5‘.

 

I. st o I: t...»
.11... Karat? ..
2.5: ha. a . .2...
v3... :. . huh .%.I 5..
. 54c
_ « sx .34
it).

w..- .....10 3

Ar,

A mm... .. xvi} .. .

.3 .meikaﬁ 1.. -

gag-ML. I.” .

.ﬂnﬁuﬁkﬂ. :5

3 $63.}
a“, 5.

“Twit
n u Brady 3
s .. in: . ..
.55....sn ...‘ =..:...w§3
3:5...
. ‘ '1‘
q .
u... .. , .
_. a
$0334. 4%...”
V l‘-'. 3
5' :1“.

.2:-

.. ii.w.w..ﬁ 4M: ,
, .llt.t-|
, 8..nt has:
l.II. 4
it.)

'1‘!!!

5'3

.3...“ in? xv» .
3...: {its .31...-

. :i 3.:

.2 .. .3..i.1.t£

05.... 2.,
.
ﬁrst}

1.9.5. I

 

dun

ff 1 a!
f...

 

,‘Rrutllfa. .«
is.

   

 

  

”Edsuuotony. u... .. LI... 113 i I
9%.... 5.3.: v... A .3,
\nl

 

qmﬁww‘nwﬁ
lllv

 

 

 

 

0170‘?

 

LIBRARY
Michigan State 3
University
_._l

This is to certify that the
dissertation entitled

ESSAYS ON MUTUAL FUND PERFORMANCE AND
ORGANIZATION

presented by

Iordanis Karagiannidis

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Finance

 

 

(Li/w, 3W

/ Maﬁor Professor’s Signature
\l Una. Z ‘7 #2 cc7

Date

 

MSU is an afﬁrmative-action, equal-opportunity employer

o--.-.--.-.-.-.---.-.-.-.-.—.-n--c-<--.-.-u--o-o-c—--o-.-.-.—.—-—-—.------.-o-.----;— -

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATEDUE

DATEDUE

DATEDUE

 

MWJWOQ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRC/DateDue.indd-p.1

ESSAYS ON MUTUAL FUND PERFORMANCE AND
ORGANIZATION

By

Iordanis Karagiannidis

A DISSERTATION
Submitted to
Michigan State University

in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Finance

2007

 

 

r1.

 

 

ABSTRACT
ESSAYS ON MUTUAL FUND PERFORMANCE AND ORGANIZATION
By

Iordanis Karagiannidis

This dissertation examines how the management team structure, management
team characteristics and individual manager characteristics of mutual fund portfolio
management teams relate to performance, risk taking and other characteristics of mutual
fund portfolios. We utilize a unique data set on over 1,200 mutual fund managers and
management teams of more than 2,000 distinct open-end mutual fund portfolios over the
1997 to 2004 period.

In the ﬁrst essay we ﬁrst analyze differences in performance and risk taking
between sole-managed and team-managed mutual funds. We ﬁnd that teams under-
perfonn single managers in terms of risk-adjusted returns in the bear market period 2001 -
2004. This underperformance is more evident among growth-oriented funs. Further, we
focus on team-managed ﬁmds and examine how team-level characteristics such as team
size, age and diversity relate to performance. We ﬁnd that teams having diverse levels of
managerial experience exhibit superior performance. However, when one of more of the
fund’s manager(s) works for multiple funds, performance deteriorates. Overall, our
results suggest that, in contrast to what conventional wisdom suggests, more heads are
not better than one when it comes to managing a mutual fund.

In the second essay we focus on the manager characteristics of sole-managed
ﬁmds. The ﬁndings suggest that, consistent with Chevalier and Ellison (1999), managers

who attend high SAT-score institutions outperform other managers; however, the

signiﬁcance of the SAT score decreases after controlling for the quality of the manager’s
MBA degree. Managers who attend highly ranked business schools perform much better
than other managers. We fail to ﬁnd evidence that managing many mutual fund portfolios
affects mutual fund performance negatively.

Finally, the third essay examines the determinants and consequences of team
management by focusing on 503 mutual funds that have switched their management team
structure during the period 1997-2004. We ﬁnd that team-managed funds switch to sole-
managed after poor performance while sole-managed funds switch to team-managed after
signiﬁcant over-performance. When a fund becomes sole-managed performance
improves signiﬁcantly (184 basis points in terms of 4-factor alphas) while a switch to
team management leads to deteriorating performance. Sole-managed funds that add more
managers exhibit a decline in performance of 190 basis points. Sole-managed funds that
switch to team-managed experience above normal increases in size (total net assets) the
year before the change. Weak evidence suggests that risk taking considerations also relate
to the decision to change a fund’s management team structure. In general, our ﬁndings
conﬁrm evidence from the ﬁrst essay that team-managed funds do not offer superior risk-

adjusted performance.

Copyright by
Iordanis Karagiannidis

2007

DEDICATION

To my parents Georgios and Anastasia

 

 

ACKNOWLEDGEMENTS

I would like to acknowledge many people for helping me during my doctoral
work. I am grateﬁrl to my committee chair and mentor Professor G. Geoffrey Booth for
his excellent guidance, support and commitment. Throughout my time at Michigan State
University he encouraged me to develop independent thinking and research skills and
greatly assisted me with scientiﬁc writing.

I would also like to thank my committee members, Professor Charlie Hadlock,
Jun-Koo Kang and Steven Dimmock for their help and encouragement. I extend many
thanks to my fellow doctoral students of the Finance Department at MSU.

Finally, my deepest gratitude goes to me family — Georgios Karagiannidis,
Anastasia Malamatidou, Alexandros Karagiannidis and Soﬁa Malamatidou — for their
patience and continued support. I also thank Elizabeth Booth, Mike Booth and Matt

Booth for treating me like family and making me feel at home throughout my time in

East Lansing.

vi

II
II.
ill a. A

 

mill-

7.1.,

TABLE OF CONTENTS

LIST OF TABLES ix
ESSAY 1. Portfolio Management Team Structure and Mutual Fund
Performance
1 .1 Introduction .................................................................. l
1.2 Institutional Background and Description ............................... 6
1.3 Related Literature ............................................................ 10
1.4 Method and Data ............................................................ 20
1.4.1 General Method .................................................... 20
1.4.2 Data Description ............. ‘ ...................................... 23
1.4.2.1 Mutual Fund Data ....................................... 23
1.4.2.2 Manager Data ............................................ 25
1.4.2.3 Team-level Variables ................................... 26
1.5 Hypotheses and Results .................................................... 27
1.5.1 Single Managers versus Management Teams .................. 27
1.5.2 Management Team Characteristics .............................. 33
1.6 Summary and Conclusion .................................................. 39
APPENDIX 1. TABLES OF ESSAY 1 .................................................... 43
ESSAY 2. Manager Characteristics and Mutual Funds
2. 1 Introduction .................................................................. 70
2.2 Method and Data ............................................................ 74
2.2.1 Method ............................................................ 74
2.2.2 Data Description ................................................... 75
2.2.2.1 Mutual Fund Data ....................................... 75
2.2.2.2 Manager Data ........................................... 77
2.2.2.3 Benchmark Portfolio Returns Data .................. 79
2.2.3 Variables ............................................................ 79
2.2.3.1 Distinct Fund Portfolios ............................... 79
2.2.3.2 Performance and Risk Measures ..................... 80
2.2.3.3 Manager-level Variables ............................... 82
2.3 Hypotheses and Results .................................................... 85
2.3.1 Discussion of Hypotheses ......................................... 85
2.3.2 Results ............................................................ 90
2.4 Robustness Checks ......................................................... 96
2.5 Summary and Conclusions ................................................ 97
APPENDIX 2. TABLES OF ESSAY 2 .................................................... 99

vii

ESSAY 3.

Portfolio Management Team Changes

3.1 Introduction .................................................................. 133
3.2 Data ........................................................................... 136
3.2.1 Mutual Fund Data .................................................. 136

3.2.2 Performance and Risk Measures ................................. 139

3.3 Hypotheses and Results .................................................... 140
3.3.1 Hypotheses .......................................................... 140

3.3.2 Results ................................................................ 141

3.3.2.1 Univariate Tests ......................................... 141

3.3.2.2 Probit Models ............................................ 144

3 .4 Conclusion ................................................................... 146
APPENDIX 3. TABLES OF ESSAY 3 .................................................... 148
BIBLIOGRAPHY ............................................................................. 158

viii

Table 1.1

Table 1.2

Table 1.3

Table 1.4

Table 1.5

Table 1.6

Table 1.7

Table 1.8

Table 1.9

Table 1.10

Table 1.11

Table 1.12

Table 2.1

Table 2.2

Table 2.3

Table 2.4

Table 2.5

Table 2.6

Table 2.7

LIST OF TABLES
Dataset Summary ......................................................... 44
Summary Statistics ....................................................... 46

Regressions for Portfolio Characteristics of Sole-managed
and Team-managed Funds ............................................... 49

Regressions for Portfolio Risk of Sole-managed and
Team-managed ﬁmds ..................................................... 50

Regressions for Investment Style / Risk Factor Loadings
of sole-managed and Team-managed Funds ........................... 51

Regressions for Performance of Sole-managed and

Team-Managed Funds ................................................... 52
Correlation Matrix of Team-level Variables ........................... 57
Portfolio Characteristics and Tearn-level Characteristics. . . . . . . . ....58
Portfolio Risk and Team-level Characteristics ........................ 60

Investment Style / Risk Factor Loadings and Team-level
Characteristics ............................................................ 62

Performance and Team-level Characteristics ......................... 64

Regressions of Performance and Team-level Characteristics

with Interaction Terms ................................................... 67

Manager Characteristics .................................................. 100
Descriptive Statistics of Characteristics Variables ................... 101
Correlation Matrix of Manager Characteristics Variables ........... 102

Regressions of Fund Characteristics on Manager Characteristics. 103

Fund Risk and Manager Characteristics ............................... 112
Investment Style and Manager Characteristics ........................ 119
Performance Logit Regressions ......................................... 126

ix

 

Table 3.1

Table 3.2
Table 3.3
Table 3.4

Table 3.5

Table 3.6

Dataset Summary — Management Structure and Investment

Advisor Changes .......................................................... 149
Performance Characteristics — Univariate Analysis .................. 151
Risk Characteristics — Univariate Analysis ............................ 152
Portfolio Characteristics — Univariate Analysis. . .................... 154
Probit Regressions — Change in Management Company

(Investment Advisor) ..................................................... 156
Probit Regressions —- Change in Management Structure ............. 157

 

 

f)

I ‘ ’7', ‘T’ I I a.
. I f? ’I .‘r L’ ‘7'." [:7 a) If! If.» 'rr; , ..

 

ESSAY 1. Portfolio Management Team Structure and Mutual
Fund Performance

1.1 Introduction

Although historically it was common for a fund to have only one person as the
portfolio manager, things have changed dramatically in recent years. For instance,
according to Momingstar’s database of domestic equity funds, in 1997 only 32.5% of all
funds and 20% of total mutual fund assets were managed by a team of managers rather
than a single individual, whereas in 2005 the corresponding percentages were 58.5% and
60%, respectively. In terms of dollar amounts, teams managed more than $1.2 trillion in
2005 compared to only $250 billion in 1997, while single managers managed $888
billion in 2005 compared to $686 billion in 1997. 1Further, many mutual funds advertise
their team-managed approach to portfolio management as an “edge” and investors seem
to prefer team managed funds.2

In spite of the increasingly important role of management teams in the portfolio
management industry there is little empirical evidence on the differences in performance
and trading practices of sole-managed and team-managed mutual fund portfolios.

Moreover, to the best of our knowledge, there is no study that investigates the importance

 

' An alternative view is that this is just a change in reporting by mutual fund companies. If this is true, we
should not see any differences in trading practices or performance. We document the opposite. Even in the
case where the increase in team-managed funds is just a change in reporting manager names, we would
expect managers to behave differently when their name is reported and tied to portfolio performance. In
any case, just the fact that mutual fund companies advertise their team managed approach as superior,
makes it interesting to evaluate differences in performance between sole-managed and team-managed
funds.

2 Many mutual funds underline the importance of the team-managed approach on their funds’ prospectuses.
For example, consider the following two quotes taken from the website of Brazos Funds: “. . .the Brazos
Fund’s team approach results in the constant interaction and contribution of the entire team of portfolio
managers. No action is taken until the team has had the opportunity to scrutinize the potential investment”,
“The Brazos Funds view the team-based approach as an important component in creating less risk for
clients and increases their long-term returns.”

rust-1.1.: m
)

 

 

J

of team-level characteristics in explaining differences in performance and risk taking.
This paper attempts to ﬁll this gap.

Recent literature recognizes the importance of the organizational structure of
portfolio management teams. Prather and Middleton (2002), Chen et al. (2004) and Baer
et a1. (2005) compare the performance of sole-managed and team-managed ﬁrnds. Prather
and Middleton (2002) ﬁnd that, consistent with the classical decision making perspective,
there is no difference in the performance of sole-manager and team-managed mutual
funds. Chen at al. (2004) and Baer et a1. (2005) do ﬁnd evidence of underperformance by
teams of managers of 5.5 and 4 basis points per month respectively.

In another paper, Qiu (2004) examines the risk-taking behavior of mutual fund
managers in response to incentives they are given. He divides funds into two groups:
funds managed by single managers and funds managed by multiple managers. He ﬁnds
that single managers adjust the risk of their portfolio to a much greater extent than
multiple managers do in the second half of the year. Further, he ﬁnds that loser single-
manager funds are more aggressive than loser multiple-manager funds.

In this paper, we hand—collect a unique and much more comprehensive dataset of

2,031 US. open-end, domestic-equity mutual fund portfolios (7,713 ﬁrnd-year

observations) in the period between January 1997 and January 2005.3 Our analysis is

‘

3 Prather and Middleton (2002) use a sample of 162 open-end mutual ﬁmds (147 managed by individuals
and only 15 managed by a team). Their requirement of data availability for 156 consecutive months
l'eslnicts the sample signiﬁcantly and introduces a serious concern of survivorship bias. Baer et a1. (2005)
use a much bigger sample (14,848 fund-year observations), however, they include a wide range of mutual
funds in their sample, including global and international, utility, balanced as well as sector ﬁrnds.

 

.. L . . a..." . .. a .r. Fl. .C .C .. .. If I . .
S «L 18 Wu s s C... Pt .43 . C E nu . m .6 wk. .N‘

 

conducted in two parts and provides several new insights regarding the role of portfolio
management team structure.4

We begin our investigation by using cross-sectional variation to examine whether
performance, risk taking and portfolio characteristics depend on whether the fund is sole-
managed or team-managed. Our sample period (1997-2005) spans both a bull and a bear
market. In contract to other papers, we report results not only for the whole sample
periods, but also for the bull and bear market sub-samples (1997-2000 and 2001-2005,
respectively. We do ﬁnd evidence of underperformance by team-managed ﬁmds, but this
underperformance is only present during the bear market period (45 basis points a year).
In “good” market conditions there is no evidence of differences in the performance of
sole- and team-managed mutual fund portfolios. Furthermore, evidence of signiﬁcant
underperformance is only present among growth funds and reaches 61 basis points a year
in terms of 4-factor alphas.5

In the second part of our analysis we delve deeper in to the structure of team-

managed funds. We collect data on all managers of team-managed funds and construct

 

4As Sharpe (1981) suggests, we can identify two types of multi-manager team structures. In the most
common type of multi-manager team structure, each co-manager is assigned only a part of the fund’s assets
and has independent decision-making authority on the assets under his or her management. In the other
structure, managers make decisions as a committee, collectively deciding on trades after reaching a
consensus. However, in reality these two types of teams are not that distinct. Management is not completely
diverse in the case of independent sub-managers, since all the managers usually belong to the same
management company, share the same pool of analysts, and communicate with each other, and even though
consensus has to be reached in committee-type teams, individual members may be held accountable for
speciﬁc recommendations. That said, one major difference between the two team types is that individual
results are more formally and directly observable when each manager has his or her own share of assets. In
addition, the compensation and incentive system is probably different across team structures.
Unfortunately, detailed information about exact team structures is not available to us.

5 Funds that have a prospectus objective of “Growth” or “Aggressive Growth” are categorized as growth-
oriented funds. Funds with prospectus objective “Equity-Income” or “Growth & Income” are considered as
income-oriented funds. Wermers and Ding (2005) examine the relation between manager characteristics
and performance and also ﬁnd that their results hold only for growth-oriented mutual funds. They posit that
the difﬁculty of accurately forecasting earnings growth requires higher experience and ability, so this might
be one of the reasons that manager characteristics are important only for those ﬁrnds.

 

9.3

n.3,

several team-level characteristics such as team size, team tenure, diversity of experience
and other forms of diversity. We then relate team-level characteristics to performance,
risk-taking and portfolio characteristics. To the best of our knowledge, this is the ﬁrst
paper that gathers and analyses data on fund managers of team-managed funds. In terms
of performance, we ﬁnd that age diversity, which serves as a proxy for experience
diversity, is positively associated with performance in the bear market period. This
ﬁnding suggests that pooling experienced managers together with younger managers that
want to prove themselves, leads to superior performance. Another signiﬁcant ﬁnding is
that the common practice of fund families to use the same managers for many of their
funds has a negative effect of the funds’ performance. Finally, we ﬁnd evidence that,
larger teams which have been working together for a long time, exhibit superior
performance in terms of l-factor and 4-factor alphas. An explanation for this is that when
people have been working together for a long time, they learn how to work with each
other, and the advantages of the team based approach outweigh its disadvantages such as
the possibility of disagreements and conﬂicts.

Another empirical study, Massa et a1. (2006), also examines the performance of
single managers, teams of managers as well as anonymous teams of managers.6 However,
they posit that all funds are more or less team managed and the way names are reported
has to do with the fund family’s decision on who gets credit for the fund’s performance.
They ﬁnd that the underperformance of tearn-funds is solely due to anonymous teams.
We also ﬁnd very weak evidence that underperformance comes from anonymous
management teams but this underperformance does not totally explain the superiority of

single managers.

 

6 Some funds report the fund as team-managed but do not disclose the manager names.

 

"V!

(*i

”1

‘7‘)

f1.

9'.”

 

 

We argue that the declaration of a fund as sole- or team-managed is not just a
reporting issue. An analysis of portfolio characteristics reveals that team-managed fund
portfolios are signiﬁcantly different that those of sole-managed funds. Speciﬁcally, we
document that team-managed funds hold more stocks in their portfolios, turn their
portfolio often less frequently and invest lesser amounts of money in their 10 top
holdings.

In a multivariate regression setting, we fail to ﬁnd evidence that single managers
take on more total (standard deviation of monthly returns) and systematic (market beta)
risk.

Our study relates to literature in other disciplines such as management and
psychology that examine teams and group decision making and therefore is of broader
academic interest. The mutual fund arena is an ideal place to test a general list of theories
of individuals versus team decisions and performance.

Our ﬁndings, taken together with the increased popularity of the team approach in
portfolio management, suggest a puzzle. Namely, why do investors and fund families
prefer team-managed funds if they do not offer superior risk-adjusted returns?

The remainder of the paper is organized as follows. Section 1.2 provides a brief
description of the US. mutual funds industry. Section 1.3 presents related literature. We
describe our method and data sources and the creation of variables in Section 1.4. In
Section 1.5 we describe our hypotheses and present results. Finally, Section 1.6

concludes.

1.2 Institutional Background and Description

The US. mutual fund industry has observed explosive growth, especially during
the past decade. From $2.8 trillion in 1995, assets managed by mutual funds grew to a
record-breaking $8.1 trillion in 2004 (Investment Company Institute (2005)). As the scale
of the mutual fund industry has changed, so too have the funds themselves. For example,
funds have introduced additional share classes to attract more investors and new sales
channels to reach the investment public. Further, the structure of portfolio management
teams has changed over time. This section brieﬂy describes the mutual fund industry,
setting the background for our paper.

A mutual ﬁrnd is a corporation or business trust that belongs to all its individual
shareholders that purchase shares issued by the fund. Mutual funds are also referred to as
open-end ﬁmds, as they can continuously offer new ﬁrnd shares to the public and they are
required to buy back outstanding shares when shareholders request that they do so. Like
any other company, a mutual ﬁrnd has a board of directors and its shareholders have
voting rights, including the right to elect directors. However, unlike other traditional
companies, mutual funds do not have their own employees; instead, they rely on third
parties or service providers to carry out their business activities. The board of directors is
responsible for administrative decisions (pricing the shares, setting fees, etc.) and
negotiates contracts with the following entities: a) the management company (or
investment adviser), who runs the ﬁrnd’s portfolio, b) the fund’s custodian, who is
usually a bank or trust company that holds the fund’s assets for safekeeping and handles
payments and receipts of the fund’s investment transactions, c) the transfer agent, who

performs recordkeeping and reporting services, (I) the Fund Distributor, who arranges for

 

 

 

,e'

 

the sale of shares, and e) the legal counsel, who provides the fund legal advice. One of
the most important functions of the board of directors is to monitor the management
company and renew or reject their contract every year. To protect investors, the
Securities and Exchange Commission requires that 75% of a board’s directors be
independent, that is not having a signiﬁcant relationship with the adviser or the
underwriter of the fund.

Even though mutual funds are legally regarded as stand-alone companies,
effectively they are not stand-alone entities. Instead, they belong to a broader
organizational structure known variously as the “fund family,” “fund complex,” or “fund
sponsor.” The fund family appoints the set of directors that oversee the fund and
generally manages all the activities needed to start, run, and even close a fund. Big
players in the industry include Fidelity, Vanguard, and American Funds, each of which
offers dozens of mutual funds.

Investors can buy mutual ﬁmd shares through a variety of channels. In addition to
the traditional channels of buying mutual fund shares through ﬁnancial advisers or
directly from the mutual fund company, investors can use newer sources such as
retirement plans and fund supermarkets. Fund supermarkets offer one-stop service to
investors, giving them the chance to buy fund shares from an extensive range of fund
families and to easily switch their money between funds.

Another recent development in the mutual fund industry is the introduction of
multiple share classes, which allows investors to choose how they want to pay for
advisory service commissions paid to brokers. Shareholders pay for ﬁnancial advisory

services through load charges (front- or back-end) and 12b-1 fees. Front-load charges are

charged to investors when they buy new shares and are calculated as a percentage of the
initial investment. Back-end (or deferred) load charges are assessed when investors leave
the ftmd, declining as time progresses (the longer the investor stays with the fund, the
lower the back-end load charge), eventually disappearing. Most back-end load fees start
at 5 percent and decline by one percentage point annually. By law, total front-end and
back-end loads cannot exceed 8.5 percent of the initial investment, though competition in
the industry has forced fund families to lower loads to an average of 3 percent to 6
percent. 12b-1 fees are part of a fund’s annual expense ratio, and include administrative
and management fees. Such [fees are used to pay marketing and distribution expenses and
cannot exceed 0.75 percent of the ﬁrnd’s average net assets per year. To suit the needs of
different investors, fund families offer various classes of shares. The most common share
classes include class A, B, and C shares, where class A shares generally have a front-end
sales charge, no deferred-end sales charge, and a low 12b-l fee, class B shares usually
have no upfront fees but they have back-end charges and 12b-1 fees, and class C shares
do not have any type of load charges but they assess a higher 12b-1 fee. Fund families
have also developed share classes suitable for investors that hold a large number of shares
(institutional, retirement), though the characteristics of these classes vary widely across
fund families. Finally, funds offer a no-load class, which gives investors the option to buy
funds without using, and hence without paying for, the advice of a ﬁnancial professional.
Due to varying 12b-1 fees, different share classes have different annual expense
ratios (note that administrative and management fees are the same), and consequently
different net returns. The annual expense ratio reflects the annual operating costs of

running the fund. Unlike load charges, it is not charged directly to the investor, but is

deducted from funds assets. It includes the distribution fee, the management fee, and
other administrative expenses. The management fee is paid to the management company
for managing the fund’s portfolio. Administrative expenses include money paid to the
ﬁmd’s other service providers such as the transfer agent.

The management company (investment adviser) hired by the board of directors of
the fund has its own employees and chooses the managers and analysts that will be
involved in making investment decisions. The investment adviser could be an internal
management ﬁrm, which is afﬁliated with the fund family, or an outside professional
portfolio management ﬁrm that manages ftmds from many different ﬁmd families. Chen,
Hong, and Kubik (2004) investigate mutual ﬁrnds’ make-or-buy decision and ﬁnd that
the decision depends on client demand for and the ﬁxed cost of offering investment styles
that are beyond the management company’s expertise. They also ﬁnd that performance is
harder to extract from outsourced funds and hence that externally managed funds are
more likely to be closed down for poor past performance than comparable internally run
funds.

In many cases management of a fund’s portfolio is assigned to more than one
investment advisor. Vanguard, for example, has advisory contracts with 24 outside
management ﬁrms, while for about one-third of its funds, multiple ﬁrms split the fund’s
stock-picking duties.7

Funds are managed by either a single portfolio manager, who has sole
responsibility over investment decisions, or a team of portfolio managers, who share
stock-picking decisions. Portfolio management teams can be organized in two ways.

First, a portfolio management team can be organized as a committee, where managers

 

7 Forbes magazine. December 13, 2004, pp 191-192.

together make consensual decisions about which stocks to buy or sell. Teams can also be
organized as a portfolio of individual managers, where each manager is assigned a certain
share of the fund’s assets and can make decisions over its share without having to agree
with the recommendations or ideas of the other co-managers.

Empirically, it is difﬁcult to distinguish between those two types of teams as in
practice the two types might not be that distinct. In the case of independent co—managers,
management is usually not completely diverse, since in general all the managers belong
to the same management company, share the same pool of analysts, and exchange ideas
with each other. In the case of committee-type management teams, even though a
consensus has to be reached, members may be held accountable for speciﬁc
recommendations and a decision can be inﬂuenced more by the most dominant team
members. However, one major difference between the organization types is that when
individual co-managers have their own share of assets, their performance is more
formally and directly observable. The compensation and incentive structures that
managers face might also be different between the two management team types. Today,
the team-of-managers approach is much more common than the single-manager approach
in the mutual fund portfolio management industry. Indeed, as of January 2005, 58.5% of
all our sample funds and 60% of total assets under management were managed by teams

of managers rather than a single individual.

1.3 Related Literature

While only a few empirical studies consider the team-based approach to portfolio

management, several theoretical papers posit why team management might be a superior

10

 

 

 

.J.
v—.—._._ v..-

F

I I I ll
.. . ... x. I r

strategy. Distinguishing between cases in which one hires several managers to analyze
one subset of securities (diversiﬁcation of judgment) and cases in which different
managers are hired to analyze different subsets of securities (diversiﬁcation of style),
Sharpe (1981) shows that the beneﬁts of multiple managers relate to diversiﬁcation
effects. Bad picks by one manager can be offset by better picks of the others in the case
that they manage independently; in the case that the managers manage together the
diversiﬁcation of judgment effect protects against the possibility that a particular
manager will make a serious error. Thus, if investors want to diversify their investments,
multiple-manager funds allow them to easily achieve this goal by investing in a fund that
is run by several independent managers rather than by investing in several different
funds. Sharpe also argues that specialization may be another determinant of the multiple-
manager phenomenon. If managers have specialized knowledge in a particular area, then
it would be reasonable to employ many managers to offer specialized knowledge in many
areas.

As Sharpe (1981) suggests, we can identify two types of multi-manager team
structures. In the most common type of multi-manager team, each co-manager is assigned
only a part of the fund’s assets and has independent decision-making authority in relation
to the assets under his or her management. In the other structure, managers make
decisions as a committee, collectively deciding on trades after reaching a consensus.
However, in reality these two types of teams are not that distinct. Management is not
completely diverse in the case of independent sub-managers, since all the managers
usually belong to the same management company, share the same pool of analysts, and

communicate with each other, and even though consensus has to be reached in

11

committee-type teams, speciﬁc members may be held accountable for speciﬁc
recommendations.

Even in the absence of specialization or diversiﬁcation beneﬁts, however, Barry
and Starks (1984) show that the use of multiple managers may still be justiﬁed and even
optimal in some cases. They develop a model that is based on agency and risk-sharing
considerations. An intuitive way to look at their argument is the following. If we suppose
that managers are prone to taking on more risk as the amount of money they manage
increases, then risk-averse investors who prefer the risk level that the manager would take
on if he managed half the money might choose to invest in a two-manager portfolio
rather than a single-manager one.

Picher (2004) characterizes the optimal organizational forms and incentive
contract for a team of money managers. In his model, the investor (principal) is risk
averse and each manager’s actions affect both that manager’s expected return and the
correlation of returns between managers, depending on the risk tolerance of the
managers. If the managers are risk tolerant, then a non-cooperative team structure and a
contract in which each manager is rewarded both for doing well and for doing better than
the team is the most efﬁcient way to encourage managers to exert effort in diversiﬁed
activities. As the investor and the managers become more risk averse, cooperation among
managers becomes the optimal organizational structure.

Hohnstrom (1982) is concerned with the moral hazard problem of inducing agents
to supply the proper amounts of productive inputs when their actions are not observed.
More speciﬁcally he studies the moral hazard problem in teams when many agents

(members) are involved. Moral hazard is also present in the single-agent case when there

12

 

 

is uncertainty in output. However, in teams, problems can occur even if there is no
uncertainty in output. If only the joint output of the team is observable, members of the
team that cheat or do not supply enough effort cannot be identiﬁed. Hohnstrom focuses
on this free-rider problem and shows under what conditions it can be alleviated. His
conclusion is that the free-rider problems can be largely resolved if there is separation of
ownership and labor and if proper incentives (such as penalties or bonuses) are given to
the team. More speciﬁcally, he shows that no contract that is “budget balancing”, that is it
allocates all of the team’s output to its members, can induce the team member to choose
the efﬁcient effort levels. The efﬁcient output is obtained by a contract that gives each
member a payoff of zero (the output goes to the principal-owner) if the output is lower
than if all the members had chosen the efﬁcient effort levels. Moreover, when individual
output is observable, relative performance evaluation of agents can help reduce the moral
hazard costs. Holmstrom’s theorem depends on the agents’ utility functions’ being linear
in money.

Rasmusen (1987) extends Holmstrom’s model to allow for risk averse agents.
With risk averse agents there exists a ﬁrst-best contract that is “budget balancing”. This
contract is similar to the non-budget-balancing contract in Holmstrom in the sense that
less than efﬁcient team output triggers a punishment. However, with risk averse agents
this punishment takes the form of a deviation lottery rather than loss of the whole output
by the team. Rasmusen describes two forms of this lottery: a) the scapegoat lottery, where
only one member is punished and the others take his share, and b) the massacre lottery,
where only one agent is rewarded by getting the shares of all the other members. The

massacre lottery seems to attain the ﬁrst-best efﬁciency level for less risk-averse agents

13

 

or more tightly bounded punishments, even though all of the deviation lotteries perform
reasonably well.

Sah and Stiglitz (1988) study the decision making of committees and contrast it to
certain forms of centralized versus decentralized organizations. In committees each
member evaluates every project available to the company and a project is accepted if
approved the number of members required to reach a consensus. The second type of
organization studied is called hierarchy. In hierarchical organizations the project is
evaluated (and either accepted or rejected) by a higher level individual only if approved
by the lower levels. Finally, in polyarchical organizations a project is accepted if
approved by any one member. Their analysis focuses on two economic trade—offs. The
ﬁrst trade-off is between the errors of not approving good projects and the errors of
accepting bad projects. The second trade-off is between the gain from a more extensive
evaluation of projects and the extra resources spent on evaluating projects. They derive
results concerning the optimal design of each of the three organizational forms and
compare the performance between them. For committees they provide a framework that
derives the optimal committee size and the level of consensus. For hierarchies they
characterize the optimal level of levels depending on the underlying economic conditions.
Finally, the also provide a framework on the optimal number of member in a polyarchy.
More importantly however, they analyze the relative performance of the three
organizational forms under different sets of parameters of the economy. For example,
they show that when the portfolio of projects is better, a polyarchy is better than a
committee and a committee is in turn better than a hierarchy. They also show that when

the evaluations costs are large the relative performance of a polyarchy or a hierarchy is

14

 

 

; .
LL. .1' "'3'“

 

better compared to a committee. They also underline the economic costs that time delays
in making decisions might impose. Anyone who compares the relative performance of
different organizational forms should take differences in the time it takes to make
decisions into account.

Slezak and Khanna (2000) examine the effect of the organizational form
(hierarchy or team) on the collection and sharing of information. In their model each
member makes a recommendation on whether to accept or reject a proj ect, and the
project is accepted or rejected according to the majority rule. The difference between the
two organizational forms they present is that, in a team each member observes the
recommendation of the other members and the members announce their
recommendations sequentially. Under the hierarchical form, all members report their
recommendation to a central authority without listening to other people’s
recommendations. They identify a disadvantage to the team structure which has to do
with the case where recommendations some members can be inﬂuenced by what other
people think and information gathered by team members might be used inefﬁciently.
Even though such inefﬁciencies are not present in the hierarchical structure, they
recognize the case where the agents still communicate with each other informally and
might also create cascades. In that case the principal has to monitor at a cost and enforce
the hierarchical structure. When the cost of imposing the hierarchical form is greater than
the beneﬁt from its enforcement, teams are the optimal form of organization. They also
look at two types of incentives given to agents: an individual bonus, which rewards only
the agent that made the correct recommendation, and a team bonus, which goes to the

whole team when their decision is good. They show that team bonuses can solve the free-

15

rider problem only when they are too big. Individual bonuses work better, but give
incentives to the members of the team to collude and lead to inefﬁcient use of
information. They conclude that the hierarchical form is the optimal organization
structure except in the case of high hierarchy enforcing costs described above.

