Mutual Fund Performance, Management Teams, and Boards

testyechinoidΔιαχείριση

18 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

272 εμφανίσεις

1


Mutual Fund Performance, Management Teams, and
Boards


John C. Adams, Takeshi Nishikawa, and Ramesh P. Rao


September 26
, 2012



Abstract

This paper examines the performance of team
-
managed and single
-
managed funds in
the mutual funds. The recent surge in the use of team managed funds in the mutual
fund industry suggests that the benefits of team management are outweighing its costs.
Howev
er, most studies show that no superior returns are related to team managed
funds relative to single managed funds. Our analysis shows that not all teams are alike,
and
therefore,
for a team fund to be a better
performer

there needs to be a moderating
inte
rnal governance
factor. We find that teams
in funds
with the highly independent
boards charge significantly less fees to their fund holders than those team
-
managed
funds with less independent directors.


Keywords
:
M
utual funds
;

board structure
;

organizational structure

JEL Classification
: G20, G32, G34



John C. Adams is an Assistant Professor of Finance in the Department of Finance and Real
Estate, College of Business at the University of Texas at Arlington, Arlington, TX 76019.


Takeshi Nishik
awa is an Assistant Professor of Finance at school of Business University of
Colorado, Denver CO 80202


Ramesh P. Rao is
a Professor and Paul C. Wise Chair in Finance in the Department of Finance at
Oklahoma State University, Stillwater OK 74078

2


1.

Introd
uction

Are
two heads better than one
?

What about three or four?

As

an increasing number of
mutual funds have been structured as team
-
managed funds
1
, more practitioners and academics
alike have
attempted

to answer this question

in relation to the performance of mutual funds in
the U.S.

While there is still a paucity of studies that empirically examine the relation between
fund performance and fund management organizations, the emerging consensus seems to be
that team
-
managed fu
nds tend to either underperform the single
-
managed funds (i.e., Chen,
Harrison, Ming, and Kubic (2004) and Bä
r, Kempf, and Ruenzi (2011
)) or at best be no different
than single
-
managed counterparts (i.e., Prather and Middleton (2002) and
Bliss, Potter, and

Schwarz

(20
0
8
)).
The increasing popularity of team
-
managed funds is odd if indeed team
-
managed structure does not enhance the performance of funds given the ever
-
increasing
importance of mutual funds as a financial intermediary.

As of 2010, about $11.
8
tr
illion in assets were managed by various mutual funds for
nearly 25 million households with about 90 million investors in the U.S. (ICI, 2011).
Echoing
this puzzle, Han, Noe, and Rebello (2008) show that there is a positive relation between
performanc
e and team management once the selection bias where managers with lower
management ability are self
-
selected to a team management is controlled
.


However, o
verall,
our understanding of the relation between fund performance and
fund management
organization
s is scant and limited.

The objective of this paper is to examine the relation
between
fund performance and fund management organizations in the presence of the internal
governance mechanisms.

The mutual fund governance, particularly
independent director requirement, has been
hotly debated among regulators, practitioners, and academics
following

the mutual fund
industry scandal of 2003
-
2004. In the wake of the scandals, in 2006 the SEC adopted a rule that
requires 75% of mutual fund d
irectors be independent.
However, following the lawsuit by the
U.S. Chamber of Commerce, the U.S. Court of Appeals for the District of Columbia faulted the
SEC for failing to provide little evidence supporting the required governance changes would
bring b
etter and improved governance in the mutual fund industry. Recent academic evidence



1

For example, a
The Wall Street Journal
article by Eleanor Laise
, “Your fund manager’s secrets,” report that as of
2006 about 65% of funds are team
-
managed funds based on Morningstar figures, up from 49% in 2000. The article
is available at
http:/
/online.wsj.com/article/SB116077809860592420.html
. Our sample shows the similar number.
Please see Panel A of Table 2.

3


(i.e., Khorana, Tufano, and Wedge (2007) and Adams, Mansi, and Nishikawa (2010)) reports
empirical evidence that boards with higher proportion of independent directors are

more
effective monitors, suggesting the SEC’s attempt to increase of independent director presence in
mutual fund boards might be beneficial.
Indeed
, the Investment Company Institute reports on
its website that “In nearly 90 percent of fund complexes, 75

percent or more of fund directors
are independent.”
2


We hypothesize that the recent popularity of team
-
managed funds in the U.S. could be
related to the governance related benefits that this form of
fund
management organization
provides.
We argue that
the benefits of team management could potentially reduce the role of
the internal governance as
the nature of the team management

could be used as a supplement
to the interna
l

governance mechanisms.
In a corporate setting, Arena, Ferris, and Unlu (2011)
find that
the presence of c
o
-
CEOships
can serve as an alternative governance mechanism, with
co
-
CEO mutual monitoring substituting for board

monitoring
.

Because both corporate and
mutual fund boards
have a fiduciary duty to protect the interests of shareholders, it is possible
that team
-
managed funds also provide alternative governance mechanisms in the mutual funds.
Sah and Stiglitz (1986 and 1988) suggest that in a team setting difference of opinio
ns among
members are compromised which lead the team decisions to be less extreme than those of
individuals.
The implication of Sah and Stiglitz’s argument for fund management is that team
would have an ability to diversify style and judgment (Sharpe (198
1)).
Consistent with Sah and
Stiglitz (1988),
Bär et al. (201
1
) find evidence
that

team
-
managed

funds follow less extreme
investment styles. Similarly, Han et al. (2008) report that
the investment style of team
-
managed
funds is less eccentric than that o
f

individually managed funds. Another potential benefit of
team management is that team setting would broaden the skills and knowledge which could
lead to better abilities to process more information (Hill (1982) and Herrenkohl (2004)).
Combin
ing

all the

benefits of team management, we expect that

the
existence and
benefits of
team management should be more pronounced in a fund where internal governance
mechanisms are relatively weak

While there
are

potential benefits, team management also has its own dra
wbacks.
One
of the drawbacks of team management is

the possible “moral hazard in teams” problem
,
highlighted by Holmström (1982).
In
team management where
it is difficult to identify



2

See at their website
http://www.ici.org/idc/policy/go
vernance/overview_fund_gov_idc


4


individual members’ contributions to the success or failure of the
team (Almazan, Brown,
Carlson, and Chapman (2004)), the potential remedies to reduce agency problems, such as the
labor market disciplines (Fama (1980)), optimal dynamic managerial contracts (Holmström
(1999)), and career concerns (Chevalier and Ellison (1
999)), become less effective. Also, team
management could lead to delays in decision making (Sah and Stiglitz (1988)).
Therefore, i
n a
sharp contrast to the potential benefits of team management where it could be used as a
supplement to governance mechan
isms, these drawbacks of team management suggests that
there needs to be
an
existence of strong internal governance mechanisms for team management
to be successful.
As such,

we hypothesize

that the existence and benefits of team
-
managed
funds
are more pre
valent with those funds that have strong internal governance mechanisms.

Employing a sample of
3,002

U.S. offered mutual fund from 99 investment companies
covering the period from 1999 to 2007, we examine the relation between fund performance and
fund mana
gement organizations in

a framework that controls for endogeneity between fund
management structure and

internal governance mechanisms.
Our study starts with an analysis
of the relation between management structure and internal governance mechanisms.
We

find
that some internal governance mechanisms are important factors to influence the choice of fund
management structure even after controlling for the percent of team funds in an investment
company, suggesting that there is a potential effect of endogeno
us choice between team and
single
-
management (i.e., Han, et al (2008)).

We also find that
larger funds and younger funds
tend to be more team managed.

Next, we examine the relation between fund performance and fund management
organizations while carefu
lly controlling for an effect of endogenous choice between team and
single
-
management. We use three different measures to capture fund performance: Ca
r
hart
four
-
factor alpha, expense ratio, and managerial performance measure of Ca
r
hart four
-
factor
alpha p
lus expense ratio (Elton, Gruber, and Busse (2004)).
Overall, our results from the OLS
regressions show that regardless of the performance measures we use, there is no evidence that
team
-
managed funds either outperform or underperform the single
-
managed c
ounterparts.
However, more interesting results emerge when we interact the team dummy with internal
governance measures.
When Ca
r
hart

four
-
factor alpha is employed as a performance measure,
the team
-
managed funds with higher proportion of independent directors perform significantly
better than those team
-
managed funds with less independent director representation.
In
5


contrast, the effe
ct of board size to fund organizations show that the performance of team
-
managed funds with large boards is significantly worse than those team
-
managed funds with
fewer board

member
s.

These results from the analysis of Ca
r
hart four
-
factor alpha show that

performance of
team
-
managed funds is highly influenced by the presence of internal governance mechanisms.
Specifically, higher proportion of independent directors, believed to
result in

better
monitor
ing
,
leads to better performance of team
-
managed funds
, while larger boards, believed
to be less effective monitoring (i.e., Yermack (1996)), results in worse performance of team
-
managed funds, suggesting that the drawbacks of team management dominates the benefits in
the U.S. mutual fund industry.

When we us
e the expense ratio of the funds to measure the performance of the funds,
we find that expense ratios of team
-
managed funds with larger boards are significantly higher
than those team
-
managed funds with smaller boards. However, we do not see any significa
nt
difference between team
-
managed funds with respect to the proportion of independent
directors.

While Ca
r
hart four
-
factor alpha and expense ratio measure the performance of
mutual fund
for
investors, the performance of fund manag
er is pre
-
expensed base.

Thus, we
add expense ratio back to alpha to measure the true value of the managerial pe
rformance
following Elton, et al
. (2004).
The OLS regression results using Carhart four
-
factor alpha plus
expense ratio as a dependent variable show that unlike
the
o
ther two performance measures
team
-
managed funds with either high proportion of independent directors or large board size
perform significantly better than those team funds with either less proportion of independent
directors or smaller board. However, fr
om the earlier OLS regression results, it is shown that
the reason why
team
-
funds with higher independent directors have better managerial ability is
due to the superior risk
-
adjusted return, measured by Ca
r
hart four
-
factor alpha, while for team
-
funds with

large boards are due mainly to the higher expense ratio. Given the fact that the
fiduciary duty of the boards is to protect interests of shareholders, we can conclude that team
-
funds perform better when there is a higher proportion of independent directo
rs. These results
are consistent with overall governance related findings with corporate settings.

Finally, we test the robustness of our findings. Specifically, we delete any f
u
nds that
experienced changes in organizational forms. In other words, we re
run the probit and OLS
models with the samples that did not change their management organizations within the last 12
6


months. Using these samples, we find the same results as our overall results, suggesting that
our results are not influenced by the change
s in the management organizations.

