underperform their counter
parts

by

0.6% to 9.
2%.
The
results
remain

even
when

we exclude
year 2008 to control for the effect of the

latest

financial crisis.


Houge and Loughran (2006) show that the median
mutual
fund expense ratio increases from around
0.65% in 1965 to 1.20% in 2001.

Different mutual

fund styles are associated with higher and lower fees
indicating that expenses have a significant impact on fund performance
.

For example,
Houge and Loughran
document

that
growth funds
typically

have
higher expense ratio
s

than value funds. If both growth
and value
funds have the same performance, growth funds will underperform value funds after expenses. If the
mutual fund performance in our tests is driven by expense ratio

differences
, the results could be biased.
A

cost difference
explanation
, however,
c
annot explain the inconsistent findings in Panel
s

A and C with
different sub
-
periods. Mutual funds
preferring
good corporate governance actually perform very well in the
first sub
-
period
,

but
perform poorly
in the second and third sub
-
periods. If
expenses
drive the performance,
mutual funds prefer
r
ing

good
corporate governance should
either
outperform or underperform all periods.


As a robustness test t
o confirm
the
initial
results and consider other determinants of mutual fund
performance

including expenses
, returns
are regressed
against past one
-
year
f
und
performance, fund size,
dumm
y variables
for investment style
, expense ratio
,

and dumm
y variable
s for
CG

preference. The
regressions are corrected for the clustered standard errors for fu
nd IDs and years. The regr
ession results are
reported in
Table 4. The
F
-
statistics of the differences for the coefficients of the two dumm
y variables

for
good and
poor
governance preferences
also
are
reported
.


The
results are consistent with previous
findings and show that expense ratios are negatively
correlated to fund returns. Higher expense ratios drag down fund performance. Mutual funds with
preference

for good corporate governance

still
outperform in the first period
,

and still
underperform in th
e
second and third periods. When all years are pooled, there is no
statistically significant
difference in the
performance
of

mutual funds
based on

corporate governance. The results further demonstrate that better
past
one
-
year
performance forecasts higher

future fund return and
that
fund size has a negative impact on
14


fund performance. Value funds have
poor
performance in the first period
,

but good performance in the
second and third periods.
For

all sample years, value funds do not earn
statistically
signi
ficant higher
returns

while

s
mall
cap

funds
generally
perform better.

Fo
r the entire sample period,
using mutual fund data,
preference
for
good corporate governance does
not have a consistent and
statistically
significant impact on fund performance

and t
he corporate governance
premium

disappears after 1999.
I
nstitutional
investors
, proxied by mutual funds,

cannot
earn

an economic
profit by
investing
in
and
trading well
-
governed firms.
This implies that
documentation

of
the
corporate
governance
premium
may be due to
a
data selection bias

(such as time period)

or other undocumented
factors
. The next section
directly
explores
whether

changes in
mutual fund

trading

and investment

behavior
s

are

related to

the corporate governance
premium
.



4
.

Investment
P
reference

and
the
Corporate Governance
Premium



In the previous section,
the
corporate governance
premium

is shown to disappear

after 2000. It

remains

interesting to
assess

why
the
corporate governance
premium

disappears
shortly after it is
documented.
The

disappearance
implies
the

possib
ility

that investors
incorporated G
IM

s findings
into
decision
-
making
and put more weight

on well
-
governed firms in

their investment portfolios

after 2000
.
When an anomaly is documented, practitioners
may
immediately imp
lement
new

trading
or investment
strategies
,

causing

the
anomaly
to disappear
. While this is an appealing argument
,
Schwert

(2003)
also
documents that sample selection bias
is

another

possible explanation.
This section explores possible
explanations

for

the disappearance of
the
corporate governance
premium
.



4.1

Efficient
m
arket
versus
i
nvestors’
i
nvestment
p
reference


15



In this subsection, mutual funds
are
again
used
to investigate

institutional
investor

pre
ference
s

regarding the quality of corporate governance.
A

major implication from GIM is that investors
became

concerned

with

corporate governance

in
the
1990s, which

cause
d

poorly

governed firms to be
less
valued

and
well
-
governed firms
to be

associated with incre
ased values
.

As

investors
b
egi
n to
integrate

corporate
governance

into the investment decision
,
however,
the corporate governance
premium

should dissipate and
then

disappear.
Under such a scenario
,
the corporate governance
premium

may exist
,

but
an

efficient market
causes it to disappear

as it is documented and becomes part of the investment decision
-
making process.

In
such a
case,
one

should observe that
mutual funds
prefer weak corporate governance before 2000 compared
with
the later period.


To i
nvestigate the relation between mutual fund investment behavior and the corporate governance
premium, the sample period is divided into two
sub
-
periods. The first period covers 1993

to
1999 and the
second period
covers
2002

to
2008.
W
e average the coefficients
of

each factor yearly across all mutual
funds to see how the
preference
for

corporate governance
changes

before and after GIM’s findings.
If

the

efficient market work
s
, the average coefficient of CGF in the later period should be

positive and larger than
it is in the prior period.
This
pattern would
indicate

that more mutual funds like well
-
governed firms after
the
premium

is
documented
.

T
he results

are in Table 5
.

In Panel A

of Table
5
, mutual funds have positive coefficients on

the
corporate governance
factor
(
CGF
)

in
five

out of
seven

years. However, the average coefficients of CGF are negative and significant for
all years after 2002. The differential coefficient of
the
CGF between the two periods is statistically
significant
at the
1
% level.

M
utual funds prefer good corporate governance before 2000, but poor corporate
governance after 2002.
Because

mut
ual funds actually prefer
poor

corporate governance in the later period,
i
t is
highly

unlikely that the revers
al

of the corporate governance premium is due to more investors
investing in well
-
governed firms.

Even

though GIM

s findings are
important
,

they

do not
appear to
16


persuade

investors
that
the
corporate governance premium exists and
that
the quality of
c
orporat
e
governance has a significant
independent
impact on

stock returns
.

It

is possible
, however,
that investors were

aware of the

importance of

corporate governance in
the
199
0
s
and i
nvest
ed

more capital in well
-
governed firms. T
o
investigate

this

possibility
,
we directly test how
the
CG
preference
for mutual funds
shifts
.
I
f fund

managers

think that

corporate governance
is important
,
mutual funds prefer
ring

good corporate governance should continue to have
a
preference
for
well
-
governed
firms
.

Hence,

mutual
funds prefer
ring

good corporate governance
should
retain a good corporate
governance portfolio
,
especially

during
the
1990s
since

the corporate governance
premium

generally is
positive
.
W
e first restrict our sample period

to
1993

to
1999

to test the

short
-
term

changes of CG preference
for mutual funds
.
Then, we expand the period

to
1993

to
2008 to test the long
-
term CG preference changes.


Table
6

presents
results related to
the shift of
preference for
corporate governance
over
three
-
year

period
s for mutual

funds
from

1993

to
1999, and the change of preference
over
nine
-
year

period
s from 1993
to 2008
.
From
Panel A,
while

more than
one
-
third of
mutual funds
retain

their preferences

over

three

years
,
more than 50% of funds move away from their original
good CG or bad CG
preference
groups.
For mutual
funds with neutral preference, the move
ment

of CG preference

is
almost equally
distributed among
groups
of good, neutral and
poor

CG

preference.
This indicates that most mutual funds do not
have a significant

preference
related to
corporate
governance

quality
.
In
P
anel B, t
he results show that only 37% of mutual
funds
preferring

good corporate governance in
the
1993

to 1
999
period
still
heavily weight

invest
ments

in
well
-
governed firms in
period
2002

to
2008. This percentage is even lower than the one in
Panel A
.
T
he low
ret
ention

ratio

in the group of

Good CG


and no significant movement of preference toward good
corporate
governance
in the

Neutral


and

Bad CG

groups
demonstrate
s

that although mutual

funds
as
a

whole

invest

more

in well
-
governed firms in
the
1990s
(as

document
ed

in Table 5
),
this tendency is unlikely
related to
a

preference
for
good corporate governance

since preferences generally decline over time
.