Many other research studies, especially in the management and psychology
literature, investigate the decision-making process, behavior, and performance of teams
versus individuals. The results differ across studies mainly due to the variety of tasks and
measures used in each study, which makes it difﬁcult to make valid generalizations or
comparisons. However, as conventional wisdom suggests, all studies agree that teams
behave differently than individuals, even though we do not always observe differences in
performance.

Hollenbeck et a1. (1998) argue that the ideal decision maker is either a team of
decision makers that must reach consensus or an individual manager in a hierarchical
structure where the support staff is not involved in the ﬁnal decision. Prather and
Middleton (2002) argue that under the classical decision making theory perspective we
should see no signiﬁcant difference in performance between sole-managed and team-
managed funds.

Moving on to empirical studies, results are mixed. For instance, Herrenkohl
(2004) and Hill (1982) discuss the advantages and disadvantages of teams when it comes
to decision-making tasks. Teams, as opposed to single individuals, have a broader range
of relevant skills and knowledge. They can also acquire and process a larger amount of
information by subdividing responsibilities. On the other hand, teams might not be able

to exploit the full range of skills and knowledge of its members because some members

16

 

 

\

may not be motivated to contribute, some members may inﬂuence the ﬁnal decision more
than others, or the diversity of team members’ views can sometimes be so broad that they
are difﬁcult to reconcile.

In terms of risk taking common sense would suggest that single managers are
prone to taking more risk. However, Vinokur (1977), Herrenkohl (2004), Janis (1984)
and many other researchers suggested that teams are subject to the risky shift
phenomenon. When people are in groups, they make decisions about risk differently ﬁom
when they are alone. They are likely to make riskier decisions, as the shared risk makes
the individual risk less.

Further, while Sah and Stiglitz (1988) argue that teams may be associated with
delays in decision making, and thus they suggest that anyone who compares the relative
performance of different organizational structures should take differences in the time it
takes to make decisions into account, Schmidt et al. (2001) ﬁnd that teams are more
effective decision-makers than individuals. In terms of accuracy in judgment, most
studies suggest that group judgments tend to be more accurate than the judgments of
typical individuals, and approximately equal to the mean judgments of their members.

However, all teams are not the same. They vary in terms of size and member
characteristics. For this reason, a great deal of research studies investigates the effect of
team-level characteristics on performance, risk taking and decision making process.

For instance, Herrenkohl (2004) suggests that larger teams may have a broader
range of skill and knowledge, but team members might not be as motivated to contribute.
Further, the larger the team, the longer it may take to reach a consensus, increasing the

possibility of time delays in decision-making (Sah and Stiglitz (1988), Ley and Steel

17

(1995)). Recognizing that there are advantages when people with varied backgrounds and
abilities co-exist in a team, but that larger teams are less cohesive and members of large
teams are less likely to cooperate and perform to the maximum of their abilities,
Thompson (2004) argues that it is wise to compose teams using the smallest number of
people that can do the task.

According to Herrenkohl (2004), the effect of team size depends on the nature of
the task relative to the combined knowledge and skill of the team members. In additive
tasks, where one person’s work is added to the work of others to arrive at a team product,
increasing the size of the team has a positive effect until an upper limit is reached, at
which point there is no extra beneﬁt from the addition of new members. When any single
member can supply the team product (disjunctive task), success depends on the most
competent member of the team. Team size has a positive effect in this case too, since the
larger the team the greater the likelihood that at least one member will be able to do the
job. Finally, in conjunctive tasks, relatively high performance is achieved by very small
teams, but decreases rapidly as size increases, since some members might be less able
than others and can slow the whole team down, limiting its performance.

Team diversity is the second most frequently studied team—level characteristic.
However, research on the effect of team diversity on performance has produced mixed
results. Smith et al. (1990) ﬁnd that educational diversity in top management teams is
positively associated with performance. However, in the same study they report that
experience diversity is negatively associated to performance. Simons et a1. (1999)
examine diversity in functional background, education, tenure, and age. The ﬁrst three

measures are considered to be more job-related forms of diversity because they largely

l8

capture experiences, information, and perspectives relevant to cognitive tasks. They ﬁnd
that job-related forms of diversity are positively associated with performance, while age
diversity does not have any signiﬁcant effect.

LePine et a1. (2002) examine gender diversity and its effect on decision-making
accuracy. In their study they ﬁnd that decision-making inaccuracy is an exponential
function of the number of males on a team. Women-only teams or teams with a balanced
gender composition invariably record the highest accuracy, even though, in their
experiment, teams competed in a traditionally masculine task by taking the place of a
military command-and-control team in a simulation program designed by the US. Air
Force. Teams in which men constituted the majority performed poorly and all-male teams
were worse than any other conﬁguration.

Krishnan et a1. (1997) examine whether performance is improved by merging
similar or dissimilar members in a team. They present plausible advantages of both cases.
More homogeneous teams whose members have similar functional backgrounds might
communicate and cooperate better, and thus demonstrate improved performance. On the
other hand, different backgrounds and skills can complement one another and lead to
improved performance too. They ﬁnd that differences in functional backgrounds have a
positive impact, that is, value-adding synergies are created when dissimilar top
management team members come together. They also ﬁnd that differences in background
are negatively related to manager turnover, meaning that those differences are more
easily integrated into the new organization while similarities might lead into

redundancies and conﬂict.

19

Janis (1984) argues that high levels of group cohesion can often result in
groupthink. Groupthink is a type of thought exhibited by group members who try to
minimize conﬂict and reach consensus without critically testing, analyzing, and
evaluating ideas. Groupthink may cause groups to make hasty, irrational decisions, where
individual doubts are set aside, for fear of upsetting the group’s balance.

Studies that investigate the relationship between demographic diversity and team
performance make the implicit assumption that demographic diversity is associated with
cognitive diversity, which in turn has an important effect on team performance. Cognitive
diversity refers to the variability of relatively unobservable characteristics such as

perceptions, values, attitudes, and beliefs.

1.4 Method and Data

1.4.1 General Method

We conduct our analysis as follows. We ﬁrst obtain management team structure
and team-level characteristics at the beginning of each year t for all the years in our
sample period (1997 to 2004). We then relate management team characteristics at time t
to portfolio attributes, risk, investment style, and performance over the course of the next
year (from t to t+1).

We mainly use ordinary least squares (OLS) regressions of portfolio
characteristics and risk attributes on management team characteristics. Similar to
Chevalier and Ellison (1999), we also use instrumental variable estimation for some of
our regressions, using lagged observations as proxies for variables (such as turnover) that

appear to be endogenous. In all regressions we estimate clustered. standard errors by fund

20

and include prospectus objective and time dummies even if we do not explicitly show
that when we present our regression speciﬁcations.

We use all the standard performance metrics to measure fund performance in a
given year: raw annual frmd returns, style-adjusted returns, one-factor alphas from the
market model, and alphas from the Carhart (1997) four-factor model.8 To estimate the

one-factor and four-factor alphas, respectively, we estimate the following models:

Rit ‘ th = at + ﬂilEMRt + ﬁrzSMBr + .Bi3HMLr + ﬂi4UMDt + git , (2)

where R, — R f, is the month-t excess gross return for fund 1' , EMR, is the excess market

return, SMB, is the difference in returns across small and big stock portfolios, HML, is
the difference in returns between high and low book-to-market portfolios, and UMD, is

the return on a momentum portfolio as computed by Farna and French.

We use the value-weighted NYSE/AMEX/Nasdaq composite index as our market
return, and the one-month T-bill rate from Ibbotson Associates as our risk-free rate in
calculating excess market returns. Returns on the HML (high minus low book-to-market
returns) and SMB (small minus big stock return) zero-investment portfolios, as well as

returns on a momentum portfolio (UMD), come from Kenneth French’s website.9

 

8 We also calculate abnormal returns from the 3-factor model and use them in our regressions. Results are
very similar to results when using abnormal returns from the Carhart 4-factor model.
9 See Kenneth French’s website for the deﬁnition and calculation of the factor portfolio returns.

21

 

 

 

 

When estimating the one-factor and four-factor alphas we use monthly gross ﬁmd
returns for the 12 months in the year. To calculate the gross monthly return, we divide the
annual expense ratio by 12 and add this to the monthly net returns. We use gross returns
because we want to measure the performance differences between various forms of
management team organization and characteristics. If better managers or organizational
forms receive rents through higher expenses, then the performance superiority of
manager characteristics might not show up when using net returns. However, we repeat
our analysis using fund returns net of management fees and the results are qualitatively
the same.

To evaluate the riskiness of the portfolio we use the market betas (the coefﬁcients
ﬂu) from equations (1) and (2), as well as the standard deviation of monthly returns,
throughout the cOurse of the year. We also examine differences in the estimated factor
loadings (the coefﬁcients ﬁg, ,6“, and ﬂi4_ from equation (2)) between team and sole-
managed funds.

We estimate all regressions for all ﬁmds-years in both the ﬁrll sample period
(1997 to 2004), and the two separate sub-periods (1997 to 2000, bull market; 2001 to
2004, bear market). We also report results separately for growth oriented (prospectus
objective of growth and aggressive growth) and income oriented funds (prospectus
objective of growth—income and equity-income). Wermers and Ding (2005) examine the
effect of manager characteristics and report that their ﬁndings are signiﬁcant only for
growth-fund managers. They posit that the reason could be the difﬁculty in accurately

forecasting earnings growth for growth stocks that requires more experience or

22

specialized skills. It is interesting to see if differences between growth-funds and income-

funds are also present in our study.

1.4.2 Data Description
1.4.2.1 Mutual Fund Data

All of our mutual ﬁmd data come from the nine January CDs of Morningstar,
Inc.’s Principia Mutual Funds Advanced database from January 1997 to January 2005.10
The January CDs report data as of December 31St of the previous year. Morningstar, Inc.
started the Principia database on January 1996. The Principia Mutual Funds Advanced
version contains more information, especially regarding managers and monthly fund
returns, than the basic version of the database. Using the nine CDs, we extract
information for all ﬁrnds in operation every year from 1997 to 2005.

We start with all the funds in existence in January 1997 and we follow them
through 2005 or until they disappear from the database. We also include in our sample all
the funds that started their operations after 1997 to minimize concerns about survivorship
bias.

Data are gathered for all domestic equity funds with a self-declared investment
objective of growth, aggressive growth, growth-income, or equity-income. We exclude
index funds, balanced ftmds, funds of funds, as well as other types of ﬁrnds that are

restricted in some sense in their investment decisions.11

 

’0 Morningstar, Inc. used different names for this database throughout our sample period. The three
different names are: a) Principia Mutual Funds Plus, b) Principia Mutual Funds Pro Plus, and c) Principia
Mutual Funds Advanced.

” These include socially conscious funds, life cycle funds, target retirement funds and tax managed funds.

23

For each fund we obtain annual and monthly returns, annual expense ratios and
loads, net asset values, total net assets, fund inception dates, mutual fund family names,
portfolio characteristics such as turnover, total number of holdings, percentage of assets
invested in the top 10 holdings, stock, cash, and bond holdings, as well as manager
names. In the “manager name” ﬁeld the database lists the name of the manager if the
fund is solo managed, the names of the multiple managers if the fund’s total assets are
divided among more than one manager, or the term “Management Team” when more
than two people are involved in the management of the fund and they manage together.12

From the advanced analytics view of the database we hand-collect each fund’s
management fees, which are the fees that the management company charges to manage
the fund’s portfolio. For most ﬁrnds the management fee on the database is taken from
the fund prospectus. For other funds a minimum and maximum management fee range
appears in the database; for such funds we calculate the midpoint and use the resulting
ﬁgure as the fund’s management fee.

The funds that appear in the Principia CDs represent fund offerings that the
investor can choose from but do not represent distinct investment portfolios.l3 However,
while various share classes offer investors different ﬁrnd choices, they are based the same
underlying portfolio and consequently the same before-fee performance. Our unit of
observation is the fund. We therefore aggregate multiple share classes into one fund

observation. We are careful to cumulate the total assets from all share classes to obtain

 

‘2 The exact description of what the term “Management Team” means, reads are follows: “This is used
when there are more than two persons involved in fund management, and they manage together, or when
the fund strongly promotes its team-managed aspect”.

'3 As Nanda, Wang, and Zheng (2005) document, in the 19905 many mutual funds introduced additional
share classes as a way to offer investors more choices about the timing of load payments, or to provide
lower expenses to investors with big holdings. They show that by the end of 2002 more than 50% of mutual
funds offered more than one share class.

24

 

 

 

the total assets of the underlying portfolio. In order to identify different share classes of
the same fund we match different share classes by four portfolio characteristics: turnover,
number of holdings, percentage invested in stock, and percentage in the top 10 holdings.
We also verify our matching by looking at the ﬁrnd names.14 Table 1.1 has all the funds

reported in the Principia database as well as the total distinct fund portfolios we identify.

1.4.2.2 Manager Data

From the advanced analytics view of each CD, we hand-collect additional
information about all portfolio managers that are members of a portfolio management
team in our sample. The Principia CDs contain a brief biographical sketch for each fund’s
manager(s). For each manager, we collect data on the starting date at the fund, gender,
undergraduate and graduate institutions attended, degrees received (including the year in
which the degrees were received), whether they are a Certiﬁed Financial Analyst (CFA),
the name of the management company for which they work, and other assets managed.
Note that for the database’s “Management Teams,” we can extract starting dates and
management companies’ names, but not manager-speciﬁc names or other information.

After collecting manager-level information from Morningstar, we turn to other
sources to complete missing information. We ﬁrst turn to the 2004 CD of Nelson’s
Directory of Investment Managers. Nelson’s 2004 CD-ROM has information about most
of the management companies and managers in the portfolio management industry as of
March 2004. Thus, matching manager names and management companies from the 2004

January Principia disk with those from Nelson’s CD, we try to retrieve as many missing

 

'4 Multiple share classes of the same fund have basically the same name. Their names differ only by the
name of the share class. Example: “Vanguard Growth A,” “Vanguard Growth B,” etc.

25

data as possible. We then turn to each fund’s prospectus, which we locate on the fund
family website. After completing as much information as possible for the managers of all
the funds that appear in our data set in 2004, we track those managers in earlier years and

complete their missing information.

1.4.2.3. T eam-level Variables

Using the manager characteristics data of managers that work in teams, we create

the following team-level characteristic variables:

1) Team size: We measure team size as the number of managers in the team.

2) Team tenure: We deﬁne the team tenure variable as the time (in years)
managers have been working together as a team. For teams whose managers
joined the fund at different dates, team tenure is calculated from the time since
the latest team member was added to the team.

3) Team Diversity variables: We create four diversity variables: 1) Gender
diversity, 2) MBA diversity, 3) Age diversity, and 4) Tenure diversity. Gender
diversity is the standard deviation of the values of the dummy variable Gender
for all managers in the team. The same method is used for the other three
variables (MBA, Age, and Tenure). Each of these variables take the value of
zero if the members are exactly similar in the corresponding dimension (for
example, all managers are male, have the same age, etc.) and take positive

values if members are dissimilar.‘5

 

‘5 There are other ways to calculate diversity. When we conduct the analyses using the coefﬁcient of
variation of manager characteristics instead of the standard deviation, the results are similar. One has to be
careful, though, with respect to what each metric actually measures. Consider the case of two three-member
teams, the ﬁrst having two managers with MBAs and the second only one. The standard deviation measure

26

4) Other team: This variable is created by averaging the values of the multiple
funds dummy variable and shows what percentage of the fund managers are
also employed by other ﬁrnds. This variable takes the value of one if all
managers work for multiple funds at the same time, the value of zero if none
work for other funds, and values between zero and one depending on how
many managers work for other funds. We use this variable to proxy for the

level of commitment the team members have to the fund.

1.5 Hypotheses and Results

1.5. I Single Managers versus Management Teams

The discussion of the research on the behavior and performance of teams versus
individuals leads to our hypotheses concerning the characteristics, risk attributes and
performance of sole-managed and team-managed mutual fund portfolios.

The portfolio characteristics we look at are turnover, number of securities in the
portfolio and the concentration of investment in the top 10 holdings. Consistent with the
superior information gathering and processing ability of teams of managers we expect to
see a signiﬁcantly higher number of securities in team-managed portfolios and at the
same time less concentration in a small number of securities. In terms of portfolio
turnover, if the theory suggesting that teams are associated with time delays due to the
time it takes to reach consensus on which securities to buy and sell is true, we expect to

see lower turnover in team—managed portfolios.

 

would record the same level of diversity for both teams, but the coefﬁcient of variation measure would
record different results.

27

To examine differences in portfolio characteristics we estimate the following

regression model in addition to presenting difference—in-the-means tests:

PortChar, = a + b, (T eam,) + szgtFee, + b3L0gAssets, + b4FundAge, + b5IntAdv, + Em , (3)

where Team is a dummy variable that takes the value of one if the fund is team-managed
and zero otherwise, MgtFee is the management fee charged by the management
company, LogAssets is a measure of fund size and is calculated by taking the log of the
average of the fund’s assets at the beginning and end of year t, and Age is the fund’s age.
IntAdv is a dummy variable that takes the value of 1 if the investment advisory company
the manager(s) work for is afﬁliated with the fund family complex and 0 otherwise. We
estimate the model for all three of our portfolio characteristic variables.

Table 1.2, panels A, B and C present tests for differences in the means between
team- and sole-managed funds. We ﬁnd that portfolios of team-managed and single-
manager funds are quite different. As hypothesized, portfolio turnover is signiﬁcantly
higher for single managers (95.56 versus 87.17 for teams), while teams hold more
securities in their portfolios. The average number of stocks in a sole-managed portfolio
is 90 compared to 102 for a team-managed portfolio. Team-managed funds also have less
concentrated holdings as indicated by the percentage of money invested in their top 10
holdings (difference of about 2.18%). All differences are signiﬁcant at the 1% level and
hold for both the full sample and the two sub-periods.

Findings from multivariate regressions are presented in table 1.3 and are very

similar. Teams of managers show lower trading propensity and hold more stocks and less

28

 

 

- .f‘....-- It. s chi. .. .h

~ 7)
It
a
"' mas-r

i.

 

___».~ .

m ..
l
|

5

EV

concentrated portfolios. In table 1.3, some other interesting ﬁndings are present as well.
Funds that are managed internally are quite different that outsourced funds. There exist
signiﬁcantly high differences in turnover, number of holdings and concentration of
internal and external funds. Speciﬁcally, investment advisors that are affiliated with the
fund family trade more aggressively (higher turnover), hold more stocks in their
portfolios and concentrate less in their top holdings.

Another interesting ﬁnding is that investment advisors that charge higher
management fees seem to be more “conﬁdent” in their abilities. They turnover their
portfolios signiﬁcantly higher and concentrate in very few stocks.

The above ﬁndings suggest that the two distinct forms of organization (single
manager versus management teams) exist and the distinction between team-managed and
sole-managed funds is not just a reporting issue as suggested by Massa et a1. (2006).

As discussed in the literature review research on the performance and risk taking
of teams and individual decision makers has produced mixed results. There are
advantages and disadvantages associated with team performance so our empirical tests
will capture the net effect.

Mutual fund families, in their prospectuses, suggest that beneﬁts of the team
management approach are higher diversiﬁcation, less risk and better risk-adjusted returns.
However, our discussion of empirical studies implies that sometimes teams exhibit risk
increasing behavior when group polarization and groupthink are present. We look at three
measures of portfolio risk: a) the standard deviation of monthly returns, a) the beta from
the market model and c) the market beta from the 4-factor model. We use those three

measures at the dependent variable in the following model:

29

 

 

1:

F undRisk, = a + b 1 (T eamt) + bZMgtFee, + b3L0gAssets, + b4 F undAge, + b51ntAdv, + 81-, t , (4)

where Team is a dummy variable that takes the value of one if the fund is team-managed
and zero otherwise, MgtFee is the management fee charged by the management
company, LogAssets is a measure of fund size and is calculated by taking the log of the
average of the fund’s assets at the beginning and end of year t, and Age is the fund’s age.
IntAdv is a dummy variable that takes the value of 1 if the investment advisory company
the manager(s) work for is afﬁliated with the fund family complex and 0 otherwise. We
also estimate regression 4 using three more measures of risk and/or investment style as
the dependent variable: 1) the four-factor SMB beta, 2) the four-factor HML beta, and 3)
the four-factor UMD beta as calculated from the performance regressions described in
Section 1.4.1. Results are reported in tables 1.4 and 1.5.

Table 1.2 reports differences in the means. Total fund risk, as measured by the
standard deviation of monthly fund returns, is lower for team-managed portfolios and the
relationship if most signiﬁcant during the bear market period 2001-2004. However,
results from the multivariate regression results in table 1.4 show, that after we account for
some control variables, the team dummy coefﬁcient not signiﬁcant, even though its sign
is negative. The same is true for our measures of systematic risk. Therefore, we fail to
ﬁnd evidence that teams of managers hold less risky portfolios.

We ﬁnd that management fees and the internal investment advisor dummy
variables have signiﬁcant and positive coefﬁcients. Advisors afﬁliated with the fund

family hold much riskier portfolios both in terms of total and systematic risk. The

30

coefﬁcient for the internal investment advisor dummy is 0.254 and is signiﬁcant at the
1% level when standard deviation of monthly returns is the dependent variable. When we
look at the bull and bear market periods separately, we ﬁnd that most of the signiﬁcance
comes from the bull market period. Advisors that charge higher fees also take on more
risk and results are signiﬁcant for all sample periods.

Turning to results reported in table 1.5, the most signiﬁcant ﬁnding is that team-
managed mutual ﬁrnd portfolios have higher HML loadings during the bull market
period.

To examine differences in performance between teams and single managers we

estimate the following model for each performance metric:

Perf, =a + b 1 T eam, + b2Turnovert__1 + b 3mgtfee, + b4LogAssetst

, 5
+ b5FundAget + b6lntAdvt + a, ( )

where Team is a dummy variable that takes the value of one if the fund is team
managed, Turnover is the fund’s turnover over the last year, LogAssets is the logarithm of
the average fund size, Age is the fund’s age, and i is the index for an individual fund.
IntAdv is a dummy variables that takes the value of 1 if the investment advisory company
the manager(s) work for is afﬁliated with the fund family complex and 0 otherwise.

Results from performance regressions are reported in table 1.6. The two
performance variables we look at are Jensen’s alpha and Carhart’s 4-a1pha. We also run
the analysis for style adjusted excess returns and for 3-factor alphas. We do not get any

signiﬁcant results when we use the style adjusted excess returns and results for 3-factor

alphas are very similar to those of the 4-factor alphas.

31

In panel A of table 1.6, which presents results for the whole sample period, we get
a negative coefﬁcient for the team dummy for all funds and especially for grth
oriented funds though signiﬁcant only at the 10% level. The most interesting ﬁndings
appear when we focus on the bear market period. Team-managed ﬁrnds perform worse
than single managers (46.3 basis points a year) in terms of 4-factor alphas. For grth
oriented funds the underperformance of team-managed funds is even more signiﬁcant
and reaches 61 basis points annually. Both coefﬁcients are signiﬁcant at the 5% level.

Before February 2005 SEC regulations did not require funds to report the
portfolio managers’ names as long as the fund was team managed. Many critics of the
team-managed approach have argued that teams often are the training ground for young
inexperienced managers. They also complain that “without a clear portfolio leader nobody
can be held accountable for poor performance or rewarded for higher retums. This
depends, of course, on how the team is organized. Single managers might be more
concerned about their fund’s performance since they bear sole responsibility and stand to
receive the entire management fee, whereas managers that are members of a management
team might not work as hard (free-rider problem). This problem might be more important
than one may initially think if we take into consideration the fact that many of the
managers that are members of a team usually manage multiple funds.

In order to check whether the common belief that when funds do not disclose the
names of the managers in their teams it is because those managers are young,
inexperienced and have inferior investing skills is true, we perform an additional test. We
re estimate the performance regressions this time replacing the team dummy variables

with two other dummy variables. The former takes the value of 1 when the fund is

32

managed by a team with known manager names and 0 otherwise. The latter takes the
value of 1 when the fund is managed by a team with undisclosed manager names and 0
otherwise. The omitted category is the sole-managed fund.

Results are also reported in table 1.6. Again, we only get signiﬁcant coefﬁcients
for the bear market period. Teams of managers that do not disclose manager names
under-perform by 83 basis points a year (p-value of 0.018) compared to
underperformance of 38 basis points for teams of managers that disclose manager names.
Focusing only on growth oriented funds underperformance of that type of teams is even
greater (almost 91 basis points signiﬁcant at the 5%). Those results provide some support
that anonymous management teams perform worse, but also suggest that there is still
underperformance but teams with known managers after we account for anonymous
teams.

Taken together, ﬁndings reported on table 1.6 suggest that sole-managed funds
produce better risk-adjusted returns than team-managed funds, especially in poor market
conditions. This opposes investors’ beliefs and fund complexes’ claims about the

advantages of the tearn-based approach in portfolio management.

1.5.2 Management Team Characteristics

The internal organization of portfolio management teams might also have a
signiﬁcant effect on how the mutual fund is run. For instance, are the characteristics of
managers in a team usually correlated? Is it better to form a team of managers with

similar characteristics or should we look for diversiﬁcation of characteristics and skills?

33

Do larger teams perform better than smaller ones? Since more than half of today’s mutual
funds are team managed, the answers to these and similar questions have important
investment implications. Accordingly, we investigate how team—level (team size,
diversity) and manager-level (education, age, experience) characteristics of management
teams relate to performance and investing behavior.

The ﬁrst team-level characteristic we consider is team size. A team’s size can
affect many aspects of team behavior and performance. As we discuss above, larger
teams may have a broader range of skill and knowledge, but team members might not be
as motivated to contribute Further, the larger the team, the longer it may take to reach a
consensus, increasing the possibility of time delays in decision-making.

Next we look at team diversity. As described in section 3, we measure diversity in
four dimensions: education, age, gender, and tenure. The discussion of related research
implies that bringing managers with diverse backgrounds, specialized knowledge, and
unique experiences together in the same team can increase the beneﬁts of diversiﬁcation,
but can also lead to more conﬂicts. On the other hand, including people with too similar
characteristics in the same team might not lead to the desired diversity a team is supposed
to achieve.

Since research on the effect of team diversity on performance has produced mixed
results, our empirical tests will show what is true for or sample of mutual ﬁrnd
management teams. We posit that education and experience diversity will have a positive
effect on performance and that tenure diversity to have a negative effect, as it my lead to
more conﬂicts. Managers that work at the fund for a long time might not agree with the

recommendations of new managers, especially when it comes to changing decisions that

34

the former made in the past. We use age diversity as a proxy for experience diversity;
however, age diversity might also have a negative effect to the extent that managers of
different ages do not get along well.

The next team-level variable we analyze is team tenure. Team tenure is deﬁned as
the amount of time the team has been working together. Longer team tenure is an
indication that the team members get along well and are producing satisfactory results;
otherwise, a change in the composition of the team would occur. Team tenure might also
affect the importance of other team level-variables. For example, we would expect the
negative effects of an extremely diverse team, such as disagreement and time delays, to
be more signiﬁcant in newly formed teams than in teams with longer tenure. We
hypothesize that teams with long tenure and high diversity levels to perform the best,
since they have members with diverse backgrounds that seem to get along well. In sum,
we estimate coefﬁcients in the following regressions to investigate the impact of team

characteristics on portfolio characteristics, risk, and investment style:

Dependent, = a + b I (T eamSize,_ 1) + b2 (GenderDiv, ) + b3 (A geDivt ) + b4 (T enureDiv, ) 5
+ b5 (MBADiv, ) + b6 (Team Tenuret) + b; (Other, /1 00) + m, + 3,, ’( )

where T eamSize is the number of managers in the team, GenderDiv is the standard
deviation of the values of the gender dummy variable for all managers in the team,
AgeDiv is the standard deviation of the managers’ ages, T enureDiv is the standard
deviation of the managers’ tenure, MBADiv is the standard deviation of the values of the

MBA dummy for all managers in the team, T eamT enure is the amount of time the team

35

 

 

 

has been together, and Other is the percentage of managers employed in other funds. X
includes all the control variables included in equation (3).
To examine differences in performance between teams and single managers we

estimate the following model for each performance metric:

Perft = a + bITeamSize, + bZGenderDiv, + b3 AgeDiv, + b4TenureDiv,

, 6
+ b5MBADiv, + b6Team Tenure, + b70ther, + ng, + 8” ( )

Variable deﬁnitions are the same as in equation (4) and X corresponds to all the control
variables included in equation (3). We also re-estimate the performance regressions
including some interaction variables of team-level characteristics

Table 1.7 reports the correlation coefﬁcients between all the team-level variables
we examine. The highest correlations always involve the team tenure variable.
Speciﬁcally team tenure is negatively correlating with team size (-0.1466) and tenure
diversity, suggesting that smaller teams and teams of member that started working
together approximately at the same time are the ones that are more likely to survive
longer.

Results on the relationship between portfolio characteristics and team-level
characteristics are presented in Table 1.8. Team members that have been working
together for a long time turn their portfolio over much less (coefﬁcient for team tenure is
-4.106, signiﬁcant at the 1% level). This result holds even after we look at the two sub-
periods separately. One possible explanation for this ﬁnding could be that such teams

chose which stocks to buy when they initially formed their portfolio, and they do not

36

 

 

. 1* an

nk‘: '

.I..l. ‘L...’ II PILU. I._r .

easily agree on changes to their securities choices. This explanation may also shed light
on why long-tenured teams have more assets concentrated in their top 10 holdings.

Teams that exhibit high gender diversity also show signs of lower portfolio
turnover. The coefﬁcient for gender diversity is negative and statistically signiﬁcant (-
27.263, with a p-value of 0.015). Mixed gender teams also hold signiﬁcantly more
securities in their portfolios during the bull market period. The coefﬁcient for gender
diversity is -37.809 for the 1997 to 2000 sample period.

We do not ﬁnd that any of the other diversity variables are signiﬁcant, except for
age diversity, which has a positive and signiﬁcant relationship with investment
concentration in the top 10 holdings of the portfolio. Finally, we ﬁnd evidence in support
of the hypotheses that larger teams have more resources and can follow more stocks. An
additional team member increases the number of securities in the portfolio by 16 (p—value
0.000) and decreases the concentration of invested assets in the top 10 holdings by
1.571%.

In terms of risk taking team size seems to have an effect (table 1.9). Teams with
more members exhibit a lower standard deviation of monthly returns for the full sample
and for the bear market period; we do not ﬁnd a signiﬁcant relationship for the years
1997 to 2000. Long-tenured teams also take on less risk as measured by the standard
deviation of returns, and this holds for all sample periods. Finally, when managers work
for multiple teams, the returns of their funds tend to have a higher standard deviation.

Long-tenured and larger teams also have lower market betas. The coefﬁcients for
team size and team tenure in Table 1.9 are both negative and have p-values of 0.070 and

0.002, respectively. Managers working for many funds hold higher beta portfolios too.

37

In Table 1.10 we present results on ﬁrnds’ investment style (factor weights). We
do not see any striking differences. The only variable that is consistently signiﬁcant is
age and gender diversity; teams with high age and gender diversity have lower HML
weights. Teams with high gender diversity also have higher UMD weights.

Table 1.11 reports results on the performance regressions. Looking at the full
sample period we can see that age diversity is very important for performance, especially
for grth oriented funds. Coefﬁcients of the age diversity variables are highly
signiﬁcant regardless of the performance metric used (1-factor or 4-factor alpha). We
ﬁnd lower 1-factor alphas for teams with high gender diversity. Another interesting
ﬁnding is that funds run by internal investment advisory ﬁrms under-perform their peers.

We do not ﬁnd any signiﬁcant relationship between team characteristics and
performance for the bull market period. In the bear market period, however, we ﬁnd
evidence that age diversity is very important for performance (all coefﬁcients for age
diversity positive and signiﬁcant at the 1% level). Gender diversity and team size are
negatively associated with performance. Finally, in terms of l-factor alphas, funds that do
not have managers that work for multiple ﬁmds do better. Results, again, hold only for
growth-oriented funds.

In table 1.12 we present results from another set of regressions. We re-estimate
the performance regressions, but this time we include a set of interactions variables. We
interact team size with some diversity variables, fund size, participation of members in
the management of other funds and team tenure. The logic is that the negative, or
positive, effects of some variables may become more or less important depending on the

number of people in a team. We report only the coefﬁcients for the interaction terms as

38

well as other variables that show signiﬁcance. However, the regressions include all
variables used in the performance regression in table 1.11.

The most signiﬁcant ﬁndings is that the T eamSize*Team Tenure has a positive and
signiﬁcant sign, especially for growth oriented funds. This implies that larger teams that
have been working together for a long time perform better. An explanation for that could
be that larger teams have a broader range of skills and when they have been working
together for a long time people in a team get along well, learn how to work with each
other and consequently the advantages of the team dominate the disadvantages such as
disagreements and conﬂicts.