This study contributes to the literature in two different ways.
First, to the best of our
knowledge

this is the
first empirical evidence that
provides
a link between
fund performance

and
fund management organizations
in the presence of internal governance mechanisms
.

Because existing literature suggests that team management can be either beneficial as an
alternative governance mechanism or detrimental unless with proper governance mechanisms,
the influence of internal

governance is critical when the performance of team management is
examined
.

Second, this paper sheds light on the effectiveness of independent director in
governing mutual funds. Given the dramatic increase in team
-
managed funds, our results
show the in
creasing importance of independent directors.

Our paper is organized as follows: Section 2 describes the data and methodology. Section 3
presents the empirical results. Finally, Section 4 concludes.


2
.

Data and
Variables Measures

2.1

The
Sample

We use s
everal databases to build our sample. We obtain mutual fund returns, total net
assets, expense ratios, 12b
-
1 fees, management fees, fund age, institutional ownership data, load
s, and other fund characteristics from the Morningstar Direct mutual fund data
base. We hand
collect end of year board of director information from the Statement of Additional Information
(SAI), which is located in each fund’s prospectus (e.g. Form 485a). We also record from each
fund’s Form 485a filing whether the investment compa
ny that sponsors the fund is a publicly
traded corporation or a subsidiary of a publicly traded corporation. We cross check this
sponsor level ownership structure against the Dun and Bradstreet, Hoover, and CRSP
databases. Since most funds offer multiple

share classes of the same underlying portfolio in
order to accommodate the preferences of varying investor clienteles regarding fees, sales
charges, and account sizes, we compute the fund level value weighted average of each share
class level variable.


We
next
sum the total net assets of each fund to obtain the size of the assets under
management of each sponsor and include the 55 largest in our sample. We avoid a large
sponsor bias by including data for small and medium sized investment companies. We

then
7


focus on actively managed funds by excluding index and exchange traded funds. The resulting
sample contains funds from 99 investment companies

over the years
1999
to 2007
.


2.
2

Team vs. Single Managed funds

The
Morningstar

database report
s

the identity of fund managers
.

For fund
s

with
individual or single managers
, Morningstar reports the
ir

full name
s.

When funds are managed
by more than one individual, Morningstar

often only

provide
s

the

last

names of managers
with
the longest tenured or lead managed listed first.

In some instances, Morningstar uses
“Management team,” “multiple managers,” and prior to 2001
the
name of
the lead

manager
followed by “et.
a
l

.

In these case
s
, we classify a fund as single
-
managed w
hen

only
one
manager name is given in the database whereas multiple names or
anonymous reporting of
“Management team” or “Multiple managers”

are classified

as team
-
managed funds.
3

However,
following the SEC’s
2004
release of
Disclosure Regarding Portfolio

Managers of Registered
Management Investment Companies
, 17 C
F
R Parts 239, 249, 270, and 274, Release No. 33
-
8458; 34
-
50227; IC
-
26533; File No. S7
-
12
-
04

funds are obliged to disclose the identities

of all team
members.
In our sample
,

the instances of Morningstar reporting anonymous team have
become
increasing less frequent and disappeared in the years following the SEC’s disclosure
requirement.




2.3 Board Characteristics


We follow Tufano and Sevick (1997) and focus our analysis o
n board structure and
composition (e.g. board size and independence).
We

collect calendar year end
board structure
data from the statement of additional information (SAI), which is included in each fund’s
prospectus (Form 485).
The SAI lists all director
s and det
ails how

a director is affiliated with
the fund sponsor and if a particular director serves as the board chairperson.
We compute
Board Size

as the natural logarithm of the number of trustees serving on each board as our
measure of board size. We use the definition of director independence in SEC (2004)
regulations and compute
Independent Directors

a
s the
ratio

of independent
(also known as



3

Massa, et al. (2010) argue that named managers exert more effort than anonymous ones. In
unreported analysis we eliminate anonymously managed funds and find similar results.


8


d
isinterested
)

trustees to
the
total
number of directors
.
SEC regulations classify as independent
those directors who are not
employee
s, not

employee family member
s, not

employee
s or
5%
or
more
shareholder
s

of a registered broker
-
dealer, and
are

not affili
ated with any recent legal
counsel to the fund.
Independent Chair

is determined using manual collection based on Form 485
to ascertain whether the board has a chairperson who is an independent director. Finally, we
gather information if a single board

ove
rsees

all of the

funds of the sponsor (Unitary Board
Structure)
and designate
a dummy variable that takes
on a value of
one if it is a unitary board
and otherwise zero.


2.4

Performance
Variables



This study employs

three

fund performance

measures.

The first measure is

each fund’s risk
-
adjusted return using Ca
r
hart (1997) four factor model
’s alpha
.

Computed as

t
i
t
t
t
t
i
t
f
t
i
e
YR
PR
HML
SMB
RMRF
R
R
,
4
3
2
1
,
,
)
1
(
)
(
)
(
)
(
)
(












,



(1)

where
R
i


is the mutual fund return

net of expenses
,
R
MRF
t

is the
excess
return on
the CRSP

value
-
weighted aggregate market proxy
,
R
f,
is the 30
-
day T
-
bill return,
SMB
t

(small minus big) is
size factor that captures the stock return performance of small firms relative to large firms,
HML
t

(high minus low)

is the relative return of value and grow
th stocks
, and
PR1YR
t

is a
momentum factor computed as the difference in returns of prior year high and low return
portfolios.
We use the Kenneth French data base to acquire values for each of the factors.
Similar
to Bliss

et al. (2008) and Almazan et al
. (2004) the
alpha measure

for each fund is

computed
each year

using monthly data
.

Second, we use the expense ratio of the fund as a way to capture fund performance.
Although the risk
-
adjusted return is important for mutual fund investors, mutual fund
inv
estors can enhance their returns by focusing on mutual fund
s with lower expense ratios
within

given fund objectives. The extant literature show
s

a

negative relation between fund
returns and expense ratios (i.e., Blake, Elton, a
nd Gruber,
1993); Malhotra

and M
acLeod, 1997;
Wermers, 2000
). More importantly, the fee negotiations are a primary role that mutual fund
directors, particularly independent directors, undertake to protect mutual fund investors’
interests. As such,
mutual fund directors could have

direct influence on the fund expense,
9


while they are less capable of controlling the fund returns (Kong and
Tang, 2008
). Adams et al.
(2010)
report that independent directors are associated with lower expense ratios
.



Third
,
we follow Elton et al (2004)

and
add back
each fund’s
expense ratio to
its

four factor
alpha to

capture managerial
performance
.

Cash polic
ies and portfolio trading activities

incur
costs to the fund that negatively impact the fund returns

so
even if there were

no

expense
s

returns among funds could

vary considerably.

Also
,
expense ratios are negotiated annually by
fund sponsors and mutual fund boards so fund managers have little input on fee levels.
Likewise,
the
custodian and distribution policies are likely set by the s
ponsor and not the fund
manager.


Bär et al. (2011) report that
team managed funds

follow less extreme investment
strategies and are

less likely to achieve extreme performance outcomes.
Likewise, Han et al.
(2008) find evidence to support their claim tha
t team managed funds follow generic trading
strategies that result in higher average returns but also make returns less i
nformative of
managerial performance
.
If
behave differently than individuals
and team
-
managed funds show
different risk
-
taking and tra
ding activities it is critical
to examine the manag
erial performance
of

team versus single funds.
A brief description of the sample variables and the data sources
used, to obtain or compute them, is presented in Table 1.



[
Insert Table 1 About H
ere]


2.5 Descriptive Statistics

Our final sample consists of
3,002

U.S. domiciled

mutual fund
s sponsored by

99 investment
companies
and covers

the period from 1999 to 2007.

Table 2 reports the distribution of single
-
managed and team
-
managed funds
in our
sample. Panel A shows the distribution

of fund
management structure
s

for each year.

In
the earlier years of the sample
,
more funds utilized a
single
-
manager structure. Beginning in 2001 and continuing for the remaining years in the
sample, the majority
of funds, about 53%, favored a team
-
management structure.
The number
of funds employing management teams increa
ses in most years

and f
rom 1999 through 2007 the
proportion of team
-
managed funds increases by about 31%.
T
his trend is consistent with

earlier studies that report
growing popularity of team
-
mana
ged funds (i.e., Bliss et. al., 2008;
Han et al., 2008
). This
value is

similar to Morningstar
’s
estimate quoted

in a recent
Wall Street
10


Journal

article
.
4

This similarity in
the
proportion
s

of tea
m
-
managed funds suggests that our
sample reflects

patterns in

fund management structures across the
universe of mutual funds
.


[
Insert Panels A and B of
Table 2 About H
ere]


Panel B o
f Table 2 reports the incidence of single and tea
m

managed funds by
investment
objective
.
It is notable that

funds with less complex investment objectives, such as sector and
municipal bond funds, are more likely than the average fund to utilize a single
-
manager. In
contrast,
funds with more complex investment objectives

(i.e.
aggressive gro
wth, balanced, and
total return funds) tend to be team
-
managed. In general, t
he

distribution

of single versus team
managed funds

seems

to indicate that team
manag
ement is more prevalent in investment
objectives that can be

characteriz
ed as

complicated and disparate, while
funds that
follow a
more focused investment strategy tend to be managed by
individual
s
.

In this respect
, o
ur
sample is similar to the one

employed by Bliss et al. (2008).

Panel C of
Table 2

provides summary
statistics for fund and family, performance, and board
of director characteristics

segmented by fund management type
.
In addition, Panel C reports
the results for tests of differences
in

the mean and median values for each variable

across the
single and t
eam managed subsamples
.
Single managed funds are sponsored by sponsors with
more assets under management than the sponsors of team managed funds. Panel C

reports
mean and median Family level TNA values are economically and statistically greater in single

managed funds (mean value of around $175 billion) than in team managed funds (mean of
about $83 billion).

The large difference in the mean and median vales reflects the nature of the
mutual fund marketplace which features a few very large fund families.

This
skewness
is also
reflected in the number of funds offered
by fund families, especially for single managed funds
where the mean number of funds offered is about 80 and the median value is 58. Of course,
interpreting family level data is problematic s
ince many families utilize both fund management
structures, some families
do not sponsor any team
-
managed funds, while others offer only
team
-
managed funds.
5





4

Please see footnote 1.