17


We also average the coefficient
s

of
the
CGF across mutual
funds

prefe
r
r
ing

good
or

poor

corporate
governance
for the

periods. Results are reported in Table 7.
T
he change of
CG

preference
toward

poor

corporate governance after 2002 for all mutual funds is driven by mutual funds with good
CG

preference.
The average coefficient of CGF is 0.28
61

in Panel A and 0.13
15

in Panel B for mutual
funds

prefe
r
r
ing

good corporate governance
. The difference is 0.1
5
46
and

is
statistically
significant at the 1% level.
Concurrently
, the
CG

preference
actua
lly increases in the second period for

mutual funds prefer
ring poor

corporate governance. The
differential

coefficient of CGF between Panel A and Panel B for mutual funds
with preference toward weak corporate governance is
-
0.03
89, but
is

not
statistically
significant.

The
pattern in preferences remains.
Mutual funds do not shift investments to firms with strong governance as
might be expected from GIM.


The results from Table
s

5, 6
,

and 7
indicate
that the efficient market
story
cannot explain

why the
corporate governance
premium

disappears.
Furthermore, investors do not
find

corporate governance
quality
to be an independent factor
impacting

stock holdings.
The
se

finding
s imply that the
appearance of
the
corporate governance
premium

may be

due
to

sample selection bias

or other undocumented factors

and
that trading based only on corporate governance metrics does not generate economic returns
.
W
e also find
that the corporate governance
premium

has
a positive relation with mutual funds


prior

three
-
year

CG

preference
s. In the period

1990

to
1999, the corporate governance
premium

generally

is

positive and the
preference of mutual funds tends to be positive. When the corporate governance
premium

becomes
negative
after 2000
, the
governance
preference o
f mutual funds is
statistically
significant
and
negative

too
.
This leads to a question
of
whether institutional investors


preference
s

predict stock price
s

and
whether a

change
in
investor
CG

preference causes the corporate governance
premium

to disappear. In the next
subsection, we explore
the

possible connection between mutual funds’ preference
for
good corporate
governance and stock performance.


18


4.2

Investment
p
reference
and
s
tock
p
erformance



Gompers and Metrick (2001) use supply and de
mand theory to argue that
the
demand for stocks


especially from institutional investors


drives stock performance. Froot and Teo (2008) argue that
investors shift investments among industries or
firms
based on

desired
firm
specific
characteristics.
These
change
s

can be manifested in
allocations related to
mutual fund investment styles
. The argument is
that
portfolio
reallocat
ions
impact stock prices and expected returns.

Biktimirov (2004) provides similar results
when investigating demand for stocks
dropped from exchange traded fund indices.
R
esearch by
Yang and
Zhang (2009)
focused on institutional ownership shows that

stock performance
may

be
driven by
institutional investors
having
greater information which implies
that the investment behavior of
institutional investors predicts, or signals, future stock performance (Wermers (1999)). Hence,
it is possible
that
a change in corporate governance preference by institutional investors could
influence the appearance
of
the corporate governance premium do
cumented in GIM.

In the previous subsection, the shift
in overall

mutual fund
preference for corporate governance is
associated

with movement
in

the corporate governance premium. In Figure 2, we plot the average monthly
corporate governance premium
and

th
e overall CG preference over the prior
three

years for all mutual funds.
The

prior CG preference for mutual funds moves in the same direction as the corporate governance
premium. When the prior preference reaches its highest point in 1999, the well
-
governed firms earn the
ir

highest abnormal returns. These figures provide initial l
inks related to a positive correlation between prior
period c
orporate governance
preference
and the corporate governance

premium
.


To examine this relation, we categorize the 16
data
years into
both
two and three groups based on each
fund’s prior corporate

governance

preference
. At the beginning of each year, the five
-
factor model is used to
estimate loadings of the CGF for each mutual fund over the prior three years. The mean
s

of the coefficients
are
used as the prior CG preference
s

for the year. The top a
nd the bottom groups are the years with the
19


highest and lowest prior preference
s

for corporate governance. Firms with a G
-
index


5 and with a
G
-
index


14 are classified
into

Democracy and Dictatorship portfolios. The monthly value
-
weighted
returns for ea
ch portfolio are used with Carhart’s (1997) four
-
factor model to estimate the abnormal returns.

Table 8 reports the abnormal returns for each portfolio categorized by the corporate governance
quality for the firms under different magnitudes of CG preferenc
e. The years are sorted into two groups in
Panel A and tercile groups in Panel B
based on
prior overall preference for corporate governance.
When

the
prior
corporate governance
preference is hig
h
, well
-
governed firms generate higher abnormal returns than
p
oorly

governed firms. When the prior preference for good corporate governance
by
mutual funds is low,
poorly

governed firms earn higher abnormal returns than well
-
governed firms, but the difference is not
statistically significant. The evidence is consistent in both Panel A and B. The findings are stronger when
the prior preferences are classified based on tercil
es.

The evidence is clearer when the same
firm
groups

are compared across years with high and low prior
overall
CG

p
reference. Firms in the Democracy portfolio always outperform when the prior
overall

corporate governance
preference

is high and underperfor
m when the prior
overall

preference

is low.
The
return
differential is
statistically
significant for well
-
governed firms between
the
two periods and is

consistent in both Panels. Firms in the Dictatorship portfolio have opposite return behaviors. They gene
rate
significantly higher abnormal returns in the years with low prior preference

for good governance

than the
years with high prior preference. The difference
is especially

significant at conventional level
s

when the
years are grouped in terciles
.

The pos
sibility that funds change their preference for good corporate governance due to the better
performance of well
-
governed firms
also
must be assessed.

A
n additional test of whether better stock
performance related to well
-
governed firms impacts subsequent
mutual fund preference for corporate
go
vernance is provided. Y
early observations are separated into groups (either two or three) based on the
stock performance of well
-
governed firms (firms in the democracy portfolio). The subsequent three
-
year
20


mutual fund

corporate governance preferences are then compared.

Results are in Table 9.
In the two
-
group
case, there is no statistical difference in subsequent corporate governance preference by mutual funds.
Positive firm performance does not precede corporate gover
nance preference. When the observations are
separated into terciles,
subsequent
mutual fund corporate governance preferences for both over and
underperforming firms are negative and statistically significant (
-
0.056 and
-
0.055, respectively). The
between g
roup differential corporate governance pr
eference
s

are
not statistically significant
.
Overall, the
stock performance associated with good corporate governance (the democracy portfolio of firms) is not
positively associated with the corporate governance pre
ference of mutual funds. Firm performance in the
previous year is unlikely to be the primary reason mutual funds invest in well
-
governed firms.

R
esults indicate

that
the
disappearance of the
corporate governance
premium

is

not
due to
practitioner
s

implement
ing

investment

strateg
ies where investors
buy firms with strong
shareholder

rights
and

sell firms with weak
shareholder

rights. Instead,
the disappearance of the
premium

is
associated

with

a

change
in corporate governance

preference
by mutual fund
s
from
well
-
governed

firms to
poorly

governed
firms
.

Our

evidence

support
s

Gompers and Metrick
’s
(2001)

view

that the demand for
certain types of
stocks can have a meaningful

i
mpact on stock price. When demand for
firms with poor corporate
governance

increases, poorly

governed firms have higher
stock
performance than well
-
governed firms.
O
ur
results indicate that investors do not
inherently
value corporate governance

quality

and
conflict with

GIM

s
argument
that quality corporate governance
is a uniqu
e
factor, which

should

attract

investors.