We also ﬁnd that the interaction variable of T eamSize and AgeDiversity has a
positive coefﬁcient, but is only signiﬁcant (at the 1% level) for the bull sample period

when l-factor alpha is the performance metric.

1.6 Summary and Conclusion

In this paper, we have examined the effect of the structure and characteristics of
mutual fund portfolio management teams on mutual fund portfolio performance, risk and
characteristics. Speciﬁcally, we have compared the performance of sole-managed and
team-managed mutual ﬁmd portfolios, as well as the effect team-level characteristics
have on performance. Our study uses a unique and comprehensive dataset that covers the
period between 1997 and 2005 and enables us to investigate whether differences in
performance and trading practices between different types of funds depend on the

underlying market conditions.

39

The analysis of team- versus sole-managed funds indicates at ﬁrst that single
managers behave differently than teams. Single managers hold fewer stocks in their
portfolios concentrate more in their top ten holdings and have higher trading propensity.
Even though the mean portfolio risk, as measured by the standard deviation of the ﬁrnd’s
returns, appears to be higher for single managers, in a multivariate regression setting we
fail to ﬁnd evidence that there are differences in risk between sole-managed and team-
managed funds.

In terms of performance we ﬁnd that team-managed fimds under-perform their
sole-managed counterparts in terms risk-adjusted returns. Even though there are not
differences in performance in the bull market period 1997-2000, team managed funds
under-perform by 46 basis points. For growth oriented funds (prospectus objective of
growth and aggressive growth) the under-performance is most severe (61 basis points
annually). Interestingly, most of the under-performance seems to come from teams of
managers that do not disclose their managers’ names, supporting the View that non-
disclosure of manages names indicates inexperience and low quality of the management
team.

We next focus on team-level characteristics of portfolio management teams. We
test whether a team’s size, tenure, and diversity affect performance and trading behavior.
The results Show that teams of managers that have been working together for a long time
exhibit lower turnover and higher concentration of investment in their top 10 holdings.
Further, larger teams hold signiﬁcantly more stocks in their portfolios, as an additional

team member increases the number of securities in the portfolio by 16.

40

Team size has also an effect on risk taking. For the bear market period (2001 to
2004), larger teams take on less total risk as measured by the standard deviation of
monthly returns and hold lower beta portfolios.

Turning to performance, we ﬁnd that age diversity, which is also a proxy for
experience diversity, is very important for performance. The coefﬁcient of age diversity
is positive and very signiﬁcant especially in the bear market period, regardless of the
performance metric used (l-factor or 4-factor alphas). Finally, we ﬁnd that it is not a
good practice to have managers working for multiple ﬁmds as this reduces the fund’s
performance (in terms of 1-factor alphas), supporting the hypothesis that those managers
are less committed to the fund.

In general our study provides new insights on the performance and organizational
structure of portfolio management teams and opens up possible new research dimensions.
We ﬁnd that teams do not out-perform single managers. This cannot explain the choices
of ﬁrnd families and investors who seem to prefer team-managed funds.

Maybe there are other beneﬁts to teams of managers that make them a more
popular choice. One of the proposed advantages of management teams are consistency in
performance and stability of management. Stability of management refers to smooth
transitions in the fund’s investment approach when there is management turnover. In
teams, when one manager retires only a small portion of the portfolio changes hands. In
contrast, single-manager turnover can lead to a complete change in investment style.
Investors prefer a more rather than less stable fund management. When the fund has only
one “star” manager, investors are likely to follow him if he switches to another fund.

Anecdotal evidence from informal discussions with investment professionals indicate that

41

 

 

the fear of losing investors following manager turnover is the most important reason for
adopting a team-of-managers approach. We believe future research should address those

issues in depth.

42

APPENDIX 1.

TABLES OF ESSAY 1

43

 

$.me

 

 

 

 

 

 

 

8&8 $58 on»? 9.3 8.85 8.23 9.2m 5.8m 85m: amazes oz
428; 8.83 2.82 8.83 2.23 8.83 8.2} 8.33 Seem «can: 388:
8.8m... 8.83 8.83 cosmos «~08; 8.85 8.83 .8. F 3.. mm. Km mace 8855-58»
8.83 8.83 8.83 3.83 8.33 3.83 9.25 2.53.. $.83 85.. ummmcmetasw

$5.0: do 955:5 35 2.0.3.5.“. 39024

SE 8.8 8.5 mm. 8 5% 2.2: 3. 5 3.8 $62 was? .829: oz
8.8 E 2.95 888 8.08 8.25 3.8m 8.84 w P .mmm 8.88 852 5852
5.33 8.33 8.48 ocean 8.5m as: $.36 8.8». 8.8m 38k
8.8m No.45 8. .8 B. 5m 8.3a 8.8: 2.. .8 6.2m $.88 «3885.055

Esau do 22:2: £82 .98
as? as? as $9 03 as as» an as? $52 $ng oz

s9 as? $9. $8 $8 s8 can as 8 $8 852 588.2

s8. s8 s3 $8 as... a? some s8 s8 Asamcmﬁemﬂ

*8 some $9. #8 s8 s8 $8 $8 $8 3°C commemeéow
an; o 5; 83 SN; 2:; So; one one 92 8:0an 25.. .252:
85. an}. new.” 83 33 83 $5 33 new; $88.0 Base 8%“.

38 38 88 88 38 88 82 83 82

 

once 3‘ .< Emu

82.388 on He: 2:8 85 £83

new as 888 8080388 2: mm :03 we :8 .80» 8.5 $88 8 N :3» 88m £83 £088 03 St: 888% 2: 8 9:33 85 $83 2: mo cab
8:8 808owm§8 05 use 858:: 2: 888qu m 38$ .88: ~58ome§8 mo 25 some new 36 Eaton omen—o: 08 use comes—«8 30mm.“ :38
05 3:88: < Esau we «won 2.: .888— comecaE 8505— 5.5, $83 womu§-8m8 mo owﬁaoobm 2: we :95 mm Bw::mE-8aB no comma—«8-03m
8a 85 888% 05 8 828.com :83 88:8 2.: mo owﬁeoouoa 05 387. 9m 36% .mpmmﬂo Beam 282:8 8.8 8:88 03 Home :8» some 8:880:
88¢ 38:8 8886 me 838:: 05 gm: < 38:: mo 38 @888 BE. .Amoom $85. I 33 8:85 RC Deana 8838832 08: 05 :o
32E: 85 £83 88:8 mo 888:: 05 $8808 < 38$ mo 38 «we 2:. :3va 8o 8 3338 £83 2: we moratocogno @888: £93 8:.

Ea88:m «83.5

ﬁn ﬂash.

44

 

$5

$mm

o\oom

“we;

*2.”

excum

08mm

#8 8C

 

 

 

m2 8 8 8 we we 8 8 5882-232
$8 $8 $8 $8 $8 s8 $8 #6 8C

5 8 8 3 8 8 a. 8 $885.28....”
88 at m? 8 m2 8 8 8 8...... 8685::
88 can 88 as. an» .8» s8 88 8C
28 N8 «8 8m. 84 «8 EN 8 8885-222
.8» 8.. son s8 s8 s8 s8 s8 3
«8 m8 8m NE 88 8m. 84 .8 8:3 3885.285
2: 5: 8: 82 8 m3 «8 m8 8%". 866.2

8.88 3.88 8.88 8.88 3.88 8.82 8-82 8.82

 

“+8 cook 5.3. Emmi 93088 corn wheat .m Ecol

 

3.83 8 938.

45

 

 

 

46

«8.: 288 m E.» 828288 B 8:52

Fooo «88 8.3 83 83 23:50 as. 852:.
88 ooo.o 2o.o Koo 2o.o 9mm 92:
Boo 8o.o- 2o.o ooo.o. ~oo.o sum 4.2...
2o.o Soo- 83 83 83 sum 88
8o.o 2o.o. 83 ooo.o goo 88:. 368$ 93
8o ~oo.o. 83 8o.o ooo.o 88.: .988 Sam
ooo.o 288 88.? 2 to Boo. 8:8 Boos-v
~8.o :88 83 8.8 o8.o 8%.. 9:88..
ooo.o o8.o 8o...- 23 82:- msae 25:9: Em
o8o 89o- :88 8.8 2o.o 382 .o 8..
ooo.o 83. 88.8 88.8 88.8 888.0 28w .6 8:52
89o o8.o 8.2 :82 8:2 8< 8:“.
ooo.o 82.8 8o. 8 88.8 88.8 8:66: 2 :2 o. 989..
ooo.o 892- 2o.~2 8&8 8:8 2.2.3: :_ 88:08 .o .8532
ooo.o 83 8 Em 898 Km. 8 .9653 2.2.3.".
83 too- vino 83 83 mo. 8:58:92
o8.o 82o- 8o.o ooo.o. ooo.o :58. 888 8.3.8 8.8
o So at. 2 ooo.o 892 8.8 532 8:8 2mm
cams-o gawk-295v ﬂow-amen: Emmb Cmmgmz-Qmswv Amoco“. =$ w_om_._m>

855:5 :85. 2mm: :82

 

Room-mam» botml ..< 3th

 

.uooﬁEon 08 .5qu .8 _o>o_ $2 05 8 Ema-«Ema 2&3 33:88-28 28 comma-Emu“ 50385 8058wa 5038088 mgr—om
gem-Sam 28 88-33 05 :8 83288 088 2: E805 U 28 m 238 BE? wotum 28:8 =2 2: Sm 8:288 gm 2885 < .ocam
233 05 «o 5.508 83 2: E ova-woo 8 08 83893 .2882. uﬁ E 32830on Bum-cash 2 ho =8 pom 88.58.0- otatomow 2:805 038 2.3-

oowmzﬁm b.2555
«A «Bah.

 

 

 

5 F0... 000. 5 mm _. .0 0000030000 00 000052

005.0 000.0 005.0 005.0 N¢5.0 >009000 WOT—2 _0E0E_
03.0 05.0 0 50.0- 000.0 000.0- 0.00 0.23
000.0 050.0. 000.0 0 5.0 03.0 0.00 4.2....
..00.0 000.0 up 5.0 0.. ...0 N_. 5.0 0.00 0.20
0 50.0 VN0.0. 000... 000.0 ..00.0 20008 00.00040 0.00
V000 QN0.0 30.0 000.0 000.0 00005 .mvtmE. 0.0m
$0.0 050.0. 00044. 000. F 00.0 0050.0 00000:.
03.0 000.0. 000d 000... 000d 020.0 0.000000
35.0 000.0 500.0 30.0 000.0 0530.. 5.500.: 00.0
v ..0.0 #5 v.0. 005.0 0 50.0 500.0 $0004. 00 00..
50.0 03.0. 000d 000.? 000.? 000005 000:0 00 000832
00 .. .0 005.0. 050.3. 03.9 v0 v.3. 00< 00:“.
..00.0 00*... 0 50.00 00.10 0 50.00 0050.0... 0.. 00. 0. 30004.
30.0 53.0. 000.00 005.00 v00. 50 0:00.000 c. 000.5000 00 000052
000.0 Q0 _..0 000.N0 50N.00 000.50 .9053 0:0an
000.0 30.0. 0 ..5.0 0 _.5.0 5 55.0 000 50500000.).
8.00 33. awed- m 36 ooo.o 522 385 8.2.8 30
000.0 F00.N 00109 00?: 000.0 .. 530.. $0000 30m.
029-0 Emoﬁoﬁgg €000.52 £00.... 000052-039 30:2“. =$ 0_nm_.m>

00:90.50 :00: c005. 0005.

 

000m .5 00 5 Eton. ..m 50th

 

3:80 2 as:

47

 

 

 

«3.0 S 00 50... 220238 S .8632

80... 8o... 83 .000 55.0 38:50 05500205. .0525
~85 086 v8... 030 58.9 0.00 0.2:
080 «8.... «8.... 03¢. 08.0. 0.00 .__2_._
02.... mood. «2.0 00.... .2... 0.00 0.20
«8... .80 08.. 08.. «8.. .085 3.00-... 0.00
08... v8.9 .03 83 08.. .08.: 09:05. 0.00
to... 08.0 80. 7 .3. .. mm... 7 2290 5.00.-..
80.0 03.0 08.9 «and- 08.? 2% 0:35..
ooo.o can... 80... com... 3.3. 9:29 258.: 90
$06 08.? 80... EB 50... 032 .o 03
ooo.o 82.. 80.0 3.0 80.0 $320 220 3 025:2
83 20o 35.0. 05.9 «5.2 0? 2:“.
08.0 36.0 08.8 cm .00 «no. 5 .3529. o. no. 2 032.
ooo.o 36.9. 30.8. 30. a 80.8 26.08 s 02.580 .o 085:2
ooo.o 3.0.2 08.0.. 80.8. 30.3 .9652 26an
«a... 08.0. 003 53.... 005.0 8.. 20500202
.80 50.... N. to onto- ooo.o 529 385... 02.3.60 0.20
.8... «2.0. .30 08.... mum... 539 0:20 sum
0:_0>.0 .0000 50610.0. .m00002 E00 .... 000.000.20.050. .0003“. .2. 0.0000>

0000.005 000.2 000.2 000.2

 

3:8. 3 2%...

VO0N-500N 00.201 .0 .0001

48

 

 

 

 

 

 

0.0.0 0. .0 0050.0 5 .0 .0 00.0 00500 0 .N.0 000 .0 00500 00.05000
.00.. .00.. .00.. mm .0 mm .0 mm .0 0 .55 0 .55 0 .55 0000050000
0000 0000 0000 0000 000.0 0000 0000 0000 0000
0 .0.5~. 505.00 050.3 N0 .00 000.3 000.50 0 .00N 000.05 0505* 80.0000
0000 0 .00 .000 0N00 0000 0000 0000 0000 .000
v .00- 0000 0000. 000. .. 00m. . . 05 .6 . 500. .- 030 . 000.0. .0033. .0808.
5 .N0 .000 0N00 0000 03.0 5000 «.00 .000 .000
000.0 .000- ~.00- 0500 NON0- 0000 03.0 30.0- 0000- 00< 00:“.
0000 0000 0000 0000 0000 N500 0000 000.0 0000
9.0.0- 0.0. . . 000.7 000. .- var-.0. 00.0. «N... .. 50.0 . 5 .00. 0.000... .0 00..
0000 0000 000.0 .000 500.0 0000 0000 0000 000.0
050.0 .800. 0 .5...” 35.0 005.0? 00...:- 0000 000.0..- 00060 00“. .05.
0000 0000 0.00 0000 000.0 0000 0000 .000 0 .00
1N0. .- «mud. «00.5- 000. .- «3.0 30.5. 0N5. .- 0000. «~15- 58050 E00...
000.20.. 0:0..th 0. .0505... [000.200 2.0an .0505... 000.20.. 2.00.0... 0. .0305...
0.00. 0. 000.5000 0 .00. 0. 0. 000.5000 0 .00. 0. 000.5000
00.008. 0.. .0 .0852 00.008. o\o .0 .0852 00.008. 05. .0 .00052
0000.“. .0000... .000 0000.... .0005. .50 0000.“. 0.00.00 =5“.

 

0808.00 0.0 .80.. .3 8080 05.80... 00.0.8.0 082000000 00.0808 0... 30.00 .0009. 000.85 .80 000000.00. 0.0 .0000 .0 .26.
$0. 0... .0 800.0806 080.0500“. .3080. .00 05 080600000 .00... 80 80.80.00. 0... 8 00.8.08 05 00.8820 08.. .80 03.00.30 0800085
.505 0... .8 00500 0.... 00.6 0.080 .00 .08. 00802. 0.080 0... .0 00. 0... .80 000 0.08.... 0... 5.80.. 0.8.. 0... 5..» 00.0.10... 0. .8060 808.008.
0.... 0. . .8 00.0.. 0... 000.0. .0... 0.008..» 5880.. 0 A5280 803.00 808.008. 00.. 5.88 0 00 3200 00.03. 08.500 808.38. 0... .3 000.000
00. 808000008 0... ”000 00.008... .8800 00... 000.20.. 0. 00. 0... 8 00.008. 0.0000 0.800 .8 00080800 0... .80 0:00.80 0... 8 000.8000
.8 .0088 00.688 2.0.0.50 505-00.30.80 ”0.8 8.00.08 8000000.. 00... 0030.00.» .8800 .80 000808-28 m. .80.. 0... 0. 0 .80 00088-800.
w. 080 0... 0. . .0 00.0.. 0... 000.0. .0... .0508... 588.6 800. 0 :0 800.00.00.80 0:00.000 .8 80.80.00. 0.5 0... 800 0.800.. 08000... 0.00. 0......

0.80m 000088-800... .80 000088-080 .0 000.020.00.000 000.80.— ..0. 80.80.03—
n.. 0.005.

49

 

 

 

 

0050.0 000.0 00 .00 0000.0 0000.0 5000.0 0550.0 0 .5 . .0 «5000 00.00.80-..
.000 .000 .000 mm .0 mm .0 mm .0 0 .55 0.55 0.55 0000020000
0000 000.0 000.0 000.0 000.0 000.0 000.0 000.0 0000
«N00 .050 000. . 000.0 000.0 050.0 .000 000.0 0.00 80.0000
0.0.0 0000 000.0 000.0 .000 000.0 000.0 .000 000.0
000.0 000.0 00 ..0 000.0 000.0 000.0 000.0 000.0 000.0 .00.>0< .0808.
000.0 000.0 000.0 .000 00.0 000.0 .000 000.0 000.0
.000. .000. 000.0. .000. .000- 000.0- .000. .000. 0000. 00< 00:...
000.0 .500 050.0 000.0 000.0 050.0 000.0 500.0 000.0
000.0 000.0 ..0.0 000.0 N .00 .000 000.0 000.0 000.0 0.000< .0 00..
000.0 000.0 000.0 0000 N000 000.0 0. ..0 000.0 000.0
000.0. 0000 0.0.0 0 .00. 0000 N0 ... 000.0- 000.0 000.0 00“. .05.
000.0 505.0 000.0 000.0 000.0 .000 000.0 000.0 .000
500.0 000.0 000.0- «000 0.00- 000.0- 0 .00 .000- 000.0- 5.0050 0.00..
..0.00. 00005. 005.0. .0 $0.00. 00005. 005.0. .0 €0.00. @005. 005.0. .0
-5 0.00 .9002. 8.5500 -... 0.00 .9002. 8.5500 -0. 0.00 .9002. 8.5500
.0005. 0.00 .0005. 05000.0 50.0.2 0.00 .0005. 05000.0 .9005. 0.00 .0005. 05000.0
00000 .0005. .000 00000 .0005. .50 00000 0.00.00 =0”.

 

00.08.50 05 08... 5.. 0.0.... 0505.0 00.0.5.0 080.0503 0058000 0... 30.0.. 50000
002.3-.. 0.5 000000.00 05 5.8.. .0 .26. $0. 0... 5 8000.005 08805000 08.000. .00 05 080.0508 ..0.0 8.. 88000.00. 0... 8 0000.08 05
00.8800 08.. 0.5 08.00.30 00.00000... .505 0... ..0 00.000 0... .26 0.0000 .00 5.0. 000.05 0.0.5”. 0... ..0 00. 0... 0.5 000 0.08... 0... 5.85.. 0......
0... ......» 005.3,... 0. 800.00 808.008. 0... .. . ..0 00.0.» 0... 08.0. 5... 0.5.5.. 58800 0 ..00.>00 80.8008. 0... 5.. 00055 00,. 80800058 0...
”05 0.0.5.5.. .8800 0...... ..0008 .8500 0.5050 80... 0.00 50.58 0... 0.5 .0008 .0858 0... 80... 0.0.. .0858 0... 0.800. 000.0 53808 0.08..
0... ..0 00.50.00 05000.0 505-08.52.00 “05 0.0.5.5.. 80000000 2.... 00.5.5... .2800 0.5 000058-200 0. 0.5.. 0... .. 0 0.5 000058-800. 0.
08... 0... .. . ..0 55> 0... 00.5. 5... 6.5.5.. 58800 050. 0 ..0 00.85.5500 ...... 8.0.0.00 ..0 88000.00. 0.0 0... 80... 0.800. 08000... 0.5. 0......-

0000. 000808-800... 0.8 000808-080 ..0 v.0.”— 0..0....0m .0. 000.000.03—
04 030,—.

50

 

 

 

 

«ooo.o 99o ~83 «ooo.o ooooo Nvﬁo otoo mouto 38o nemoowé
S3 53 53 «a; «N; «NE 9: or: 8:. 8262030
ooo.o ooo.o ooo.o ooo.o ooo.o Ergo Boo ooo.o ~oo.o
F Foo. o Eo «So. 53. ooo.o ooo.o. ooo.o. oouo auto. 3988
oomo ooo.o ooo.o Boo 53 «So woo «woo vomo
oFoo zoo. zoo. ooo.o ooo.o- ooo.o Koo Sod. 2o.o .832 REE...
ooto ooo.o ooo.o ooo.o «moo ooo.o vooo ooPo ooo.o
ooo.o ooo.o. ooo.o. ooo.o Pooo. Boo. ooo.o Fooo. Boo. oo< BE
ooo.o ooo.o oSo ooo.o ooo.o oNoo Eoo ommo «moo
ooo.o. Fooo. ooo.o. ooo.o ooo.o. «Foo. ooo.o. Boo. Boo. £32. oo oo..
33 Boo ooo.o Rho outo ooo.o ooo.o ooo.o ooo.o
moo? ooo.o and too moo? coho Eoo owoo. oo..o won 52
«So Soo ~33 «Bo ooo.o ooo.o mto ooo.o ooo.o
ooo.o. Boo- ovoo ooo.o. ooo.o :3 «So. «woo woo 5:50 E8»
388$ e923 .8.“ Cassi cosmoi :28“. i cooomoi c2034; asap. -3 353$
98 92: 9mm 42: 9mm £6 93 92: 3mm ...2: 98 25 3mm 0.22 9mm .__2_._ 9mm 9%

 

Boon. Ea: .QO 8.5a aims. :8 8:8 238 .5“.

 

.338th who was... 3 88.5 @8883 33320 858508
38833 05 323 889% mos—STAN 88 30823 08 Eton 8 3%: $3 05 “a 885%6 $86580 68.582 8: 8o 358508 :05 :5
82383» 05 8 39:28 08 8883 08¢ 88 338.30 £50385 .8?» 05 mo 8.88 uﬁ 850 33.3 on: :38 owﬂgo £83 2: .«o wE at
can own Mtg 05 588 85¢ of £5 Boa—Ea a 8833 8283>8 05 .2 _ mo 2:? 05 3&8 35 032.5» 8886 o .8833 808538 Eu
3 @0920 wow 808%888 05 ”So 3335, 35:8 BF .mmcuawwok 58o.“ QED 25 ASE .mzm 30>-w832—8 Ba $53.23 “cognac 2F
onion? .888 088 v5 8&8?“me a 28¢ 05 .2 o 88 Emacs—8-988 .8 28¢ 08 a _ mo 02? 05 8&8 35 .0585, 8886 83“ o no
3on ooooomé @8285 $888 8.3% one some 8.“ 38890 $832 “Bongo—mt ozomtoq «o 8838on mAO 05 88m £38 £588 £98 out.

...—EE— Bwa=a8.8aoh ...—a coma—88.9.8 you managea— uoauah ama \ 3.3m 89839:.— ..8 ouoﬁmouwom
m._ 033.

51

 

 

 

 

 

 

 

ond hood . 3.0 N Ed nood 0 5d omod mowd mood «mod odwd mood
ooo.o mood. 30d. ooo.o mood. 30.0. 000.0 mood. .ood- hood mood. Food. .03th 00000..
000.0 oomd mm .d mwmd .‘ .od vmod
.mod. 000d. oovd- owmd- Nomd- onod- 00E02 02 - E00...
womd «Ed 03d 03d unto «mud
03d nwmd. 50d- omwd Nowd. 000.0. 008.02 .00.). - 800..
mood wood ooo.o 000d mid noto
Nvod 000d. mvnd. ooo.o mid- mood- .5800 E00...
_ 0. .0 0 00.5”. _ .0 .0 0 00:0“. _ 0 .0 0 00.5". . 0 .0 0 00:30.
.0 0< ..< .0 0< ..< .0 0< ..< .0 04. _.<
.0. .0. .0. . 0.
02000-0. 0002 000.0 00.00.... 0002 00000

 

0080.0 00.: 00:00. 0.8000 .30 ..< .800.

 

00.08.00 000 0.08 0.0. 000000 00000.0 0000.020 080.00....08 00.08000 00. 30.00 0000.00 000.0010 0800 00000.00 000 00.000
00 .0>0. 0x00. 00. .0 8000.030 35.05000 00.000.800.000 000.000 voo~-.oo~ 080 003-80. 00. 00.. 0.0000 80000.. D 080 0 0.0800. 0.0.3 6.0.800
=0". 00. 00.. 0.0.000 0.00000 < .0800. 8800840800 080 080080.308 ..0 08.00.30 00.000000... 0080.. 00.80.00 08008 080 3.30% 0>.000.0wm0 080
0.30% .0 08.00.30 000.000.0000.. 008.. 0080.00 0.30% 00.. 30.00000 00.0 .00 008.. ..0 00.. 00.00000 00 0:003. 00.00000 .0: 000 080.00....000 0.00.
.00 000.0000w00 00. 8 0000.08 00 00.8800 08.. 000 08.00.30 080000000. .000» 00. ..0 000000 00. 00>0 0.0000 .0: .0.0. 0w000>0 0.08.... 00. ..0 w0. 00.
080 0m0 0.080. 00. $.80. 08... 00. 5.3 00.0.5.0 0. 000.>00 808.0020. 00. .. . ..0 0:.0> 00. 000.0. .00. 0.00.00> 0.8800 0 $0.03.. 0000000 808.0908
..0 09800. 0 0. 0000 000.800 808.008. 00. 00 00300.0 00.. .8080w0808 00. .000 00.00.00> .00800 00.. ..0008 00.00..-.“ 00. 800. 0080.00 .0080.
8800 .08000000 00. 080 .0008 00.008 0... 800.. .0080. 80000 .0800800 00.. “000 00.00.00> 800800.00 00... ..0.0 00 0000.00.0 000 800. 00. 8 000m0808
00. .0 0080: 00.. 000.003 0.00.08 .00. 00.8800 800. 03. AN 080 .00.00.00> .8800 0800 0000 .00w00008-0.00 0. 08... 00. .. o 0.00 00m0808-800.
0. 08... 00. .. . m0 0:.0> 00.. 000.0. .00. .0.00.00> 0.8800 800. 0 A. 000 008080.000 8.0.0000 ..0 080.0000w00 mAO 00. 800,. 0.8000 0.800000 0.00. 0.0..

0080,... 00000087800... 0:0 00a0:08..0.0m ..0 00.008.00.00.— 00. 080.000.003—
0.. 0.00..

52

 

 

0.000 000.0 0.0.0 300.0 0000.0 ..0.0 0.000 000.0 000.0 0000.0 0000.0 00000 020000-”.

0000 00.0 0.00 0000 000.0 0.00 0000 00.0 0.00 0000 000.0 0.00 08.002800

..00.0 000.0 0000 0.0.0 .000 000.0 000.0 0.00 000.0 000.0 000.0 000.0

000.0. 000.0 000.0 000.? 000.0 000.. 000.0- .00..- 000... 000.0- 000... 000... 000.0000

00.0 000.0 000.0 000.0 000.0 000.0 000.0 0.00 000.0 000.0 000.0 000.0

000.0 0.0.0 00.0 000.0 03.0 0.00 000.0 000.0. 000.0. 00.0.0 030. 0000. 0000500
.02 0595

0.0.0 .000 0 00.0 000.0 .000 0.0.0 0000 000.0 000.0 0 3.0 000.0 000.0

80.0. 000.0. 000.0. 000.0 000.0. 0.0.0. 0000. 000.0. 000.0. 000.0. 000.0. 000.0. 00< 0.5“.

000.0 000.0 000.0 000.0 000.0 000.0 000.0 000.0 000.0 000.0 000.0 000.0

..0.0 000.0 0000. 00.0 000.0 000.0. 00.0 000.0 000.0 00.0 000.0 000.0 0.002 .0 03

000.0 000.0 0.0.0 0000 000.0 000.0 000.0 000.0 000.0 0.00 0.00 000.0

000.0 000.0. 0. ..0- 0.0.. 000.0. 000.0- 000.0 000.0 000.0 000.0 000.0 000.0 00”. 00.2

.0080 0.0 0.000

53

 

 

 

 

 

 

 

ooo.o wo.d omod vood oowd 0ood 0o~d NN_.d ooo.o oomd om..d ooPd
mood. mo0d w .wd 000d. oo0d 0owd ommd ooo.o m 00.0 Nomd m Ed oo0d 0.0000000
.92 .00—000.0.

omod wwod .o.d o.od wwod 0o0d o0wd mood wmod o0wd mood .mod
o0od ..mod. omod- 03d «mod. omod Sod ooo.o. 000d. :od ooo.o. omod. 004. 0:0”.

ooo.o ooo.o 0ood ooo.o oowd . 0od wmmd Food Food wNNd Food mood
ooo.o. 0o.d .ood ooo.o- 0w 0d ooo.o E. .d oowd ooo.o w0 0d 000d ooo.o 0.0000. .0 00..

ooo.o ond ooPd omod momd word ooPd med med 000d wowd oo..d
owod wo0d woo. . ..oo.o oo0d ..Noé ooNN woo. . 00o. . owod wood wmo... 00“. 00.2

0wod «mod 03d wood .Nod o.od ooo.o ooo.o ooo.o omod ooo.o ooo.o
.ood ooo.o 0ood Food ooo.o 0ood ....od o ..od 0 ..od 30.0 o ..od 0. rod 003:5... 00000..

NNOd mo0d owod o0od ooo.o oo0d
tad- womd 0ood. omwd. mood oomd- 0.0802 02 - E00...
om0d Nomd ..omd wood owod moNd

oo0d ooo.o- oomd. 03d- momd. ommd- 000002 .003. - E00...

.ood .Nod w0md ~m0d o.wd .mmd
ooo.o .omd. 30d 03d- owwd- oowd. >00an E0.0...

. .o .0 0 000:“. ..o .0 0 000:“. . o .0 0 00:0". . .o .0 0 000:”.

.o 0< ..< o 0< ..< o 0< ..< .o 0< __<
.0. . o .0. . .0
.00000.-w. 00.02 00000 €0.00..- ... 0002 00000

 

Soomemt notol .9062 0.05 ..m 3th

 

.0080 0.0 0.000

54

 

 

 

 

 

 

 

 

 

 

000.0 000.0 000.0 00.0 3.00 :00 .000 000.0 000.0 000.0 000.0 000.0
000.0 000.0. 000.0- 00 0.0 000.0. 000.0- 000.0 000.0. 03.0. 000.0 000.0. 03.0. 982 .0 03
000.0 000.0 000.0 000.0 000.0 000.0 .000 000.0 000.0 000.0 000.0 000.0
30.0. 005. 000..- 000.0. 000..- 000.0. 000.0- 000.0 E0 :00. 000.0 50.0 00”. .02
000.0 000.0 000.0 000.0 000.0 000.0 000.0 000.0 000.0 00.0 000.0 000.0
00.0. 0.0.0. 000.0. .000. 0.0.0. 0.0.0. 000.0 000.0. 000.0. 000.0 0.0.0. 0.0.0. .0553 00003
000.0 0.00 00.0 000.0 000.0 0000
00.0.0. 000.0. 000.0. 30.0. 000.0. 000.0. 00802 02 - E8»
0000 000.0 000.0 000.0 000.0 000.0
000.0 03.0. 000.0. 50.0 3.0. 000.0- 00502 .00: - 5000
000.0 000.0 000.0 000.0 02.0 000.0
05.0- 000.0. 000.0. 000.0 000.0- 000.0- 0:500 5000
_ .0 .0 o 000:“. _ .0 6 o 0000". _ .0 _o o 00.5“. _ .0 _o o 0000".
.0 o< _.< 0 0< __< .0 o< __< .0 o< __<
000 3 .0o . :
Coooﬁwo 050?. 009.0 A8600: 020?. 0090
0000.800. 0000a. .9002 .000 ..o .0000
002.0 000.0 005.0 02.0 0000.0 000.0 03.0 0000.0 :20 0000.0 0000.0 0500 0200000
0000 030 0020 0000 030 0020 000. 0000 0000 000. 000 0000 20002080
.000 000.0 000.0 000.0 000.0 000.0 :00 000.0 00.0 :00 000.0 000.0
02.0 000.0 000.0 02.0 000.0 000.0 003. 000.0 000.0 003. 000.0 000... 00000000

3080 0.0 0.00.0

55

 

 

NPwod Rood umnod «owod ooood ownod owmﬁo oomod Fmood 000d momod owood 0000000-”.
mom? Nooo Fomw mow... wooo Fomw oomp wooo Fomw mom? Nooo vomw 0000020000
«and ooo.o ooo.o mood ooo.o ooo.o ooﬁo ooo.o ooo.o oomd Food ooo.o
50d woo.w ..ohw 000d oom.w ooow omme moo... oumo mwué Foo... Fowo 0:000:00
wuod nowd Food omod ow_..o mmNd mwod omod ooo.o ooo.o omod owod
Fwwd mwmd- mwod- omFd oomd. mwod- ooo.o. ..ood. oohd. mmod- ooo.o. mid. €0an0

6.2 .0590.

ovod rwod nood oFod wwod ooo.o wood ooo.o ooo.o ooo.o ooo.o ooo.o

omod. wuod. owod. 2o.o. wnod. omod. omod. ..mod. hood. ooo.o. pmod. hood. 0o< 0:00.