5

American funds and

Cox & Dodge are some of the examples that have been using team
-
managed funds
extensively. Janus and Fidelity have adopted team management style more recently. See the following article from
U.K. Reuters:
http://uk.reuters.com/article/2007/10/25/fidelity
-
mutualfunds
-
idUKN2522199020071025


11



[
Insert Panel
s

C & D
of Table 3 About H
ere]


Consistent with the notion that bigger funds require more managerial
resources
,
Panel C
reports that
team
-
managed funds are, on average, larger with

mean

and median

TNA

values
of

approximately $1.6 billion and $0.38

billion,
respectively. These values are

economically and
statistically greater than the TNA of single
-
managed funds.
Also, team
-
managed fund families

offer more
share
classes than single
-
managed families

suggesting their marketing and operating
policies are more complex than those of single m
anaged funds.
6

Consistent with the results

presented in Panel A that
team managed funds
are more prevalent in
later

years
, team managed
funds tend to be younger than single managed funds.
Average and median fund age
s are
higher

for the single
-
managed fun
ds than for team
-
managed funds, suggesting recently offered
funds tend to be team
-
managed

(i.e., Bliss et al. (2008) and Massa et al. (2010)).

In terms of trading activities, mean turnover ratio is greater for single
-
managed funds,
but the median value i
s higher for team
-
managed funds. However
, team
-
managed fun
ds tend
to hold more stocks and less cash

than single
-
managed funds.
Single
-
managed funds offer
significantly fewer share classes, lower front and rear loads (sales charges), and lower
institution
al ownership. Overall, the univariate statistics in Panel C suggest that single and
team managed funds differ in terms of their clienteles and that clientele effects are important
factors in deciding fund management type. This finding is supported by ane
cdotal evidence
that some investors prefer certain fund management structures. For example, it has been
reported that

Janus started to
employ
team
s in response to

institutional investors’ demand
s
.
7


Panel C also summarizes the

performance metrics employe
d in this study. Consist
ent
with Elton et al. (2004), the mean and median Carhart four factor alphas
(Alpha)
of the sample
funds are negative. Panel C also indicates there are no significant differences in the mean and
median
alphas

of single and team managed funds.
The mean and median expense ratios are
about 3 and 5 basis points lower for single
-
managed funds, differences that are significant at the
one percent level
.

However, our measure of fund manager performance, alpha plus
expense
ratio, is similar for single and team
-
managed funds. Overall, the performance characteristics in






6

We use the terms single and team managed families to indicate to families of funds offering each management
structure
and acknowledge that many families offer both single and team managed funds.

7

See the link provided for an article in footnote 5

12


Panel C do not suggest meaningful

difference
s

in performance between single
-
managed and
team
-
managed funds.

Finally,
Panel C reports
distributional

statistics for funds’ boards of directors. Panel C
reports that

boards of team managed funds are smaller, more independent, and more often
oversee all of the funds within the family complex (Unitary), results that a
re

statistically
significant at the one

and five percent levels. However,
in terms of economic significance the
differences appear small.

For example,
the differences in the median values for all of the board
characteristics are zero and the differences in the board size and independent direc
tor means
are negligible.


Panel D of Table 2 reports Pearson

correlations for our key variables of interest.
Consistent with the univariate statistics there is no significant correlation between team
management and fund performance,

measured by
alpha

o
r alpha plus expenses
. However,
team management is a positive and significant correlation with expense ratio, fund TNA,
independent director proportion, and unitary board presence. In contrast, family TNA, the
number of funds offered in the family, board

size, and independent chair are negatively and
significantly correlated with team management. Consistent with earlier studies (i.e., Adams et
al. (2010), Elton et al. (2004)), the Carhart four factor alpha is significantly and negatively
correlated with
fund’s expense ratio. Overall, t
here is a significant variation

in the fund, family,
and board characteristics between team and single funds.


3
.

Multivariate Analysis



We provide multivariate OLS regression analysis to examine the relation between fund

management structure, board governance, and performance while controlling for fund
characteristics. To test which effect, substitution or free rider, dominates we apply the
following specification while using fund level clustered standard errors

Performance
i,t

=

0
+
4
1


j
β
j
(
Board

Characterisitics
i,
t
) +
10
5


j
β
j
(
Fund
Characteristics
i,t
)

+
11
11


j
β
j
(
Public
,t

)
+ β
12

(
Investment Objective
i,t
)
+
22
13


j
β
j
(
Time
t
)

+

i,t
,



(1
)

13



where
Performance

is either the annualized Carhart

four factor alpha (Alpha) computed over the
preceding twelve months, the annual expense ratio (Expense) , or the sum of each fund’s alpha
and expense ratio (Alpha + Expense). Board Characteristics include board size, independent
directors, independent ch
air dummy, and unitary board dummy.
8

Fund Characteristics
include fund TNA, fund age, cash holdings, stock holdings, portfolio turnover, and institutional
ownership. Public is a dummy that takes on a value of 1 if the fund sponsor is publicly traded.

Th
e variables
Investment Objective

and
Time

represent
control
dummies for each fund’s
benchmark investment objective category (e.g., growth and income, aggressive growth,
international equity, etc.) and time dummies, respectively.




3.1 Single and Team
-
Ma
naged Sub
-
Sample Analysis



Table 3 presents results from regressing each of the fund performance measures (Alpha,
Expenses, and Alpha + Expenses) on b
oard and fund characteristics for

sub
-
samples of single
and team
-
managed funds in Panel A while Panel B r
eports
Chow
tests for differences in
selected estimated coefficients.
Models 1, 3, and 5 report regression coefficients for single
-
managed funds while Models 2, 4, and 6 report results for team
-
managed funds.
The
dependent variables are alpha (Models 1 a
nd 2), expense ratios (Models 3 and 4), and manager
performance, measured as alpha plus expense ratios (Models 5 and 6).


Model 1 reports the estimated coefficient for Fund TNA in the single
-
manager sample is
positive and statistically significant at the o
ne percent level. Model 2 reports similar results
Fund TNA in the team
-
managed sample. Models 3 and 4 report that Fund TNA is negatively
related to fund expense ratios, while Models 5 and 6 report that Fund TNA is positively and
significantly related to
gross expense ratios (Alpha + Expense
s). Institutional ownership
estimated coefficients are negative and significant for expenses (Models 3 and 4). In terms of
portfolio characteristics, Table 3 reports that cash holdings are positively

related to performance



8

In unreported analysis we construct a composite board structure variable. This assigns values of one when a fund’s
board is

smaller or more independent than the sample means. These values are then added to the independent chair
and unitary board dummy to create a board composite measure that ranges in magnitude from one to four. The
results from these alternative specificati
ons are similar to those reported.

14


in team
-
managed funds only (Models 2 and 6) while stock holdings are negatively re
lated to
performance for single
-
managed funds (Models 1 and 5).



Table 3 also reports that board size is negatively and significantly related to perf
ormance
for team
-
managed funds (Models 2, 4, and 6) but insignificantly related to performance for
single
-
managed funds (Models 1,3, and 5).
Independent directors are negatively related to
performance for single managed funds in Models 1 and 5 and positiv
ely related to performance
in team managed funds in Models 2 and 6. Interestingly, the estimated coefficients for
independent directors are positive and statistically significant (at the one percent level) for both
single and team
-
managed funds

in the exp
ense ratio specification (Models 3 and 4)
.

These
results differ from Adams, Mansi, and Nishikawa (2010) who report a negative relation between
board independence and expenses, however they employ a sample of index mutual funds while
we are concerned with
actively managed funds. Furthermore, the economic impact of board
independence on expenses in our sample of actively managed appears small.

The remaining
governance variables, Independent Chair, Unitary Board, and Public sponsor ownership, are
generally

significant for expenses only and have similar signs and significance levels in single
and team
-
managed funds.


Panel D lists the results from Chow tests of

differences in the estimated coefficients
(team


single) for each perfo
rmance measure
.

Panel

D
reports the

estimated coefficients for
board size and independence are significantly larger for alpha and alpha plus expenses

(e.g.,
Model 2 vs Model 1 and Model 6 vs Model 5 in Panel A)
. The tests also indicate significant
differences in
the independe
nt chair and unitary board coefficients for the expense ratio
specifications (Models 3 and 4 in Panel A). Overall, the results presented in Table 3 are
more
consistent with
boards acting to mitigate problems associated free rider problems inherent in
team
-
managed funds (i.e., the free rider hypothesis) than the hypothesis that management
teams are substitutes for board governance (i.e., the substitution hypothesis). Of course, both
effects could be present and free rider effects dominate substitution effe
cts.



3.2
Self Selection,

Teams

and Boards of Directors



Table 3 provides compelling evidence th
at the impact of board structure

on fund
performance varies significantly across single and team
-
managed funds. However,
s
election
15


bias is a concern as
managers may self
-
select single or team management structures. For
example, better managers (those whose expected performance is high) have career incentives to
favor single
-
managed funds so controlling for self
-
selection is critical. We control for thes
e
issues by

employing
Heckman (1979) selection correction and instrumental variable
approaches.


We implement the Heckman selection correction by first modeling team management

using
fund, family, and governance characteristics
.
Fund characteristics

are

the natural l
og
fund TNA
under the premise that as

fund size

increases so does the requirement for managerial
resources

and

the natural log of fund age as

Bliss et al. (2008) and Mas
sa et al. (2010) argue that
younger fund
s tend to be structured as team
s.

Our models also include

the number of

share
classes offered by a fund

and
the proportion of institutional ownership to capture clientele
effects,
cash

and stock holdings as well as portfolio turnover to
account for

differences between

single
-
managed and
team
-
managed funds
(i.e., Bär et al.
, 2011
)
. F
or completeness
, our
Heckman correction models
also include front

load and deferred load dummies

as in

(Almazan,
et al., 2004
)
.

Family c
haracteristics

include t
wo variables
; the

percentage

of team
-
funds in a family
and the numb
er of funds offered in a family. Unreported analysis
indicates that

the choice of
single vs team
-
management type reflects policy decisions at the

family level
.
Anecdotal
evidence supports this finding.
For example,
American Fund
s

and Cox & Dodge

are reportedly
strongly advocates of team management structures.
9

The number of funds offered in a family is
included to capture the potential peer monitoring effect
in Almazan et al. (2004) that occurs on
large

fund fa
milies.