4.3 Institutional
i
nvestment
,
i
ndustry
, and
m
utual
f
und
c
orporate
g
overnance
p
reference



Froot and Teo (2008) explore the relation between institutional ownership and future stock return
s

and
find that
institutional
investors invest strategically
based on investment
style

or purpose by

investing, for
example, in
technology
or

non
-
technology stocks, value or growth stocks, and small or large
cap
stocks.
21


Also, institutional investor portfolio reallocations are shown to
significant
l
y

impact stock prices and
expected returns.

In the present case, i
f mutual funds prefer
technology and healthcare stocks

in
the
1990s
and invest more in
these and
related industr
ies, it is possible that firms in these industries have a better
return

because of this allocation
.

G
iven the
strong
corporate governance structures of these firms
, it
also
is
possible that
mutual fun
d
holding preference toward good corporate governance
is

subsumed
by an
industry
effect.


To explore

whether
fund allocation

change
s

impact
the

corporate governance preference
s

of
mutual
funds
, we calculate
the
mutual fund ownership
of
each firm in the Democracy and Dictatorship portfolios at
each quarter
-
end
from 1990

to
2007.

The ownership data are from Thomson
Reuters

Mutual Fund

Holdings

as CRSP mutual fund ownership data are not available before 2003.
V
ariable
CG Pref

is

set to 1 if the firm
is in the Democracy portfolio, and 0 if otherwise.
Control variab
les include firm size, book
-
to
-
market ratio,
past three
-
month return, and past nine
-
month return before the end of
the
last quarter, turnover ratio
,

and
firm age. Fama
-
MacBeth’s
(1973)
approach is implemented and coefficients
are averaged across the whole
sample
period

with r
egressions
estimated quarterly

as reported in
Table
10
.


In Panel A of Table
10
,
a general
mutual fund
preference for w
eak corporate governance

is shown
. The

coefficient of

CG Pref

is negative and significant
when

all quarters are pooled
. When the period is separated
into two sub
-
periods, mutual funds demonstrate
statistically
significant different preference
s

for
corporate
governance

quality
. In the period 1990

to
1998, the

CG Pref

estimate

is negative
,

but not
st
atistically
significant. However,
in the later period
,

mutual funds invest more in firms with weak

corporate governance.
The coefficient of
CG Pref

is
statistically significant, negative, and
almost three times
greater

in the later
period th
a
n in the previous period.
The results are consistent with our previous findings.


In Panel B, industry dummies are added to control for
firm
-
level
industry effect
s

on holdings.
Firms

are group
ed

in industries
using two
-
digit
and
three
-
digit SIC code
s
. To m
ake the table more concise, only
22


two
-
digit SIC code
results are reported
.
4

The results show that
inclusion of an
industr
y classification has

an
important effect on the
empirical assessment of mutual fund
corporate governance preference. The
CG Pref

coefficients
are negative and
statistically
significant
for

the
whole period
,

and
in
both of
the two sub
-
periods.
In addition, the preference related to good corporate governance is no
t

statis
tically
significant in any quarter

(0/25)
.

T
he coefficients on t
he
CG Pref

variable

in Panel B

while retaining statistical significance are
substantial
ly

greater

and again negative
when compared to results without industry controls
.
5

O
ur results
support the industry explanation o
f

the relation between good corporate governance and better stock
performance.
Furthermore
, the better performance
associated with
good corporate governance firms may
be
related
to
shifting investment allocations between
industries
by

institutional investor
s
,

as Froot and Toe
posit
.

Finally, the results hold even when analysts’ growth forecasts are included in the models (not
presented).
In this latter case, m
utual fund ownership is positively related to forecasted growth.


5
.

Conclusion



The
appearance

and disappearance

of the corporate governance
premium

documented by Gompers,
I
sh
i
i and Metrick (2003) and
the ability of

investors
to
capture a corporate governance return

are

investigated
.
While i
t is widely believed that good corporate governance provides management discipline
and reduces agency costs, it is
not
clear
that
investors can
actually

benefit
from good corporate governance
(
for example,
by
earning

higher stock returns)
.
If investors can
not make a
bnormal

profit
s

by investing in
and trading well
-
governed f
irms,
it can be argued that
the

corporate governance
premium
does not
provide

economic
benefits to
investors
.




4

The results are qualitative
ly
similar using two
-
digit and three
-
digit SIC code
s

to group industries.

5

In an unreported
t
able, we also add analysts’ forecast
s

for firm growth to control for the effect of analysts on the
holding intention. Although the coefficients on these variables are positive and significant, mutual funds still show
significant preferen
ce toward good corporate governance in some quarters.
Again, when the industry dummies are
added in addition to the analysts’ growth forecasts, the preference toward good corporate governance for mutual funds
disappears. Industry has a stronger effect than

analysts on mutual fund corporate governance preference.

23


A

related research question
,

given
ambiguity over the
persistence
of the
corporate governance
premium
,
concerns the prospective causes for
the premium’s

appearance and
disappearance.
Subsequent to
the publication of
Gompers, Ish
i
i and Metrick
’s

findings, the corporate governance
premium

is no longer
consistently
evident

in empi
rical studies
.
According to
Sch
wert (2003)
,
s
ample selection bias and market

efficiency

are two possible reasons
proffered for
the disappearance of

anomalies. Gompers and Metrick
(2001)
argue
that institutional investors’ trading behavior causes anomalies
to disappear

while
Cremers and
Pareek (2010) posit that anomalies are related to the trading actions of short
-
term institutional investors.
While

it is difficult to disentangle
these
complementary

theories
, w
e provide empirical evidence
supporting

Gompers
and Metrick’s view that
changes in
the

investment
preference
s

of institutional investors
affect

stock performance and
can

cause anomal
ies to

disappear.

Our evidence shows
that well
-
governed firms do outperform their peers
in the period
1990

to
1999.
However,
if
the
two years 1998 and 1999 are excluded from the sample period,
the outperformance
disappears
. In addition,
well
-
governed firms
do not
generate
greater returns

than poorly

governed firms
after 2000.

More importantly, w
hen all years are p
ooled, firms with good corporate governance do not
evidence a return premium.
In short, the corporate governance premium

does not persist over the extended
period of
assessment
, which

implies that other factors
are
of greater importance

in determining

stoc
k
performance
.

T
he relation between m
utual fund performance

and corporate governance

also
is investigated

to
determine if institutional investment in the form of mutual funds impacts the corporate governance
premium
. M
utual funds with
a preference for
good corporate governance earn higher returns before 2000
,

but
lower
returns after 2
000. When all years are pooled,
m
utual funds with
a
preference toward good
corporate governance

do not
have better performance.
The

results are robust even after
controllin
g

for
expense ratio
,

past
mutual fund
performance
,

and other
determinants

of

fund performance
.
The
corporate
24


governance
premium
does not
provide

an
economic
impact
for
institutional
investors

(proxied by mutual
funds)
.


When we examine the shift in
corporate governance preferences of mutual fund
s

across years,
w
e
find that mutual funds had a preference for good (strong) corporate governance in the period 1990

to
1999,
but shifted their preference to weak corporate governance after 2000.
This move
ment

is driven by
mutual

funds with
an
initial
preference for
good corporate governance reducing th
eir

holdings o
f

well
-
governed
firms
.
When the relation between the
corporate

governance premium and fund

preference
for

corporate
governance is examined, w
e

find

that t
he corporate governance
premium
can be predicted by mutual funds’
prior three
-
year
preference
s
. If mutual funds have a higher positive average loading on the
corporate
governance factor (CGF) for the
previous
three
-
year period
, the corporate governa
nce
premium

will be
positive in the following year; otherwise, the corporate governance
premium

will be negative in the
following year.



The importance of shifts

in
institutional investor allocations is documented as well as the interaction
between corpor
ate governance quality and industry classification.
Over the period studied,
m
utual funds
have a general preference for firm
s with weak corporate governance as measured by percentage of firm
ownership. In short, good governance quality
at the firm level

is negatively related to the holdings of mutual
funds.