3000 0.0 200...

56

 

 

- 0000.0 0000.0. 0000.0 0000.0 0000.0 0000.0 0000.0. 0000.0 0000.0 8.. 000“. .0 0.
- 0000.0. 0000.0 0000.0. 0000.0 0000.0 0000.0. 0000.0 0000.0 0.0000 0002 .0 00.. .0.

- 0000.0. 0000.0 0000.0. 0000.0. 0000.0 0000.0. 0000.0. 000 0005000002 .0.

- 00.0 0.0. 0000.0 0 000.0. 0000.0. 0000.0 0000.0 00000 20050. .0. 000:0; 00 E

- 0000.0 0000.0. 0000.0 0000.0- 0000.0. 0500.0 £000 .0.

- 0000.0 0000.0 0000.0 0000.0 000% <02 .0.

- 0000.0 0000.0 00 00.0. 00.020 0.0000 .0.

- 0000.0 0000.0. 00.9.0 00< .0.

- 000 0.0 0.0020 000000 .0.

- 000 E80 .0.

.0 0. .0. .0. E .0. 00. .0. .0. .0. .0. 0500;

 

57

0000.22. 000 000000. 00 .900. gm 2.0 00 0:00....0w.0
0000205000 00.00.2000 0.005... 003.005-8000 00.. 0.009.650. no... ..0 00.. 00.90.0000 $578000 :0 .00 00.00.0580 08.50.0080 000600.00. 0500 0.0....

00.0—00.000» 3.6—-800“. .00 00.0003. 000.00.00.80
0.. 030a.

 

 

 

 

.LNd ..ood oowd ooo.o w Foo oowd \Lwd howd o wod
omod- ..No. .. ooN. .. ..- owwd oom.w.. oowd mom. 0.. mooK oNN. ..- b.2020 (m2
FNNd Nowd oo0d o_.od ooo.o owod oowd o..od on ...o
wood- oo ... ..- noo. 0- 5 ..od ..oN.w. oond. ..Nod wood. onw. 0- 3.2020 0.500...
ooo.o hood ond NNod omod oo0d ooo.o ooﬁo wo 0d

wood woud F. o ..o.m- o rod ooo.w ooo.o 0- «mod. woo .o- too. 3.00020 090.
hood sood o 0od hood ooo.o omod oowd _. ..md o ..od

Foo... wwoo ooo. ..o. oooN oooﬁo. ooo.oN. o 5.0 oooK- wooNN. 3.00020 000000
wood ooo.o omnd wowd wood hood ooo.o ..ood m Ed
oow. .... ..oo.o.. omw. 0.. ..ood. oow.o—. NNNd. ooN. ... ..Nod .. Now. ..- 0N5 .000...

09.0.00 30.0.00 .05th 0001.20: 03.0.00. .0505... $0.20... 3.0.0.0.. 003th
0009 000.08 0009 0.0.050 0000. c. 0.2000
0. 00.095 00 0 c_ 02002. 00 0 02095 .0 .0 0
0\0 0\0
00000.. .9002 .000 00000 09.0.2 .30 00000 20500 =0".
0000000000

000 000.. 03 0000.00 20.00000 00000030 0000.058”. 000000.000 00.0 30.00. 000000 00:.0>-& .000 0000.00.00. 000 00.00.. 00 .26. 00o. 05 00 0002.00.30
005.00....000 000.0000.— 000 000 003.00....000 0.00.0 0.5 000.03%": 00.0 0. 00.00.00. 000 00.00.0000 000.. .000 2500.30 000000.005 .0000. 00.0 .00 000.000
2.0 00.00 000000 000 .0000 0o000>0 0.000.. 00.0 ..0 o0. 2.0 .000 0&0 0.050.. 00.0 5.000.. .05.. 00.0 5.3 0000......00 0. 000060 0000500000. 00.0 .0. . .00 00.? 2.0
08.00 000.0 0.0.0.0000 b00800 0 5:000 0000000 000050200. 00.. .0800. 0 0. .000 000060 00.00.000.00. 00.0 .3 00300.0 00.. 0880m00000 00.0 ”0.00 00.00.0000
.0008 00.... .005... 000.00 .3 09.0.0000 00.0 0.00 000w0008 0.00000 0... .00 030000000 000.? 032.0 .000 0000. 00.0 0. 000900000 ..0 00.. 0030.000, 0.80000
0.00:". 0.00.08 05 ..0 000.0> 00.0 o5w000>0 3 0000000 0. 8000 02.00 .8000 0 00 000.0030 $0.003 0000. 0200. 000m0008 000.0 2.0 0. 0.000000 0000...
A052 .0000 000000... .0900. 0030.000, b.0830 000.00 00.0 00.. .000: 0. 00508 00000 0...... .8000 00.0 0. 00030000 ..0 00.. 00.0000 0.0.0.00> >883 00.0 00
000.0.» 00.0 .00 00.00300 30.00000 0... 0. 3.00205 00.0000 .800. 000§0o0008 0:00.000. 05 0. 00030000 .0 02.0.5: 00.. 0. 00.0 0000... 090.200. o. 000
0... 0. 00000.00. 000000 0000.. .0 0300000000. 0... .000 2.0.0002. 0... 00. 00.000000 ..0 000.0000 000000.50 2.000000. 000%-w0.30..00. 000 0050.09» 0000.00.00.00
00.... 0030.00.» .0008 0:000 .000 00.005.000.000 .0>0..E000 00 00.005.000.000 0:00.000. ..0 00.000030 mAO 2.0 0000. 00.00000 00.000000. 0.0.00 0......

00300000000000 SET—0009.. .000 00.00.580.030 2.0.0.5..
o.. 0.0.0...

58

 

 

030.0 0000.0 0000.0 030.0 0000.0 300.0 0000.0 0000.0 300.0 0000 $0
000 000. 000 000 000 000. 000 0 000 0 00: 080020000
80.0 000.0 08.0 08.0 000.0 03.0 80.0 000.0 08.0
000.00 000.00 02.30 30.0.0 30.0. «00.3 0.00.00 00.0.00. 00.0.00 0:000:00
000.0 000.0 0:3 000.0 000.0 000.0 000.0 000.0 000.0
000.0- 00.0.0 000.00 000.0- 000.0 000.00 03.0- 000.0 000.00 0:858
00.2 0500:.
000.0 000.0 000.0 000.0 000.0 000.0 000.0 03.0 000.0
000.0 000.0 000.0 000.0. 30.0 000.0. 30.0 000.0 03.0- 00< 0050
000.0 08.0 30.0 03.0 08.0 000.0 08.0 80.0 30.0
000.0. 000.2 03.0. 03.0. 000.00 30. 0- «00.0. 000.00 0.0.0.0. 00002 00 03
0.00.0 0.00.0 000.0 08.0 30.0 03.0 80.0 000.0 000.0
08.3 03. 00- 000.00- 0 00.0 N 00.00 000.8 000.00 000.00- 03.0 000 00.2
000.0 30.0 000.0 000.0 000.0 000.0 000.0 ~30 000.0
03.0- 000.0 03.3 000.0- 000.0 000.0 000.0. 003 000.00 0050 0050
08.0 000.0 08.0 08.0 000.0 80.0 08.0 000.0 80.0
000.0 000. 0- 000.0. 30.0 000.0. 30.0. 000.0 000. 0- 02.0.. 05:00 5000

3:8. 0.0 0.0.0.0

59

 

 

 

$55 535 5~m5 585 2.55 535 5:5 535 ~85
2.55 .555 m~.5- -55 ~85 mm~5 ~85 :55. ~85 £225 <92
3 .5 55 ~85 5555 E55 5~m5 5-5 :5 855
5555- 555. 255. 5555 ~85. 5555 855- 855. 355- £995 95:3
525 585 -55 555 355 525 ~35 535 355
535. 855. m P F5 ~85. 255- $55- :55. 5555. 855 35.95 mm<
585 5:5 5555 -~5 $55 585 5.35 .35 -m5
2.55 255 535- $55- 585 585. m~55 555 mo~5- £395 .850
5-5 ~85 3~5 .85 ~85 585 83 585 $55
5555. 355- 855. 5555- $55. 585. ~55- «~55. ~25. 35 camp
353$ 558.2 2:22 ho €58-55 558.2 552...: .5 e255; 5822 mes». 3
95m .3525 8:355 Sam 59...sz 85538 55m 59.5.25 8:955
$8 2856 95m 288$ 98 935%
8:8 .352 50m 8:55 59.5.2 :3 8%“. 29:8 =5“.

 

.wougumo 25 83 3 88k... 838% 38885
5558508 685858 05 323 85255 mo:_~>-& 98 @8823 85 Stop .8 _o>o_ £52 05 8 88$:me $806580 dot—camp Ho: 85 5806580
:05 :5 82552on 2: 8 cows—ca 85 50.88% 08: 2.8 258?. 530385 .39» 05 .8 08:8 05 5.6 gamma 8: :38 09896 $83
05 (8 m2 05 ES own £88 2: .388 .83 05 55, woamzmma 8 88.555 558582: 2: a ﬂ mo 02? 2: 583 $5 03~t~> >888 a 5:86
88.53 828538 88 >88 5 £ 23 85.53 808538 05 3 3320 new 80803858 05 6.8 833.8.» 8.805 23. 595m 850 3 coho—80
025 0.8 80w8~8 5.853 oﬁ mo 038083 “an? $65 98 858 of E 80w~8~8 :5 8m moﬁmcg b.8885 $25k. 38935 on“ ,«o mos—S, oﬁ
wﬁmﬁga .3 @888 ﬁ :88 .850 8~8 a mm 850on wan—83 soon 26: 80w~8~8 08: 05 8 93:8 Suﬁ .352 88 9:80... .om<v woﬁatg
3836 850 05 .5.“ wow: 8 @0508 088 25. .858 05 8 ﬂow—~88 :5 8m .5980 oB~t~> >88 05 mo 339, 05 90 823.5% 83835
05 8 53.395 .5586 .88: 808%888 0:883 05 E 80388 mo 5988 2: mm mm? “SEN .6008 538% 5.5850 8cm Sun How—88
on. was 3on 35:58 2: 88m 83 How—.88 05 .5882 30% 38:08 5.25..“ 05 mo 8035.535 vaccSm 89¢w83o=£ 25 33589, “58:33
05. 525388, .888 “.88 98 8.588585% _o>o_-8~3 no 8558535 x8 020.383 mo 82.0.58on mAO 2: 89¢ 83%.: $880.8 058 mg

85:32:85 _o>o_.8~oh ES “.52 9:32...—
a.— «Sub

6O

 

 

52:5 5585 ~5~55 52:5 8555 885 285 -5~5 885 595555.”.
8~ 8~ 8~ 55¢ 55¢ 555 8. F 52 F 58 F 8852580
855 555.5 855 585 85 855 585 855 585
~85 :3 53.5 585 5555 85.5 535 855 ~3.~ €5.88
8.5 555 ~85 8~5 -F5 8~5 5-5 855 E55
585 585- 585- F 85 585 88 585 5 85 585 2880
as. .5558.
5.55 85 585 5555 5555 ~F~5 5E5 ~85 ~85
585 ~85. ~85. 85- 85 ~85- 85- 585 ~85- 55¢. 58“.
585 5555 585 535 ~85 5555 5555 535 855
85 85- 585- ~85 2.55 585 855 855 585. 2552 8 8..
85 585 v-5 :85 585 -55 5.55 855 ~85
585- 585- 585- 855- 585 355 5:5. ~85- ~85- 55“. as.
585 855 585 5555 585 8~5 5~55 855 $55
585 585 855 585- 8 .5 855 -55 585 855 88". .550
855 585 5. F5 8~5 855 585 ~85 ~85 585
85- ~85. 585. 585- ~85. 585. 855- ~_.55- 585- 558.- 553

383 .2 5.8.5

61

 

 

 

~5.5 5.~5 55.5 585 5555 5555 5.55 55.... ~85
~85 ~85 5555. 5.55. .85 ~85- 5.55 585 5.55- 8.258 5.2
5. ..5 55~5 ..5~5 ~55... 5...... ~85 5555 .85 88
585- ~85 ~85 .55.? .85 ~85 ~85. 5...... ~85 8.5.525 585.
.85 ~85 5~55 $8 .85 ~25 8~5 8.5 5-5
355- -5.9 5.55- 585 ~85. .555- 5.55- .85. 58.9 5.2525 55<
585 .85 855 ~85 5.55 .85 .85 ~85 5.5
.85 8.5. .85 5555 ~85. 8.5. 585 55.5. -55- 8.2525 .5850
5555 ~85 855 5555 55 .5 5 .~5 ~. .5 5~55 8~5
585. ~85 ~85. .85. .85 ..55 5.55- ~.55 585- 55.5 E55.
5955..-... 55.55.-.. 8.55.-.. 8.55.-.. 8.55.... 8.55.55 55.55.-.. 855.-... 8.55....
5.55 5.2: 5.55 5.55 5.25 5.55 5.2.. 5.55 5.55 5.25 5.55 5.2: 5.55 ...2... 5.55 5.25
...2... ...2...
5855 .5855. .555 8:55 .585... .55 8555 53555 ....u.

 

55.58.85 5.5 .8... .... 5.585 5.5.85.5 5585.530 88.55.58 55.58855 55 30.5.. 855885 5525.13.85 585-.20.. 5.5 .582. .o .55.
£5... 55 .5 855.-.8»... 5.85.5558“. .5835. .88 5.5 5.85.5558 8.5... ...... 585.8585. 55 8. 55.5.58. 55 55.88.... 58.. ..85 53.55.38 58.55858...
.55.. 55 .8 55.855 55 .58 5.5555 .58 .58. 555.55 5.58.... 5... .8 mo. 55 .585 5w5 5.58.... 55 5.85. 58.... 55 5.3 ..5.5...-...5 5. 85.55 858.558.
55 .. . ..o 58.5.. 5... 58.5. .5... 5.5.5.85.» .88.... 5 $8.58.. 8853.... 858.552.. 80.. .08... 5 5. 585 .85-.55 858.552.. 5... .... 5595.... 55-. .8585w5858
5... .585 555585.. .8888 5.... 5.588.. .558 .... 558.885 85.5 5.5 5.555858 5.855. 55 .0 5888585.. .5...» 52.8.5 ..85 855. 55 8. 585w5858 ..5
.0.. 835.85.. 288.... 5.588.. 5......88 5... .o 5525.. 55 w8.w5.5>5 .... 55.55... 5. 855. .550 855.5 55 .5555... 38.83 855.. 5>5.. 5.5w5858 58.. 55
5. 58.85. 855 ... .55.). .585 58.85... 6?... 55.85.85.» b.8525 .558 55 .0.. .5558 5. .5558 5855 5....- 855.55 8. 5.5w5858 ..5 85.. 855850 555.85..
.88.... 55 ..o 5525.. 55 .0 8.8.5.5.. 5.5.85.5 55 5. 5.5.5.89 .5585O .855. .8585w5858 2.8.8.8.. 55 8. 5.5m5858 ..o .5888 55 5. 55.5 855 ...
558.535.... 5.55.. QED ..85 ...)... .mEm 855258.323. 5.5 555585.. 8558585.. 5....- .55...5...5> .8888 588 .85 5585.85.55.55. .5>5..855. 8c
.5588 8855...... 5.85550 8.88:. 5.5.5 .58.... ..555 85.. 558.538 $8.550. 8.55.5.8. 2.5.8.... .0 88.85.55. m-.O 55 88.. 5.355. 5.8555... 5...... 5...-.-

55..5...5.55..5..U .55—855... ...—5 38.55.... 88.55... 5.5.”. \ 5.55 858.55.»...
a... 5.5...-

62

 

 

N. . ...o m .NNd o .m .d omood moo . .o NNm . .o woood NnoNd N .m ..o 55.5:om-m
mos no» no. oww mow mww mm. . mo. . mo. . 5555525550
ommd mood moNd mm ..o ..ood oN .d omod .ood ..N ..o
ooo.o oomd oN_..o B ..o- owod ooNd wNod msmd 59d 85.5850
Nnod Nood mNod m .md mm .d owmd w. ..o mood mwmd
smod om ...o. hood. wmod wmod. mmod omod oo ...o. oNod- 855.550
.92 .582...
N .Nd owod ooo.o .owd mwod o. ..o mm .d nmod mood
.ood Nood. mood. _.ood mood. Nood- .ood Nood. Nood. 59.. 58......
w. ..o oowd mwod ooﬁo Emd wood mwmd omwd mm .d
o .od. wood .ood. wood m .od. 53.? wood- Nood- N .od- 5.555.... .5 55..
mmod mood de owod mm .d uood mmNd ...mNd .owd
ooo.o. Nwod. wNod Nmod mm ..o. oNod mwod- Nmod- . .od 55". .92
hood mood wmwd momd N .od omwd Nde ooo.o oRd
ooo.o .... ...o. omod. oNod w» ...o. m .od m .od Nm ..d. . .od- 55:5... .550
med o .md Nwod w. _,d wmod mwod Nmod mNod N ....d
mood. mood ooo.o mood- 5 ..od ooo.o mood. o Ed .ood 55:5... 855.5

3:55. o... 535,—.

63

Table 1.1 1
Performance and Team-level Characteristics

This table presents results from the OLS regressions of portfolio performance on team-level
characteristics and some control variables. The dependent variables are: the abnormal return
(alpha) from the market model and the abnormal return (alpha) obtained from the 4-factor model.
Team size is the number of managers in the portfolio management team. Gender Diversity is the
standard deviation of the values of the dummy variable Gender for all managers in the team. The
same method is used for the other diversity variables (Age, Tenure and MBA). Team tenure is the
time managers have been working together as a team. Other team is created by averaging the
values of the multiple funds dummy variables for all managers in the team and shows what
percentage of the team’s managers are also employed by other funds. The control variables are:
the management fee charged by the investment advisor and is a proxy for investment advisor
quality, a dummy variable that takes the value of 1 if the investment advisor is afﬁliated with the
fund family, the fund’s age and the log of the fund’s average total net assets over the course of the
year. Prospectus objective and time dummies are included in the regressions but their
coefﬁcients are not reported. Results are reported for all ﬁinds but also separately for growth
oriented funds (prospectus objective of growth and aggressive growth) and income oriented funds
(prospectus objective of growth-income and equity-income). Panel A reports results for the full
sample, while panels B and C present results for the 1997-2000 and 2001-2004 periods
respectively. Coefﬁcients signiﬁcant at the 10% level or better are boldfaced and p-values appear
below the estimated coefﬁcients. Clustered standard errors by fund are estimated.

 

Panel A: Full Sample Period (1997-2004)

 

 

All Funds AG & G GI 8|
1- factor 4-factor 1- factor 4-factor 1- factor 4-factor
Alpha Alpha Alpha Alpha Alpha Alpha
Team size -0.293 -0.106 -0.241 -0.227 -0.181 0.356
0.332 0.675 0.552 0.502 0.616 0.278
STD Gender -2.903 -1.164 -3.711 -1.229 -1.343 2.161
0.006 0.210 0.005 0.291 0.470 0.137
STD Age 1.654 1.444 1.714 1.683 0.839 0.234
0.005 0.003 0.013 0.005 0.231 0.687
STD Tenure -0.039 -0.037 -0.056 0.158 0.103 0.267
0.711 0.709 0.694 0.193 0.412 0.032
STD MBA 0.087 0.368 0.344 0.885 1.040 -0.515
0.930 0.646 0.789 0.398 0.412 0.596
Team Tenure -0.036 0.019 -0.095 -0.027 0.117 0.117
0.705 0.755 0.411 0.719 0.379 0.243
Other Funds -0.872 0.057 -1.154 0.579 -0.342 0.993
0.194 0.921 0.165 0.417 0.736 0.303
Mgt Fee -0.658 -0.679 -0.759 -0.995 -2.656 0.072
0.630 0.608 0.638 0.520 0.182 0.965
Log of Assets -0.042 -0.100 -0.227 -0.197 0.301 0.236
0.838 0.493 0.390 0.269 0.284 0.278

 

64

Table 1.11 (cont.)

 

 

 

 

 

Fund Age -0.027 -0.012 -0.028 -0.006 -0.061 -0.052
0.189 0.475 0.428 0.837 0.005 0.001
lntemal Mgt
Company -1 .51 9 -0.967 -2.451 -1 .389 1 .371 0.484
0.018 0.046 0.002 0.023 0.154 0.528
Constant 1 .739 2.013 3.369 4.085 2.498 -2.097
0.396 0.244 0.172 0.058 0.339 0.308
Observations 1 193 1 193 872 872 321 321
R-Squared 0.1598 0.1696 0.1831 0.202 0.3974 0.1738
Panel B: Bull Sample Period (1997-2000)
All Funds AG 8. G GI & I
1- factor 4-factor 1- factor 4-factor 1- factor 4-factor
Alpha Alpha Alpha Alpha Alpha Alpha
Team size 0.594 0.265 0.946 0.112 -0.262 0.485
0.384 0.638 0.339 0.890 0.754 0.488
STD Gender -2.622 0.088 -3.416 0.700 -2.239 -3.349
0.240 0.962 0.264 0.782 0.453 0.155
STD Age 1.456 0.843 1 .345 1 .186 1 .265 0.000
0.169 0.349 0.345 0.342 0.197 1.000
STD Tenure -0.146 -0.076 -0.152 -0.254 0.063 0.305
0.392 0.657 0.564 0.253 0.751 0.115
STD MBA 0.783 0.769 0.410 0.671 2.201 0.328
0.652 0.601 0.867 0.745 0.246 0.846
Team Tenure -0.209 -0.043 -0.234 -0.053 -0.040 0.023
0.233 0.713 0.253 0.676 0.887 0.935
Other Funds -0.812 0.628 0.577 0.265 -2.950 0.196
0.719 0.604 0.726 0.866 0.143 0.923
Mgt Fee -0.812 -1.276 -1.553 -2.233 -1.382 1.157
0.719 0.575 0.589 0.424 0.644 0.654
Log of Assets -0.104 -0.091 -0.430 -0.280 0.565 0.538
0.774 0.738 0.387 0.415 0.242 0.202
Fund Age 0.004 -0.009 0.038 0.002 -0.055 -0.059
0.898 0.764 0.479 0.968 0.161 0.047
Internal Mgt
Company 0.221 -0.522 -1.031 -1.215 2.584 1.001
0.863 0.554 0.553 0.303 0.080 0.426

 

65

Table 1.11 (cont.)

 

 

 

 

 

 

 

 

 

Constant 6.220 5.981 13.351 1 3.322 6.369 -2.027
0.083 0.063 0.010 0.002 0.183 0.655
Observations 488 488 333 333 155 155
R-Squared 0.2313 0.1989 0.2282 0.2400 0.4415 0.1545
Panel C: Bear Market Period (2001-2004)
All Funds AG & G GI 81
1- factor 4-factor 1- factor 4-factor 1- factor 4-factor
Alpha Alpha Alpha Alpha Alpha Alpha
Team size -0.622 -0.246 -0.740 0.389 0.136 0.370
0.051 0.363 0.081 0.281 0.708 0.256
STD Gender -3.535 -2.146 -4.1 21 -2.51 5 -2.199 -1 .973
0.003 0.042 0.004 0.060 0.248 0.168
STD Age 1 .786 1 .850 2.082 2.109 0.673 0.705
0.010 0.002 0.008 0.004 0.484 0.293
STD Tenure 0.079 0.027 0.096 -0.035 0.234 0.244
0.567 0.824 0.577 0.820 0.326 0.078
STD MBA -0.329 0.020 0.266 0.823 -0.763 -1.621
0.770 0.981 0.846 0.464 0.560 0.102
Team Tenure 0.099 0.067 0.080 0.024 0.093 0.149
0.329 0.450 0.533 0.840 0.493 0.169
Other Funds -1.592 -0.334 -2.526 -1 .121 2.1 49 1 .682
0.067 0.610 0.015 0.165 0.075 0.044
Mgt Fee 0.600 0.351 -0.642 -0.346 -4.563 0.929
0.722 0.800 0.727 0.822 0.064 0.644
Log of Assets -0.01 1 -0.1 13 -0.077 —0.166 0.022 0.008
0.964 0.537 0.804 0.481 0.946 0.975
Fund Age -0.068 -0.019 -0.102 -0.014 -0.083 -0.051
0.005 0.269 0.025 0.671 0.002 0.000
lntemal Mgt
Company -2.993 -1.501 -3.757 -1.727 -0.066 -0.282
0.000 0.019 0.000 0.027 0.956 0.759
Constant 6.245 4.003 5.305 3.945 5.489 0.078
0.023 0.032 0.064 0.071 0.036 0.976
Observations 705 705 539 539 166 166
R-Squared 0.0986 0.1 173 0.1299 0.1 160 0.3469 0.2452

 

66

 

This tab]
character
11. The r
Team_ Tc
other 51 gr
but thetr
grouth (
oriented

results f0
periods r:
appear be

m

Team size
Team size
Team size

Assets

Team 5,26

STD Gen d

STD AQe

 

STD Tenbr

 

Table 1.12
Performance and Team-level Characteristics

This table presents results from the OLS regressions of portfolio performance on team-level
characteristics and some control variables. The dependent variables are all the same as in table
11. The new variables are the interactions of the T eamSize variable with ST D_Age, STD_Tenure,
T eam_Tenure, LogAssets and OtherFunds. We report the coefﬁcients of the interaction terms of
other signiﬁcant variables. Prospectus objective and time dummies are included in the regressions
but their coefﬁcients are not reported. Results are reported for all funds but also separately for
grth oriented funds (prospectus objective of growth and aggressive growth) and income
oriented funds (prospectus objective of growth-income and equity-income). Panel A reports
results for the full sample, while panels B and C present results for the 1997-2000 and 2001-2004
periods respectively. Coefﬁcients signiﬁcant at the 10% level or better are boldfaced and p-values
appear below the estimated coefficients. Clustered standard errors by fund are estimated.

 

Panel A: Full Sample Period (1997-2004)

 

All Funds AG 8G GI & I
1- factor 4-factor 1- factor 4-factor 1- factor 4-factor
Alpha Alpha Alpha Alpha Alpha Alpha
Team size * STD Age 0.847 0.235 0.401 -0.049 2.845 1.387
0.218 0.685 0.618 0.947 0.014 0.131
Team size * STD Tenure 0.126 0.080 0.326 0.249 -0.193 —0.125
0.376 0.545 0.101 0.140 0.300 0.415
Team size * Team Tenure 0.353 0.133 0.499 0.282 0.074 -O.117
0.001 0.178 0.001 0.039 0.432 0.166
Team size * Log of
Assets -0.266 -0.073 -0.459 ~0.288 0.127 0.342
0.192 0.666 0.108 0.211 0.576 0.067
Team size * Other Funds 0.710 -0.190 0.789 0.599 -0.067 -1.531
0.294 0.773 0.363 0.473 0.954 0.091
STD Gender -2.974 -1.207 4.028 -1.485 -1.100 -1.880
0.005 0.202 0.002 0.222 0.562 0.192
STD Age 0326 0.888 0.928 1.957 -5.816 -2.963
0.854 0.567 0.668 0.330 0.028 0.171
STD Tenure -0.341 -0.231 -0.838 -0.768 0.538 0.568
0.374 0.527 0.129 0.111 0.221 0.143
Team Tenure -0.907 -0.307 -1.276 -0.697 -0.076 0.451
0.002 0.222 0.001 0.037 0.817 0.119
Other Funds -2.785 0.495 -3.191 -2.125 -0.029 5.013
0.153 0.797 0.209 0.383 0.992 0.036

 

 

 

 

67

 

 

Internal M

Constant

Observatu

R-Sguare<

PanelE

Team size
Team size
Team size

Team size
Assets

Team size
STD Gendr

STD rem”

 

TeaanenL

OmerFunc

 

Table 1.12 (cont.)

 

 

 

 

 

 

 

 

 

Internal Mgt Company -1.330 -0.891 -2.105 -1.213 1.421 0.636
0.039 0.073 0.009 0.054 0.132 0.403
Constant 3.599 2.210 2.675 3.396 8.258 0.307
0.323 0.496 0.616 0.470 0.095 0.934
Observations 1193 1193 872 872 321 321
R-Squared 0.1683 0.1716 0.1946 0.2079 0.4086 0.1902
Panel B: Bull Sample Period (1997-2000)
All Funds AG 8 G GI 8 l
1- factor 4-factor 1- factor 4-factor 1- factor 4-factor
Alpha Alpha Alpha Alpha Alpha Alpha
Team size * STD Age 5.773 2.524 5.731 3.073 4.334 1.814
0.004 0.105 0.027 0.128 0.078 0.374
Team size * STD Tenure 0.446 0.322 0.867 0.655 -0.326 -0.144
0.232 0.210 0.151 0.031 0.386 0.633
Team size * Team Tenure 0.575 0.044 0.523 0.060 0.368 0.260
0.013 0.813 0.049 0.759 0.712 0.685
Team size * Log of
Assets -0.441 0.199 -0.618 -0.214 -0.206 0.730
0.274 0.557 0.270 0.643 0.664 0.141
Team size * Other Funds 0.431 -2.049 -1.923 -0.906 0.969 -3.918
0.831 0.192 0.592 0.732 0.768 0.201
STD Gender -2.561 0.172 -4.007 0.297 -1.115 -2.114
0.248 0.927 0.186 0.908 0.733 0.358
STD Age -11.302 -4.746 -1 0.998 -5.344 -8.701 -4.206
0.012 0.185 0.056 0.253 0.106 0.357
STD Tenure -1.092 -0.760 -2.062 -1 .689 0.800 0.605
0.193 0.225 0.138 0.035 0.312 0.374
Team Tenure -1.504 -0.126 -1.404 -0.175 -0.755 -0.534
0.013 0.784 0.053 0.731 0.716 0.708
Other Funds -1.023 5.569 4.881 2.540 -4.625 9.859
0.832 0.173 0.545 0.695 0.566 0.163
lntemal Mgt Company 0.852 -0.391 -0.166 -0.950 2.937 1 .159
0.498 0.659 0.924 0.426 0.049 0.362
Constant 13.202 9.644 13.718 14.776 9.132 4.829
0.053 0.125 0.291 0.187 0.378 0.568
Observations 488 488 333 333 1 55 1 55
R-Squared 0.2542 0.209 0.2547 0.2523 0.4596 0.1827

 

68

IL

 

R—s ..
\0%

m

Team size
Team size

Team size
Tenure

Team size
Assets

Team size ‘
STD Gend
STD Age

STD TGDUH

Team Tenu

 

Other Fund

Internal Mg

COnstam

Obser‘v’allo'

 

Table 1.12 (cont.)

 

Panel C: Bear Market Period (2001-2004)

 

 

 

 

 

 

All Funds AG & G GI &l
1- factor 4-factor 1- factor 4-factor 1- factor 4-factor
Alpha Alma Alpha Alpha Alpha Alpha
Team size * STD Age -0.663 -0.669 -1.047 -1.117 2.162 1.062
0.350 0.339 0.225 0.185 0.053 0.158
Team size * STD Tenure 0.060 0.035 0.133 0.105 .0201 -0.128
0.683 0.820 0.480 0.580 0.343 0.332
Team size * Team
Tenure 0.288 0.184 0.432 0.385 -0.053 -0.1 69
0.008 0.1 13 0.008 0.023 0.668 0.050
Team size * Log of
Assets -0.205 -0.185 -0.357 -0.269 0.418 0.272
0.367 0.341 0.209 0.287 0.156 0.094
Team size * Other Funds 1.252 0.481 1.608 1.150 -0.160 -0.558
0.095 0.490 0.085 0.183 0.892 0.502
STD Gender -3.560 -2.215 4.396 -2.795 -2.386 -2.134
0.002 0.036 0.002 0.040 0.206 0.121
STD Age 3.521 3.580 4.914 5.070 4.480 -1.804
0.095 0.070 0.053 0.035 0.116 0.398
STD Tenure -0.092 -0.075 -0.240 -0.306 0.694 0.594
0.832 0.865 0.666 0.583 0.294 0.137
Team Tenure -0.666 -0.412 -0.996 -0.934 0.230 0.677
0.027 0.171 0.015 0.023 0.636 0.035
Other Funds -5.141 -1.726 -6.955 4.362 2.533 3.407
0.036 0.405 0.021 0.084 0.486 0.186
lntemal Mgt Company -3.012 -1.545 -3.647 -1 .660 0.076 -0.027
0.000 0.015 0.000 0.034 0.950 0.976
Constant 7.153 2.593 4.247 2.913 1 3.522 1 .838
0.139 0.497 0.466 0.545 0.034 0.681
Observations 705 705 539 539 166 166
R-Squared 0.1102 0.124 0.1472 0.1336 0.3606 0.2666

 

69

 

 

 

bemg in I
Some Den

account in

2.1 lntr

T
performa
determinc
returns. \
fund man
this anicl
perfonna;
Pleﬁous r

01
131%” dat
Particular_
bull mark
beansh pt
This alter.
and bear :
different I.

mOdels ”'1.