Increased peer monitoring

suggests that there is less incentive for larger
fund families to employ team management

structures
.
10
,
11


The g
overnance
variables are
the natural log of board size,
the

proportion of
independent

directors,
a uni
tary board dummy,

independent chair

dummy
, and
a
dummy for
publicly
-
sponsored funds.
We also include

fund
-
objec
tive and year fixed effects
. We
then



9

Need a article cite here for the dodge and cox and American funds sentence……

10

Indeed, Massa et al. (2010) argues that the fund family uses team management as an effort to avoid falling victims
to “star” who leave the fund.

11

Because of very high correlation between log of family TNA and the number of funds offered in a family,
fo
llowing Han et al. (2008), we use the number of funds in a family in the model. We also use log of family TNA
instead of the number of funds offered in a fund family. The results do not change.

16


calculate the inverse Mill’
s ratio

(
Lambda
)
for each fund

and include it in

regression models to
control
for self
-
selection effects
.




3.2 Empirical results


The results from OLS regression a
nalyses

employing the Heckman approach

are
presented in Table 4

which
includes three panels to separately report the results for each of the
three performance measures
. Panels A, B, and C report regression results when Carhart’s four
factor alp
h
a (
Alpha
), expense ratio

(
Expense Ratio
)
, and the sum of alpha and expense ratio

(
Alpha + Expense
)

are the dependent variables, respectively.
Model 1 of each panel is the base
model while Models 2 through 5 interact the team dummy with board size, the percentage of
independent directors, the independent chair dummy, and the unitary board dummy
.
In all
models, the inverse Mill’s ratio

(
Lamba
)

is significant, suggesting that the
consideration of self
-
selection bias is important.

Panel A of Table 4

reports estimated
coefficients when alpha is the dependent variable.
Model 1 reports that the team managed dummy is not significantly related to alpha.

This result
is consistent with
Bliss et al. (2008) and Prather and Middleton (2002).
Our results show that
overall fund performance is not significantly related to the governance issues (i.e., Kong and
Tang (2008)).
Model 1 also reports positive and statistically significant (at the o
ne percent level)
results for Fund TNA suggesting economies of scale benefits. Alternatively, larger funds may
employ managers with superior security selection skills. Fund Age is negatively related to
alpha, results that are consistent with Evans (2010)

who notes new funds ar
e often incubated in
that they not offered for sale until they are successful.

T
he estimated coefficient for
institutional ownership is positive and significant (at the five percent level) suggesting that
institutional investors pr
ovide monitoring that improves performance and/or institutional
investors are performance sensitive and seek better performing funds.
Cash holdings and stock
holdin
gs also significantly affect

fund performance. Consistent with Khorana (2001), larger
turno
ver by funds leads to underperformance, suggesting managers of underperforming funds
tend to trade excessively.
None of the governance estimated coefficients are statistically
significant

suggesting that on average select optimal governance schemes.

17


Mor
e interestin
g findings are presented M
odels

2 through 5 that include
interactions of

governance measures and
the
team management dummy.
The estimated coefficients for the
fund characteristics Models 2 though 5 have similar signs and significance to those reported in
Model 1. Model 2 notes that the
board size and team management dummy interaction term is
negative and significant (at the fi
ve percent level), results that are consistent with those reported
in Table 3. This finding indicates that smaller boards are associated with superior performance
in team managed, but not single managed funds. The estimated

coefficient for the independen
t
director


team managed dummy is positive and significant in Model 3, suggesting that more
independent boards are associated with better performance in team managed funds but not in
single managed funds. The remaining governance measure interactions wit
h team
management, independent chair dummy and unitary board dummy, are not significantly
related to alpha. Overall, the results in Panel A suggest that board factors commonly associated
with improved monitoring (e.g. smaller and more independent boards)
appear to be effective in
team managed funds where free rider
problems can be

severe.

Panel B of Table 4

summarizes the resul
ts using expense ratios
as
the
performance
measure. The mutual fund investors can enhance their returns by lowering expense of t
he
funds. Since the primary role of fund boards is to negotiate the fees with investment
companies, the expense ratio is an important
operational
performance measure.
Model 1
reports that w
hile the team dummy is negative, it is not significant, indicatin
g that the team
-
managed
funds do not charge less fees

than single
-
managed funds.
The
governance
characteristics are all significantly

related to expense ratio
s.
Interestingly, Model 1 reports a
positive relation between board independence and fun
d

expens
e ratios (significant at the one
percent level), results that differ

from Adams et al (2010) who

document an inverse relation
between independent directors and expense ratios. However, our study focuses on actively
managed funds while they examine passive
ly managed funds. In addition, although the
estimated coefficient on independent directors is positive and statistically significant it is
economically very small.

Model 2 reports the interaction term between board size and team management is
positive a
nd statistically significant at the one percent level. The board size

team management
coefficient is economically significant as well with a one standard deviation in board size
resulting in a
19 basis point

increase in fees for team managed funds. Thi
s is in line with the
18


notion that bigger boards are less effective monitors (i.e.
,

Yermack, 1996), and evidence that
team
-
managed funds’ performance is critically impacted by governance structure. In contrast
to Panel A, the interaction term between indep
endent directors and teams is insignificant on
Model 3. Also unlike Panel A, the independent chair and team interaction term is positive and
significant.

Finally, following Elton et al. (2004), we ex
amine the fund management skill

by adding
back

the expense

ratio to
the
Ca
r
hart four factor alpha
(Alpha)
fo
r each fund in Panel C of Table
4. In general, the signs and significance levels are similar to those reported for alpha in Panel A.
For example,
Model 1 reports that the team managed dummy co
efficient is insignificant. Model
2 shows the interaction term between board size and team management status in negative and
significant at the 10 percent level. Model 3 reports a positive and significant estimated
coefficient on the director independenc
e and team ma
nagement dummy interaction term
.

Overall,

our findings in Tables 3 and 4 show
s
ignificant influence of board

governance
mechanisms on
performance in
team
-
managed

but not single
-
managed funds.

The free rider
hypothesis predicts that strong g
overnance mechanisms are necessary to alleviate potential free
rider problems among fund management team members. The substitution hypothesis argues
that team members serve to moderate extreme investment decisions and enhance managerial
skills so that eff
ective board governance in single but not team
-
managed funds.

Our
findings
,
that small and independent boards
are
positively associated with performance in team but not
single
-
managed funds, are more
consistent with
the free rider hypothesis than the sub
stitution
hypothesis.

Of course, both effects could be present and potential free rider problems in team
-
managed funds outweigh any substitution in governance benefits they provide.
Give
n the
trend in recent years towards more

team managed
-
funds
our ex
amination

highlights the
importance of boards in
protect
ing

the interests of the mutual fund investors.



3.3 Robustness Check
s


Our results from the previous section show the importance of internal governance on
team
-
managed funds’ performance. However, it is possible our findings are driven by the
changes in the management organization through time.
For example, sponsors commonl
y use
management teams to mentor new managers prior to planned departures of senior managers.
19


This practice results in some funds
temporarily

changing
management structure

(e.g. single
managed funds become team managed during a mentoring period)
. To mak
e sure our findings
are not driven by fun
d

organizational changes, we exclude
funds that change management style
in the preceding year
.
12


Table 5 employs the restricted sample to repeat the analysis presented in Panel C of
Table 4.
The sample is reduced
by almost 3,000 fund
-
year

observations
, suggesting that
changes in management structure are
common in the mutual fund industry
.
Unlike Panel C of
Table 4, the interaction term between board size and team management structure is not
statistically significa
nt.
However
, the

estimated coefficient on the independent director and
team management interaction term is positive and statistically significant at the one percent
level. Furthermore, this finding is consistent with the results presents with the unrestr
icted
sample in Table 4. Although not reported, we repeat the specifications in Panels A and B of
Table 4 (e.g. employing alpha and the expense ratios as separate dependent variables) and find
similar results. Overall, the results presented in Table 5 su
ggest that frequent changes in
management structure are not driving our results.



Our analysis suggests the choice of board composition and structure causes returns in
team managed funds to be relatively high or low. It is also possible that causality runs in the
other direction, that prior performance leads funds to adopt certain gover
nance structures (see
e.g. Hermalin and Weisbach, 2003).
13

The level of monitoring effort exerted by individual
directors and the propensity of both fund sponsors and directors to actively promote
shareholder interests are important but unobservable determ
inates of mutual fund returns. If
board structure is correlated with these unobservable factors the board size and independence
variables are endogenous. I
n addition to employing year fixed effects,
in Table
6 we

employ a

two stage least squares
approach

and use
the percentage of a sponsor’s funds that are team
managed as an exogenous first stage instrument. The percentage of sponsor funds that are
team managed is a plausible instrument since it satisfies both the relevance (e.g., the decision to
employ

teams is typically made by the sponsor) and the untestable exclusion (only influences
returns via its effect on the likelihood of a fund to be single or team managed) conditions.




12

We also consider two
-
year restriction. Though not repor
ted, using the funds that do not change management
structure in the prior two year period, we find the same results.

13

However, mutual fund boards typically oversee multiple funds (a single board oversees all of the funds within an
investment trust and ma
ny investment companies sponsor multiple trusts) making it unlikely that the performance of
individual funds causes changes in board structure.


20


Table 6 repeats the analysis presented in Panel C of Table 4. The estimat
ed coefficient on the
board size and team management structure interaction term is negative and statistically
significant at the ten percent level. This finding is similar to Panel C of Table 4 and favors the
hypotheses that free rider problems are more p
roblematic in teams where board myopia is more
likely (e.g. with larger boards). Table 6 also reports that the impact of independent directors on
fund performance is more pronounced in team managed funds (the independent
directors/team management interact
ion term is positive and statistically significant at the one
percent level). Alternatively, team managed funds benefits more from independent director
monitoring than do single managed funds. Either interpretation favors the free rider over the
substitu
tion hypothesis. Unlike Table 4, the estimated coefficient on th
e independent director
and team interaction term is negative and statistically significant at the five percent level.
However, the results suggest an economically small effect.

Finally,
a
s an alternative approach to examine the effect of fund governance characteristics

and team management structures on

fund performance,
in the spirit of
Gompers, Ishii, and
Metrick (2003)

we create an index score for the governance strength of each fund.

First, a value
of 1 is assigned to a fund whose board has greater independe
nt director

representation than the
overall
median

or otherwise 0. Second, a value of 1 is assigned to fund whose board
size is
smaller

than the overall median value, or otherwise

0. The third value is the indicator variable
used to capture the presence of an independent chair, while the fourth and the last measure is
the indicator variable used to capture the presence of a unitary board. We sum these indicator
variables to creat
e an index that has the maximum number of 4 and the minimum of 0, whereas
the 4
being

the strongest
hypothesized
governance
measure
and 0 being the weakest. Using
this index, we rerun the Heckman’s approach, and report the second stage OLS model in Table
7
. The results are consistent with the evidence from
our primary specifications.