Furthermore, t
he effect of the preference toward well
-
governed firms is substantially
impacted

when

industry controls are m
o
deled.

In essence, industry level corporate governance
preference subsumes m
uch
of
the relevance of firm specific governance.


The present

work
highlights the relation between
corporate governance
and
stock performance

including

whether
institutional
investors

heavily focus on
corporate governance
when making investment
decisions
.
The corporate governance
premium

does not persist and
the quality of corporate governance
alone
is
not
likely
an
independent
factor
used by

investors
to
make investment decisions.

We

find
little

evidence show
ing

that investors
are specifically i
nfluenced

by

the quality of corporate governance before or
25


after
documentation of the
corporate governance
premium
.
O
ur findings suggest that the disappearance of
the corporate governance
premium

reflects sample selection bias and change
s

in

investment

preference
from institutional investors.

The efficient market
argument based on practitioners adjusting investment
preference that eliminates the corporate finance premium is not support
ed

since
practitioners actually
shift
to

weakly

governed firms after
2000.

26


Appendix A

Top 30 Growth Firm
s

in
the
Democracy Portfolio

during 1998 to 1999

This table shows the top 30 firms
with

the highest performance during
the period
1998

to
1999

in the
Democracy portfolio (G
-
index


5)
.
The return is calculated as the
two
-
year
cumulative

return and is
expressed in decimal
s
.
Yahoo Finance defines the sector and industry of each firm
.


Ticker

Company Name

Return

Sector

Industry

AMZN

A
mazon Com Inc.

12.13

Services

Catalog & Mail Order Houses

PMCS

PMC S
ierra

I
nc

9.62

Technology

Semiconductor

VLNC

V
alence Technology Inc

5.70

Industrial Gds

Industrial Electrical Equipment

MEDI

M
edimmune Inc

5.37

Healthcare

Pharmaceutical

TSNG

T
seng Labs Inc

4.37

Technology

Computer Hardware

CSCO

CISCO S
ystems Inc

4.21

Technology

Networking & Communication Devices

INTU

I
ntuit

Inc

3.84

Technology

Application Software

NSCP

N
etscape Communications

3.82

Technology

Application Software

XIRC

X
ircom Inc

3.67

Technology

Networking & Communication Devices

RATL

R
ational Software Corp

3.60

Technology

Application Software

IVX

I
vax Corp

3.43

Healthcare

Drugs

NVLS

N
ovellus Systems Inc

3.09

Technology

Semiconductor

VTSS

V
itesse Semiconductor

3.01

Technology

Semiconductor

ATML

ATMEL C
orp

2.84

Technology

Semiconductors

CY

C
ypress Semiconductor

2.39

Technology

Semiconductors

JBIL

J
abil circuit Inc

2.36

Technology

Printed Circuit Boards

WAT

W
aters Corp

2.20

Technology

Scientific & Technical Instruments

SANM

S
anmina Holdings Inc

2.16

Technology

Printed Circuit Boards

AVX

A
VX C
orp

1.99

Technology

Diversified Electronics

PRGS

P
rogress Software Inc

1.93

Technology

Application Software

MXIM

M
axim Integrated products

1.90

Technology

Semiconductors

SIII

S3 I
nc

1.80

Technology

Computer Hardware

WMT

W
al
-
Mart

S
tores

1.78

Services

Discount, Variety Stores

SWC

S
till Water Mining Co

1.71

Materials

Industrial Metals & Minerals

HNCS

H
NC S
oftware

1.70

Technology

Application Software

TLB

T
albots

I
nc

1.62

Services

Apparel Stores

PRIA

PR
I A
utomation

1.55

Technology

Semiconductors

JMED

J
ones Medical Industries

1.51

Healthcare

Pharmaceutical

SCLN

S
ciclone Pharmaceuticals

1.48

Healthcare

Drug

CCU

C
lear Channel
Comm.

1.24

Entertainment

Advertising

27


References


Agca, S. and S.A. Mansi, 2008.
Managerial ownership, takeover defenses, and debt financing,
The Journal
of Financial Research

31, 85
-
112.


Bebchuk, L., A. Cohen, and A.

Ferrell, 200
9
.
What matters in corporate governance?
Review of Financial
Studies

22, 783
-
827.


Bebchuk, L.
,

A. Cohen, and C.C.Y. Wang, 2010.
Learning and the
d
isappearing
a
ssociation between
g
overnance and
r
eturns
,

Discussion paper

667, Harvard Law School.


Biktimirov, E.N., 2004. The effect of demand on stock prices: evidence from index f
und
rebalancing
,
The
F
inancial Review

39,
455
-
472
.


Borokhovich, K.A., K.R. Brunarski, Y. Harman, and J.B. Kehr, 2005. Dividends,
c
orporate
m
onitors and
agency c
osts
,

Financial

Review

40,
37

65.


Carhart, M.M., 1997.
On persistence in mutual fund performance,
Journal of Finance
52, 57
-
82.


Chan, L.K.C., H.
Chen,
and J. Lakonishok, 2002.
On mutual fund investment styles,

Review of Financial
Studies

15, 1407
-
1437.


Core, J.E., W.R. Guay, and T.
O. Rusti
cus, 2006.
Does weak governance cause weak stock returns? An
examination of firm operating performance and investors’

expectations,

Journal of Finance

66,
655
-
687.


Cremers, M. and V.B. Nair, 2005.
Governance mechanisms and equity prices,
Journal of Finance

60,
2859
-
2894.


Cremers, M., V.
B. Nair, and
K
.
John
, 200
9
.
Takeovers and the cross
-
section of returns,
Review of Financial
Studies

22, 1409
-
1445.


Cremers, M
.

and A
.

Pareek, 2010
.

Institutional
i
nvestors’
i
nvestment
d
urations and
s
tock
return anomalies
:
momentum
,
reversal
,
accruals
,
share i
ssuance and R&D
increases,

Working Paper
, Yale University.


Davis, J.
L., 2001
.
Mutual fund performance and manager style,
Financial Analysts Journal
57, 19
-
27.


Fama, E
.

and J
.

MacBeth, 1973
.
Risk,
return
, and
equilibrium
:
empirical tests
,
Journal of Political
Economy

81,

607
-
637.


Fields, M.A. and
P.Y. Keys, 2003.
The
e
mergence of
c
orporate
g
overnance from Wall St. to Main St.:
o
utside
d
irectors,
b
oard
d
iversity,
e
arnings
m
anagement, and
m
anagerial
i
ncentives to
b
ear
risk,
Financial Review

38,
1

24
.


Froot, K
.

and M
.

Teo,

2008
.

Style
investing
and
institutional investors
,
Journal of Financial and
Quantitative Analysis

43, 883
-
906.


28


Gompe
rs, P.A. and A. Metrick,
2001.
Institutional
i
nvestors and equity prices,
Quarterly Journal of
Economics

116, 229
-
259.


Gompers, P.A., J.L. Ishii, and A. Metrick, 2003.

Corporate governance and equity prices,
Quarterly Journal
of Economics

118, 107
-
155.



Harford J., S.
A.
Mansi, and W.
F.
Maxwell, 2008. Corporate governance and firm cash holdings,
Journal of
Financial Econom
ics

87, 535
-
555.


Houge, T.

and T
. Loughran, 2006.
Do investors capture value premium,
Financial Management

35, 5
-
19.


Jensen, M
.
C., 1978
.
Some anomalous evidence regarding market efficiency,
Journal of Finance
6,
95
-
101.


Jiraporn, P. and K.
C.

Gleason,
2007. Capital
s
tructure,
s
hareholder
r
ights, and
c
orporate
g
overnance,
The
Journal of Financial Research

31, 21
-
33.


Jiraporn, P., J.
-
C
.
Kim, and Y.S. Kim, 2011. Dividend payouts and corporate governance quality: an
empirical investigation
,
Financial Revie
w

46
,

251

279


Johnson, S
.
A., T
.
C. Moorman, and S
.
Sorescu, 2009
.
A reexamination of corporate governance and equity
prices,
Review of Financial Studies
22, 4753
-
4786.