 

 

 

ESSAY 2. Manager Characteristics and Mutual Funds

2.1 Introduction

Traditionally, mutual fund research has focused on the measurement of fund
performance and its persistence over time. The goal of most papers in this literature is to
determine whether some funds are able to consistently produce positive risk-adjusted
returns. While the results thus far are mixed, there seems to be a consensus that mutual
fund managers are not able to outperform benchmarks and produce consistent returns. In
this article, we add to this literature by examining how manager characteristics relate to
performance, using a much more comprehensive dataset on mutual ﬁind managers than
previous research.

Our analysis extends the literature in at least three ways. First, we employ a much
larger data set than previous work on the characteristics of mutual fund managers. In
particular, in contrast to Chevalier and Ellison (1999), who focus on the 1988 to 1994
bull market period, and Gottesman and Morey (2005), who focus on the 2000 to 2003
bearish period, our sample period (1997 to 2005) spans both a bull and a bear market.
This allows us to investigate our hypotheses separately on the full sample and the bull
and bear sub-samples (1997 to 2000 and 2001 to 2004, respectively), and thereby take
different market conditions into account. Second, we evaluate performance using logit
models that show how manager and team characteristics affect the probability of a fund
being in the top (or bottom) performance quartile. Finally, we include in our analysis
some new manager characteristics and control variables that have not been taken into

account in prior literature.

70

r ‘ '1 mwn‘wv-nr- -~_.‘- -

 

 

A
managers
is the ﬁrs
managem

Se
relation
characteri
characteri.
and whet}
longer ten
Note that
530 funds
equal. the
With low .
Recessam}
(30166 (192
length of [1'

Che
pel’formani

 

Whmher [b
u"hOSe in“ '

pen-0d. Th
I

 

A recent practice of investment advisory ﬁrms is to assign the same manager or
managers to the management teams of multiple funds. To the best of our knowledge, this
is the ﬁrst paper to provide evidence on how a manager’s participation in the portfolio
management team of another fund affects his risk-taking behavior and performance.

Several studies focus on individual fund managers’ characteristics and their
relation to performance outcomes. Golec (1996) tests whether fund manager
characteristics help to explain a fund’s performance, risk, and fees. The manager
characteristics he analyzes are manager age (years), manager tenure, years of education,
and whether the manager has an MBA degree. He ﬁnds that younger managers with
longer tenure at their funds and MBA degrees extract better risk- adjusted performance.
Note that he also analyzes a team size variable. However, as he observes, his sample of
530 funds from 1988 to 1990 includes very few team-managed funds; indeed, all else
equal, the coefﬁcient for the team size variable is insigniﬁcant. He also ﬁnds that funds
with low expenses realize better performance, but that large management fees do not
necessarily imply poor performance. In summary, out of all the manager characteristics
Golec (1996) studies, the most signiﬁcant predictor of poor performance seems to be the
length of time a manager has been with the fund (tenure).

Chevalier and Ellison (1999) also examine the relation between mutual fund
performance and fund manager characteristics. In particular, they focus on the manager’s
age, the average composite SAT score at the manager’s undergraduate institution, and
whether the manager has an MBA. Their study focuses on 492 single-manager funds
whose investment objective is growth or growth and income over the 1988 to 1994

period. Their strongest result is that managers who attend higher-SAT undergraduate

71

 

institutic
that you
report, t1
Chevalie

G
the relath
the qualii
score, the
average (
education
CFA desi.
that for a :
Who hold
raﬂklngs (
institution‘
performan

Bli

The)” ﬁnd

 

men. HO“.

ortradl‘nQ -.

w,
l

more than

 

institutions produce systematically higher risk-adjusted excess returns. They also ﬁnd
that younger managers fair better than older ones. As Gottesman and Morey (2005)
report, this last result has been criticized as being the product of the bull market period
Chevalier and Ellison (1999) study.

Gottesman and Morey (2005) extend the work of Chevalier and Ellison (1999) on
the relationship between fund performance and educational characteristics. In addition to
the quality of the undergraduate institution, as measured by the program’s average SAT
score, they investigate the quality of the MBA program, as measured by the program’s
average GMAT score, and its effect on performance. They also relate performance to
education variables such as whether the manager attends a liberal arts institution, has a
CPA designation, or holds any other graduate-level degree (masters or PhD). They ﬁnd
that for a sample of 518 single-manager funds tracked between 2000 and 2003, managers
who hold an MBA degree from a school ranked in the top 30 of the Business Week
rankings outperform managers with no MBA or with an MBA from a lower ranked
institution. The other education variables seem to be unrelated to mutual fund
performance.

Bliss and Potter (2002) examine the effect of the gender in portfolio management.
They ﬁnd that women fund managers hold portfolios with marginally more risk than
men. However, they do not ﬁnd any other signiﬁcant differences in terms of performance
or trading behavior.

We use a unique data set on over 1,200 individual mutual fund managers from

more than 500 management companies and 2,660 fund-year observations for the 1997 to

72

 

2005 per
bull (199

C
can explz
older ma:
especial]:
bearish SI
and inves
true for 0]

Ti
ﬁnding th
managersl
Ellison‘s I:
high ax‘er'

Interesting!
I
tier busint

buSirtess s

PIOgramS
to explain

it“
other mm

managers

 

2005 period. We report results for the whole market period, but also for the two separate
bull (1997-2000) and bear (2001-2004) market periods.

Our main results are as follows. First, we ﬁnd that single-manager characteristics
can explain differences in risk taking and investment style. Portfolios of long-tenured and
older managers exhibit higher levels of turnover, whereas managers with MBA degrees,
especially graduates of highly ranked business schools, exhibit lower turnover during the
bearish sub-period. Managers with MBA degrees also hold more stocks in their portfolios
and invest a smaller part of the fund’s assets in their top holdings, while the opposite is
true for older managers.

The results fail to conﬁrm Golec (1996) and Chevalier and Ellison’s (1999)
ﬁnding that MBAs hold more systematic (beta) risk. Instead, we ﬁnd that long-tenured
managers hold the lower-beta portfolios. On the other hand, we conﬁrm Chevalier and
Ellison’s (1999) ﬁnding that managers that attended an undergraduate institution with a
high average SAT score outperform other managers, even during a bear market.
Interestingly, however, the SAT variable is positively correlated with the dummy for top-
tier business schools and negatively correlated with dummies for third-tier and unranked
business schools, which implies that those managers who attend high SAT undergraduate
programs also attend more prestigious MBA programs, and thus the SAT score’s ability
to explain performance decreases once we take a manager’s MBA program into account.

We fail to ﬁnd evidence that participation of managers in management teams of
other mutual fund portfolios causes any underperformance. We do ﬁnd, however, that

managers employed by many mutual funds turn their portfolios over much often.

73

 

method.
and resr

concludc

2.2 Me
2.2.] file
V
all funds
2004). \\
im’estmei
mainly us
attributes
We also u
Observatio
CVaIUate
PerfOrman
Performan
dependem
qUalTlle (I

Van'ables'

 

 

The remainder of the paper is organized as follows. Section 2.2 describes our
method, data sources and deﬁnition of variables. In section 2.3 we present our hypotheses

and results. In sections 2.4 we perform some robustness checks. Finally, section 2.5

concludes.

2.2 Method and Data
2.2.1 Method

We conduct our analysis as follows. We ﬁrst obtain manager characteristics for
all funds at the beginning of each year t for all the years in our sample period (1997 to
2004). We then relate those manager characteristics at time t to portfolio attributes, risk,
investment style, and performance over the course of the next year (ﬁ'om t to t+1). We
mainly use ordinary least squares (OLS) regressions of portfolio characteristics and risk
attributes on management team characteristics. Similar to Chevalier and Ellison (1999),
we also use instrumental variable estimation for some of our regressions, using lagged
observations as proxies for variables (such as turnover) that appear to be endogenous. To
evaluate performance we follow a novel approach compared to other mutual fund
performance studies and estimate logit models. For each year we rank all funds into
performance quartiles. We then estimate the coefﬁcients of a logit model where the
dependent variable takes the value of one if the fund belongs to the top performance
quartile (top 25%) and zero otherwise. Manager characteristics are the independent

variables. Rather than reporting the actual logit estimates, we report the marginal change

74

 

in probabi
is the rat
example, .
if the mat
degrees h;
MBAs. Fe
fund. To
models us
one if tht
otherwise;
\Vt
2004), ant
market). l

Characteri,

 

 

in probability for each independent variable as well as the log-odds ratio.1 The odds ratio
is the ratio of odds for two different observations of one explanatory variable. For
example, an odds ratio of 1.2 for the MBA dummy variable, which takes the value of one
if the manager has an MBA and zero otherwise, would mean that managers with MBA
degrees have a 20% higher probability of being in the top performance quartile than non-
MBAs. For all estimated models (OLS and logit) we estimate clustered standard errors by
ﬁrnd. To check the robustness of the results, in untabulated tests we re-estimate all
models using an alternative speciﬁcation for the dependent variable, giving it the value of
one if the fund belongs to the worst (bottom 25%) performance quartile and zero
otherwise; the results are available upon request.

We estimate all models for all funds-years in both the full sample period (1997 to
2004), and the two separate sub-periods (1997 to 2000, bull market; 2001 to 2004, bear
market). In this way we can determine whether the behavior of manager-speciﬁc

characteristics depends upon underlying market conditions and levels of uncertainty.

2. 2.2 Data Description
2. 2. 2. 1 Mutual Fund Data

All of our mutual fund data come from the nine January CDs of Morningstar,
Inc.’s Principia Advanced database from January 1997 to January 2005.2 The January

CDs report data as of December 31St of the previous year. Morningstar started the

 

I We also repeat the analysis by running probit models. The marginal change in probability is almost
identical. We choose to report the logit estimation mainly because of the intuitive interpretation of the log-
odds ratio.

2 Morningstar, Inc. used different names for this database throughout our sample period. The three different
names are: a) Principia Mutual Funds Plus, b) Principia Mutual Funds Pro Plus, and c) Principia Mutual
Funds Advanced.

75

 

TC .55 ‘— ‘—

Principia
informatio
version ot
operation
January 1
database.
1997, so a
equity fun
gromhq'm
funds. as \
decisions:
and loads.
names. p0;
aSsets inn
manager P.
if the ﬁmd
are dividek
than RVQ P
We also 0‘.
Morning“

caIEgOriZe

 

Principia database on January 1996. The Principia Advanced version contains more
information, especially regarding managers and monthly fund returns, than the basic
version of the database. Using the nine CDs, we extract information for all funds in
operation every year from 1997 to 2005. We start with all the funds in existence in
January 1997 and we follow them through 2005 or until they disappear from the
database. We also include in our sample all the funds that started their operations after
1997, so as to minimize concerns about survivorship. Data are gathered for all domestic
equity funds with a self-declared investment objective of growth, aggressive growth,
growth-income, or equity-income. We exclude index funds, balanced ﬁmds, funds of
funds, as well as other types of funds that are restricted in some sense in their investment
decisions.3 For each fund we obtain annual and monthly returns, annual expense ratios
and loads, net asset values, total net assets, fund inception dates, mutual ﬁrnd family
names, portfolio characteristics such as turnover, total number of holdings, percentage of
assets invested in the top 10 holdings, and stock, cash, and bond holdings, as well as
manager names. In the “manager name” ﬁeld Morningstar lists the name of the manager
if the fund is solo managed, the names of the multiple managers if the fund’s total assets
are divided among more than one manager, or the term “Management Team” when more
than two people are involved in the management of the ﬁrnd and they manage together.4
We also obtain data on the Morningstar equity style box designation for the fund and the
Morningstar category to which the fund belongs. These two variables are similar, as they

categorize funds among the nine different investment styles according to the style (value,

 

3 These include socially conscious funds, life cycle funds, target retirement funds and tax managed funds.

’ The exact description of what the term “Management Team” means, reads are follows: “This is used when
there are more than two persons involved in ﬁmd management, and they manage together, or when the fund
strongly promotes its team-managed aspect”.

76

growth, and blend) and size (small, medium, large) of their underlying portfolio holdings.
The only difference between the two variables is that the equity style box is calculated
using the latest portfolio information for the ﬁrnd, while the Morningstar category is
based upon the trailing 36 months. From the advanced analytics view of the database we
manually obtain each fund’s management fees, which are the fees that the management
company charges to manage the fund’s portfolio. For most funds the management fee that
appears on Morningstar is taken from the fund prospectus For other funds a minimum
and maximum management fee range appears in the database; for such funds we

calculate the midpoint and use the resulting ﬁgure as the fund’s management fee.

2. 2. 2.2 Manager Data

From the advanced analytics view of each CD, we hand-collect additional
information about all portfolio managers, regardless of whether they manage a fund
individually or as part of a team. The Morningstar CDs contain a brief biographical
sketch for each fund’s manager(s). For each manager, we collect data on the starting date
at the fund, gender, undergraduate and graduate institutions attended, degrees received
(including the year in which the degrees were received), whether they are a Certiﬁed
Financial Analyst (CFA), the name of the management company for which they work,
and other assets managed. Out of all distinct managers in our sample, we are able to
collect complete information for all of the manager characteristics variables for about
37% of them. This yields complete information for 46% of qll manager-year

observations. For 20% of the managers we have available data on all variables except

77

graduation date and for 40% of them we are missing data on several manager
characteristic variables.

After collecting manager-level information from Morningstar, we turn to other
sources to complete missing information. We ﬁrst turn to the 2004 CD of Nelson’s
Directory of Investment Managers. Nelson’s 2004 CD-ROM has information about most
of the management companies and managers in the portfolio management industry as of
March 2004. Thus, matching manager names and management companies from the 2004
January Morningstar CD with those from Nelson’s CD, we try to retrieve as many
missing data as possible. We then turn to each fund’s prospectus, which we locate on the
fund family website. After completing as much information as possible for the managers
of all the funds that appear in our data set in 2004, we track those managers in earlier
years and complete their missing information.

We obtain data on all manager variables for 42% of the managers, which results
in complete information for 56% of all manager-years. It is worth mentioning that for
31% of the managers (28% of manager-years), we are missing only their date of
graduation from college. Retrieving those graduation dates would lead to complete
information on more than 80% of the manager-years in our sample. We use the data on
those managers to perform robustness tests on our ﬁndings in Section 2.4. In summary,
we create a unique data set, which, to the best of our knowledge, provides the most
comprehensive set of mutual fund manager characteristics to date and is the ﬁrst to

include characteristics on managers that work in teams.

78

2. 2. 2.3 Benchmark Portfolio Returns Data

To evaluate the risk-adjusted performance of funds we employ returns on
benchmark portfolios. The performance metrics we use are described in Section 4.3.2.
We use the value-weighted NYSE/AMEX/Nasdaq composite index as our market rettu'n,
and the one-month T-bill rate from Ibbotson Associates as our risk-free rate in calculating
excess market returns. Returns on the HML (high minus low book—to-market returns) and
SMB (small minus big stock return) zero-investment portfolios as well as returns on a

momentum portfolio (UMD) come from Kenneth French’s website.

2.2.3 Variables
2. 2. 3.1 Distinct Fund Portfolios

The funds that appear in the Morningstar CDs represent fund offerings that the
investor can choose from but do not represent distinct investment portfolios. As Nanda,
Wang, and Zheng (2005) document, in the 19905 many mutual funds introduced
additional share classes as a way to offer investors more choices about the timing of load
payments, or to provide lower expenses to investors with big holdings. They show that by
the end of 2002 more than 50% of mutual funds offered more than one share class.
However, while various share classes offer investors different fund choices, they are
based the same underlying portfolio and consequently the same before-fee performance.
Thus, in the present study our unit of observation is the fund. This choice has
implications for our data collection.

The Morningstar Principia Mutual Funds Advanced Database offers the user the

option to extract data about distinct portfolios only. However, this option shows data only

79

for a fund’s oldest share class. For variables such as the portfolio characteristics of a
management team, such information is the same across share classes. In contrast, net
returns are different across share classes while gross portfolio returns, which are
calculated by adding back the annual expenses to the net annual returns reported by
Morningstar, are the same for all share classes. Further, using data on only one asset class
underestimates the total assets (size) of the portfolio, because it does not takes into
account the amount of money invested in the other share classes. We therefore perform
our own aggregation of multiple share classes into one fund observation. We are careful
to cumulate the total assets from all share classes to obtain the total assets of the
underlying portfolio. In order to identify different share classes of the same ﬁmd we
match different share classes by four portfolio characteristics: turnover, number of
holdings, percentage invested in stock, and percentage in the top 10 holdings. We also

verify our matching by looking at the ﬁrnd names.5

2.2.3.2 Performance and Risk Measures

Momingstar’s calculation of returns is determined each month by taking the
change in the fund’s monthly net asset value (NAV), reinvesting all income and capital
gains distributions during that month, and dividing by the starting NAV. Since fund
expenses are usually calculated daily and are reﬂected in the fund’s NAV, rettu‘ns
reported by Morningstar are net returns and are different for each share class. To derive
the gross (before- expenses) return of each fund portfolio, which is the same for all share

classes, we use a method similar to Gottesman and Morey (2005). We divide the annual

 

5 Multiple share classes of the same fund have basically the same name. Their names differ only by the
name of the share class. Example: “Vanguard Growth A,” “Vanguard Growth B,” etc.

80

expense ratio by 12 and add the monthly expense ratio to the monthly returns. 6 We use
those gross portfolio returns to evaluate and compare the performance between the
different types of management teams. We also calculate managers’ net returns by
subtracting the management fee they charge (same for all share classes) from the gross
portfolio returns. Again, we divide the annual management fee by 12 and subtract it from
the monthly gross returns to arrive at the monthly net returns.

We use three performance metrics to measure fund performance in a given year:
the gross annual fund return, the market model one-factor alpha, and the Carhart four-
factor alpha. To ﬁnd the gross annual return we add fund expenses to the net annual
return reported by Morningstar. We then run the following two time-series regression

models to estimate the one-factor and four-factor alphas, respectively:

Rit ‘ R ft = at + ﬂi/EMRz + ﬁrzSMBz + ﬂi3HMLr + ﬂi4UMDt + 3i: , (2)

where Ri, — R f, is the month-t excess gross return for fund i , EMR, is the excess market

return, SMB, is the difference in returns across small and big stock portfolios, HML, is
the difference in returns between high and low book-to-market portfolios, and UMD, is

the return on a momentum portfolio as computed by Fama and French. When estimating
the one-factor and four-factor alphas we use monthly gross fund returns for the 12

months in the year. To calculate the gross monthly return, we divide the annual expense

 

6 The annual expense ratio includes administrative, 12b-1, and management fees.

81

ration by 12 and add this to the monthly net returns. We use gross returns because we
want to measure the performance differences between various forms of management
team organization and characteristics. If better managers or organizational forms receive
rents through higher expenses, then the performance superiority of manager
characteristics might not show up when using net returns. However, we repeat our
analysis using fund returns net of management fees and the results are qualitatively the
same.

To evaluate the riskiness of the portfolio we use the market betas (the coefﬁcients
,6”) from equations (1) and (2), as well as the standard deviation of monthly returns,
throughout the course of the year. We use the estimated factor loadings, the coefﬁcients

ﬂu, [3,3, and ,8“, from equation (2) as measures of the fund’s investment style.

2. 2.3.3 Manager-level Variables

Using the information we collect from the biographical sketches, we construct
additional variables about the characteristics of individual managers for every manager in
our sample, including those that manage in teams. More speciﬁcally, for each individual
manager we construct the following variables:

1) Manager age: We calculate the manager’s age, which we use as a proxy for
experience, by starting with the manager’s undergraduate graduation date, and
assuming that the manager is 21 at the time of graduation.

2) SA T/I 00: We record the average SAT score of students at the institution where
the manager earned his undergraduate degree. Similar to Gottesman and Morey

(2005), we obtain up-to-date SAT scores for the undergraduate school the

82

manager has attended by searching collegeboard.com.7 We obtain maximum and
minimum bounds for the verbal and math sections of the SAT. The bounds are
constructed such that the middle 50% of students at the institution fall between
those bounds. For some institutions the composite ACT score is also reported, for
others only the composite ACT score is reported, and for a few schools (less than
10) neither the SAT nor the ACT lower and upper bounds are reported. For
schools for which SAT scores are available, we calculate the average SAT score
as follows: We calculate the midpoints of the upper and lower bounds of the
verbal and math SAT ranges and then take the average of those midpoints. For
schools for which both the SAT and composite ACT scores are available, we
regress mean ACT scores (midpoint of lower and upper bounds) on average SAT
scores (as calculated above) and predict average SAT scores from the mean ACT
scores. We ﬁnd SAT scores for 422 undergraduate institutions. Finally, for
sealing reasons, we divide the SAT score assigned to each manager by 100. The
fact that we gather SAT scores for the undergraduate institutions ﬁom 2005 and
business school rankings from the 19905 might raise some concerns, as the quality
of the schools might have changed over time and might be different relative to
when the manager actually attended the school. This issue is also raised in
Gottesman and Morey (2006), who test for possible changes in the relative quality
of schools and ﬁnd that the rankings of undergraduate institutions and graduate
business schools in the early 1980S are very similar to those in 2003. More
speciﬁcally, they ﬁnd that the correlation coefﬁcient between the 1983 and 2003

SAT relative rankings is 0.86 and the correlation coefﬁcient for business school

 

7 The data are gathered in the summer of 2005.

83

3)

4)

5)

6)

7)

quality (as measured by the GMAT score) is 0.82. Also, Dechev (1999) reports
that changes in the Business Week rankings are mostly transitory.

MBA: We construct an MBA dummy that takes the value of one if the manager
has an MBA and zero otherwise.

CFA: We create a CFA dummy that takes the value of one if the manager has a
CF A and zero otherwise.

Manager tenure: We calculate a manager’s tenure with the fund by subtracting
the date the manager started at the fund from the date for which we wish to
measure tenure.

Gender: We create a Gender dummy, which takes the value of one if the manager
is male and the value of zero if the manager is female.

MBA program rankings (B WI to BW5): We create ﬁve dummy variables that
correspond to the quality of the MBA program each manager attended. To do so,
we use the last ﬁve Business Week business school rankings as reported on
Business Week magazine. Business Week rankings come out every two years. We
obtain the 1990, 1992, 1994, 1996, 1998, and 2000 rankings. We assign a school
to the ﬁrst ranking category (BWl) if it is present in the ﬁrst 10 places of the
rankings in at least four out of the six Business Week rankings. We assign schools
to the second category (BW2) if they are present in the ﬁrst 15 places in at least
four out the sox rankings. Schools are assigned to the second-tier category (BW3)
if they are consistently ranked by Business Week (at least four out of six
rankings) and to the third-tier category (BW4) if they were ranked in any of the

rankings. Finally, we assign all remaining business schools to the “never ranked”

84

(BW5) category. The ﬁve dummy variables (BWl to BWS) take the value of one
if the school belongs to the corresponding category and zero otherwise.

8) Others: This variable takes the value of one if the manager works as a portfolio
manager for more than one fund at a time. We obtain this information from the
advanced analytics view of the Morningstar database.

Manager characteristics in the ﬁrst (1997) and last (2004) year of our dataset appear
in table 2.1. We do not see any signiﬁcant changes of managers characteristics over this
period. About half of the managers have a CPA designation, around 60 percent hold
MBA degrees and the majority of managers are male The most signiﬁcant difference
between 1997 and 2004 is that in 2004 more than half of the managers (56%) are
employed by more than one mutual ﬁrnds compared to 46% in 1997. Descriptive
statistics of manager characteristics appear in table 2.2 and correlations between variables

in table 2.3.

2.3 Hypotheses and results

2.3. 1 Discussion of hypotheses

The effects manager age has on performance and portfolio characteristics are
ambiguous. Manager age is highly correlated with experience and therefore superior
performance. However, Golec (1996) points out that manager age is also a measure of
stamina, which is negatively correlated with age and, assuming that investment
management is a highly demanding job, could have a negative impact on performance.
Moreover, if an older manager is closer to retirement his future earnings from his job

might not be that important to him. Chevalier and Ellison (1999) make a similar

85

argument. Younger managers who want to advance their careers and increase their future
income will work and try harder than older managers. We can only measure the
combined effect of those two competing hypotheses. If the experience effect dominates,
we expect to see superior performance by older managers. If the stamina and career
concerns hypothesis is true, we expect younger managers to work harder and better. Both
Chevalier and Ellison (1999) and Golec (1996) ﬁnd that younger managers perform
better than older ones. They also ﬁnd that older managers hold higher beta portfolios.

Tenure is a measure of fund-speciﬁc experience but could also be a measure of
success. On the one hand a manager’s long presence at a fund could be an indication that
the fund organization is happy with his performance; however, a manager with a long
tenure might have run out of new investment opportunities and ideas. Long tenure is
likely to make the manager more risk averse since he probably not get ﬁred unless he
signiﬁcantly underperforms. Moreover, since the existing portfolio is mainly his
construct, he would be less willing to turn it over even if some if his holdings are not
performing up to par. We expect long-tenured managers to hold more diversiﬁed
portfolios, have lower trading propensity (turnover), and worse performance than short-
tenured managers that have to prove and establish themselves.

The quality and nature of manager education can also prove important for
investment management. Golec (1996) underlines the importance of an MBA in
identifying well—managed companies and understanding basic investment principles.
Further, Chevalier and Ellison ( 1999) and Gottesman and Morey (2005) suggest that an
MBA degree provides superior information beneﬁts, given the connections the manager

builds with other members of the ﬁnancial and business community while attending the

86

MBA program. Such networking beneﬁts are likely to be greater the higher the prestige
of the business school. Accordingly, Gottesman and Morey (2005) also look at the
rankings of the business school in which the managers receive their MBA as well as the
average GMAT scores of those institutions. They ﬁnd that managers from high-ranked
institutions outperform managers from low-ranked institutions. Of course, it could be the
case that managers from more prestigious schools receive a better business education or
are hired by better management companies with superior resources and support staff.
Note that a CFA designation is also a measure of business-speciﬁc (mostly ﬁnance-
speciﬁc) knowledge.

The average SAT score of the undergraduate institution the manager attended can
also determine manager performance. In Chevalier and Ellison (1999), the SAT score is
the only manager characteristic studied that can explain differences in the risk-adjusted
returns (alphas) of mutual funds. Similar to the MBA score above, the SAT score is a
measure of the quality of the undergraduate institution, and therefore the quality of the
education the manager enjoyed, as well as a measure of the network connections that can
result in better access to information.

The next variable we include in our study relates to whether a portfolio manager
is a member of the management team of multiple funds. Investment management is a
demanding job and managing more than one fund at a time could have negative effects
on performance. In addition, a manager who is evaluated on overall performance across
the funds he manages might be willing to take more risky positions in the portfolios of

funds that are not doing well since he has little to lose in doing so. On the other hand, the

87

most obvious beneﬁt from managing several funds is the extended network the manager
might have access to.

In order to evaluate the effect of manager characteristics on portfolio composition
and risk, we estimate each fund portfolio characteristic using OLS equations (3), (4), and
(5), where the dependent variables are as follows: portfolio turnover, number of
securities in the portfolio, percentage of assets invested in the top 10 holdings, percentage
of assets invested in cash, standard deviation of monthly returns, market beta, market beta
from the four-factor model, SMB beta, HML beta, and UMD beta. Speciﬁcally, we

estimate:

Fund Characteristic” = a + b1(Gender;-,,_1)+ b2 (Agem / 10) + b3(ll/IBA,-,,)

+ anIiJ + b8 (Log _ Assets” ) + 8L!

71

Fund Characteristic i, t = a + b, (Genderl-J_1) + b2(Age,~,, / 10) + [)3 (WAorCFAm)
+ b4 (Othersi, , )+b5 (T enure, ) + [)6 (SA 72,, / 100) , (4)

+ anIiJ +b7(Log ._ Assets”) + 5w

n

Fund Characteristic” = a + b1(Gender,-,,_1) + b2(Age,-,, /10) + 2kain
k
+ [)3 (Othersu) + b4 (Tenure, ) + b5 (SA 7},, /100) , (5)

+ 21)” x [U + [)6 (Log _ Assets”) + 31,:

n

88

where i is the fund index, Gender is a dummy variable that takes the value of one if the
manager is male and zero otherwise, Age/10 is the manager’s age divided by 10, MBA
and CFA are dummy variables denoting whether the manager has an MBA or a CFA
designation, respectively, Others is a dummy variable that takes the value of one if the
managers works for multiple funds, and SA T/100 is the average SAT score of the
undergraduate institution the manager attended divided by 100. The variable MBAorCFA
takes the value of one if the manager has either an MBA or a CFA and zero otherwise.
The intuition behind this last variable is that the business- (and ﬁnance-) speciﬁc
education an MBA and a CPA provide are similar and therefore people that have either
might behave similarly. The set of BW dummy variables (BWl to BW5)is a set of
dummies that take the value of one if the managers attended an MBA program of the
respective quality ranking. The set of I dummy variables (11 to 13) accounts for
differences in the prospectus objectives of each ﬁrnd. These dummy variables rake the
value of one if a fund has the respective prospectus objective, and zero otherwise. 11
corresponds to the “Growth and Income” objective, 12 corresponds to the “Growth”
objective, and 13 corresponds to the “Aggressive Growth” objective. “Equity-Income” is
the omitted prospectus objective category.

To evaluate the effect of manager characteristics on portfolio performance we
estimate three logit models for each performance measure described in Section 4.3.2. We
take all the single-manager funds and rank them in performance quartiles. We then
investigate how manager characteristics affect the probability of a fund being in the top
performance quartile. Speciﬁcally, we estimate and report results for the following three

models:

89

Prob(TopQuartile) i, , = F (b, (Genderj,,_1) + b2 (A361,; / 10) + 173 (MB/41,: ) + 1’4 (CF/1%,)

, (6)
+b5 (OthersL, ) + b6 (Tenure, ) + b7(SA 7}, / 100) + b8(Log _ Assets” ) + 8“)

Pr0b(T0PQuartile)i,r = F (b1(Ge"derr,r—1)+ 1920186,: / 10) + 53(WAOVCFAI,1 ) +

, (7)
b4 (Others,,,)+b5 (Tenure, ) + b6 (SA Ti, /100) +b7(Log _ Assets” ) + 5,3,)

Prob(TopQuartile),-,, = F(b1(Gender,-,,_ ,) + b2(Age,-,, / 10) + ZkaWU
k , (8)
+b3 (Othersu ) + b4 (Tenure, ) + b5 (SA 71;, / 100) + b6 (Log _ Assets,“ ) + 81L! )

where all independent variable deﬁnitions are the same as in OLS regressions (3) to (5).

2.3.2 Results

Tables 2.4 though 2.7, present results of the regression for single managers and
reveal a number of interesting ﬁndings that can be compared and contrasted with
previous literature.

Table 2.4, panels A to D present the results for turnover, number of securities held
in the portfolio, and percentage of the fund’s assets invested in the top 10 holdings and in
cash, respectively. In panel A we ﬁnd that the manager’s age, tenure, and employment in
other funds signiﬁcantly affect turnover, though signiﬁcance varies with underlying
market conditions. Speciﬁcally, unlike Chevalier and Ellison (1999) and Gottesman and
Morey (2005), we ﬁnd that older managers turn their portfolio over less often than

younger managers, who prove to be more active traders; the coefﬁcient on the age

90

variable is around -9 for the full sample period, depending on the model speciﬁcation,
and is signiﬁcant at the 1% level, whereas the coefﬁcient for older managers is higher for
the bear market period (coefﬁcient ranging from -1 1.391 for the ﬁrst speciﬁcation of the
model to -13.545 for the third speciﬁcation) and still statistically signiﬁcant at the 1%
level.

However, similar to Gottesman and Morey (2005), we ﬁnd that manager tenure
has a signiﬁcantly negative effect on turnover and that this is true across all sample
periods; a manager with ten more years of tenure is expected to have almost 22% lower
turnover. With respect to employment in other funds, we ﬁnd that managers who work
for multiple funds turn their portfolio over more, but this is true only during the bear
market period. Our ﬁndings for the MBA dummy variable are consistent with past
research. In line with Chevalier and Ellison (1999), whose study focuses on a bull market
period, we do not ﬁnd a signiﬁcant relationship between an MBA degree and turnover in
the ﬁrst subperiod, whereas we do ﬁnd a positive and signiﬁcant relationship during the
bear market period, conﬁrming Gottesman and Morey’s (2005) results. We also ﬁnd that
higher turnover is associated with MBA degrees from lower-ranked institutions. This
result is also consistent with Gottesman and Morey (2005), who ﬁnd that managers from
top-ranked MBA programs tend to record lower turnover.

Panel B shows that in terms of number of security holdings in the portfolio, age
again seems to have an effect. Older managers have more concentrated holdings,
especially during bad market conditions. The opposite holds for managers with MBA
degrees -- they signiﬁcantly hold more securities in the portfolios, especially in the

second subperiod (coefﬁcient of 21 .738, signiﬁcant at the 1% level). However, this is not

91

true for all managers with MBA degrees. Results from the third speciﬁcation of our
regression indicate that managers from top business schools hold more securities
(coefﬁcient of 31.186 for managers from the top 10 business schools), while the
coefﬁcients for managers from third-tier and unranked schools are not signiﬁcant at any
level.