4
.

Conclusion


The mutual fund industry has witnessed an increasing popularity of team managed
funds in recent years. While the benefits of team management have been recog
nized, there also
have been potential drawbacks related to team management. Indeed, much of the finance
literature examined the performance of team managed funds has failed to reveal relative
outperformance of team managed funds over single managed funds.

If there is no apparent
21


superior performance, why has the mutual fund been experiencing this trend in shifting to
more team managed funds?
This paper sheds some light on this apparent puzzle why there has
been an increasing trend in the use of team mana
gement in mutual fund management, while no
superior performance is recognized for this type of management.


Using the funds offered by 99 largest fund families for the period from 1999 to 2007, we
test the performance of team managed funds with single m
anaged funds. The results are
similar to the earlier studies that there is no significant difference in performance between these
two organizational forms. However, when we control for potential governance moderating
factors in our models, we find evidenc
e that is more consistent with the notion that the
drawback of the team management (e.g., free
-
rider problem) dominates the benefit of team
management. This means that rather than being a potential governance mechanism, the team
management could pose a pr
oblem on its performance, measured by the alpha and expenses in
our study, suggesting that not all team managed funds are alike. Our results show that the
team managed funds with highly independent boards
have significantly lower expense than
team
-
manage
d funds with
less independent
boards
. Our results are robust to alternate
specifications of governance measures.


Our finding complements a growing literature that examines the performance of team
management in organizations through its benefits and dra
wbacks.

Although there has been an
increasing number of

research

exploring

the team phenomenon, few have succeeded to provide
a direct analysis of teams

because “team” can be defined loosely in the literature
.

Unlike these
earlier studies, we use team more firmly by using the mutual fund teams, thus our results are
cleaner. However, we acknowledge that there needs to be more research to enhance our
understanding of team management.




22


Reference

Adams, J., M
ansi,
S.,
Nishikawa,
T.,
2010. Internal governance mechanisms and operational
performance: Evidence from index mutual funds. Review of Financial Studies 23, 1261
-
1286
.

Almazan, A., Brown, K., Carlson, M., Chapman, D., 2004. Why constraint your mutual fund

manager? Journal of Financial Economics 73, 289
-
321.


Arena, M., Ferris, S., Unlu, E., 2011. It takes two: The incidence and effectiveness of co
-
CEOs.
Financial Review Forthcoming.


Bär, M., Kempf, A., Ruenzi, S., 2011.

Is a team different from the sum of its parts? Evidence
from mutual fund managers. Review of Finance 15, 359
-
396.


Blake, C., Elton, E., Gruber, M., 1993. The performance of bond mutual funds. Journal of
Business 66, 370
-
403.


Bliss,
R., Potter, M.,
Schwarz, C., 2008. Performance characteristics of individually
-
managed vs.
team managed mutual funds. Journal of Portfolio Management 374, 110
-

119.


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

Chen, J.,
Har
rison, H., Ming, H., Kubik, J., 2004. Does fund size erode mutual fund
performance? The r
o
le of liquidity and organization. American Economic Review 94,
1276
-
1302.


Chevalier, J., Ellison, G., 1999. Career concerns of mutual fund managers. Qua
r
terly Journa
l of
Economics 114, 389
-
432.


Elton, E.
, Gruber, M., Busse, J., 2004.
Are investors rational? Choices among index funds.
Journal of Finance 59, 261
-
288.


Fama, E., 1980. Agency problems and the theory of the firm. Journal of Political Economy 88,
288
-
307.


Gompers, P., J. Ishii, and A. Metrick, 2003, Corporate Governance and Equity

Prices, Quarterly Journal of Economics 118, 107
-
155.


Investment Company Institute (ICI), 2011. 2011 Investment Company Fact Book. Investment
Company Institute, Washington, DC.


Han, Y., Noe, T., Rebello, M., 2008. Horses for courses: fund managers and organizational
structures. Working paper, available at SSRN
http://ssrn.com/abstract=1100676
.


Hill.,

G. 1982. Group vs. individual performance: are N+1 heads better than one? Psychological
Bulletin 91, 783
-
810.


23


Hermalin, B. and M. Weisbach. 2003. Boards of Directors as an Endogenously Determined
Institution: A Survey of the Economic Evidence.
Economic
Policy Review

9: 7
-
26.



Herrenkohl, R., 2004. Becoming a team. Cincinnati, OH, South
-
Western.


Holmström
, B., 1982. Moral hazard in teams. The Bell Journal of Economics 13, 324
-
340.


Karagiannidis, I., 2010. Management team structure and mutual fund perf
ormance. Journal of
International Financial Markets, Institutions & Money 20, 197
-
211.


Khorana, A., 2001. Performance changes following top management turnover: Evidence from
open
-
end mutual funds. Journal of Financial and Quantitative Analysis 36, 371
-
39
3.

Khorana, A
., P. Tufano and L. Wedge, 2007
. Board structure, mergers, and shareholder wealth:
A study of the mutual fund industry. Journal of Financial Economics 85, 571
-
598.

Kong, S., tang, D., 2008. Unitary boards and mutual fund governance. Journal o
f Financial
research 31, 193
-
224.

Massa, M., Reuter, J., Zitzewitz, E., 2010. When should firms share credit with employees?
Evidence from anonymously managed mutual funds. Journal of Financial Economics 95,
400
-
424.

Malhotra, D., MacLeod, R., 1997. An emp
irical examination of mutual fund expenses. Journal of
Financial Research 20, 175
-
190.

Petersen, M., 2009. Estimating standard errors in finance panel data sets: Comparing
approaches. Review of Financial Studies 22, 435
-
480.

Prather, L., Middleton, K., 2002. Are N+1 heads better than one? The case of mutual fund
managers. Journal of Economic Behavior & Organization 47, 103
-
120.

Q
iu, J., 2003. Termination risk, multiple managers and mutual fund tournaments. European
Finance Re
view 7, 161
-
190.

Sah, R., Stiglitz, J., 1988. Committees, hierarchies and polyarchies. The Economic Journal 98,
451
-
470.


Sharpe, W., 1981. Decentralized investment management. Journal of Finance 36, 217
-
234.


Sirri, E. and P. Tufano, 1998. Costly search

and mutual fund flows
.

Journal of Finance, 53, 1589
-
1622.

Stock, R., 2004. Drivers of team performance: what do we know and what have we still to learn?
Schmalenbach Business review 56, 274
-
306.

24


Wermers, R., 2000. Mutual fund performance: an empirical dec
omposition into stock
-
picking
talent, style, transactions costs, and expenses. Journal of Finance 55, 1655
-
1695.

Yermack, D., 1996. Higher market valuation of companies with a small board of directors.
Journal of Financial Economics 40, 185
-
202.



25


Table 1

Variable d
efinitions

and primary s
ource*


Variable

Data Source

Explanation




Managerial Organization




Team Managed

Morningstar

Dummy
that equals one if the fund has more than one manager




Performance and Risk




Carhart 4 Factor Alpha

Morningstar/CRSP

Computed using 1 year of monthly data. See Carhart (1997) for details.


Expense Ratio

Morningstar/CRSP

P
ercentage of assets used
to pay for operating expenses, including 12b
-
1 fees,

management

and
administrative fees, and other asset
-
based costs incurred except for sales charges.


Alpha plus expense

Morningstar/CRSP

Carhart 4 factor alpha plus expense ratio




Fund
&

Family
Characteristics




Fund and Family
TNA

Morningstar/CRSP

Log of total net assets of fund
or family.


Fund Age

Morningstar/CRSP

The number of years since the fund’s oldest class inception


Front Load

Morningstar/CRSP

Dummy that equals one
if a fund has a sales charge at initial fund purchase, not included
in expense ratio in percentage


Rear Load

Morningstar/CRSP

Dummy that equals one if a fund has a redemption charge, not included in expense ratio
in percentage


Institutional Holding

Morningstar/CRSP

Percentage of institutional class holdings in fund



Portfolio

Turnover

Morningstar/CRSP

Trading activity/change in portfolio holdings computed as the lesser of sales or purchases
divided by average monthly t
otal net assets
in percentage


Funds in a Family

Morningstar/CRSP

The number of funds under a mutual fund family


Percentage of Team

Morningstar/CRSP

The percentage of funds in a family organized as a team


Stock Holdings

Morningstar/CRSP

Percentage of TNA held in Common Stocks


Cash Holdings

Morningstar/CRSP

Percentage of TNA held in Cash


Classes in a Fund

Morningstar/CRSP

The Number of Classes offered in a fund







Governance Characteristics




Board Size

Form 485

Log of
number of directors on fund board


Independent Directors

Form 485

Proportion of directors who are classified as
outsiders (
independent
)


Independent Chair

Form 485

Dummy that equals one if the board chair is an independent director


Unitary Board Structure

Form 485

Indicator variable if a single board oversees all funds managed by sponsor.




Note
: For fund with multiple share classes we compute the weighted average value (using TNA of each class), where the reported fu
nd TNA
is the sum of the
TNA from all classes. All Morningstar data are cross
-

checked or recomputed with the CRSP Mutual Fund database.


26


Table
2

Distribution of team managed funds


Panel A
:
Distribution of fund management structures by year



Number

%

Years

Single

Team

Single

Team






1999

776

683

53.19

46.81

2000

891

886

50.14

49.86

2001

932

1,037

47.33

52.67

2002

1,054

1,213

46.49

53.51

2003

1,138

1,439

44.16

55.84

2004

1,053

1,401

42.91

57.09

2005

1,063

1,577

40.27

59.73

2006

936

1,490

38.58

61.42

2007

925

1,454

38.88

61.12






Total

8,768

11,180

43.95

56.05

Note
: Panel A

reports the
number and
percentage of
samples
by year for each
type of fund management
structure
.




27


Panel
B
:
Distribution of team
-
managed funds by objectives


Objective Title

Single

Team


Obs.

(%)

Obs.