Kim,

M
.
, R
.

Shukla, and M
.
Tomas, 2000
.

Mutual fund objective misclassification,
Journal of Economics
and Business

52, 309
-
323.


Lewellen,
S.

and
A. Metrick, 2010. C
orporate
g
overnanc
e and
e
quity
p
rices:
a
re
r
esults
r
o
bust to
I
ndustry
a
djustments?

Working paper
, Yale University
.


Masulis, R.W.,
C.
Wang,
and F. Xie,
2007
.

Corporate
governance and a
cquirer
r
eturns
,

Journal of Finance

62
, 1851
-
1889
.


Ng,
L.
, and
Q.
Wang, 2004
.

Institutional
trading
and the
turn
-
of
-
the
-
year
effect
,
Journal of Financial
Economics

74, 343
-
366.


Phalippou, L
.
, 2004
.

What
drives
the
value premium
?

Working
paper
, University of Amsterdam
.



Robicheaux, S.H.,
X. Fu,
and
J.A. Ligon,
2008
.
Lease
f
inancing and
c
orporate
governance,
Financial
Review

43,
403

437.


Schwert, G.W.
,
2003
.

Anomalies and market efficiency,
” in G.
Constantinides
,

M. Harris, and R.M.
Stulz
,
Eds.,
Handbook of the Economics of Finance
,
North
-
Holland, London
.

).


Starks, L.T.
,
2009.
EFA
k
eynote
s
peech: “Corporate
governance and corporate s
ocial
r
esponsibility:
w
hat
d
o
i
nvestors
c
are about? What
s
hould
i
nvestors
care about?”
,

Financial
Review

44,
461

468.


Wermers, R., 1999. Mutual fund herding and the impact on stock prices,
Journal of Finance

54, 581
-
622.


29


Yang, X. and Z.

Zhang
, 2009.
Institutional investors and equity returns: are short
-
term institutions better
informed?

Review of Fin
ancial Studies

22, 893
-
924.

30


Table 1

Average Monthly
Returns for Portfolios Based on
Corporate Governance

This

table reports the average value
-
weighted

monthly

returns

in percent

by year

for stock portfolios
formed
based
on

the

quality of
corporate
governance

as
measured by the
G
-
index.

Four portfolio
classifications are created with
Democracy

and Dictatorship

portfolio
s

representing portfolios
of
firms with
the lowest and highest G
-
index scores. The Democracy portfolio
contain
s

firms
having a

G
-
inde
x


5
. The
Dictatorship portfolio contains firms having a
G
-
index


14.
Firms with
a
G
-
index between
5

and 9

(includ
ing

9)

are

allocated to Port
folio

2

and
f
irms
with

a
G
-
index between 9 and 1
4

are
allocated

to

Port
folio

3
.
The
Corporate G
overnance
F
actor (CGF) is calculated as the
difference
in

average
monthly
value
-
weighted returns between

the
Democracy and Dictatorship portfolio
s
.

The G
-
index is compiled
following

Gompers, Ishii and Metric
k

(2003).
The s
ample period is 1990

to
2008.


Y
ear

Democracy

Port
folio 2

Port
folio 3

Dictatorship

CGF

1990

1.40

1.13

0.91

1.05

0.34

1991

2.33

2.53

2.60

2.44

-
0.11

1992

0.58

0.86

0.68

0.63

-
0.06

1993

0.56

0.95

0.58

0.62

-
0.07

1994

0.38

0.16

0.11

0.07

0.31

1995

2.62

2.71

2.75

2.35

0.28

1996

1.59

1.84

1.80

1.70

-
0.11

1997

2.70

2.55

2.56

2.34

0.36

1998

2.81

2.17

2.26

1.14

1.67

1999

3.21

1.94

0.97

-
0.61

3.83

2000

-
0.14

-
0.42

-
0.12

1.92

-
2.06

2001

-
1.36

-
0.96

-
0.36

-
0.77

-
0.60

2002

-
1.63

-
1.88

-
1.76

-
0.24

-
1.39

2003

1.94

2.24

2.40

2.37

-
0.42

2004

0.57

1.13

0.89

1.36

-
0.79

2005

0.06

0.55

0.79

0.82

-
0.75

2006

1.49

1.22

1.02

1.42

0.07

2007

0.33

0.54

0.45

0.53

-
0.20

2008

-
2.80

-
3.37

-
3.47

-
3.84

1.04

Mean

0.88

0.84

0.79

0.81

0.07


31


Table 2

Performance Regressions for Democracy and
Dictatorship Portfolios

This table reports results from regressions of
the
value
-
weighted monthly returns
using the

four
-
factor model

from
Carhart

(
1997)
. Regressions are
run for the entire period and for four sub
-
periods.

The
Democracy portfolio contains
firms with
a
G
-
index


5

and
the
Dictatorship portfolio
includes firms with
a
G
-
index


14
.
The portfolio of Democracy
-
Dictatorship buys
the
Democracy portfolio and
short
s
the
Dictatorship portfolio.

The
return (A
lpha) reported in this table is the monthly

abnormal return

in percent
.
The
sample period
is from 1990 to 2008.

The

t
-
statistics
for the
regression variables are presented

with their corresponding coefficients.



Factors

Period

Portfolio

Alpha

t
-
Stat


RMRF

t
-
Stat


SMB

t
-
Stat


HML

t
-
Stat


Mom

t
-
Stat

1990
-
1999

Democracy
-
Dictatorship

0.69

2.72*
*
*


-
0.04

-
0.51


-
0.22

-
2.5
0**


-
0.53

-
5.19
***


-
0.01

-
0.14


Democracy

0.30

2.19*
*


0.99

25.60*
*
*


-
0.24

-
5.17*
*
*


-
0.21

-
3.77*
*
*


-
0.05

-
1.56


Dictatorship

-
0.40

-
2.15**


1.03

19.40*
*
*


-
0.02

-
0.33


0.33

4.41*
*
*


-
0.04

-
0.94

2000
-
2008

Democracy
-
Dictatorship

-
0.21

-
0.66


0.01

0.20


0.04

0.42


-
0.46

-
4.75*
*
*


-
0.01

-
0.16


Democracy

-
0.05

-
1.00


0.95

16.39*
*
*


-
0.12

-
1.84
*


0.04

0.48


0.02

0.54


Dictatorship

0.15

0.79


0.94

20.28*
*
*


-
0.16

-
2.98*
*
*


0.49

8.25*
*
*


0.03

0.95

1990
-
2008

Democracy
-
Dictatorship

0.26

1.30


0.02

0.34


-
0.08

-
1.31


-
0.51

-
7.36*
*
*


0.00

0.07


Democracy

0.16

1.16


0.99

27.86*
*
*


-
0.17

-
4.26*
*
*


-
0.06

-
1.22


0.02

0.79


Dictatorship

-
0.10

-
0.73


0.96

27.55*
*
*


-
0.10

-
2.39**


0.45

9.48*
*
*


0.02

0.69

1990
-
1996

Democracy
-
Dictatorship

0.24

0.91


-
0.08

-
1.08


-
0.07

-
1.00


-
0.49

-
5.37*
*
*


0.00

0.02


Democracy

0.18

0.92


0.93

16.96*
*
*


-
0.23

-
4.37*
*
*


-
0.07

-
0.97


0.02

0.70


Dictatorship

-
0.06

-
0.29


1.01

16.89*
*
*


-
0.16

-
2.76**
*


0.43

5.75*
*
*


0.02

0.62

1997
-
1999

Democracy
-
Dictatorship

1.07

2.09*
*


-
0.03

-
0.28


-
0.17

-
1.24


-
0.88

-
4.71*
*
*


0.16

1.50


Democracy

0.52

2.44*
*


0.99

19.63*
*
*


-
0.18

-
3.04**


-
0.26

-
3.37*
*
*


0.06

1.44


Dictatorship

-
0.55

-
1.34


1.02

10.61*
*
*


0.00

-
0.04


0.62

4.13*
*
*


-
0.09

-
1.12

*
**
, **, * indicates statistical
significance at the 1
%,
5
%, and 10%
leve
l respectively
32


Table 3

Average Annual Returns for Mutual Fund Portfolios Formed on Fund Styles

This table reports
equal
-
weighted post
-
one
-
year returns for
fund portfolios

formed
based
on fund
styles
using

prior three
-
year data
.
At the beginning of e
ach year, equity mutual funds are
independentl
y sorted into
fund styles
based on

size,
book
-
to
-
market ratio

and corporate governance
according to their loadings on
the
SMB,

HML
and CGF

variables

from
the
five
-
factor model over
the
past

three

years.
The five
-
factor model includes
the four
factors used in C
arhart’s (1997) model
plus the Corporate Governance Facto
r (
CGF
)
.