In panel C, managers with MBA degrees also have less concentrated holdings in
their top 10 holdings, which seems to be true for all MBAs, and the same holds for
managers with a CFA designation; the coefﬁcient for both dummy variables is negative
and signiﬁcant at the 1% level for all model speciﬁcations and sample periods. In
contrast, managers with longer tenure behave in the opposite way; the coefﬁcient of the
tenure variable is positive and highly signiﬁcant in all regressions.

Finally, in Panel D, we ﬁnd that long-tenured managers hold more cash in their
portfolio, especially in the bull market period. Managers for hi gh-SAT institutions exhibit
the same behavior. There is no relationship between percentage of assets invested in cash
and either of these variables for the 2001 to 2004 period.

Previous literature finds a relationship between risk and manager characteristics
such as age and MBA education. Table 2.5 reports results for the standard deviation of
fund returns and portfolio beta. Focusing on the standard deviation of returns, the only
signiﬁcant variable is manager tenure. In particular, managers that have been working
with the fund for a long time exhibit a lower standard deviation of returns during the bull
market. There is no signiﬁcant relationship for the relatively bearish period of 2001 to
2004. In terms of market beta, both Golec (1996) and Chevalier and Ellison (1999) ﬁnd

that managers with an MBA hold more systematic risk. Gottesman and Morey (2005) do

92

not ﬁnd any such relationship. Our results are similar to Gottesman and Morey (2005), as
we do not ﬁnd any relationship between quality of education (MBA, CFA, and SAT
score) and systematic risk. We do ﬁnd, however, that older managers hold less systematic
risk during a bear market. We also ﬁnd a negative coefﬁcient on tenure for the whole
sample period, especially for the bull market period of 1997 to 2000. In summary, we
ﬁnd that managers that have been with a fund for a long time take on less risk, probably
to minimize the probability of losing their position, though the result is not robust in the
bear market period of 2001 to 2003.

Table 2.6 shows that our results provide some evidence of differences in terms of
investment style. Older managers have higher HML factor weights. This result is
signiﬁcant at the 1% level for the bear market period. Depending on model speciﬁcation
the coefﬁcient ranges from 0.055 (speciﬁcation 1) to 0.061 (speciﬁcation 2). Managers
with an MBA also have higher HML factor weights, however, this result is due to
managers from top business schools (coefﬁcients around 0.10, signiﬁcant at the 1%
level). We do not ﬁnd evidence of this relationship for managers that graduate from low-
ranked business schools.

Performance results are presented in Table 2.7. We report the marginal change in
probability instead of the logit estimates, as well as the odds ratio for each independent
variable. Similar to Chevalier and Ellison (1999) we ﬁnd that the most signiﬁcant
manager characteristic with a positive effect on performance is the SAT score of the
undergraduate institution. The coefﬁcient of the SAT variable in model speciﬁcations 1
and 2 is positive and signiﬁcant at the 1% level when we use gross portfolio returns as

the performance measure. When we account for market risk, the coefﬁcient on SAT is

93

still positive, but signiﬁcant only at the 5% level (p-value around 0.03) this time. Finally,
using four-factor alpha as our performance variable we continue to obtain positive and
signiﬁcant coefﬁcients, but signiﬁcance decreases to the 10% level. The SAT variable is
signiﬁcant during both the full sample period and the bear market subperiod. These
results are in line with Chevalier and Ellison (1999) and provide evidence that this
relationship is present even during a bear market. When we repeat the analysis using the
third model speciﬁcation, in which we include the business school ranking, a very
interesting result emerges: the SAT variable is no longer signiﬁcant, though we the
business school rankings dummies continue to show signiﬁcance. More speciﬁcally, we
ﬁnd highly signiﬁcant negative coefﬁcients for the third—tier and unranked business
school dummies for both the full sample period and the bear market subperiod, and a
signiﬁcant positive coefﬁcient for the top 25 business school dummy during the bull
market period in terms of four-factor alphas. Managers with an MBA from a third-tier
school have 67% less chance of being in the top performance quartile than other
managers (including those that do not have an MBA), while managers from unranked
business schools have 56% less chance of being in the top performance quartile in terms
of four-factor alphas. This result suggests that managers from top business schools
perform better than managers from lower-ranked business schools, but not better than
managers with no business education. Moreover, our results suggest that managers from
low-ranked business schools also perform worse than non-MBAs. Gottesman and Morey
(2005) ﬁnd similar results for the 2000 to 2003 period.

The above results regarding SAT scores and the business school ranking dummies

are not that surprising if we take into account the correlations reported in Table 2.3

94

between the SAT score and the business school ranking dummies. The correlation
coefﬁcient between the SAT score and the BWl dummy is 0.331, and that between the
SAT score and the BWS dummy is -0.328. These correlations suggest that managers that
graduate from high-SAT undergraduate institutions go to better business schools and in
turn perform better than others. We cannot distinguish whether this superior performance
is because these managers are smarter because they graduated from a high-SAT score
institution or because they went to a better business school. Both could be the case. And
of course other explanations may exist. For instance, managers with high SAT scores and
with MBA degrees from top business schools might be hired by better investment
companies with more resources and experience, leading in turn to superior performance.

In terms of other manager characteristics the only signiﬁcant variable is manager
gender. Male managers have a higher probability in being in the top performance quartile
than women. This result is highly signiﬁcant for one-factor and four-factor alphas, and
for bearish market conditions. Note that we only ﬁnd a positive and signiﬁcant
relationship during the 1997 to 2000 period when the one-factor alpha is the performance
metric used.

In summary, we conﬁrm a number of past ﬁndings and we provide several new
insights on the role of manager characteristics in mutual fund performance. In terms of
performance, we conﬁrm Chevalier and Ellison’s (1999) ﬁnding that the SAT score of a
manager’s undergraduate institution has a positive effect on performance, and we show
that this result is robust to bear market conditions. Also, consistent with Gottesman and
Morey (2005) we ﬁnd that the quality of the business school matters, but we show that

this variable is highly correlated with the SAT score, suggesting that managers with high

95

SAT scores are the same managers that attend highly ranked MBA programs. In addition,
we ﬁnd that male managers outperform female managers, especially during bad market
conditions, and that managing multiple funds does not negatively (or positively) affect
the manager’s performance.

For robustness, we reestimate all models focusing on the effect manager
characteristics have on the probability of the fund being in the worst performance quartile
and ﬁnd similar evidence. The only exception is the Others variable. Even though we do
not ﬁnd evidence that when the manager works for multiple funds the ﬁrnd has a lower
probability of being in the top 25%, we do ﬁnd evidence that a manager working in
multiple funds increases the probability that a given fund will be in the worst

performance quartile.

4. Robustness Checks

When we perform our analysis for single-manager characteristics, we exclude a
signiﬁcant number of managers because we do not have information about their age.
However, for many of them other characteristics information is available. As reported in
Table 2.4, if we disregard the manager age variable we have complete information for
72% of managers. In terms of management team-years, disregarding the age variable
would lead to complete data for 84% of all management team-years in our data set. To
investigate whether our results change when we utilize this larger data set, we reestimate
all models, this time dropping the age variable from all regressions. We ﬁnd that all our
results regarding the other manager characteristics are similar. In particular, we obtain

slightly different coefﬁcients, but the sign and signiﬁcance levels are unchanged.

96

We also perform an additional check. In all our logit regressions we rank the
funds by performance quartile and then estimate logit models in which the probability of
a ﬁrnd belonging to the best (or worst) performance quartile is the dependent variable. To
determine whether the results are sensitive to our choice of performance categories, we
reestimate all logit models, but this time we rank the funds in performance terciles and
quintiles. In each case, the results are very similar to those that obtain when we use
quartiles. We do not see major changes in the signiﬁcance and sign of the relationships

for any of our results.

5. Summary and Conclusions

In this paper, we study the effect of the structure and characteristics of mutual
fund portfolio management teams on mutual fund portfolio formation, risk, and
performance. We build a unique data set that contains information about the management
teams of more than 2,000 distinct mutual ftmd portfolios and 1,200 mutual fund
managers. Our analysis consists of three parts. We ﬁrst focus on sole-managed funds and
relate individual manager characteristics to risk taking, investment style, and
performance. We then compare the performance and portfolio characteristics of team-
managed and sole-manager funds. Finally, we examine team-level characteristics of
team-managed funds. We conduct our analysis over the 1997 to 2004 period and also for
the two 1997 to 2000 and 2001 to 2004) subperiods to determine whether the full-sample
results are robust to differing market conditions (bull and bear markets).

We ﬁnd that single-manager characteristics can explain differences in risk taking

and investment style. For instance, portfolios of long-tenured and older managers exhibit

97

higher levels of turnover, whereas managers with MBA degrees, especially graduates of
highly ranked business schools, turn their portfolios over less frequently during the
relatively bearish 2000 to 2004 period. Managers with MBA degrees also hold more
stocks in their portfolios and invest a smaller part of the fund’s assets in their top
holdings. The opposite is true for older managers.

We fail to conﬁrm Golec’s (1996) and Chevalier and Ellison’s (1999) ﬁndings
that MBAs hold more systematic (beta) risk. Instead, we ﬁnd that long-tenured managers
are associated with lower beta portfolios. In terms of performance we obtain some
interesting ﬁndings. We conﬁrm Chevalier and Ellison (1999), who report that managers
that attend an undergraduate institution with a high average SAT score outperform even
during a bear market. However, we show that this is mainly because those managers
attend more prestigious MBA programs; the SAT variable is positively correlated with
the BWl dummy and negatively correlated with the BW4 and BWS dummies. One
possible explanation for these ﬁndings is that high-SAT managers that attend top-ranked
MBA programs are smarter, receive a better education, and enjoy a better network of
connections. Of course it could be the case that those managers are hired by better

management companies with more resources and experience in portfolio management.

98

APPENDIX 2.

TABLES OF ESSAY 2

99

Table 2.1
Manager Characteristics

This table summarizes the characteristics of mutual fund managers in the ﬁrst (1997) and the last

(2004) year of our sample period.

 

 

 

 

1997 2004
Siggle Managers Single Mamers
N (%) N (%)
Funds with Complete Manager
Information 381 344
Managers managing the above funds 311 100% 241 100%
Managers with CFA designation 166 53% 123 51%
Managers with MBA 196 63% 146 61%
MBA (Rankcode=1) 81 26% 65 27%
MBA (Rankcode=2) 37 12% 23 10%
MBA (Rankcode=3) 30 10% 20 8%
MBA (Rankcode=4) 4 1% 5 2%
MBA (Rankcode=5) 44 14% 33 14%
Managers with other Masters 20 6% 23 10%
Managers with J.D. degree 8 3% 5 2%
Managers with Ph.D. 9 3% 6 2%
Average Manager Age 47.13 - 47.76 -
Average Manager Tenure 5.98 - 6.28 -
Average SAT Score 632.23 - 639.59 -
Number of Male Managers 288 93% 216 90%
Number of Female Managers 23 7% 25 10%
Managers Managi_ng Other Funds 142 46% 135 56%

 

 

100

Table 2.2
Descriptive Statistics of Characteristics Variables

This table presents summary statistics of manager-level variable for the 2,660 single manager-
years for sole-managed funds. GENDER is a dummy variable that takes the value of 1 if the
manager is male and 0 otherwise. Age is the managers age in years. MBA is a dummy variable
which takes the value of 1 if the manager has an MBA. CFA takes the value of one if the manager
has a CFA designation and 0 otherwise. Others is a dummy variable that takes the value of 1 if
the manager works for other funds at the same time and 0 if he only is employed by one fund.
Tenure is the manager’s tenure with the fund in years. SAT is the average SAT score of the
manager’s undergraduate institution. The ﬁve business school rankings variables are dummy
variables taking the value of 1 if the manager’s graduate school belongs to the corresponding
ranking and 0 otherwise.

 

 

Std
Varriable Name N Mean Deviation Minimum Maximum
GENDER (1 =Male) 2660 0.9113 0.2844 0 1
Age 2660 47.7192 9.3627 26 84
MBA (1 =has MBA) 2660 0.6361 0.4812 0 1
CFA (1 =has CFA) 2660 0.5526 0.4973 0 1
Others (1 =managers other funds) 2660 0.6462 0.4782 0 1
Tenure 2660 6.0852 5.1037 0.1667 25.9167
SAT 2660 640.7871 71 .0796 412.5 745
Gr. School Rank (1 -10) 2660 0.2842 0.451 1 0 1
Gr. School Rank (11-25) 2660 0.1000 0.3001 0 1
Gr. School Rank (second-tier) 2660 0.0917 0.2887 0 1
Gr. School Rank (third-tier) 2660 0.0218 0.1461 0 1
Gr. School Rank (unranked) 2660 0.1383 0.3453 0 1

 

101

 

 

.8... :3. and. E3. 88.? 88.? :88 ~88. ~88. ooo.o. 83 mwod 85$ 5:
83 «8.8 23- «8.? 88.? «8.8. 83 «8.... ~88 mwod. 2.2. :8 28...: w:
83 Ra... 83. Rod. 83. as... a; 88 omod «8... «8.8- 9.86 :5

83 8.8. 5.8. 33. «Rd. 83 m8.o 88 80.8- 88.8 85 a:

83 88.8. 88.9 83. {.3 m3... 2:. n8... 8...? 25m 5

88.. 88. 88. 33 23 «3.8 8.8. .8... 35 E

83 38. $3 Sod 88 88. 28.8. «26 E

83 «8.8 9.8.8. 8.3 83 .13 :sm .3

83 «mm... «8.8 88. #3. Eu a $2 E

88.. «8.: 83. Bed. E0 E

83 83. ~88 <92 E

83 28... 8:02 E

83 222 E

a a = a a a E E E E E E E E E 8.8_a>

102

Became—op one .833 .8 BR: Axum 05 we “:82an 356508 cone—8.80
mesa 3358-28 he mcmozdommemﬁ ooc.~ :3“ H8 83559» 85589820 homage =m mo £58508 :ouﬂoboo 05 3535 038 mat.

833.5.» 85239229 Sauna:— ue 3.5:: zeta—9.30
ON «Sah.

 

88.8 :88 82.8
08.8 25.. 03.: $0882
:88 88.8 88.8
8m.» $8. 83 <8
88.8 88.8 88.8
08.2 83.- 83 <02
:88 :88 88.8 88.8 :88 :88 88.8 :88 68.8
03.8- 8.:- 888- 088.- :3. «Rs- 8::- 23. 83. 8:02
~88- 68.8 88.8 :88 88.8 :88 85.8 32.8 :88
08.0. ~88- 88.2- 80.» 83 2: 808. 83- oar-.m- 88:00
8 8 E 8 a 8 8 8 :8
v8~88~ 8838: 88-82 0.8.2.8
88a 88: 80 8:0“. 88.2 .30 88a 2980 =3“.

 

030.89. 2.80:ch 05 0.. .8853. ..< 8ch

608850 08 65 >0 8080 686008 60.8820
886888 E 808% 823-8 6:0 6006200 08 8300 8 88— $3 06 8 6006630. 838% 6080800 8: 08 3885000 086 :5 00280800
06 E 606306 08 86.556 08C. 0380.30 2500808 @656 06 .8.“ 6580 MN 8 E 80> 06 .8 8800 05 .8>0 8.080 8: ~80”. 0w08>0
P606 06 («0 w0_ 06 2 80884.8 M04 065.860 080 6:0 mac—:8 36009-0800 05 8 8880 80:8 0868a P8868: 06 .2 0:0 .80 028»
06 0x8 86 83089 >886 08 8533 mace“.— _00:0m 80:83 0>m 05- .2: >9 606.56 006355 08689068 988288 06 .«0 088
Had 08.88 06 ﬂ 83R Va. .m80> 6 6:6 06 68> 08:00 9.8me 06 ﬂ 0.328-m 6:6 000 >0 608380 .04 >60 0: .6 08a 6:0 0:5 088 06 8
6:3 03208 .8.“ 8803 88288 06 a 0:0 80 028» 06 0.068 86 032...? >886 a mm 8.850 083.860 08a 600 0063986 <m0 0 .8 <m2 :0
868 was 88:08 06 .8 0:0 80 0.68 06 80.8 86 03098 >886 a 3 34086.38 083.860 o 600 8388806 .35 0 0.0: 88288 06 .2 0:0
.80 020> 06 863 N06 .<m2 :0 8: 8888 06 .2 0:0 80 033 06 803 86 050ta> >886 a 2 “$3 .3 >0 6063.6 88> E 0mm 808288 06
2 383. 0.065.860 08m 6:0 0.08 ...: 88:08 06 a 0:0 .80 020> 06 803 86 03088, >856 0 mm 86:08 .>_0>600%00 £80 E 60803: $0.080
0:06.88 .80 08300088 06 600 886—0: 2 QB 06 E 60802: 8080 0:80-88 .80 0838088 06 8:823 06 5 8.0680000. .80 800.5: 06 58> 06
80 08000 06 8>0 8853 mm 0308? 08600806 06 .Q 68 ,U .m .< 2058 E .Amv 68 A8 A9 8880800 mAO :86 $380 $888 058 mE-H

 

838808320 83.82 :0 85:08:20 60:..— .«0 80088.80:
v." 03:.

103

880.8
30.00

808.8
02.00

83.8
«F F0

:88
03.0-
:88
08:-
$088
08.0

808.8
000.3

.058
{0.00.

888
330

80088
08.2

38.8
000.0

88.8
03.8

88.8
00.0.00

:58
0.00. F F

8008
5.?

83.8
02.0.

$008
000.:

88.8
000.8-

880.8
30.8

8000.8
80. F F

:88
03.0-

$88
08.0.

:08
F0;

888
30.2-

88.8
02.00

28.8
0.00.:

8008
F80-

88.8
83.
88.8
000.3

8000.8
002-.

808.8
03. F-

:88
3.0.2-

8.0.8
:03 F-

68.8
08.0-

88.8
000.00-

80.8
0050

:88
03.2

300.8
000.0.

88.8
000.0-

808.8
08.2

808
F8:-

88.8
«8.00
800.8
000.:

:88
08.0.

88.8
000.0.

0.88
08.0..

3080 0.0 20:

88.8
03.00

880.8
2.0.00

808.8
002..

800.8
08.~-
88.8
30.0-
63.8
000.2

880.8
0 F F.0-

A0008
000.00

800.8
000.0

8000.8
000.0.

A088
20.0

88.8
08.00-

88.8
F0000

:88
08.2

800.8
08.0-
88.8
F03-

:88
000.8

88.8
000.2

88.8
80.8

88.8
80.2

803.8
000.0-

880.8
000.0.

:58
FF0.FF

m.

N.

2

oo _-\._.<m

05:08.

0.550

m>>m

v>>m

m>>m

N>>m

{Sm

104

 

 

 

 

 

$.38 :88 38.8

.83? 8...... 8% E9935.

$88 :58 $28
«3.8. 2.3. 8%. $0

888 38.8 :88
8». K 83 «3. P P 5.2

88.8 :68 :88 $8.8 $8.8 68.8 88.8 88.8 38.8
83. 23.. 882. 83. 83.. «.88. 88¢. 83. 2.3. o :03

88.8 688 3:8 88.8 38.8 88.8 :38 838 8:8
82: {.88 ~88 $8.”- 8:. 31m- 85” was.” 9.3 .880

E E A 8 E E 3 a Q A 8
88-88 88-82 «8382 osmzm>
Bron. 582 8% Eton. 88m: :5 88m 238 =3“.

$3.29. Embamqmb m5 8 0.80qu 05 3 8850mm *0 30:32 ..m $ch
82.0 :88 22.8 38 82.8 82.8 82.8 $8 .o 8.88 Baamé
82 83 83 NR; 93 NR; 83 83 83 88338

88.3 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8
88.83 5.x: 3.8.5 88.": 882 88.34 852 $33 28.3 80285

:88 88.8 88.8 38.8 38.8 38.8 88.8 88.8 888
SS. 83. 83. S3. :3. $3.. «.88- 3%- $3. 9328 84

388 ...N ~35

105

388
m8. 8 F-

$88
88.8-

88.8
88. F
39.8
83-

a. .8
88.2-

:88
88. v

$8.8
am:-

88.8
33....

88.8
88.8

68.8
85”

£88
83;-

88.8
83.8-

A 8.8
83

:88
88.

$88
82-..-

38.8
88.2-

88.8
.88-

$88
8:.

68.8
mono-

:88
98.8-

88.8
88.8-

38.8
88.2-

33.8
83.

A858
«8.8-

88.8
88.0

88.8
82..-

£88
82:

£88
83-

85.8
88.8

828.8
m 5.9

88.8
8%.

88.8
82:-

338
8a

888
:88.

$88
83m

68.8
~88

88.8
82:-

88.8
2.3

688
two-

888
83.

388 ...N 28,—.

68.8
E:-

89.8
08.8.

:88
~88

88.8
8nd.

38.8
«88-

82.8
83.

38.8
t 2

CR8
53

88.8
3::

38.8
2&8

38.8
83-

3. :8
$8.3.

68.8
83

83.8
www.c-

E88
mam-.m-

88.8
www.m-

32.8
8me-

82.8
8%

£88
Eno-

838
888-

N.

2

00 {Cum

9:8;

225

m>>m

v>>m

m>>m

N>>m

_.>>m

106

 

 

 

 

 

$8.8 88.8 88.8
88.- 888. 8.8. <80
88.8 88.8 88.8
:88- 88. F. 888. <8.2
888 88.8 E :8 88 :8 88.8 A-~8 888 am :8 88
88.. <88 :88 888 8 58 888 888 888 888 8 to?
:88 88.8 38.8 388 :88 88.8 38.8 88.8 88.8
888 88 888 88.: 83 8.: 8:8 .8: .~8.~ .880
:8 Q A 8 8 E E Q E A 8
88-88 88-82 88-82 8.88,;
8888 898.2 .88 888“. .9822 .58 .888 8.858.0- =8".
83%? “cotton-Eb ms macioc 2 QB- 85 S bow-828.. 3mmm< \o gﬂzmogol ..0 $ch
88 8 28:8 :38 888 :88 888 8:28 5:8 888 8988.8.
88.: 88.: 88.: 88.: 88.: 88.: 88.~ 88.~ 88.~ 88883880
38.8 G 8.8 :88 828 A888 358 88.8 A 88.8 828
2.8.2. 8888 :88 F888 8888: £88 .88.... ~88 88.8 888.88.
88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8
88.8 88.2 8.8 288. 88.: 88.: ~88 88.3 88.2 2882 .8 8..
6:8 388 E88 888 88.8 :88 88.8 :88 888
8:8 88.~- v88- 88 :~. : 88.8 8 58 ~38 88.~ 8.

3:88 .8 8.8.:-

107

A888
88....

A888
5.8

A88
8:. :-

A888
888

A808
888

A288
.88.

A888
«88-

A 88
88.?

A808
88. .-

A288
88...".

A888
8.8.8

:88
$88
:88
38...
A288
«88

A888
$88

38.3
”3.”.

A888
83.

A888
88.8

A888
88. :-

A288
~88

A838
88

A888
83

A888
88. .
A888
888
A .88
88
A828
88. :-
A888
8.8."-
A888
888.

:88
888.

A888
888-

A888
88. .-

A888
8.8.8.

A858
888

A2 :8
888

:88
888

:88
88. .-

:88
28.8.

:88
«88

:88
888

A88
:88

A888
88

Aoomdv
5N. ..-

A888 .8 28,8

A888
83

A8: :8
888

A288
888.

A888
888

A888
888-

A858
«88--

A888
:88-

A888
888-

A828
8:. :-

A888
88.8-

A888
83

:38
8.8..

A 888
.88-

A888
888

:38
888-

:88
8:..."-

A888
88.»

A88
85

A888
:88-

A888
838

:88
888-

N.

2

oo {ham

23:2.

9050

m>>m

¢>>m

m>>m

N>>m

_.>>m

(n. 0.5455.

108

 

 

 

 

 

A888 :88 A888
888. 888. 8.88- Eu
A888 A888 A888
83.8. «88 8:8- .82
A888 A888 A888 A888 A888 :88 A888 A888 :88
888 88 ~28 8 :8- 88. .88 88 ~88. 888- 8:82
A888 388 A888 A888 :88 A888 A888 :88 A888
88 $88- 888. .88- 88..- St .- 88... 88... 88..- .880
A8 A8 A: A8 A8 3 A8 :8 A 5
88-88 88-88 88-82 288>
888 898.2 .88 88.”. .982 :8 8:8 8.858 =8“.
oBmtm> 8:905qu 2: m.‘ ammo E P8835 «.886 m5 *0 mmmucmogml ..Q 3ch
:88 888 8888 8:8 888 888 :88 828 8: 8 888.1
88: 88: 88.: ~8: ~88 ~8: 888 888 888 8882080
A888 A888 A888 A888 A888 A888 A888 A888 A888
v8.8 «2.8 ~88 88.8 8.8 ~88 88.8 88.8 88.8 .8288.
A888 :88 A888 A888 A888 A888 A888 A888 A888
88. 388 88. ~88. v8..- 83. 888- 888. 888. 982 8 8..
A888 A~88 :88 A888 :88 A888 A888 A888 A388
v8... 83. 888 888 v8.» 83. 8....- 85. ~88 8.

3.58 YN 039—.

109

3&de
mwod-

A808
888

A888
~28

A888
8.88.

:88
88

A838
38.

A888
5......

A888
e-...

A888
888

:88
888.

A-~8
~28-

6:48
mwod.

A888
88

A888
~58.

Am-8
8.8

:88
~88

A838
23.

:88
888.

A-v8
58

A828
388.

:88
~88

A888
888

:-8
~88

A888
3.88
A888
838
A888
888

A888
888.

A888
.8. 7

A888
8:...

:88
8::

A8~8
~28

A888
388

A888
888
:88
o-8
:88
828

$88
888

A858
2:8

:38
5.8

:88
8~88

A888
3:8

:88
38

:88
888

A888 v.~ as?

Am-8
828

A838
288

:88
~88
:88
88

Am: :8
888

A8-8
888.

:88
33-

A888
~85.

A838
888

A888
~:~8-

A888
~88

A858
~88

:88
~28

A :88
888

:28
~88

A858
~88-

A888
388.

A888
.888

:88
288

A888
~88

A 588
888

N.

2

oo :._.<m

825...

38.0

m>>m

v>>m

m>>m

N>>m

P>>m

<n_0._o<m_2

110

 

 

858 888 8:88 888 888 8.88 888 858 888 885.1
83 88.: 88.: 88.: 83 88.: 88.~ 88.~ 88.~ 8882880
:88 AA: 88 A888 :88 A888 A888 A888 :88 :88
8 :8 ~88 . :88 £88 888 888 88.: 8 a: v8~ 88.88.
:88 :88 A888 :88 A888 A888 A888 A888 :88
858 ~88 888- 888 2:8 888 888 888 ~88 982 :8 8..
A888 A888 A838 A88 A88 A888 A888 :88 A38
888- 888- 888- 8:- 8:- .88.? 888. .38. 88.8- 2

3.33 1N oEah.

111

 

:88 A888 A888
88 88- 888 Eu
:88 :88 :88
888- 888. 88- 82
A8-8 A888 A888 A888 :88 :88 A888 A888 A888
88- 88- 88- 888 888 88 ~88- 888. 888- 8:8,:
A8~8 A828 :28 A388 888 :88 A888 :88 A888
88 E8 88 88 88 ~:~8 88 88 888 880
8 E A: A8 A8 A: A8 A8 A:
88888 888.82 88.88: 288>
88m. .982 .88 8:8 88.2 .88 88.". 8.85.8 ...:

 

030.28. Embtmimb 05 M: 02.35.: >520: hA0 20.80.30Q b86205. ..< A0201

508880 0.8 5:5 >n 80:5 25:88
505830 8080558: :A 80:98 m0:A8>-nA 600520: 08 552A :0 A0>0A 00:85:38 880A 05 8 5:805:38 £80& 50805: :0: 08 8555000
:55 :5 3280.80: 05 :A A5522: 08 805830 08:. 8500.30 8300805 95:5 05 :5 A0850 3-: 80> 05 .50 08:00 05 .5>0 8088 5:
A80: 0858 5:5 05 .3 woA 05 8A 8088 .3 meg 85,550 0 5:8 mg $5582.80 05 3 8:050 A0058 0:858» 8.58:8: 05 .AA A .3 03A?
05 win: 8032:; >555: 08 8038:? 8:28: A0058 80:83 25 05. .03 >nA A5035 8853585 082089055. 8.58:9: 05 .3 0.88 .A.<m
08.58 05 mA :38 Km. 88> 5 8:5 05 5.; 05:0: 8.58:9: 05 8 085A. 5:5 0:0 >AA 50>0AQE0 mA >A:0 0: .AA 0 5:8 0:5 088 05 8 85:5
550 :5 8103 53:9: 05 t A .3 28> 05 8028 :85 05858:, >883. 8 8 E059 083550 0 A08 8588505 <qu a .8 <82 :8 .555 8:
53:8: 05 50 A .3 028.» 05 853 85 038:? >85 8 mA 8:890:38 583.550 0 5:: 5588605 55 8 88A 53:88: 05 .5 0:0 .3 03A? 05
8003 N85 .<mAA>A :8 8: 58:8: 05 .AA A .3 029 05 528 £053 038:? >685 8 8 V3: .3 >2 A5535 88> 5 0w: 853:8: 05 8A 353..
.08.:5550 0 5:8 03:: 8 53:88 05 .AA A .3 028» 05 8023 85 033:? >85 8 mA memeV .0385? 5:095:05 05 8A A0508 555-: 05
805 80: 50:8: 05 0 20:8: 5 .0322? 5:055:05 05 8A A0508: 50:8: 05 805 80p 05 m A08: 5 .0585; 5:095:05 05 8A 80> 05 .8 08:00
05 ~26 8:30.: >A5:0E .3 858305 5:85:88 55> 8:80: 8:805 .4. A28: .Amv 5:8 Ob Amv €280.80: mAO 05 805 8:50.: 8:082: 058 85.

8058205285 53:8: 1:: 0A8: A055
Wm 0.53

112

888
an:

:88
.23

A888
:88

:88
2o.o-

A838
83

A 88
mwod-

888
888

A338
81o

A888
mwod

A838
33.

888
85

:88
8......

:88
«86

A88
2o.o.

:88
83

:28
83

A808
25

:88
«.8...

88:8
mwod

A828
8a.?

A388
88o

A088
33

A808
o8...
A888
88o.
A :88
838
:88
.88...

A888
88.?

A888
«2.8

:88
BE.

:88
«80.

AA: :8
88.0-

888
88..

A88
88...

A888
88.0.

:88
838

A838
83

:88
888

688
88..

888
:8...
$88
888
A 88
:38

A .88
88...

3:03 m.N 03,;

888
83

888
8......

A388
88.0

A F 38
:38

A888
89o

A808
~58

A838
«and.

A888
83

A338
888

A88
888

688
83

A808
88...
A :88
«88

A388
880.

:88
83.

A868
89o

888
:3...

A088
:8...

3mde
mwod.

A858
.88

A888
Sod

N.

Z

oo :55

05:8.

2050

m>>m

¢>>m

m>>m

N>>m

F>>m

<u_0._o<ms_

113

 

 

 

 

 

 

Aoomoo Auvooo Aooooo
omoo ooo.o- ooo.o Eu
88.8 A Shoo Aooooo
Eoo ooo.o. 2o.o. <92
A~oo.oo Aooooo Amoooo 85.8 38.9 85.3 88.3 Aoopoo Amoooo
«So. ooo.o. 89o. ooo.o ooo.o mooo ooo.o. 2o.o. 2o.o. otoo<
85.8 ASN.9 some cute Aooooo Aoovoo 85.8 83.8 Aoﬁov
ooo.o ooo.o ooo.o ooo.o. ooo.o- ooo.o. ooo.o Boo Boo 528
A9 A3 A 3 A3 As A: A9 As A :
ooomooom ooomgooo 38-32 «3%.;
Boom Ea: 58 Econ. .252 :8 Soon. macaw =3".
m3mtm> “cobcmqmb 05 MA 8mm Bin: ..m 3ch
oomoo o ooo.o omooo A ooo.o woo to Soto ooooo ooooo ooooo 8.38-1
ooo.o mom; ooo.o «5; «5.. who; ooo.o ooo.o ooo.~ 293238
88.9 88.9 88.9 88.8 88.8 38.3 88.8 88.8 889
R3 ooo.o ooo.o 5:. :3. ooo.o 3o.» 33 «one .828...
ASN.9 Aooooo 85.9 Ammooo Aoooov A819 839 Aﬁvov A203
ooo.o ooo.o ooo.o 2o.o. ooo.o- ooo.o- ooo.o ooo.o ooo.o 232 o o3
88.8 88.3 88.3 88.9 88.3 88.9 88.3 88.8 88.9
so." can." ooo.o 9.3 85. ooo.o £2” 83 own...” o.

0.33 m.N 03.?

114

Aooo.8
onto

Aooo.8
ooo.o

Anot8
Koo

A308
ooo.o.
Aono8
Boo.
Anoo.8
ooo.o
85.8
o3.o
Aono.8
noto

35.8
2o.o

€38
2o.o.

Aooo.8
nnno

A Foo.8
ooo.o

ASN.8
Eoo

A808
nooo.

ASN.8
ooo.o-

$2.8
Soo

Aooodv
wand

A .88
ooo.o

€38
Eoo

An~n8
«So.