(%)

Aggressive Growth

729

37.29

1,226

62.71

Balanced

187

21.27

692

78.73

Global Bond

14

4.18

321

95.82

Global Equity

174

29.59

414

70.41

Government Security

326

40.50

479

59.50

Income

149

47.60

164

52.40

International Equity

759

41.84

1,055

58.16

Ginnie Mae Fund

165

43.88

211

56.12

Growth and Income

545

38.06

887

61.94

High Quality Bond

553

40.93

798

59.07

High Quality Municipal

576

52.65

518

47.35

High Yield Bond

230

46.28

267

53.72

Long Term Growth

1,276

44.46

1,594

55.54

Sector Fund

770

57.72

564

42.28

Single State Municipal

1,739

56.50

1,339

43.50

Total Return

230

29.60

547

70.40

Others

146

58.40

104

41.60






Observations

8,768

43.95

11,180

56.05

Note
: Panel B

summarizes

the
number and
percentage of
samples
by
investment objectives

for each
type
of fund management organization
.

Others include precious metal funds, utility funds, and special funds.






28


Panel C
:
Descriptive statistics











Single

Team

Tests of Difference


Mean

Median

St
and. Dev.

Mean

Median

St
and.
Dev
.

Mean

Median










Family and Fund
Characteristics










Family TNA

174,544

49,673

283,531

83,212

41,330

159,180

91,332
***

8,343
***



Number Family Funds

80.219

58.000

66.810

53.302

49.000

31.622

26.917
***

9.000
***


TNA

1,485

320

4,813

1,649

376

6,108

-
164
**

-
56
***


Fund Age

14.632

12.340

10.90

14.424

11.414

11.64

0.208

0.926
***


Portfolio

Turnover

93.739

57.000

127.520

92.228

62.000

116.382

1.511

-
5.000
***


Stock Holdings

50.524

76.915

46.819

55.673

83.075

45.116

-
5.149
***

-
6.160
***


Cash Holdings

3.674

2.200

6.360

3.456

2.000

6.698

0.218
**

0.200
***


Number of Class

2.880

3.000

1.705

3.477

4.000

2.010

-
0.597
***

-
1.000
***


Front Load

0.476

0.000

0.499

0.613

1.000

0.487

-
0.137
***

-
1.000
***


Rear Load

0.465

0.000

0.499

0.582

1.000

0.493

-
0.117
***

-
1.000
***


Institutional Holdings

15.867

0.000

30.947

24.114

1.405

36.554

-
8.364
***

-
1.405
***


Public

0.674

1.00

0.469

0.779

1.000

0.415

-
0.105
***

0.000
***










Performance
Characteristics










Alpha

-
0.649

-
0.805

9.917

-
0.775

-
0.864

8.386

0.126

0.059


Expense Ratio

1.118

1.038

0.487

1.147

1.090

0.502

-
0.029
***

-
0.052
***


Alpha + Expense

0.469

0.142

9.915

0.372

0.113

8.374

0.097

0.029










Board
Characteristics










Board Size

9.188

9.000

2.722

9.099

9.000

2.477

0.089
**

0.000


Independent Directors

80.044

80.000

11.412

80.516

80.000

10.831

-
0.472
***

0.000
***


Independent Chair

47.183

0.000

49.923

43.184

0.000

49.535

3.999

***

0.000
***


Unitary Board

36.234

0.000

48.070

42.710

0.000

49.468

-
6.476
***

0.000
***











Observations


8,768



11,180





Note
: This table
report
s

descriptive statistics

for the overall fund
samples from 1999 through 2007, segmented by fund management
organizations
. All variables
are gathered or computed
a
t the end of each year from 1999 through 2007
.

Definition of variables presented in this table can be found in Table 1. The last two
columns report the tests of difference. The notations ***, **, *

denote statistical significance at the 1%, 5%, and 10% levels respectively.



29


Panel D
:
Pearson Correlations (need alpha +expenses in here)



Team
Managed




Alpha

Expense
Ratio



Alpha +
Expense

Family

TNA



Number
of Funds

Fund

TNA

Board

Size


Indep
.

Directors

Unitary
Board












Alpha

-
0.007





















Expense Ratio

0.029
***

-
0.0
43
***




















Alpha + Expense

-
0.005

0.999
***

0.012
*



















Family TNA

-
0.200
***

0.0
31
***

-
0.178
***

0.021
**


















Number of Funds

-
0.257
***

0.026
***

-
0.073
***

0.022
***

0.792
***

















Fund

TNA

0.015
**

0.0
22
***

-
0.122
***

0.015
**

0.301
***

0.094
***
















Board Size

-
0.017
**

0.008

0.098
***

0.013
*

0.293
***

0.33
1
***

0.090
***















Independent

0.021
***

0.015
**

0.043
***

-
0.013
*

0.019
**

0.152
***

-
0.017
**

0.008














Unitary Board

0.065
***

-
0.009

-
0.08
3
***

-
0.014
*

-
0.118
***

-
0.262
***

-
0.036
***

0.038
***

0.045
***













Indep. Chair

-
0.04
0
***

-
0.009

0.027
***

-
0.007

0.221
***

0.348
***

0.029
***

0.217
***

0.389
***

0.14
8
***












Note
:
This table

reports correlation statistics for our sample.
The notation
***,**,*,

represent significance at the 1 and 5% levels, respectively
. Variable definitions are
provided in Table 1


30


Table 3

Performance in Team and Single Managed Funds


Dependent Variables

Alpha


Expenses


Alpha + Expenses



Single

Team


Single

Team


Single

Team



(1)

(2)


(3)

(4)


(5)

(6)


Intercept

1.396

(0.481)

-
1.748

(0.145)


1.433
***

(0.000)

0.970
***

(0.000)


2.829
*

(0.099)

-
0.778

(0.509)


F
und TNA

0.556
***

(0.000)

0.385
***

(0.000)


-
0.088
***

(0.000)

-
0.063
***

(0.000)


0.468
***

(0.000)

0.322
***

(0.000)


Fund A
ge

-
0.722
***

(0.000)

-
0.257
*

(0.079)


0.001

(0.935)

-
0.050
***

(0.000)


-
0.720
***

(0.000)

-
0.307
**

(0.034)


Institutional

0.002

(0.381)

0.003
*

(0.069)


-
0.003
***

(0.000)

-
0.004
***

(0.000)


-
0.001

(0.610)

-
0.001

(0.660)


Cash Holdings

0.006

(0.633)

0.042
***

(0.000)


-
0.002

(0.106)

-
0.001

(0.555)


0.004

(0.727)

0.041
***

(0.000)


Stock Holdings

-
0.015
**

(0.040)

-
0.005

(0.391)


-
0.001

(0.687)

-
0.001

(0.853)


-
0.016
**

(0.037)

-
0.005

(0.382)


Turnover

-
0.002
*

(0.093)

-
0.001

(0.131)


0.001
**

(0.023)

0.001
***

(0.000)


-
0.001

(0.120)

-
0.001

(0.312)


Board S
ize

-
0.049

(0.875)

-
0.637
**

(0.019)


0.016

(0.564)

0.173
***

(0.000)


-
0.033

(0.915)

-
0.464
*

(0.088)


Indep. D
irector
s

-
0.018
**

(0.019)

0.011
*

(0.095)


0.003
***

(0.009)

0.003
***

(0.000)


-
0.015
*

(0.051)

0.015
**

(0.028)


Independent C
hair

0.003

(0.121)

-
0.002

(0.187)


0.001

(0.526)

0.001
***

(0.001)


0.003

(0.106)

-
0.002

(0.302)


Unitary B
oard

0.001

(0.898)

-
0.001

(0.888)


-
0.001
***

(0.009)

-
0.001
***

(0.000)


-
0.001

(0.959)

-
0.001

(0.548)


Public

-
0.212

(0.268)

0.036

(0.835)


0.150
***

(0.000)

0.119
***

(0.000)


-
0.067

(0.731)

0.154

(0.362)












Objective dummy

Yes

Yes


Yes

Yes


Yes

Yes


Year dummy

Yes

Yes


Yes

Yes


Yes

Yes












Ajdusted
-
R
2

0.069

0.068


0.486

0.484


0.069

0.068


N

8,768

11,180


8,768

11,180


8,768

11,180



B: Chow tests for differences in coefficients in Single and Team

managed funds. (p
-
values)



Alpha

Expenses

Alpha + Expenses

All

0.002
***

0.000
***

0.001
***

All Governance
-
related

0.001
***

0.000
***

0.002
***

Board Size


0.013
**

0.000
***


0.046
**

Independent Directors

0.001
***


0.399

0.001
***

Independent Chair


0.497

0.003
***


0.656

Unitary


0.994



0.092
*


0.901





Note
: This table presents ordinary least square regressions of fund performance on fund management organizations.
The data covers the period from 1999 to 2007 for
3,002

funds. The dependent variable
s are
Ca
r
hart four factor alpha

(Alpha), the annual expense r
atio (Expenses), and the sum of Carhart alpha and expense ratio (Alpha + Expenses)
,
all
are
calculated each year. The independent variables include log of fund TNA, (log fund TNA), log of fund age, (Log
fund age), the number of classes offered in a fund (
Number of class), the percentage of institutional class in a fund
31


(Institutional), the percentage of cash a fund holds (Cash holdings), the percentage of stock a fund holds (Stock
holdings), approximate percentage of fund holdings that changed over the pre
ceding year (Turnover), proportion of
outside directors (Independent Directors), log of number of directors on fund board (Board Size), a dummy variable
that equals one if the board is comprised as unitary board (Unitary board), and a dummy variable that e
quals one if
the board chair is an independent director (Independent Chair). P
-
values derived from fund
-
level clustered robust
standard errors are reported in the third column. The notations ***, **, * denote statistical significance at the 1%, 5%,
and 10
% levels respectively.





32


Table
4

Teams and Boards of Directors


Panel A
:

Depend
e
nt variable =
Alpha


Base

Board Size

Indep. Dir.