The
CGF is defined as

the difference
in the

monthly
value
-
weighted returns between
the
Democracy

(
G
-
index


5
)

and
the
Dictatorship

(
G
-
index


14)

portfolios

in the month
t
.
The model used
is
:


𝑥 𝑒
𝑡
=

+

1
×
 
𝑡
+

2
×

𝑡
+

3
×

𝑡
+

4
×
 
𝑡
+

5
×

𝑡
+
𝜀


where
𝑥 𝑒
𝑡

is the excess return.
Mutual funds
with loadings in the

top (bottom) half of
S
MB

and

HML

loadings are classified in
to
Small Cap (Large Cap)

and
Value (Growth)

funds. Mutual
funds with
loadings in
the top (bottom)
third

of loadings on CGF are classified as funds with
a
preference
for
good (
bad
) corporate governance.
O
ther mutual funds are treated as neutral.
The
differential returns between funds preferring good and bad corporate governance
also
are

reported
with their corresponding

t
-
statistics.

The

s
ample period is from
1
993 to 2008
with
different
sub
-
periods
in Panels A, B, and C
.
All returns are

expressed

in percent.




CG preference





Good

Neutral

Bad

Good
-
Bad

t
-
Stat


Panel A: 1993
-
1999

Growth Funds

Large Cap

24.87

20.76

22.88

1.99

2.64*
*
*


Small Cap

26.00

27.73

20.03

5.98

5.96*
*
*

Value Funds

Large Cap

18.48

18.36

13.12

5.35

9.65*
*
*


Small Cap

15.54

14.13

13.73

1.81

2.23*
*


Panel B: 2000
-
200
1

Growth Funds

Large Cap

-
16.27

-
13.24

-
7.88

-
5.36

-
9.67*
*
*


Small Cap

-
18.14

-
17.95

-
13.00

-
5.14

-
4.79*
*
*

Value Funds

Large Cap


-
6.68

-
4.55


0.20

-
6.88

-
9.36*
*
*


Small Cap


-
0.95


1.77


7.25

-
8.20

-
9.05*
*
*


Panel C: 2002
-
2008

Growth Funds

Large Cap

-
4.02

-
0.92

1.62

-
5.63

-
9.17*
*
*


Small

Cap

-
0.89

-
3.07

-
0.24

-
0.65

-
1.11

Value Funds

Large Cap

-
4.56

1.27

4.67

-
9.23

-
17.39*
*
*


Small Cap

2.66

3.28

6.02

-
3.36

-
5.55*
*
*

***, **, *, indicates statistical significance at the 1%, 5%, and 10% level, respecitevely.
33


Table 4

Regressions of Mutual Fund Return
s

on Past Performance, Fund Size, Style Dummies, and Expense Ratio

This table shows the regression estimates of the following model,


𝑒 𝑛

,

=

+

1


,

+

1
𝑖𝑧𝑒

,

+

2
𝑉

,

+

3


,

+

4


,

+

5


,

+

6
𝑥𝑝𝑒𝑛𝑒

,

+
𝜀


where
Return

is the
annual
return for fund

i

in year

j
.

The control variables includes past performance (
PR
) in log of (1+prior one year return),

fund
size

(
Size
)

in

log of TNA,
dummy variables

(
VD, SD, GCG
, and
BCG
) for fund styles of value, small
-
cap
, good corporate governance and bad
corporate governance, and expense ratio

(Expense)
. Fund returns are annual returns net of
expense.
Fund styles are estimated by using the
methodology described in Table 3.
Th
e regressions are
error
-
corrected
at fund and year levels. The

t
-
statistics are reported along with their
corresponding coefficients as well as the
F
-
statistics for the hypothesis

of


0
:

4
=

5
. Sample period is from

1993 to 2008.













H
0
:


4
=

5

呩T攠健ei潤

Al灨p





卩Se






䝃G

䉃B

䕸灥湳E

A摪⸠
R
-



F
-
Stat

1993
-
1999


0.280



0.345

-
0.002


-
0.070

0.007

0.003

-
0.028


-
0.282

0.25


34.73
*
**


34.23
*
**


15.25
*
*
*

-
2.59
*
*

-
24.79
*
*
*

2.32
*
*

0.66


-
6.76
*
**


-
1.21

















2000
-
2001


-
0.150


-
0.001

-
0.007


0.140

0.028

-
0.028


0.048


-
2.108

0.37


133.74
*
*
*


-
21.98
*
*
*


-
0.16


-
8.39
*
*
*

34.84
*
*
*

8.02
*
*
*

-
7.17
*
*
*

11.42
*
*
*


-
7.64
*
*
*
















2002
-
2008



-
0.402


0.038

-
0.001


0.025


0.015


0.001


0.018


-
0.780

0.86


145.47
*
*
*


-
126.41
*
*
*


4.62
*
*
*

-
5.27
*
*
*

29.19
*
*
*

15.73
*
*
*


1.18


16.43
*
*
*


-
4.12
*
*
*
















Al l Ye ar s


-
0.393


0.146

-
0.001

-
0.001


0.011


0.003


0.005


-
0.609

0.7
7


1.15


-
167.88
*
*
*


18.61
*
*
*

-
3.62
*
*
*

-
1.22


11.33
*
*
*


2.76
*
*
*


3.65
*
*
*


-
5.11
*
*
*




*
**
, **, * i ndi c at es s t at i s t i cal
s i gni f i c anc e at t he 1
%,
5
%, and 10%
l e ve
l r es pe ct i ve l y.
34


Table
5

Average Coefficient
Estimates

for Mutual Funds

At the beginning of each year, the five
-
factor model is used to estimate the loadings on each factor
for every equity fund over the prior three years. The five
-
factor model includes the four factors
used in Carhart (1997) and the Corporate Governance Facto
r (CGF). The CGF is defined as

the
difference
in the

monthly value
-
weighted returns between
the
Democracy

(
G
-
index


5
)

and
the
Dictatorship

(
G
-
index


14)

portfolios in the month
t
.

In Panels A and B, the average coefficient
estimates are reported yearly across all equity funds. At the bottom of Panels A and B, the average
coefficient estimates across years are also reported. Panel C reports the difference in coefficient
estimates fo
r each factor between the period
s

1993

to
1999 and 2002

to
2008.
Alphas are reported
as monthly abnormal returns in percent.

The differential coefficients between two periods are
calculated by first taking the differences of coefficients for each fund in a

given year in the first
period to every year in the second period. Then, the average coefficient differences are calculated
across all mutual funds in a given year and the time series
standard errors are used to compute
t
-
statistics. The

t
-
statistics are
reported along with their corresponding coefficients. The sample
period is from 1993 to 2008.