:38
So?

888
nnno

88.8
on to

83.8
2o.o-

88.8

25° 36

A208
«Foo

888
Foo?
A398
«Bo.
Ao~o8
noo.o
A. 8.8
ooo.o.

$3.8
ooo.o-

88.8
33

88.8
onto

Ao$8
nnoo.

Aooo.8
2o.o.

Aoot8
zoo

88.8
Boo

Aooo.8
mono

Aooo.8
onto

Aont8
nnoo.

Aooo.8
ooo.o.

63.8
2o.o

3:63 m.~ 033—.

38.8
2nd

888
noto

Ao~n8
noo.o

AAoo.8
ooo.o.

$8.8
ooo.o

:88
noo.o-

38.8
onoo.

Aoo~.8
ooo.o

Gm ...ov
mwod-

Aonn.8
«no.8

88.8
33

88.8
n. to

$3.8
ooo.o-

Anoo.8
ooo.o.

Aoon.8
ooo.o

:38
ooo.o

88.8
2nd

300.3
62.6

Ao~o8
wood.

:88
ooo.o.

$3.8
ooo.o

N.

I

oo E._.<w

05cm...

99.6

m>>m

¢>>m

m>>m

N>>m

F>>m

<..._ 03(9).

115

 

A318 Aoon.8 ASN.8
:oo ooo.o nooo E0
82.8 88.8 23.8
ooo.o too 2o.o <92
Anno.8 $8.8 Aono.8 888 $8.8 $8.8 Aooo.8 Ao~o8 ASo.8
ooo.o. 2o.o. ooo.o. oto. 2o.o. ooo.o. ooo.o. 2o.o. 2o.o. 83?.
A38 688 Aonn.8 Anvo.8 82.8 32.8 88.8 83.8 83.8
A2o.o ooo.o ooo.o ooo.o Boo Boo Boo n.o.o :oo 5.88
A8 A8 A: A8 A8 A8 A8 A8 A:
ooon-ooo~ ooonnoo. voongooo ozooo>

netmm 89.5.2 5mm

02.5.". “axons. __:m

Boon. 23mm .3“.

 

mﬁmtg Embemqmb 2: m. AmboE gooomﬁv m5 Eek Sow $ng ..0 65¢

 

 

no? to an? to no? to nooo.o ooo.o «ooo.o voono nnoto ono to ooaoow¢
ooo.o ooo.o ooo.o En; ﬁn; ﬁn; ooo.o ooo.o ooo.o 226238
88.8 38.8 88.8 88.8 88.8 88.8 88.8 88.8 888
«So too ooo.o ooo.o onoo ooo.o onto vnoo ooo.o 58.85
:88 Anoo8 82.8 Annt8 83.8 :58 Aono.8 858 88.8
ooo.o. noo.o- ooo.o- ooo.o ooo.o nooo Boo ooo.o. nooo.o 982 he o3
88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8
o8.o Sto ooo.o onto noto onto too too onto n.

3.88 no as;

116

Aooo.8
onto

Anot8
nnoo

Annnov
noo.o.

Ao. t8
noo.o-

Anno.8
ono.o

Am. 8
ono.o

Aono.8
Soo

AS 8
95o

Att8
ono.o

62.8
ooo.o

Aooo.8
nnto

Anot8
ooo.o
Avon.8
noo.o.
Anoo.8
noo.o.

Anno.8
ono.o

Anno.8
ooo.o

Aooo.8
onto

An~t8
ono.o

Aonn.8
ooo.o.

Anot8
noo.o-

£38
ono.o

:oo.8
noo.o

ARN8
ono.o

Anot8
ooo.o

A F no.8
ooo.o.

$2.8
«Foo

$8.8
ono.o
Aooo.8
ooo.o.

83.8
ooo.o-

Amon8
«no.8

Aoo~.8
ono.o

300.8
umcd

ASN.8
nnoo

A208
Boo

A$o8
noo.o.

83.8
ooo.o

Awwodv
shod

Aooo.8
ooto

Aoo~.8
ono.o

A808
Boo

A$o.8
ooo.o.

Anot8
«Foo

3.33 m.N 03.;

88.8
nnto

At t8
ono.o

Aono8
ooo.o

Aooo.8
noo.o.
Anoo.8
ono.o
A398
ooo.o
688
So?

Anot8
Eoo

Anoo.8
ooo.o-

A:n.8
ooo.o

Aooo.8
ooto

Anoo.8
oonoo

83.8
.88

68.8
ooo.o.

Aooo.8
ono.o

65.8
ooo.o

Aooo.8
tto

Anot8
Enoo

Aono.8
noo.o-
Aooo.8
noo.o.

Aooo.8
«nod

N.

2

oo Sham

05:0...

92.0

m>>m

v>>m

m>>m

N>>m

_.>>m

(u. 0565.2

117

 

 

ooo.o vooo ooo.o ooooo ono.o omnoo «ono.o ooo.o Eoo oemoomi
ooo.o oon; ooo.o ﬁn; «no; on; ooo.o ooo.o ooo.o 226238
88.8 88.8 68.8 88.8 88.8 68.8 88.8 88.8 88.8

2o.o Boo «coo 2o.o ooo.o :oo ono.o ooo.o ono.o 38.8...
:58 35.8 A3o8 Anoo.8 Aooo8 A388 A358 838 £38

oto ooo.o ooo.o ooo.o .ooo ooo.o ooo.o ooo.o ooo.o 282 6 oo..
88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8

ono.o ono.o ono.o two ooo.o End onno on~.o ooo.o n.

3.58 on 2%...

118

 

Aooo.8 Amno8 A888
oooé nooo noo.o- Eo
A-n8 88.8 62.8
ooo.o ono.o ono.o <92
Aooo.8 ASN.8 82.8 88.8 Anoo.8 Aooo.8 88.8 88.8 88.8
ono.o. 2o.o- Koo. ono.o. ono.o. ono.o. ono.o. noo.o. ono.o. o:oo<
Ao~n8 on~8 88.8 83.8 68.8 88.8 82.8 83.8 38.8
nnoo ono.o ono.o ooo.o ooo.o ooo.o ono.o o3.o too .850
A8 A8 A: A8 A8 A: A8 A8 A?
nooo-ooon 88.32 38.52 o_oo_a>
Boon. “9.3.2 58 Econ. .352 :8 Soon. mason __ou_

 

mint? Acobcmqmb 95 MA 385 88$-» 05 Set gem msAm ..< 3ch

 

.AovoEAAoo 8o AVE: 3 Echo AoAoncﬁm AoobAosAU .ooooﬁcvoAA
E 33% mosAo>AA .AooooAAvoo' 8o Stun no $3 2.: Au EmomeAm 338m 6380.30 mBoonAooa MESA 05 8A A838 2.: .80» 2: .3 8.38 2:
.85 $38 A0: A88 owobg BSA 2: Ac woA 05 a 38mm .3 mod .ooAEoﬁo o Ea weigh wEAEonAmotoo 2AA 3 $815 AconAon 896a 9.53:9:
05 .AA A A0 028, 2.: mg 853:3 banana Rm 8335, owcﬂaﬁ 30:3. 3253 u>A.A 2F .2: 3 963.6 8:8:ch EnsuﬁwBAoEA oLuwoSwE
2: .3 Boom H<m owﬁgo 05 nA ooAAH<m ﬁne» 5 BSA 05 5A3 93:8 98on 05 2 95:3. .25.: 28 A3 @0onan oA baa 2A .AA o 28 0:5
2:8 05 Am voSd 8:8 SA 8:03 Samson: 05 .AA A A0 029» on: mBAS 35 Boot? >883 n oA $050 .35550 o 98 :23:onvo <..AU n no <82
5 855 no: Smog 05 A0 A A0 02? 2: ooVAS A9: 03ng >553. o oA <,.AU oo<mA>A 623.550 o Aoco =23:onvo <qu n no: 89288 05 .AA 28
A0 028, 05 $on <nAU .<mA>A an ooAA Homage 2: .AA A .8 029» 05 oBAS :03? 033.5, A2556 n oA <32 .3 3 AovoEAo who?» 5 own £0?an
06 ﬂ oA\uw< .ooAEoﬁo o A38 vo8 mA bwoaoﬁ 05 .AA A Ac 2:9» 2: oooAS 35 Boots, @556 o E Mme/EU .oAAAntg EvocoAAvo 2: ﬂ AvooE
ooooﬂé 2AA EBA 82A Q35 05 0 228m :A dint? 33:08vo 2.: 2 AoAuoE noooﬂé 2AA EBA 82A dz: 2: m Agog :A 653.5, 8023vo
05 .04 A258 “888% 05 Bob 82A mEm 2AA can? BAA—o2 3585 < Aocnm .Amv can 98 A8 558%: mAO 2AA EBA 3:58 3:805 038 25.

85:89:35 hows—:2: can 25o 3258.»:—
e.~ nigh.

119

Aooo.8
ooto

85.8
ooo.o-

8.5.8
ooo.o

8.8.8
Aooo-
:88
o3?

ASN.8
ooo.o

38.8
ono.o

ASN.8
Eoo

ASN.8
noo.o-

63.8
too

88.8
33

88.8
ooo.o

38.8
nooo.o

35.8
Boo.

888
So?

888
2o.o

88.8
«to

88.8
ooo.o

35.8
nooo.o

88.8
38.8

38.8
ooo.o.

Aooo.8
ooto

38.8
Aooo.
38.8
$3.
$38
Aooo
82.8
ooo.o

$8.8
ouoo

:58

onto.

338
Food

88.8
too.

888
ono.o

38.8
«to

$3.8
So?

588

Boo.

38.8
Aooo

82.8
ooo.o

88.8
2o.o

Aooo.8
Eco

8.8.8
Aoooo-

33.8
Boo.

88.8
Aooo

$9.8
ooo.o

3:88 3 «so...

88.8
«to

8.6.8
~oo.o

83.8
goo-

$8.8
Aooo

$8.8
ooo.o

38.8
Boo

$3.8
Boo.

88.8
ooo.o

$5.8
ooo.o.

88.8
ooo.o

88.8
:3

88.8
ooo.o

38.8
moo?

88.8
«ooo.o

AA 58
ooo.o

$3.8
2o.o

88.8
«to

83.8
ooo.o

63.8
noo.o.

33.8
Aooo

8E8
Boo

N.

2

oo Eh<w

935...

9.2.0

m>>m

35m

m>>m

N>>m

w>>m

<u03<m2

120

 

 

 

 

 

 

A888 888 88.8
888 A Ao.o ooo.o. $6
88.8 83.8 388
Amoo ono.o 88o <82
:88 88.8 88.8 88.8 88.8 83.8 ANA-N8 838 8:8
ono.o Aoo.o ooo.o ono.o. ono.o. ono.o. 3o.o ooo.o ooo.o o :02
88.8 88.8 838 A388 A388 888 888 A888 888
«So «So 3o.o ono.o o3.o ono.o voo Auoo $3 .880
A8 A8 A A8 A8 A8 A: A8 A8 A A8
38-88 88-32 38-58 288>
38a A382 88 8:8 .982 :8 Soon. macaw =8
039:9. Aconcmmmb m5 MA EDGE 89054» 05 EDAAA 30m 421 ..m 3ch
82o A83 883 883 Am: 3 8A 3 82o 8A 3 8A 3 593$
88; o8.A o8.A NBA 83 NBA 888 888 omod 2282080
:88 88.8 8 A A8 88.8 82.8 88.8 83.8 88.8 88.8
Avo oto 83 A88 mvo 83 83 2.3 ooto 88.8...
88.8 88.8 88.8 88.8 :88 88.8 88.8 88.8 88.8
88o. 88o. 88o. 88o. 88o. 88o. 88o. ono.o. ono.o. 232 Ao 9:
88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8
38o A98 8A..o 8A..o 8A..o Auto 83 8A..o ono.o m.

8.38 o8 «Ea-n

121

88.8
2nd.

888
83.
88.8
88.8
888
A88
888
Sod.

888
A88.

888
NA A8

88.8
88.8-

88.8
88.8

888
88

88.8
and.

88.8
83.

89.8
88.8

888
A88

88.8
38.8

88.8
88.8

88.8
23.

888.8
EA...-

8A 8
88.8

888
A88

888
88.8-

88.8
88.8.

88.8
A28-

888
88.8

88.8
88.8

88.8
A 8.8

888
3A...

88.8
288

:88
A88

88.8
«8...

88.8
83

88.8
«.88.

88.8
33.
88.8
A 8.8
88.8
88.8

85.8
88.8

88.8
88.8

88.8
88.8.

88.8
83.

88.8
~88

88.8
88.8

88.8
88.8

888 8." 28h

88.8
83.

88.8
83.

88.8
88.8

888
88.8

88.8
88.8

88.8
88.8

:88
88.8

88.8
88.8-

:88
«8.8

:88
83

88.8
88.8-

88.8
83.

88.8
88.8

88.8
88.8

888
A88

:88
88.8

88.8
v.88.

88.8
3.3.

88.8
88.8

:88
88.8

88.8
88.8

N.

2

oo _.\._.<m

92cm...

286

m>>m

v>>m

m>>m

N>>m

_.>>m

<u. 05(9).

122

 

 

 

 

 

83.8 888 88.8
88.8 88.8 888- E0
888 88.8 88.8
88.8 28.? A88- <82
:88 88.8 88.8 88.8 :88 88.8 :88 :88 88.8
88.8- 888 88.8. 88.8 88 A888- 88. A88- 88.? o :8<
:88 :88 :88 88.8 88.8 88.8 88.8 88.8 88.8
88.8- 888. 88.8. 88.8- 888. «8.8. 88.8. 88.8. 88.8. .880
88 8% E 88 E 3 A8 88 E
88-88 888.82 88.82 288>
8:8 .382 88 88a 8882 :8 8:9". 238 =3”.
0385? 8:88:0qu 05 ms ‘mboE 3.68-» ms Eek 96m Q2: ..0 3:91
3.98 8A A8 883 AAmAd 8A A8 883 888 N83 8 A A8 8.88-”.
88; 88A 88; 88A NBA NBA 88.8 888 88.8 «88238
88.8 89.8 82.8 88.8 88.8 88.8 88.8 8 8.8 88.8
88.8- 38.8- \- A88- 83... 88.8 88.8 E 3 88.8 88.8 8858.
88.8 88.8 38.8 88.8 88.8 88.8 :88 88.8 888
88.8. 888.8 8.8 88.8. 88.8- 88.8- 888- 888- 888- 282 Ao 8..
88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8
8m...- 83. A38. «8.8. 8......- 83- «8.8. 83.8. «8.8- m.

8253 9N «Ea-H

123

88.8
88.8

88.8
88.8
88.8
V8.8
888
88.8-

888
A88

888
88.8

88.8
88.8

88.8
88.8

888
A88

888
88.8.

88.8
«8.8

88.8
:88

88.8
88.8.
88.8
88.8.

88.8
«88

AA 88
88.8

A888
«8.8

88.8
:88

83.8
88.8

AAA-8.8
88.8-

88.8
”8.8

88.8
A28

888
~88

88.8
88.8

88.8

88.8-

88.8
«88

88.8
88.8

88.8
88.8
88.8
88.8-

88.8
88.8-

88.8
88.8

88.8
53

88.8
~88

83.8
85.8

88.8
888.?

888
A88

888
88.8

88.8
8A...

888
88.8
83.8
88.8-

88.8
88.8

88.8
88.8

3.88 88 28,..-

88.8
83

88.8
88.8

88.8
A“-8.8
888
A88.
82.8
:88

82.8
88.8

88.8
88.8

88.8
«8.8

88.8
88.8.

88.8
88.8-

88.8
82.

88.8
28.8

82.8
«88.

8,-8.8
88.8-

83.8
88.8

88.8
Bod

88.8
83

88.8
A A88

828
~88.

88.8
88.8.

828
A88

N.

2

oo E-_-<m

05:2-

9050

m>>m

v>>m

m>>m

N>>m

_.>>m

(.... Ugo-«ms.

124

 

 

88.8 88.8 88.8 88.8 NASA. 88.8 88.8 88.8 E88 8.88-“.
88; 88; 88; 88; 88; 88; 88.8 88.8 88.8 2882080
88.8 88.8 88.8 88.8 83.8 88.8 88.8 88.8 88.8
88.8 38.8 «8.8 88.8 88.8 83 88.8 88.8 A88 .828...
89.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 82.8
88.8 88.8. 88.8- 88. 8.8. 88.8- 88.8 88.8. 88.8 982 .6 83
88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8 88.8
8:. «8.8 88.8 «8.8 88.8 88.8 «8.8 88.8 88.8 m.

388 88 28,—-

125

 

8.A 8.0 8.0
82.8 88.8 88.8
88.8 «A88. 8.8.0- <05.
8.8 8.0 8.0 8.0 8.0 8.0 8.0 8.8 8.0
88.8 88.8 88.8 88.8 88.8 88.8 88.8 83.8 83.8
88.0. 3.8.0. 88.8- 88.0- 88.0- 38.0- 888. 88.8 88.0. 8:02
8A 83 8A 8A 8A 8A 8A 8A 5A
838 83.8 88.8 88.8 88.8 83.8 88.8 A808 888
88.8 88.0 88.0 88.8 88.0 88.0 88.0 $8.0 88.0 .0080
A8 A8 E A8 A8 A8 A8 88 3
8888 088-83 88-83 0_8_a>
8:9... 09.82 .80 8:00 .085. :8 8:3 0.850 =3”.

 

03009. 2500030 05 & ESE mac-Am 0AA0¥0Q ..< 6ch

 

008850 0.8 05¢ 3 30:00 0:80:30
0000820 .0335, 30059005 £000 :8 worm-A m000 mg 05 08 80287: 05 3200 300802 8080658: :A .8088 m00Am>AA .0000-A0A00 0:0 00300
:0 $2 0.0 30 0:805:w8 3080M 088:0: 30: 08 3:22.880 :05 A00 "0:28.080: 05 E 000205 08 88.8-30 0EA-A- 038030 3300803 30:3
05 :0-A A03:00 «AAA .80» 05 .A0 0338 05 :26 3080 A0: 038 035% 0:3 05 .3 woA 05 3 3088 .3 wo-A 0330050 0 0:8 mEuASﬁ w:A0:0%0t00
05 8 $530 A0008 80:0an 3:088:88 05 AA A A0 020.» 05 393 338-5, 888:0 0.8 83888, 3:25: A0008 80:83 26 05- .03 3
000.50 8.5355 080083003 3:080:88 05 A0 0:008 -A-<m 0wﬁ0>a 05 3 ooA\-A-<m .300» 5 0:3 05 5.; 05:3 3:03:88 05 3 05:0-A- .05
0:0 3 0080380 3 3:0 0..— .AA 0 0:8 0:5 088 05 8 m0::-A 00.00 :0-« 3803 Swag 05 .AA A .A0 00A? 05 8003 :20 0508? 888:0 a 3 3050
03300.00 o 0:8 5098800 <nAU a :0 <mA>A :8 000:0 m0; Swag: 05 .A0 A .A0 02? 05 $003 305 033.5» 8:850 a 3 «£0 00452 .00-05060
0 0:8 5.583.800 $5 a 33 009052: 05 .AA 0:0 .3 0:03 05 "”0003 «ED 492 :8 32A 83:08 05 “A A A0 00A? 05 0.0003 0023 053.5, 8:850
a 3 <mA>A .3 3 00030 309» 5 0mm 303:9: 05 3 A: \0w< .0mE050 o 0:0 002: 3 83:08 05 .AA A -A0 028A 05 883 85 03088, 38:00 a
3 anAZ-HAO .0508 005800.06: 05 3 «AAnAAm 0803-0 05 0 3:3 :A .328 A0008 A0888 :88-AA 05 3 0.508 00:88.5wa 05 m A05: 5 .003
2.008 00:88:88.: 05 3 E30: 0:038: 88w :05? 3080: 3:805 < A0:mnA .va 0:8 AD .68 328080: 380A 05 :83 3:80: 3:805 058 3::-

358353— 3801— 0052:0800.—
bd 020.“.

126

00.0
800.00
00000.

00.0
800.9
0000.0.

2.0
800.8
03.0.?

03
002.00
500.0

5.0
600.00
0000.0.

00. 0
30.00
000.0

00.0 00.0
£00.00 A3000
800.0. 0000.0.

00.0
600.00
030.0-
00.0
$00.00
00000.

00.0
38.3
0000.0.

00.0
800.00
0000.0.

00.0
A3000
002.0.

00.0
c.0000
0000.0-

00.0
800.00
0000.0.

00.0
$00.00
00000.

00.0 3.0
300.00 300.00
00000. 0:0,?

000
£3.00
0000.?
00.0
800.00
0800.

3:0“: 5." 030%

00.0
600.3
020.0.

00.0
8.000
«000.0.

000
$8.00
00000.

02
800.00
800.0

00.0
800.00
0000.0.

00.0
A0800
0000.0

00.0 :00
A3000 800.00
020.? 020.0.

000
£00.00
300.?
00.0
c 8.00
0000.0.

2050

m>>m

0>>m

m>>m

N>>m

_.>>m

<u000<m2

(no

127

 

 

 

 

 

8... 8... 8... 8... 8... 8... 8... 8... 8...
889 5...... ......9 ...89 $89 £89 8.9 8.9 68.9
8.....- 88...- 8.8.... 5...... 88.9 .88...- 8...... 8...... 8.3.- 2.8...
a... R. E... 8.. 8.. 8.. .... 8.. 8..
.88.... .88.... 8....9 £89 ......9 :89 €89 8...... .N89
88... 88... 88... 3...... 8...... 8...... 3.8... 88... 88... .880
.... .9 E .9 .9 .... .9 .9 ...
.8988 .838. .838. 28...;
88a. 88: 88m 88.. .285. :8 88.. 2.88 :8
$3.29. .53....qu m... a. 33m .98... 89m o...o..tom ..m .ozmo.
$8... 8...... 2...... 8...... 8...... 3...... 8...... 88... 88... 8.88mi 888
«8.. «8.. «8.. 8.... 8a.. 88.. .8... ......N .88 22.9.88
8... 8... 8... 8.. 8.. 8.. 8... 8... 8...
$89 ......9 $89 ..89 .82.. 83.9 .859 889 839
. .8... 8...... 83.... 88... «.8... 88... R8... 88.... 88...- 8.82 ... 8.
m... 8.. 8.. 8.. 8.. 8.. .N. 3.. 3...
.....9 .....9 8...... ......9 889 889 ......9 .899 889
.8..... 82.... :8... 83... 88... 3.8... 2.8... 88... 88... 8.58
8.. 8.. 8.. 8... 8... 8... 8... 8... 8...
2.8.9 ......9 $89 .... .9 .....9 .....9 $89 ..89 .88....
88... 88... 88... 88...- 88...- $8.... 2.8...- 88...- 88...- 98....

.....8. ..N 28h

128

......
.119
8.8.?

a...
..89
82...?

8..
..89
8......

8...
389
88.?

8..
.82..
88...

......
.889
88...-

......
... .9
88.?

.... .
.889
88...

2....
.....9
28.?

3....
889
83.?

8...
.82..
8...?

......
......9
«8...?

......
889
38.?

......
889
«to?

8...
.889
38.?

8...
:39
3.8.?

.....8. 2 as;

2...
.359
«8...?

3...
$89
88.?

8...
.88....
...8?

2....
.... .9
38.?

8. .
.82..
«.8...

8...
$89
88.?

8...
£89
8...?

8...
......9
8...?

mam

v>>m

m>>m

N>>m

_.>>m

(m 0.935.

(”—0

(m2

129

 

 

 

 

 

.....N 8.. 8.. N... .... N... 8.. 8.. 8..
...89 ..89 ..89 .889 .88.... .88.... .88.... .. ...9 ......9
8. .... .8... 8...... ....8... 8 ...... ...8... 88... .88... 88... .880
.... .... ... .... a. .... .... .N. ...
88-88 88-8... 88-8... 288>
8:8 .982 .88 8:8 .38.). :8 88.. 8.2.8 =8
2.8.8.. 2686.5. m... 8... 33m .28.-.. o..ofo& ..0 ....ch
...8... 3...... 8...... 88... 88... 88... 88... 88... 88... 8.8.8-”. 888
8m. 8w. 8.4... 88.. 88.. 88.. .8.~ .8... .8.~ 82828....
8... 8... 8... 8.. 8. . 8. . 8... 8... 8...
.88.... ...89 .859 ...89 :8... .88.... .88.. .88.... .88....
8...? 8...? 8...? 88... 88... 88... 88...- 88.? 88.? 982 ... 8..
o... 8.. 8.. 8.. 8.. 8.. 8.. o... ....
.88.... .88.... .88.... .889 .88.... .889 .889 .88.... .889
8...... 88... 83... 8 ...... 8 ...... 8...... 8 ...... 88... 2.8... 8.58
.... 8.. 8.. 8... 8... 8... ...... ...... ......
.88.. ....m9 .8 .9 ..89 ...m9 ...8.... .88.... .88.... ......9
88... 88... :8... 88.? 88.? 88.? ...8... 88... 88... e8...
8... 8... 8... . ... .... .... 8.. 8.. 8..
.88.... .88. .88.... .~.~9 .83.... .88.... ...89 ..89 .889
88.? 88.? 8...? 88... 88... 88... 88... 88... 88... 98.0

.....8. ... 2......

130

8...
£89
2.3.?

8...
.....9
... ...?

......
6......
8...?

8. .
.889
88...

8...
.~.~9
8...?

......
.88....
88.?

8...
.8..9
88.?

8...
.88....
.88.?

8...
.889
.88.?

8...
..89
88.?

8...
8......
88.?

......
.82..
«8...?

8...
.88....
2...?

8...
.889
88.?

8..
.88....
88...

mod

...89
88.?

8..
.88....
88...

8...
.889
88.?

......
.88....
88.?

8. .
.889
88...

.....8. .... 2......

8...
...89
... . ..?

......
.88....
88.?

......
.8......
..8?

8...
.89
8...?

8...
.88....
88...?

......
.889
.88.?

8...
.88....
88.?

8...
.889
88.?

.8...
.88....
88.?

8...
...89
88.?

v>>m

m>>m

N>>m

F>>m

4.... 0.94m.)—

<n_0

<82

8...?

131

 

 

.63... 66...... 2...... 3...... ....o... as... 8...... 88... N. ...... 69.55-”. 2.83.
«8.. New. «3.. .....N. ....N. 6...... .66 .66 .66 2252680
6.... ...... 6.... N... 8... mm... ...... ...... ......

EN... .62.. 6...... :8... 6...... 68.... 6.8.... 68.... £8...

:8..- $8...- $8...- 6...... 6...... «6...... 3...... 38.... 38.... 33.2 .0 ......
N... .... .... .... .... .... .... .... ....

6...... 68.... 68.... 6...... 6...... .N. ..o. 6...... 6...... ...o...

3...... 3...... 38.. :8... ....No... 88... 8...... =8... 6.8... 8.56
...... mm... ...... ...... 3... ...... ...... 8... ......

6...... 66...... 68.... 66...... 66...... 68.... 6...... 6...... 62....

88...- 58.? 8...... 68...... 63...... 63...... 68.....- 68.....- 5.....- 2:5.
6.... ...... 8... 6.... ...... ...... 2.... ...... 6....

662.. 68.... .. .6... 62.... 68.... 66.... 66...... .66.... 666....

38...- 88.... 6.8.? $8... 8...... ..m ...... . . .....- 2.8...- 68.....- 92.0
3.... 8... 6...

68.... 6mm... 68....

82.... «6...... 36...... 626

.....8. 2 2......

132

ESSAY 3. Portfolio Management Team Changes

3.1 Introduction

A large body of literature has been devoted in assessing mutual fund performance.
This research has evaluated the impact of numerous factors on fund performance,
including the fund’s size, structure, expenses and risk proﬁle as well as individual
characteristics and compensation of the managers. The focus on individual managers
overlooks a recent trend in the mutual funds industry. Although historically it was
common for a fund to have only one person as the portfolio manager, things have
changed dramatically in recent years. For instance, according to Morningstar’s database
of domestic equity funds, in 1997 only 32.5% of all funds and 20% of total mutual ﬁmd
assets were managed by a team of managers rather than a single individual, whereas in
2005 the corresponding percentages were 58.5% and 60%, respectively. Motivated by the
increasingly important role of portfolio management teams recent studies have started to
examine differences in the performance and characteristics of sole-managed and team-
managed funds and provide an explanation for the preference of team management.

There are two main explanations that have been proposed in the literature so far.
One is that fund families are trying to avoid falling victims to “star” managers that leave.
A celebrated example of recent years is Elizabeth Bramwell's 1994 separation from
Man'o Gabelli's GAMCO. Elizabeth Bramwell managed the Gabelli Growth Fund to one
of the leading performers in its class. Near the fund's then peak, she leﬁ and started her

own funds. She was also successful in convincing the SEC to allow her to use her track

1 3'3

record with Gabelli on a limited basis in marketing her new funds. The second one is that
groups perform better than individuals when it comes to managing a stock portfolio.

However, evidence so far shows that management teams do not perform better
than single mutual fund managers. For instance, Prather and Middleton (2002) ﬁnd,
consistent with the classical decision making perspective, no difference in the
performance of sole-manager and team-managed mutual funds. Chen at al. (2004) and
Baer et a1. (2005) do ﬁnd evidence of underperformance by teams of managers of 5.5 and
4 basis points per month respectively.

In this paper, we examine the determinants and consequences of team
management by focusing on funds that have chosen to change the structure of the
portfolio management team. We identify 867 funds that have changed the structure of
their portfolio management team in the 1997 to 2004 period. Of those, 530 have switched
from sole-manager to a team-managed, and 337 from team-managed to sole-managed.
We collect data on the performance, risk, size and other characteristics of the fund
portfolios for the year before and the year after the change and we address two broad
questions: a) What are the characteristics of the funds that change structure and b) what
are the consequences of the change in the performance, risk and size of the fund.

Focusing on management team structure changes helps us overcome the
endogeneity problem that is present in past research that has studied the performance of
team-managed versus sole-managed funds. In all those studies it is not clear whether the
underperformance of team-managed funds is because of the team structure and because

of other reasons that also caused the fund to be team-managed. In this paper, the fund

134

portfolio is exactly the same and the only thing that varies is the management team
structure.

We ﬁnd that team-managed funds switch to sole-managed after poor performance
while sole-managed funds switch to team-managed aﬁer signiﬁcant over-performance.
When a fund becomes to sole-managed performance improves signiﬁcantly (184 basis
points in terms of 4-factor alphas) while a switch to team management leads to
deteriorating performance. Funds that switch from sole-managed to team-managed
exhibit a decline in performance of 190 basis points. Taking all this into account, results
suggest that team management is not good for performance.

So, why do ﬁmd families choose the management team structure? We ﬁnd that
sole-managed funds that switch to teams experience above normal increases in size
consistent with the view that ﬁmd families, investors, managers, or all of them believe
that a single individual manager cannot efﬁciently manage a lot of money. We also ﬁnd
slight evidence that risk considerations play a role too. Team-managed funds seem to be
taking on less risk than their sole-managed counterparts.

Finally we also examine a fund’s decision to change the investment advisor
altogether. Findings suggest that poor performance is the major reason that leads ﬁmds to
ﬁre their advisors. In general, our ﬁndings, conﬁrm past evidence that team-managed
funds do not offer superior risk-adjusted returns but also show that team management

turns hi gh-performing funds into poor performers.

135

3.2 Data

3. 2.1 Mutual Fund Data

All of our mutual fund data come from the nine January CDs of Morningstar,
Inc.’s Principia Mutual Funds Advanced database from January 1997 to January 2005.1
The January CDs report data as of December 31St of the previous year. Morningstar
started the Principia databases on January 1996. The Principia Advanced version contains
more information, especially regarding managers and monthly fund returns, than the
basic version of the database. Using the nine CDs, we extract information for all funds in
operation every year from 1997 to 2005.

We start with all the ﬁmds in existence in January 1997 and we follow them
through 2005 or until they disappear from the database. We also include in our sample all
the ﬁmds that started their operations after 1997 to minimize concerns about
survivorship. Data are gathered for all domestic equity funds with a self-declared
investment objective of growth, aggressive growth, growth-income, or equity-income.
We exclude index ﬁmds, balanced funds, funds of funds, as well as other types of ﬁmds
that are restricted in some sense in their investment decisions.2

For each fund we obtain annual and monthly returns, annual expense ratios and
loads, net asset values, total net assets, fund inception dates, mutual fund family names,
portfolio characteristics such as turnover, total number of holdings, percentage of assets

invested in the top 10 holdings, stock, cash, and bond holdings, as well as manager

names. In the “manager name” ﬁeld Morningstar lists the name of the manager if the

 

' Morningstar, Inc. used different names for this database throughout our sample period. The three different

names are: a) Principia Mutual Funds Plus, b) Principia Mutual Funds Pro Plus, and c) Principia Mutual
Funds Advanced.

2 These include socially conscious funds, life cycle funds, target retirement funds and tax managed funds.

136

fund is solo managed, the names of the multiple managers if the ﬁmd’s total assets are
divided among more than one manager, or the term “Management Team” when more
than two people are involved in the management of the fund and they manage together.3

From the advanced analytics View of the database we hand-collect each fund’s
management fees, which are the fees that the management company charges to manage
the fund’s portfolio. For most funds the management fee that appears on Morningstar is
taken from the fund prospectus. For other funds a minimum and maximum management
fee range appears in the database; for such funds we calculate the midpoint and use the
resulting ﬁgure as the fund’s management fee.