Indep. Chair

Unitary


(1)

(2)

(3)

(4)

(5)

Intercept

-
0.830

(0.398)

-
1.961
*

(0.070)

0.520

(0.627)

-
0.823

(0.402)

-
0.833

(0.399)

Lambda

0.272
**

(0.034)

0.249
*

(0.053)

0.256
**

(0.047)

0.269
**

(0.036)

0.272
**

(0.035)

Team

0.065

(0.657)

2.017
**

(0.016)

-
2.298
***

(0.004)

0.126

(0.495)

0.069

(0.701)

F
und TNA

0.463
***

(0.000)

0.464
***

(0.000)

0.464
***

(0.000)

0.464
***

(0.000)

0.463
***

(0.000)

Fund Age

-
0.455
***

(0.000)

-
0.460
***

(0.000)

-
0.457
***

(0.000)

-
0.457
***

(0.000)

-
0.455
***

(0.000)

Institutional

0.003
**

(0.039)

0.003
*

(0.053)

0.003
**

(0.039)

0.003
**

(0.038)

0.003
**

(0.039)

Cash Holdings

0.026
***

(0.001
)

0.027
***

(0.001
)

0.026
***

(0.001)

0.026
***

(0.001)

0.026
***

(0.001
)

Stock Holdings

-
0.009
**

(0.048)

-
0.009
**

(0.048)

-
0.009
**

(0.044)

-
0.009
**

(0.049)

-
0.009
**

(0.048)

Turnover

-
0.001
**

(0.039)

-
0.001
**

(0.041)

-
0.001
**

(0.047)

-
0.001
**

(0.040)

-
0.001
**

(0.039)

Indep. Director

-
0.002

(0.703)

-
0.001

(0.835)

-
0.018
**

(0.014)

-
0.002

(0.668)

-
0.002

(0.705)

Board S
ize

-
0.262

(0.182)

0.227

(0.427)

-
0.283

(0.148)

-
0.276

(0.163)

-
0.002

(0.182)

Independent
C
hair

0.001

(0.947)

-
0.001

(0.842)

0.001

(0.815)

0.001

(0.637)

0.001

(0.947)

Unitary
B
oard

0.001

(0.572)

0.001

(0.533)

0.001

(0.800)

0.001

(0.544)

0.001

(0.722)

Public

-
0.137

(0.265)

-
0.092

(0.463)

-
0.145

(0.238)

-
0.126

(0.308)

-
0.137

(0.265)

Board S
ize*Team


-
0.910
**

(0.019)




Indep Dir
*Team



0.029
***

(0.002)



Indep. Chair*Team




-
0.001

(0.544)


Unitary*Team





-
0.001

(0.975)







Ajdusted
-
R
2

0.066

0.066

0.066

0.066

0.066

N

19,948

19,948

19,948

19,948

19,948

Note
: This table presents ordinary least square regressions of fund performance on fund management organizations.
The data covers the period from 1999 to 2007 for
3,002

funds. The dependent variable is Ca
r
hart four factor alpha,
calculated each year. The ind
ependent variables include log of fund TNA, (log fund TNA), log of fund age, (Log
fund age), the number of classes offered in a fund (Number of class), the percentage of institutional class in a fund
(Institutional), the percentage of cash a fund holds (Ca
sh holdings), the percentage of stock a fund holds (Stock
33


holdings), approximate percentage of fund holdings that changed over the preceding year (Turnover), proportion of
outside directors (Independent Directors), log of number of directors on fund board
(Board Size), a dummy variable
that equals one if the board is comprised as unitary board (Unitary board), and a dummy variable that equals one if
the board chair is an independent director (Independent Chair).
The lambda, the inverse Mill’s ratio calcula
ted based
on the probit model in Table 6, is added to control for endogeneity in each odd column of the results.
All models
include year and investment objective fixed effects. P
-
values derived from fund
-
level clustered robust standard
errors are reporte
d in the third column. The notations ***, **, * denote statistical significance at the 1%, 5%, and 10%
levels respectively.




34


Panel
B
:


Dependent

variable =
Expense Ratio


Base

Board Size

Indep. Dir.

Indep. Chair

Unitary


(1)

(2)

(3)

(4)

(5)

Intercept

1.315

***

(0.000)

1.534
***

(0.000)

1.371
***

(0.000)

1.312
***

(0.000)

1.305
***

(0.000)

Lambda

-
0.093
***

(0.000)

-
0.089
***

(0.000)

-
0.094
***

(0.000)

-
0.092
***

(0.000)

-
0.093
***

(0.000)

Team

-
0.019

(0.135)

-
0.396
***

(0.000)

-
0.117

(0.178)

-
0.045
***

(0.000)

-
0.008

(0.590)

F
und TNA

-
0.077
***

(0.000)

-
0.077
***

(0.000)

-
0.078
***

(0.000)

-
0.077
***

(0.000)

-
0.077
***

(0.000)

Fund Age

-
0.023
**

(0.042)

-
0.023
**

(0.050)

-
0.024
**

(0.041)

-
0.023
**

(0.047)

-
0.024
**

(0.040)

Institutional

-
0.004
***

(0.000)

-
0.004
***

(0.000)

-
0.004
***

(0.000)

-
0.004
***

(0.000)

-
0.004
***

(0.000)

Cash Holdings

-
0.001

(0.491
)

-
0.001

(0.413
)

-
0.001

(0.488
)

-
0.001

(0.487
)

-
0.001

(0.483
)

Stock Holdings

-
0.001

(0.553)

-
0.001

(0.550)

-
0.001

(0.537)

-
0.0
0
1

(0.544
)

-
0.0
0
1

(0.
562
)

Turnover

0.001
***

(0.000)

0.001
***

(0.000)

0.001
***

(0.000)

0.001
***

(0.000)

0.001
***

(0.000)

Indep. Director

0.003
***

(0.000)

0.003
***

(0.000)

0.002
**

(0.020)

0.003
***

(0.000)

0.003
***

(0.000)

Board S
ize

0.089
***

(0.000)

-
0.005

(0.832)

0.088
***

(0.000)

0.095
***

(0.000)

0.089
***

(0.000)

Independent
C
hair

0.001
***

(0.003)

0.001
***

(0.001)

0.001
***

(0.003)

-
0.000

(0.964)

0.001
***

(0.003)

Unitary
B
oard

-
0.001
***

(0.000)

-
0.001
***

(0.000)

-
0.001
***

(0.000)

-
0.001
***

(0.000)

-
0.001
***

(0.004)

Public

0.130
***

(0.000)

0.121
***

(0.000)

0.130
***

(0.000)

0.125
***

(0.000)

0.130
***

(0.000)

Board S
ize*Team


0.176
***

(0.000)




Indep Dir
*Team



0.001

(0.241)



Indep. Chair*Team




0.001
***

(0.001)


Unitary*Team





-
0.000

(0.117)







Ajdusted
-
R
2

0.483

0.485

0.483

0.484

0.483

N

19,948

19,948

19,948

19,948

19,948

Note
: This table presents ordinary least square regressions of fund performance on fund management organizations.
The data covers the period from 1999 to 2007 for
3,002

funds. The dependent variable is
expense ratio
. The
independent variables include log of fund TNA, (log fund TNA), log of fund age, (Log fund age), the number of
classes offered in a fund (Number of class), the percentage of institutional class in a fund (Institutional), the
percentage of cash a fund

holds (Cash holdings), the percentage of stock a fund holds (Stock holdings), approximate
percentage of fund holdings that changed over the preceding year (Turnover), proportion of outside directors
(Independent Directors), log of number of directors on f
und board (Board Size), a dummy variable that equals one if
the board is comprised as unitary board (Unitary board), and a dummy variable that equals one if the board chair is
an independent director (Independent Chair).
The lambda, the inverse Mill’s rat
io calculated based on the probit
35


model in Table 6, is added to control for endogeneity in each odd column of the results.
All models include year and
investment objective fixed effects. P
-
values derived from fund
-
level clustered robust standard errors a
re reported in
the third column. The notations ***, **, * denote statistical significance at the 1%, 5%, and 10% levels respectively.




36


Panel
C
:


Depend
e
nt variable =
Alpha + Expense


Base

Board Size

Indep. Dir.

Indep. Chair

Unitary


(1)

(2)

(3)

(4)

(5)

Intercept

0.485

(0.618)

-
0.427

(0.690)

1.891
*

(0.079)

0.489

(0.615)

0.471

(0.629)

Lambda

0.179

(0.162)

0.160

(0.212)

0.162

(0.206)

0.178

(0.166)

0.179

(0.162)

Team

0.047

(0.750)

1.621
*

(0.053)

-
2.415
***

(0.002)

0.080

(0.662)

0.060

(0.734)

F
und TNA

0.386
***

(0.000)

0.387
***

(0.000)

0.387
***

(0.000)

0.387
***

(0.000)

0.386
***

(0.000)

Fund Age

-
0.479
***

(0.000)

-
0.483
***

(0.000)

-
0.481
***

(0.000)

-
0.480
***

(0.000)

-
0.479
***

(0.000)

Institutional

-
0.001

(0.466)

-
0.001

(0.406)

-
0.00
1

(0.467)

-
0.001

(0.470)

-
0.001

(0.473)

Cash Holdings

0.026
***

(0.001
)

0.026
***

(0.001
)

0.026
***

(0.002
)

0.026
***

(0.001
)

0.026
***

(0.001
)

Stock Holdings

-
0.009
**

(0.043)

-
0.009
**

(0.043)

-
0.009
**

(0.039)

-
0.009
**

(0.044
)

-
0.00
9
**

(0.
043
)

Turnover

-
0.001

(0.104)

-
0.001

(0.108)

-
0.001

(0.104)

-
0.001

(0.105)

-
0.001

(0.105)

Indep. Director

0.001

(0.819)

0.002

(0.711)

-
0.015
**

(0.035)

0.001

(0.840)

0.001

(0.810)

Board S
ize

-
0.173

(0.377)

0.221

(0.436)

-
0.195

(0.318)

-
0.181

(0.360)

-
0.172

(0.377)

Independent chair

0.001

(0.721)

0.001

(0.887)

0.001

(0.595)

0.001

(0.

633)

0.001

(0.719)

Unitary board

0.001

(0.919)

0.001

(0.881)

-
0.001

(0.821)

0.001

(0.900)

0.001

(0.874)

Public

-
0.007

(0.953)

0.029

(0.813)

-
0.015

(0.899)

-
0.001

(0.992)

-
0.007

(0.952)

Board S
ize*Team


-
0.734
*

(0.057)




Indep Dir
*Team



0.031
***

(0.001)



Indep. Chair*Team




-
0.001

(0.733)


Unitary*Team





-
0.001

(0.888)







Ajdusted
-
R
2

0.066

0.066

0.066

0.066

0.066

N

19,948

19,948

19,948

19,948

19,948

Note
: This table presents ordinary least square regressions of fund performance on fund management organizations.
The data covers the period from 1999 to 2007 for
3,002

funds. The dependent variable is Cahart four factor alpha,
calculated each year

plus expen
se ratio
. The independent variables include log of fund TNA, (log fund TNA), log of
fund age, (Log fund age), the number of classes offered in a fund (Number of class), the percentage of institutional
class in a fund (Institutional), the percentage of cas
h a fund holds (Cash holdings), the percentage of stock a fund
holds (Stock holdings), approximate percentage of fund holdings that changed over the preceding year (Turnover),
proportion of outside directors (Independent Directors), log of number of direct
ors on fund board (Board Size), a
dummy variable that equals one if the board is comprised as unitary board (Unitary board), and a dummy variable
that equals one if the board chair is an independent director (Independent Chair).
The lambda, the inverse Mi
ll’s
37


ratio calculated based on the probit model in Table 6, is added to control for endogeneity in each odd column of the
results.
All models include year and investment objective fixed effects. P
-
values derived from fund
-
level clustered
robust standard
errors are reported in the third column. The notations ***, **, * denote statistical significance at the
1%, 5%, and 10% levels respectively.