Factors

Year


Alpha

t
-
Stat



SMB

t
-
Stat



HML

t
-
Stat



CGF

t
-
Stat


Panel A: 1993
-
1999

1993



-
0.07

-
2.42*
*




0.23

10.23*
*
*



-
0.01

-
0.87





0.00

0.03



1994



0.06

2.11*
*




0.00

0.03





-
0.07

-
4.61*
*
*



-
0.04

-
4.35*
*
*

1995



-
0.15

-
4.82*
*
*



0.26

19.45*
*
*



-
0.04

-
2.61*
*
*



0.02

1.34




1996



-
0.11

-
3.94*
*
*



0.24

13.82*
*
*



-
0.02

-
0.98





-
0.03

-
4.13*
*
*

1997



2.48

2.47*
*




0.01

0.07





-
0.27

-
2.94*
*
*



0.03

4.76*
*
*

1998



-
0.18

-
6.01*
*
*



0.29

16.32*
*
*



-
0.07

-
1.24





0.01

0.42




1999



-
0.59

-
0.72





0.33

8.83*
*
*



-
0.04

-
0.34





0.04

6.23*
*
*

Avg.



0.21

0.53





0.19

3.85*
*
*



-
0.07

-
2.20*




0.00

0.38





Panel B: 2002
-
2008

2002



0.32

17.06*
*
*



0.12

24.58*
*
*



0.00

0.78





-
0.08

-
28.89*
*
*

2003



-
0.11

-
9.81*
*
*



0.15

31.70*
*
*



0.04

8.96*
*
*



-
0.09

-
27.08*
*
*

2004



-
0.25

-
15.26*
*
*



0.20

40.71*
*
*



0.02

2.11*
*



-
0.07

-
29.10*
*
*

2005



-
0.16

-
10.28*
*
*



0.21

30.97*
*
*



0.00

-
0.30






-
0.06

-
19.93*
*
*

2006



-
0.08

-
11.27*
*
*



0.16

11.84*
*
*



-
0.01

-
2.00*
*



-
0.04

-
24.35*
*
*

2007



-
0.03

-
2.55*
*




0.21

26.95*
*
*



-
0.06

-
11.96*
*
*



-
0.07

-
20.30*
*
*

2008



-
0.05

-
6.84*
*
*



0.
18

37
.
55
*
*
*



-
0.
04

-
9.86*
*
*



-
0.
04


-
26
.
72*
*
*

Avg.



-
0.
05

-
0.76





0.
1
8

13.64*
*
*



-
0.01

-
0.56





-
0.0
6


-
8.94*
*
*


Pa n e l C: D i f f e r e n c e be t we e n Two Pe r i o d s

D i f f


0.
2 6

0.6 7



0.0
2


0.3 7




-
0.07

-
1.
99*




0.07

6.06
*
*
*

*
**
, **, * i ndi cat es s t at i s t i cal
s i gni f i cance at t he 1
%,
5
%, and 10%
l eve
l r es pect i vel y.


35


Table
6

Shift of Corporate Governance Preference for
Equity

Funds

This table reports the change
s

of corporate governance preference for mutual funds from year
t

to
t
+
x
. At the beginning of each year, all equity funds are classified as funds with preference of good,
neutral and bad corporate governance according to the loadings on the CGF from five
-
fa
ctor
regressions over the past 36 months. Five
-
factor model includes four factors used in Carhart (1997)
and CGF
, and is described in Table 3
. CGF is measured as the differential value
-
weighted monthly
returns for firms between Democracy and Dictatorship p
ortfolios. Mutual funds with top (bottom)
third

loadings on CGF are classified as funds with good (bad) corporate governance preference.
The other mutual funds are classified as neutral. In Panel A, the change
s

of corporate governance
preference
for mutual

funds are

measured between year t and t+3. In Panel B, the change
s of
corporate governance preference are

measured between year t and t+9.


Funds’ Preference in
t
+
x

Funds’ Preference in
t

Good CG

Neutral

Bad CG


Panel A: 1993
-
1999 (
x
=3)

Good CG

40.43%

30.90%

28.67%

Neutral

31.79%

35.42%

32.80%

Bad CG

28.25%

30.91%

40.84%






Panel
B
: 1993
-

2008 (
x
=9)

Good CG

37.79%

29.92%

32.29%

Neutral

36.25%

33.66%

30.09%

Bad CG

28.79%

28.92%

42.30%




36


Table 7

Shift of Fund Styles
for Mutual Funds with Good and Bad Corporate Governance Preference

This table reports the relations between mutual fund investment styles and preference for good and bad
corporate governance in two sub
-
periods. Between sub
-
period c
hanges of fund
style
s

for mutual funds with
the same preference for the corporate governance
are

reported. All equity funds are
sorted yearly into
groups based on their loadings on the
factor of
CGF from the five
-
factor regression over the prior 36 months.
The five
-
factor mode
l includes the four factors used in Carhart (1997) and the Corporate Governance
Factor (CGF). Good (Bad) CG groups contain mutual funds with top (bottom)
third

loadings on CGF.
In
both Panels A and B, the coefficient estimates are first averaged across all

mutual funds in the same fund
group in a given year and the reported coefficients for a fund group are the distribution of yearly average
coefficient estimates. The differential coefficients in Panel A and B are calculated by pairing yearly average
coeffi
cient estimates from two fund groups in the same year. In
Panel C, the same methodology described in
Table 5 is used to calculate the differential coefficients for a fund group between two periods. The
t
-
statistics are reported along with their correspondi
ng coefficients. Sample period is 1993

to
2008 but
excludes year 2000 and 2001.



Fund Styles

Fund Preference

MKTRF

t
-
Stat


SMB

t
-
Stat


HML

t
-
Stat


CGF

t
-
Stat


Panel A:
1993
-
1999

Good CG

0.9506

49.22
*
**


0.3597

12.60*
*
*


-
0.1264

-
1.92



0.2861

11.06*
*
*

Bad CG

0.9688

39.19
*
**


0.2210

6.08*
*
*


-
0.0052

-
0.13


-
0.3061

-
1
5.14*
*
*

Goo
d
-
Bad

-
0.0181

-
0.58


0.1387

3.00*
*



-
0.1213

-
1.58


0.5923

18.03*
*
*


Pane l
B:
2002
-
2008

Good CG

1.0118

73
.
50
*
**


0.0751

3.75*
*
*


-
0.0022

-
0.05


0.1315

8.36*
*
*

Bad CG

1.0394

28.22
*
**


0.3845

4.17*
*
*


-
0.1193

-
1.09


-
0.2672

-
16.98*
*
*

Goo
d
-
Bad

-
0.0276

-
0.75


-
0.3094

-
3.36*
*


0.1171


1.07


0.3988

25.33*
*
*


Pane l C:
Di f f er e nc e be t ween Two Pe r i ods

Good

CG

-
0.0612

-
3.17
*
*


0.2846


9.97*
*
*


-
0.1242

-
1.89


0.1546

5.97*
*
*

Bad CG

-
0.0707

-
2.86**


-
0.1635

-
4.50**
*



0.1142

2.90**


-
0.0389

-
1.92

*
**
, **, * indicates statistical
significance at the 1
%,
5
%, and 10%
leve
l respectively.
37


Table
8

Corporate Governance Preference and Subsequent Returns

This table reports the
abnormal returns for firms in Democracy and Dictatorship groups

categorized

by prior corporate governance preference (CG
preference) of mutual funds

overall
.
At the beginning of each year,
the
five
-
factor model is used to estimate the CG preference for each individual
mutual fund over the past three years. The

me
an

lo
ading on
the

CGF
across all funds
is
used as the overall
prior
CG
preference for the year. Then, all
16 years

in
the

sample period

are sorted into two (
halves
) or
three

(
terci
les) groups
according

to the prior overall preference. Firms with G
-
index


5
or with G
-
index


14

are classified as Democracy

or Dictatorship portfolios. For each portfolio,
Carhart’
s

(1997)

four
-
factor model is again used to
estimate the abnormal
returns. T
he
return

difference
s
also
are reported along with their corresponding
t
-
statistics.