38


Table
5

Robustness check using the funds years without fund structure changes


Dependent variable =
Alpha + Expense


Board Size

Independence

Independent Chair

Unitary Board


(1)

(2)

(3)

(4)

Intercept

-
0.835

(0.
494
)

1.594

(0.
179
)

-
0.026

(0.
981
)

-
0.017

(0.
988
)

Lambda

0.260
*

(0.002)

0.259
*

(0.0
53
)

0.277
**

(0.039
)

0.278
**

(0.039
)

Team

1.697
*

(0.076
)

-
2.550
***

(0.00
2
)

0.310

(0.
127
)

0.272

(0.
173
)

F
und TNA

0.375
***

(0.000)

0.375
***

(0.000)

0.374
***

(0.000)

0.374
***

(0.000)

Fund Age

-
0.
339
***

(0.00
4
)

-
0.337
***

(0.00
4)

-
0.
335
***

(0.00
4
)

-
0.
334
***

(0.00
4
)

Institutional

-
0.001

(0.616
)

-
0.00
1

(0.677
)

-
0.001

(0.
695
)

-
0.001

(0.
679
)

Cash Holdings

0.0
32
*

(0.
000
)

0.0
31
***

(0.
000
)

0.0
31
***

(0.000
)

0.0
31
***

(0.000
)

Stock Holdings

-
0.01
0
*
*

(0.039
)

-
0.01
0
*
*

(0.035
)

-
0.01
0
*
*

(0.039
)

-
0.01
0
*
*

(0.039
)

Turnover

-
0.001

(0.
155
)

-
0.001

(0.
150
)

-
0.001

(0.
153
)

-
0.001

(0.
153
)

Indep. Director

0.001

(0.
899
)

-
0.019
**

(0.
013
)

0.001

(0.
990
)

0.001

(0.
997
)

Board S
ize

0.308

(0.
330
)

0.071

(0.
744
)

-
0.049

(0.
822
)

-
0.043

(0.
841
)

Independent chair

-
0.001

(0.
888
)

0.001

(0.
827
)

0.001

(0.
840
)

0.001

(0.
975
)

Unitary board

0.001

(0.
690
)

-
0.001

(0.
970
)

0.001

(0.
712
)

0.001

(0.
893
)

Public

0.159

(0.
246
)

0.113

(0.
398
)

0.128

(0.
346
)

0.122

(0.
363
)

Board S
ize*Team

-
0.661

(0.
133
)




Indep Dir
*Team


0.035
***

(0.
000
)



Indep. Chair*Team



-
0.001

(0.799
)


Unitary*Team




0.001

(0.917
)






Ajdusted
-
R
2

0
.072

0.
073

0.
072

0.
072

N

16,736

16,736

16,736

16,736

Note
: This table presents ordinary least square regressions of fund performance on fund management organizations

using the Heckman’s approach
. The data covers the period from 1999 to 2007 for
2,951

funds. The dependent
variable is Cahart four factor alpha, calculated each year
, plus expense ratio
. The independent variables include log
of fund TNA, (log fund TNA), log of fund age, (Log fund ag
e), the number of classes offered in a fund (Number of
class), the percentage of institutional class in a fund (Institutional), the percentage of cash a fund holds (Cash
holdings), the percentage of stock a fund holds (Stock holdings), approximate percenta
ge of fund holdings that
changed over the preceding year (Turnover),
governance index score which is a sum of the four governance
measures. If the fund has greater proportion of outside director representation than the overall sample median, it is
39


assigned with a vlue of 1, and otherwise 0. If the fund’s board has the smaller size than the overall sample median
size, then it will be assigned a value of 1, and otherwise 0. The presence of independent chair is assigned a value of 1,
and otherwise 0.

The unitary board is assigned a value of 1, and otherwise 0. All these four scores are added up to
make a governance index score where 4 is the highest (best) and 0 is the lowest (worst).

The lambda, the inverse
Mill’s ratio calculated based on the pr
obit model in Table 8, is added to control for endogeneity in each odd column
of the results.
All models include year and investment objective fixed effects. P
-
values derived from fund
-
level
clustered robust standard errors are reported in the third colu
mn. The notations ***, **, * denote statistical
significance at the 1%, 5%, and 10% levels respectively.



40


Table
6

2SLS instrumental variable approach


Dependent variable =
Alpha + Expense


Board Size

Independence

Independent Chair

Unitary Board


(1)

(2)

(3)

(4)

Intercept

-
0.744

(0.564)

3.233
***

(0.009)

0.806

(0.396)

0.585

(0.545)

Team

2.373

(0.120)

-
4.556
***

(0.001)

0.115

(0.663)

-
0.017

(0.943)

Log fund TNA

0.386
***

(0.000)

0.387
***

(0.000)

0.387
***

(0.000)

0.385
***

(0.000)

Log fund age

-
0.482
***

(0.000)

-
0.480
***

(0.000)

-
0.483
***

(0.000)

-
0.478
***

(0.000)

Institutional

-
0.001

(0.361)

-
0.001

(0.457)

-
0.001

(0.500)

-
0.001

(0.559)

Cash Holdings

0.
0
27
***

(0.001)

0.026
***

(0.002)

0.
0
26
***

(0.001)

0.
0
25
***

(0.001)

Stock Holdings

-
0.009
**

(0.043)

-
0.009
**

(0.036)

-
0.009
**

(0.044)

-
0.009
**

(0.045)

Turnover

-
0.107

(0.109)

-
0.108

(0.103)

-
0.107

(0.110)

-
0.106

(0.111)

Outside director

0.002

(0.663)

-
0.
0
28
***

(0.008)

-
0.000

(0.947)

0.002

(0.704)

Log board size

0.472

(0.263)

-
0.207

(0.293)

-
0.245

(0.217)

-
0.165

(0.400)

Independent chair

0.001

(0.949)

0.001

(0.475)

0.005
*

(0.061)

0.001

(0.629)

Unitary board

0.000

(0.882)

-
0.001

(0.612)

0.000

(0.761)

0.003

(0.280)

Public

0.038

(0.761)

-
0.035

(0.774)

0.037

(0.762)

-
0.026

(0.828)

Board size*Team

-
1.193
*

(0.086)




Outside*Team


0.054
***

(0.001)



Indep. Chair*Team



-
0.008
*

(0.051)


Unitary*Team




-
0.006

(0.204)






Ajdusted
-
R
2

0.068

0.068

0.067

0.067

N

19,948

19,948

19,948

19,948

Note
: This table presents
2SLS approach with the percent
of team fund management in the fund family as an
instrumental variable to examine the relation between

fund performance
and

fund management organizations. The
data covers the period from 1999 to 2007 for
3,002

funds. The dependent variable is Cahart four
factor alpha,
calculated each year
, plus expense ratio
. The independent variables include log of fund TNA, (log fund TNA), log of
fund age, (Log fund age), the number of classes offered in a fund (Number of class), the percentage of institutional
class in

a fund (Institutional), the percentage of cash a fund holds (Cash holdings), the percentage of stock a fund
holds (Stock holdings), approximate percentage of fund holdings that changed over the preceding year (Turnover),
proportion of outside directors (I
ndependent Directors), log of number of directors on fund board (Board Size), a
dummy variable that equals one if the board is comprised as unitary board (Unitary board), and a dummy variable
41


that equals one if the board chair is an independent director (I
ndependent Chair). P
-
values derived from fund
-
level
clustered robust standard errors are reported in the third column. The notations ***, **, * denote statistical
significance at the 1%, 5%, and 10% levels respectively.


























42


Table
7

Management Structure and Governance Index



Dependent variable = Alpha + Expense



(1)

(2)

Intercept

0.208

(0.786)

0.526

(0.502)

Team

0.169

(0.184)

0.160

(0.209)

Log fund TNA

0.038

(0.798)

-
0.425
*

(0.092)

Log fund age

0.383
***

(0.000)

0.381
***

(0.000)

Institutional

-
0.481
***

(0.000)

-
0.478
***

(0.000)

Cash Holdings

-
0.001

(0.505)

-
0.001

(0.417)

Stock Holdings

0.026
***

(0.001)

0.026
***

(0.001)

Turnover

-
0.009
**

(0.042)

-
0.009
**

(0.038)

Intercept

-
0.001

(0.105)

-
0.001

(0.106)

Public

-
0.005

(0.971)

-
0.011

(0.934)

Governance index

0.026

(0.609)

-
0.135

(0.127)

Governance index*Team


0.258
**

(0.019)




Adjusted R
2

0.066

0.066

N

19,948

19,948


Note
: This table presents ordinary least square regressions of fund performance on fund management

organizations

using the Heckman’s approach
. The data covers the period from 1999 to 2007 for
3,002

funds. The dependent
variable is Cahart four factor alpha, calculated each year
, plus expense ratio
. The independent variables include log
of fund TNA, (l
og fund TNA), log of fund age, (Log fund age), the number of classes offered in a fund (Number of
class), the percentage of institutional class in a fund (Institutional), the percentage of cash a fund holds (Cash
holdings), the percentage of stock a fund h
olds (Stock holdings), approximate percentage of fund holdings that
changed over the preceding year (Turnover), proportion of outside directors (Independent Directors), log of number
of directors on fund board (Board Size), a dummy variable that equals one

if the board is comprised as unitary board
(Unitary board), and a dummy variable that equals one if the board chair is an independent director (Independent
Chair).
The lambda, the inverse Mill’s ratio calculated based on the probit model in Table 8, is a
dded to control for
endogeneity in each odd column of the results.
All models include year and investment objective fixed effects. P
-
values derived from fund
-
level clustered robust standard errors are reported in the third column. The notations ***,
**,
* denote statistical significance at the 1%, 5%, and 10% levels respectively.