Sample period is from year 1993 to
2008.



CG Preference over Prior
Three

Years


Top

Yrs


Bottom

Yrs


Diff

(Top,Btm
)


Mean

t
-
Stat


Mean

t
-
Stat


Mean

t
-
Stat


Panel A:
Halves

Based

Democracy

0.44


2.33*
*


-
0.13

-
0.54


0.57

1.84*

Dictatorship

-
0.19

-
0.82




0.01

0.07


-
0.20

-
0.72


Di f f

(De m, Di c )

0.62

2.11*



-
0.14

-
0.45














Pane l B:
Ter ci
le
s

Based

Democracy

0.31

1.60



-
0.32

-
1.19


0.63

1.87*

Di c t at or s hi p

-
0.56

-
1.99
*



0.27

1.77


-
0.83

-
2.65**

Di f f

(De m,
Di c )

0.87


2.26**


-
0.59



-
1.91*




*
**
, **, * i ndi c at es s t at i s t i cal
s i gni f i c anc e at t he 1
%,
5
%, and 10%
l e ve
l r es pe ct i ve l y.



38


Table
9

Corporate Governance Preference of Mutual Funds Following Stock

Performance of Well
-
governed Firms

At the beginning of each year, the corporate governance premium (CGP) in the previous year is calculated. Years are categoriz
ed into two or three
groups according to the value of CGP in year t
-
1. Top (bottom) years include years with the highest (lowest) C
GP. Then,
Fama
-
MacBeth’s
methodology is implemented to find the average corporate governance preferences across all
mutual funds over the subsequent three years (t, t+2).
Results are reported in each year group. The table also reports the difference of cor
porate governance preference between top and bottom year groups
along with their
t
-
statistics by using methodology described in Table 5.


Corporate Governance Preference in Year t,t+2


Top Years

Bottom Years

Diff (Top, Bottom)


Corporate Governance
Preference
t,t+2


Mean

t
-
Stat

Mean

t
-
Stat

Mean

t
-
Stat

Halves Based

-
0.039

-
2.02
*

-
0.042

-
3.82*
*
*


0.003

0.16

Te r ci l e s Bas e d

-
0.056

-
2.99*
*

-
0.055

-
10.72*
**


-
0.001


-
0.01

*
**
, **, * i ndi c at es s t at i s t i cal
s i gni f i c anc e at t he 1
%,
5
%, and 10%
l e ve
l
respectively.



39


Table
10

Corporate Governance Preference and
Industry Effect

This table reports

Fama
-
MacBeth

estimates of

the
corporate governance preference

(

CG Pref

)

for

mutual fund portfolio

holdings

and

for firms in
Democracy and Dictatorship Portfolios
.
At the
end

of each
quarter
,
all mutual fund
s’ equity

holdings

are aggregated
at the firm level to calculate the
fraction of shares
held by
mutual funds.
The dependent variable is the total mutual fund ow
nership

in each firm
.
Variable of

CG Pref


will be set to
1 if
the
firm
is

in the Democracy portfolio, 0 if otherwise.
Control

variables include firm size

(Size)
, Book
-
to
-
market ratio

(BM)
, gross return over the
past three months

(
Ret
t
-
3,t
)
, gross return

over the past nine months before the end of last quarter

(
Ret
t
-
12,t
-
3
)
, turnover

ratio (Turnover)
, and firm age

(Age)
.
All

variables except mutual fund ownership
, ‘CG Pref’

and returns are expressed in log
.
Regressions are estimated at each quarter
-
end

an
d the
average coefficients are reported along with their corresponding
t
-
statistics.

The table also gives the number of coefficients that are positive
ly

or
negative
ly

significant at the 10% level. In Panel B, industry dummies

defined with two digits of SIC

code

are added to control for
industry effect.
Mutual fund holdings are from Thomson Reuters Mutual Fund data.
Sample period is from year 199
0

to 200
7
.



Whole Sample Period


1990
-
1998


1999
-
2007


Coef.

t
-
Stat

Pos/Neg Sig.


Coef.

t
-
Stat

Pos/Neg Sig.


Coef.

t
-
Stat

Pos/Neg Sig.



Panel A: Corporate Governance Preference without Industry Effect


CG Pref

-
0.
00
5

-
2.25
**

3/13


-
0.
003

-
1.37

1/2


-
0.
00
7

-
1.86
*

2/11

Size

0.003

3.63
**
*

13/2


0.
00
2

2.36
**

5/0


0.
00
3

2.77
**
*

8/2

BM

-
0.004

-
2.46
**

0/9


0.001

0.86

0/1


-
0.008

-
3.49
**
*

0/8

Ret
t
-
3,t

0.017

2.75
**
*

9/2


0.029

3.62
**
*

5/0


0.007

0.76

4/2

Ret
t
-
12,t
-
3

0.018

4.63
**
*

24/3


0.034

5.76
**
*

15/0


0.005

1.08

9/3

Turnover

0.033

17.20
*
*
*

67/0


0.021

14.59
**
*

31/0


0.044

19.48
**
*

36/0

Age

-
0.015

-
2.01
**

1/9


-
0.030

-
2.01
**

0/2


-
0.002

-
0.45

1/7














Panel B: Corporate Governance Preference Adjusted for Industry Effect

CG Pref

-
0.016

-
6.07
*
*
*

0/25


-
0.011

-
4.17
**
*

0/10


-
0.020

-
4.78
**
*

0/15

Size

0.006

7.44
**
*

23/1


0.006

5.15
**
*

10/0


0.007

5.42
**
*

13/1

BM

0.004

2.32
**

9/3


0.010

4.22
**
*

6/0


-
0.001

-
0.61

3/3

Ret
t
-
3,t

0.016

2.36
**

14/2


0.030

3.06
**
*

9/0


0.005

0.48

5/2

Ret
t
-
12,t
-
3

0.025

5.59
**
*

22/1


0.040

5.51
**
*

13/0


0.011

2.64
**

9/1

Turnover

0.032

16.86
**
*

66/0


0.022

12.64
**
*

30/0


0.041

16.93
**
*

36/0

Age

-
0.013

-
1.82

1/8


-
0.026

-
1.88
*

0/1


-
0.002

-
0.32

1/7

*
**
, **, * indicates statistical
significance at the 1
%,
5
%, and 10%
leve
l respectively.
40





Figure 1

Distribution of Returns for Democracy, Dictatorship and CGF
Portfolios

This figure displays the
average monthly
return distribution
s

by year for firms in the Democracy,
Dictatorship, and Corporate Governance Factor (CGF) portfolios. The CGF portfolio is the portfolio which
buys the Democracy and shorts the Dictator
ship portfolios. The sample period is 1990 to 2008.

-5.00
-4.00
-3.00
-2.00
-1.00
0.00
1.00
2.00
3.00
4.00
5.00
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
Democracy
Dictatorship
CGF
41






Figure 2

CGF Distribution and Mutual Funds


Corproate Governance Preference

CGF is the corporate governance premium which buys the Democracy portfolio and shorts the Dictatorship
portfolio. CGCoef

is the average of coefficients across all equity mutual funds
in a

year on
the factor
CGF

taken
from
the
five
-
factor model described in Table 3
. To find
the
CGCoef

for each fund
, the five
-
factor
model is used at the beginning of
each

year

in the sample pe
riod

to estimate the loading on the CGF over the
prior three years.
Each mutual fund runs one regression to get one coefficient of CGF. Then, all funds are
pooled to find the average of
CGF
coefficients in that year. These coefficients on CGF represent mut
ual
funds


corporate governance preference.
The five
-
factor model includes four factors used in Carhart (1997)
and CGF. The average of coefficients on CGF is scaled by 10 for
a
clear view in this figure.



-3.00
-2.00
-1.00
0.00
1.00
2.00
3.00
4.00
5.00
CGF
CGCoef(EW)