The funds that appear in the Morningstar CDs represent fund offerings that the
investor can choose from but do not represent distinct investment portfolios.4 However,
while various share classes offer investors different fund choices, they are based the same
underlying portfolio and consequently the same before-fee performance. Our unit of
observation is the fund. We therefore aggregate multiple share classes into one fund
observation. We are careful to cumulate the total assets from all share classes to obtain
the total assets of the underlying portfolio. In order to identify different share classes of
the same fLmd we match different share classes by four portfolio characteristics: turnover,
number of holdings, percentage invested in stock, and percentage in the top 10 holdings.

We also verify our matching by looking at the fund names.5

 

3 The exact description of what the term “Management Team” means, reads are follows: “This is used when
there are more than two persons involved in ﬁmd management, and they manage together, or when the fund
strongly promotes its team-managed aspect”.

4 As Nanda, Wang, and Zheng (2005) document, in the 19903 many mutual funds introduced additional
share classes as a way to offer investors more choices about the timing of load payments, or to provide
lower expenses to investors with big holdings. They show that by the end of 2002 more than 50% of mutual
funds offered more than one share class.

5 Multiple share classes of the same fund have basically the same name. Their names differ only by the
name of the share class. Example: “Vanguard Growth A,” “Vanguard Growth B,” etc.

137

In all, we have a total of 7,846 fund-years. Out of those, 867 represent ﬁmds-years
with a change in the management team structure during that year. We require that there is
available information for all fund-year observations for one year before and one year
aﬁer. This reduces our sample to 3,850 funds-years and 503 funds-years with
management changes. For all those funds we also collect information on whether there
was a change in the management company that is in charge of the fund portfolio. We
could have changes in the management company without having changes in the
management team structure.

Table 3.1 summarizes our dataset. Panel A shows all the fund—years with
management structure changes by prospectus objective. We can see that most of the
changes (530) are from single managers to management teams. However, a sizeable
number of funds choose to replace the team of managers with a single manager (337). For
the majority of changes, funds decide to keep the same management company (same
managers). This means that, in the case of switching from a team to a single manager,
some of the managers where ﬁred but one of the old managers stays to manage the
portfolio. In the case where the fund moves from a single manager to a team, the original
manager is still present but other managers are added to help him manage the portfolio. In
the case where the whole management company is replaced (New Managers) then the
entire management team (single manager or team) is replaced by another single manager
or team.

Table 3.1, Panel B lists all the changes by year and Panel C shows the fund-years

that survive the three year window data requirement.

138

3.2.2 Performance and Risk Measures

We use all the standard performance metrics to measure fund performance in a
given year: raw annual fund returns, style-adjusted returns, one-factor alphas from the
market model, and alphas from the Carhart (1997) four-factor model. However, we report
results only for the market model alpha and the 4-factor alpha, which is the most
comprehensive performance measure in the literature To estimate the one-factor and

four-factor alphas, respectively, we estimate the following models:

Ru ‘th =06: +ﬂimEMRt +811, (1)

Rit _ th = ai + ﬂim EMRI + ﬂismbSMBt + :BihmlHMLt + [BiumdUMDt + git ’ (2)

where R,-, - R f, is the month-t excess gross return for fund i , EMR, is the excess market

return, SMB, is the difference in returns across small and big stock portfolios, HML, is

the difference in returns between high and low book-to-market portfolios, and UMD, is

the return on a momentum portfolio as computed by F ama and French.

We use the value—weighted NYSE/AMEX/Nasdaq composite index as our market
return, and the one-month T-bill rate from Ibbotson Associates as our risk-free rate in
calculating excess market returns. Returns on the HML (high minus low book-to-market
returns) and SMB (small minus big stock return) zero-investment portfolios, as well as

returns on a momentum portfolio (UMD), come from Kenneth French’s website.6

 

6 See Kenneth French’s website for the deﬁnition and calculation of the factor portfolio retums.

139

When estimating the one-factor and four-factor alphas we use monthly gross fund
returns for the 12 months in the year. To calculate the gross monthly return, we divide the
annual expense ration by 12 and add this to the monthly net rettu'ns. We use gross returns
because we want to measure the performance differences between various forms of
management team organization and characteristics. If better managers or organizational
forms receive rents through higher expenses, then the performance superiority of
manager characteristics might not show up when using net returns. However, we repeat
our analysis using fund returns net of management fees and the results are qualitatively
the same.

To evaluate the riskiness of the portfolio we use the market betas (the coefﬁcients
[3,...) from equations (1) and (2), as well as the standard deviation of monthly returns,
throughout the course of the year. We also examine differences in the estimated factor

loadings (the coefﬁcients ﬂismb, ﬂihml, and ﬂiumd, from equation (2)) between team and sole-

managed funds.

3.3 Hypotheses and Results
3.3.] Hypotheses

Several theoretical papers posit why team management might be a superior
strategy. Many other research studies, especially in the management and psychology
literature, investigate the decision-making process, behavior, and performance of teams
versus individuals. The results differ across studies mainly due to the variety of tasks and
measures used in each study, which makes it difﬁcult to make valid generalizations or

comparisons. However, as conventional wisdom suggests, all studies agree that teams

140

behave differently than individuals, even though we do not always observe differences in
performance.

Past literature has suggested that the major advantages of teams of managers are
ability to process more information and research more investment opportunities,
diversiﬁcation and specialization beneﬁts, and, consequently, better risk-adjusted
performance. If those hypotheses are correct we should expect to see signiﬁcant
underperformance of funds that are sole-managed before the structure change and an
improvement in performance after the switch to a team of managers has taken place. We
should also expect to see underperformance prior the ﬁinds’ decision to switch
management companies.

Another proposed determinant for the switch to team management has been the
fund’s size. Single managers might not be able to handle the funds management if it
becomes very large. At the same time, investors might not feel conﬁdent that a single
manager can handle the load of a huge ﬁmd. Whatever the reason, we should expect to
see size to play an important role on whether the fund is sole-managed or team-managed.
Finally, another characteristic that could trigger a switch in management could be the

extreme risk—taking from the side of the manager(s).

3. 3.2 Results
3. 3. 2. I Univariate tests

Table 3.2, panels A, B and C present performance results. Panel A shows the
mean performance of funds in all categories. Funds that do not change their management

team structure (columns 1 and 2) exhibit positive risk-adjusted performance the previous

141

year. However, performance is signiﬁcantly lower for funds that also chose to change
their investment advisor. Moving to funds that have switched from a team-management
structure to a single manager, we ﬁnd that there is signiﬁcant underperformance the year
before the change. Underperforrnance is 72 basis points for funds that choose to stay with
the same investment advisor and reaches 123 basis points for funds that decide to ﬁre
their investment advisor too.

Surprisingly, sole-managed funds that choose to keep the same investment
advisor and add more managers have been doing extremely well the year before the
change. The mean performance for those funds is close to 200 basis points a year in terms
of 4-factor risk-adjusted returns. This means, in opposition to our hypothesis, that
underperformance is not the reason sole-managed ﬁmds switch to team management. We
do ﬁnd that sole-managed ﬁnds that also replace the investment advisor exhibit negative
4-factor alphas.

To summarize the ﬁndings in table 3.2, panel A, it seems to be the case that
underperformance leads to the replacement of the investment advisor, but when it comes
to the change in management teams structures (keeping the investment advisory ﬁrm the
same) underperformance is not a determinant of the switch to team management. On the
contrary, team-managed funds that underperformed are replaced by a single manager.
The difference in prior year performance between sole-managed-tumed-team funds and
team-managed-tumed-sole-managed funds is 393 basis points signiﬁcant at the 1% level.

In panel C we compare following year (after the change) performance with prior
year performance and present differences in performance (following year — previous

year) for all types of funds. Team-managed funds that keep the same investment advisor

142

 

and get rid of all managers except one show an improvement in performance which
reaches 326 basis points in terms of market adjusted returns (signiﬁcant at the 5% level)
and 184 basis points in terms of 4 factor alphas (signiﬁcant at the 10% level). On the
other hand, sole-managed funds that added more managers have much worse
performance compared to the year before the change. Their market model alpha goes
down by 315 basis points (signiﬁcant at the 5% level) and the 4-factor alpha goes down
by 190 basis points (signiﬁcant at the 10% level).

Table 3.3 presents the risk characteristics of ﬁmds the year before and following
the management team structure change. Sole-managed funds that switch to teams hold
higher beta portfolios and have higher total risk (standard deviation of monthly returns)
than team-managed funds that switch to sole-managed. Panel C presents the differences
in portfolio risk after and before the change. Funds that switch from team management to
single managers exhibit an increase in systematic risk (0.10 increase signiﬁcant at the 1%
level). On the other hand, when funds switch to team management total fund risk, as
measured by the standard deviation on monthly returns, goes down and so does the UMD
factor loadings. Taken together results in table 3 suggest that risk consideration might
explain why some funds prefer team management.

Results on the relationship between the change type and other fund characteristics
such as size, inﬂows, turnover, number of holdings and concentration of holdings are
presented in table 3.4. We ﬁnd that a decrease in assets leads is correlated with changes
of the investment advisor. Taking into account results from table 3.2 we conclude that
when performance is bad and investors pull out of the fund investment advisors are in

trouble. The most interesting ﬁnding in table 3.4 is that funds switching from single

143

managers to teams exhibit a tremendous increase in size the year before the change
compared to other funds. On average, sole-managed funds that switch to teams
experience an additional increase in assets of $328 million compared to funds that also
switch investrnent advisors (signiﬁcant at the 1% level) and an additional increase in
assets of $258 million compared to funds that switch from team-managed to sole-
managed the year before the change. Taking into account that the performance of those
funds is positive the year before the change, we conclude that funds switch from sole— to
team-managed because of size and not poor performance. An interesting ﬁnding in Panel
C of table 4 is that when good performing sole-managed funds switch to team-managed

their assets do not increase at the same high rate.

3.3.2.2 Probit models

In this section, we explore possible determinants of a fund’s management team
change and investment advisor change using probit models. We relate the probability of a
funds changing advisor, Prob(NewAdvisor), and the probability of a fund changing from
team-managed to sole managed, Prob(TeamToSole), to performance, risk and size

characteristics of the mutual fund portfolio. First, we estimate the following probit model:

Prob(NewAdvisorh = 1’0 + 1’1 * at—l + 1’2 * ﬂm,z—1 + b3 "' ﬂsmb,t—1 + ’94 * ﬂhmu-l

2002 (3)
+b5 *ﬂude—l + b6 *STDt_] +b7 * ln(assets),_1 + 2 Dn
1998

where 69.] is previous year’s risk-adjusted performance (4-factor), ,Bm,,_,is last year’s

portfolio market beta from the 4-factor model, ﬂSMBJ—I is the last year’s portfolio SMB

144

 

beta from the 4-factor model, ﬂHMLJ—l is last year’s portfolio HML beta from the 4-
factor model, ﬂUMDJ—l is last year’s portfolio UMD beta .from the 4-factor model,
STD,_,is last’s years total portfolio risk as measured by the standard deviation of

monthly returns, ln(assets),_1 is last year’s fund size as measured by the log of total net
2002

assets, and 2 D" is a set of year dummy variables. We also reestimate the model
1998

replacing the ln(assets),_, variable with the Inc(assets),_1which is the increase in the
fund’s assets during the year.
To investigate the determinants of the type of management team change we use

only the fund-year observations where there is a change in the team structure we estimate

the following model:

Prob(TeamToSole): = 170 + 1’1 * at—l + bz * ﬂm,z-1 + 1’3 * ﬂsmb,t—1 + 54 * ﬂhmu—I

2002 (4)
+b5 *ﬂude—I +b6 *STD,_1+b7 *1n(assets),_1+ 2 D"
1998

where all the independent variables are exactly the same as in model (3) and the response
variable takes the value of 1 if the fund changes from team-managed to sole-managed
and 0 otherwise (sole-managed to team-managed).

Results for model (3) are presented in table 3.5. We estimate the model for all
funds (columns 1 and 3), and for team-managed (columns 3 and 4) and sole-managed
funds (columns 5 and 6) separately. Similar to the results from'univariate tests, we ﬁnd
evidence that funds switch investment advisors mainly because of poor performance

(signiﬁcant at the 1% level). For sole-managed funds, poor last year’s performance seems

145

to be the only reason to change the investment advisor. For team-managed funds we ﬁnd
that high portfolio systematic risk increases the probability of investment advisor change,
while high SMB and HML factor loadings size and total fund risk seem to reduce the
probability of a change. All results are signiﬁcance at the 5% level or better.

Table 3.6 reports results from the estimation of model (4). Out main focus is on
columns 5 and 6 of table 3.6. When funds do not change investment advisors, the only
change is the structure of management team. We ﬁnd evidence that performance
negatively affects the probability of a switch from management team to single manager.
The interpretation is that when performance is low the probability of switch to a single
manager is higher, which conﬁrms our ﬁnding from the univariate tests. Fund size also
has a negative effect on the response probability again conﬁrming the ﬁnding that high

fund size increases the probability that a ﬁnd will switch from sole-managed to team-

managed.

3.4 Conclusion

In recent years team management has become extremely popular in the mutual
fund industry. This study investigates the determinants of management team structure and
its consequences in performance and risk. We focus on 503 changes of management team
structure and identify the possible reasons that make funds select one structure type over
the other.

We ﬁnd that team-managed funds switch to sole-managed after poor performance
while sole-managed funds switch to team-managed aﬁer signiﬁcant over-performance.

When a fund becomes sole-managed performance improves signiﬁcantly (184 basis

146

 

points in terms of 4-factor alphas) while a switch to team management leads to
deteriorating performance. Funds that switch from sole-managed to team-managed
exhibit a decline in performance of 190 basis points. Taking all this into account, our
results suggest that team management is not good for performance.

So, why do fund families choose the management team structure? We ﬁnd that
sole-managed funds that switch to teams experience above normal increases in size
consistent with the view that fund families, investors, managers, or all of them believe
that a single individual manager cannot efﬁciently manage a lot of money. We
also ﬁnd weak evidence that risk considerations play a role too. Team-managed funds
seem to be taking on less risk than their sole-managed counterparts. In general, our
ﬁndings, conﬁrm past evidence that team-managed funds do not offer superior risk-
adjusted returns.

Finally we also examine a fund’s decision to change the investment advisor

altogether. Findings suggest that poor performance is the major reason that leads funds to

ﬁre their advisors.

147

APPENDIX 3.

TABLES OF ESSAY 3

148

 

 

New own 3. omm mVN Nm 5mm .60th =<

 

 

 

 

 

 

 

 

 

 

mN. Nv oN we we m . om mOON-VOON
N. . om km 2. om m an VOON-MOON
mo . an Nm on «N m . on mOON-NOON
No. .m 5N mm mN m . 3. NOON- .CON
w. . mm v. No mm o. .m .ooN-OOON
8 N... m. om .N 6 mm oooN-mmm.
m. . t. mN mu N». N. 3. mam 78a .
mm nv N. an ..N m on cam .Kmm .
in... men... .92 262 __< 8.2 «Em... ...-ﬂ 262 __<
$9.ch =< 3%....2888. 2 3952-28 wean—2-20m o. vomng-Emm.
60th 3 magma”. 65.2...“ .cmEmmmcmS. 5.3. when“. ..m .280.
...... 68 8. ...... 6.8 mm .8 «9.8.30 5.
8 8 ... NV 6. . . 8 «285.355
CON mm .v VN. mm .N on 02.005-52.90
31.. EN mN. ...mm .0. av o .N 5390
8 ..N ... .... 8. .. «N 5320 8289......
I82 .....Emm 88. 262 __< .92 «Em... 82 .82 __<
momcmco =< nommcmZ-Emm. o. gng-Qom nommcmE-Qow 0. 89.52.68...

 

62.8.30 2.68.8.6. A... 8......ch 63.25% EmEmmmcmE 5.3. when”. ..< Bean.

698....
8302.6 :80. .EEowmcmE o... 8.... ...8 0.0.0.. 89» o... .0.. 9.58.88 8.8.. 8.5.8. o? 8.“... 888.. So .0 3.8.6 o... 6.58.... U .28.. 58...“.
.89. 3 moms-8.... .58... m .28... A883 no... ...8 38.88.88. ... 33.88.39... 0.30:... 28803.88 2.. 5.3 ow o. 8.. own-8.... .25 o... .8...
...8 08... z... 6. 6.88.8... 8&3... a... 8... 3.8.6:. .M: 2.8% .25.. u... gamma... 0. 3.888 m. .8... .8868 .coﬁowmcmﬁ o... .... #888836. 8
m. 0.0... .8... 880...... .m: 362 62.83.. 3.880.... .3 mowsano .53... < .28.. 6038.... :88 ...oEowmcmﬁ ..o 88.8.. 5.. 68.8883 038. "...:-

mouuagu .9223. .58835 ...—8 9.32:5 .noaowaaazl b.2555 .3825
..m ~38...

149

 

 

 

mom .5. .... ..m mm. vm Na. 000:9". =<
om KN om hm mN m mm VOON-NOON
mu .m mN ..m m. o mN NOON-NOON
om mN .N 00 NN N . 3w NOON- .OON
mm 0.. N. mm .m o. .v .ooN-OOON
om Nm m 0v .N m mN OOQN-mmm.
mm mm 0N mm AN 0. mm mam .-wmm.
.92 0580 .92 302 __< .92 0:80 .92 262 __<
3%ch __< “Egan—2-280.. 0. 83:02.28 009052-28 o. 00m0c02-E00.

 

.00. 9.2.0.0. 0:0 2.0.30.6. 0c. .0. 0.00 030.6% 0:0 000.220 0.3.02.5 .c0E0w0002 5...: 005.“. ..0 .020Q

3:00. ﬁn 030,—-

150

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N2 2 F 92 E Evm 8m Emmxécé 2
New. _‘- mv F .0 «mg; NNm. _‘ r; 35.0. omwd. 93% .9034»
... 5 v .m- van. F «baud w 3. _. ...;moad- wmwd- mcaﬁ .588-..
E9» mzoEEQ in?» :82 ..0 3ch
mac. w- wmwd ono.o 2.10 000.0 m Fwd. mwﬁw wmwd mwod sand. mcaﬁ hogan—é
:mmvN- o 5.0- wove .N 5.0 memo. ooo.o ovwd wmwd wmwd mvwd. 29m .2034
E9» :22 ..m 3ch
«immod :Nmmd momd .vmwd Nmmé oomd- wmwd. mum. F- wmwd wmwd «22¢ .9034
...Nmmd ...:mmod vad. ...mmm... N 5N v3.0- owo. w- on _‘ .0. now; 35.0 9.9m 520$:
gmmi 98.3%“; ..< 3ch
$-on 3:9 81$ :13 § § E E Q E
as. as. as. as. .67.). .62
«Sam >52 9:8 262 9:3 262
mocmoEch I m8:90t5 commcmzémmh uMmmcmZéBw @9ch oz

2 Bamamzéom

9 gangs—-83..

 

mmcmco 922:5 EmEmmmcms.

 

53533." :26. $3 can

fem .o\.._ 05 8 EmoEnwa 8:58 8365 .. ES 3,1... dosage—om 3933a 98 83-3523“ .5953 mooaﬂobﬁ 3.599— 0 .053 5.25
.omSEo 05 mEBo=8 so» 05 Ho.“ 85,585.“ :88 05 853.— m ESE .mowﬁﬁo mo womb 80:56 5:5 85a 5233 £82: 05 E moocobt€
he 383 mm :03 mm uwcﬁo 05 mo 30» 05 883 so» 05 35850me :38 3.8%: < 3.8m dam—a Bﬁﬂé @9380 38 £233 90608 “8325
88mm; 05 0.8 3:88 8.5308 oogéba 2F .owSEo mo 25 05 3 “888 .80 5 $5 =« 5m oogotoa :86 05 $893 053 mg

max—«.3. 83.52:: I macaw—82230 moans—Stem
Nd 033—.

151

 

 

 

 

 

 

0 30 «~00 N000 .....0000 0.000 000.0 30.0 :00- 30.0 ~30- 90900 ...21
000.0 :00 000.0 38.0 00.0 000.0 000.0 0.000 F30 0000 $0880 mzm
000.0 0N00 v8.0- ..0N00. 0 00. _. 000.0 000. F 0.00. _. 0000 «N0; $0900 8882
N000- 000.0. 000.0. 3.8000- 00.00 3.0; 304 «000 00.0.0 mwoé 900 8x82
88> ..me ..m 8.80
.0000 2 P0 300. 0.000. 0 00.0 F0 0.0 000.0 N000 000.0 $0.0 mEBmm Eb
\. F00 3.00 000.0. 0N00. 000.0 000.0. 0 3.0 300 0.00.0 0N00 9.880 0.2:
000.0. 0000. 30.0 ...0000 30.0 30.0 0000 v8.0 :80 «N00. E0900 .5:
v3.0. ..Eod. 85.0. .880 83 $3 83 3.3 S ..o «8.0 0098 $8
300 ~30. 000.0- x. 00.0. 0 F0. _. $0. _. N000 000. _. 000. F 0N0. P @393 8x85.
.0000 N000 0000. .0000. 000. F 000.0 30.0 0 00. _. N000 000.0 880 8x82
.59» 900.391 ..< 0251
$0.80 60.80 813 3-80 80 49 i E S E
.092 .092 as. as. a: .82
9:3 282 05$ 282 9:3 262
3:35:05 - 80:08.05 “80.85288... gmmcmzéﬁw 898:0 oz
9 080882.828 2 Bomcmzémo...

 

898:0 2:825 0.8580885.

 

5038008.— _o>o_ $00 98 fem ax; 2: 8 28200.88 £32 08068 ... 98
3.1.... can 800-820 98 800083280 08380 80:22.06 .3802 O 8.80 .3880 89800 20 838:0.“ 8o» 05 .80 xmc 82: 05 8.802
m 8.80 .wowﬁﬁo .8 80b 80.6006 5.05 850 0853 £82: 2: 5 80020008 80 v.88.“ 8 :03 ...8 omﬁﬁo 05 .8 8o» 05 20.80 .80» 2: xmt
808 2002 < 880 8522 .2588 .8 028.5% 08988 05 0.8 8on 8880-0» 05 520 880 QED E8 .02: .mEm 05 .820 8x88 8880
.v 05 .880 8x88 20on 880880 888.: 05 28 08.802 8882: 208 80053522 BE. .0288 .8 003 .3 x2.— 82: 5.802 038 0:0.

£9223. 23.83:: I 850888800 sag
Wm 038,—.

152

 

 

 

3‘ v: o? E 3% ooo 383%: 2

«famhm.on «iimmm. Fl ©rN.CI NQGVAV «itwwo. PI catQWF.—.I $51.me nun—um

..Looo. ooo.o ooo.o «moo vooo ooo.o $52 0.2:

ooo.o as? ooo.o. L o to- :Luoo. no? c398 ...2:

2o.o .oooo- ooo.o. .too. Fooo ooo.o c398 m2m

ooo.o :Eoo. ooo.o ooo.o- Boo- ooo.o c392 aim:

ono.o- o Foo ...Nowo ooo.o :m Foo .Xoo ES .35:

50> w:o.§8l - .69» ~32 ..0 $th

or to- ...otoo ooo.o. ovoo. ooo.o ooo.o ooo.o ooo.o ooo.o ooo.o «53$. Em
...Nohoo. o8? ooo.o. :3? Foo. ooo.o ooo.o ooo.o ooo.o ooo.o $92 92:

3.83 no ~35.

153

 

 

 

 

 

 

 

 

oNK wad. mmd- hams- mmdm Famsm Mwa coda maﬁa FN.moF 005201 ”.0 oz
waK- ...-madm- 1. FNNm- tuna- o F .mm mod F F Fodm NEmNF mméw n F .09 005:5...
- 3».va Fmdm omKN- mmda o F.mm mmdv m F .mm- NNKm- v :3 dem- 9002 302
mde F dem and- no.2. F swam 05mm wm.mm- Fm.mN- v F .o F F Fade- 000905 .002
...mmd ...-cod de- Fx-o F .o F. F .m #06 mad Fwd mad and 0.002 no 00..
0mm> 082 ..m 00:01
v FNo \- F. F and. :- Fv. F omdm m FNm mmNm ov.mm omdm m FNm scum-acoocoo
m6 oo. F No.w- :omd- FmNm Fm. Fm F Fdw m F.0m Fmém wYVOF 005201 F0 02
mowN F- ...-no.8- mwﬁ- Nod. hmdw om. FN F on. 2: mvdo F {mom mm.mm 005:5...
msmém 2.9V. F5 F Roda F twmﬁm mo. F F F owdm- omdm \- F .mh- hm. Fm 3d T 393.0. >.62
2. Fodmw ...-hmdwm wmdw .344 F mvdmm mvd- nwKo F06 F- N F60 #5. F0- 00090:. #002
.3de cams-m6 Rio- Nod mad Fwd mm...“ mad mod mad $002 ”.0 904
50> 000.395 ..< 00:01
$-60 3:8 8:3 :13 so so E 5 E E
.092 .092 as. .092 as. .62
9:3 302 9:8 302 083 302
00cmo§m - 000500.20 Banzai-E00... EmacmE-mﬁm 09.20 oz
2 BomcmE-Qow 9 acumen—2-80“:

 

09.050 0.3025 EoEmmmcmZ

 

$330003“: _0>0_ $2 98 fom .o\o_ 05 on 0:00:8me 3300 000065 .., 98 ......I... .320.» 0008 80%-Bra 0:0
aux-mamaéom 000303 o00=000§0 3.5000 0 Emma— .zzoﬁm 098:0 05 mE30=8 000% 05 00.“ 8530008050 083 05 00m 828' 508 05 3.8%:
m 355 .momﬁﬁo m0 8%» 05000.06 5.25 owns.“ 50300; $808 05 E 800000.006 00m $33 mo :03 mo 098:0 2: 0.8.009 30% 2: 855000030
0:03.50 .«o 829» 0008 $000.00 < Ecoméxov 3520: S 08 05 E 305602: m0 00:80:00000 05 0:0 020.2000 Muse 05 E 3520:
m0 000:5: 05 £9653 25:08 £003 05 $0008 30: 3003500 35 38mm 5 080005 mo Fan— 05 .000» 05 Scam—505 3030 5 0.80005
05 .883 00: :38 .«0 wo_ Ba ”05 000 000.5000 355008050 028:3 2:- .0mSE0 m0 09o an 83380830 0:80.80 :88 3.5000 038 2:-

ﬂobag‘ 0.02.315 I 85.02000320 0:820.—
vd 030k.

154

 

 

 

 

 

he 3 0 02 E 8% 08 «$04.23: 2

0N.OI it!” P .MI NQAVO ##NN.N& :53. PI iﬁtg. PI SCENb—Lgcoo

8.0 v3 8N- «E- 5.0 R. F- $222.. .6 oz

PN.VI mm. —. Fl NN.QI QN.®—. itﬁmm-ml ﬁttOQ-NF u0>OE:-—u

Nmﬁvt mmdo F 00. FFq F- mm. FN 00. F F- NYNV- macaw/x 252

{F F. FVN- mN. Fm mm. FOF- om.vF- dem NFN- wmmeoE “mam/x

3am F .0 mod- Nod N F .0- 150 F .0 mod- m~0wm< *0 GOA

km0> W30N>9Q I kmm> “X02 no \thn‘

Nomd *aaomﬁ m F. F Now. F NmNm 09mm 0N. Fm modm N F.Nm Fmdm cozmbcwocoo

3:80 ...m 2.3.

155

 

 

 

 

 

000N 000N VFON VFQN 000V 000V 2
m0> m0> W0> m0> m0> m0> m_O.=COO hm0>

a: m FCV. —.| 0 PDQ. PI :3 ”QMN. PI .1: NNQG.OI «.0.... MC PO. Fl in... @VQVN. Fl vamp-mu:—
0000.0 0000.0 0000.0 muwmmxx E 0000.6:—
otod .. 003.0- 025.9 0.00.-<3 03

0k. F0.0 #0 F0.0 .. NN00.0- 3 0000.0- 0000.0- 0 F000.0- 0530.. ”—0 $00 .30
003.0 003.0 0000.0 F0000 N000.0 h00000 OE: .553-Fa
00N0.0 0 F000 :3, 003.0- :3 0000.0.- .. 03 F .0- « 006v F .0- FE.— ..ouomu-v
300.0 0000.0 .2. 030.0- 3... 5050.0- 0 F0 F .0- 0 F F F0. 90.5 03000.?
05v F .0 050 F0 25 0000.0 :5 00050 at. N030 ...... NON00.0 mama houomm-v

it ”NCO.°I fa WNOC.QI it mace-OI a: NO FC.OI «it ©®OC.OI it: moo-CI NEG—W anUmhlV
80 30 E 5 E E $3000> .5238...

00:3”. 000.05.20.00 00:3“. 000052-500... 000:“. =<

 

30.2 memmno .00-.2 2,02" a
09..ch 0006000005. ”0.00_.0> 0000:0000

 

00006000 000 050 3 00800 0000030 00000020 000.5900 00: 000 30000-0000 005 80 00200000.— 0£ E 000205 00 00080.50 08E. 5020000000
_0>0~ $3 000 8m 2x.— 05 00 0:02-0:06 0:300 0000005 ... 000 3.1.... 000000 E 0000005 000 505 30000 0.0000 05 .00 02 000 000—000 0B 80 0:0 :0
A8 00060030.— 5 .3030 00: 0008 0.003 05 .00 w2 3030: 05 £00300 35:08 .00 0000300 00000.80 05 40008 08000-0 0.0000000 800.0 005000—
08000 Vic .0000 08000-0 06 ”000 A3 000 A9 .20 0020000000 E 00305.» 0000000005 05- .>_00000000 0050 003008-030 0:0 00w000E-E08 00.“
000 .0050 :0 00.0 002000000 0.0 :3 0 .3 .050 0.0 8000 00.3003 :00 000 Amvh0w0c08 080m 05 a 0000 000 0000 B0: 505 Amvu0w0=08 05 0000300
000 050 05 .0 0:0 .00 020> 05 00x00 005 030i? >583 0 ﬂ 05090, 000000000 0&- .30 020000000 0005 E90 33000 E00000 038 $5.

€003.30 0008.00.30 0005000002 5 $0000 I 0030000001 050.5
Wm 0.00.—-

156

 

 

 

mmm 00m 00.. 0.00 80 80 2
00> 00> 00> 00> 00> 00 > 0.00.000 ..00>
00000.0 .... 00000.0 Swad- v 5 V. T 3 3.0- mot- _. .0 000.000.
300.0- .. 300.0- 0.000< 0. 00000:.
.... ..oood- :- monod- 00000- 0.0.8.0- 0.000< .0 00..
$00.0- 00.00.0- 0300- 003.0- N0000- N0000- 005.0. 00 $00 0.0
003.0 «000.0 wammd 30nd 0000.0 0.09.0 005 09000-0
030.0 0000.0- 0000.0. cm 5.0- 0000.0- 53.0- .80 0.000-v
. 0000.0 .. 0000.0 0.30.0- 0.08.0- Svmd nwmmd 00.0 02000-0
0.000.0- mmmmd- 00.0-Nd 2.0m... $3.0- Roto- 0.00 02000-0
...... ovwod- ...... 003.0- 58.0- 00 5.0- : mm ..0.0. ...... cm ..0.0- 000.0 02000-0
.0. .0. E 3 .0. 3 0030...; 2005002.

.65. 00000 .55. 302 0000000 :4.

 

000000 00308.0 .00E000005. ”0.0000> 000000000

809000.). 800... 0. 00d0002-0.0mn0 009002-000 0. 00010002800 .Fu q

 

00008000 000 00a .3 00000 0000000 0000.020 00:00.00 .00
000 08005000 00... 05 000.0000w00 0.0 0. 0000.00. 000 0000800 08:- 5030000000 .0>0. $0. 000 gm ...\o. 0... 00 8000830 00.0000 000.00. ... 000
3.3... 000000 0. 0000000. 05 5.05 000000 0.08.“. 0... .00 mo. 05 000.000 03 A00 000 00 AS 000.0000m00 0. 000000 .00 .80. 0.08.". 05 .00 mo. .0500 0.0
.000300 0.5008 .00 0000300 000.0080 0.0 ..0008 08000.0 0.0000000 8000 30.000. 0800-. 0.0.0 .0080 00000.30 05 ”000 A3 000 Amv .20 002000030
0. 0050.000, 0000000000. 00... 000300 808000.00. 0800 05 0000. 00 000000 80 00000.00 800. 0.00.0 0w00..0 .05 00000 Am 000 000900 80800000.
0.0.0 000.000 00.0 0w0000 00300.00 05 5.3 005030 005 00000 3 .0m0000 00800.00 800. 0 00.08 .0.... 0000.0 ..< .0 ”0008 00 0000 000.. 000 30.000000
00.000030 05 8: 0 3 .00w0008-803 00 00w0008-0.00 8000. 0008.050 000 003 00.0 ... 0000 000 000.00.080.00 00 003008-800. 800.0 0000.050
00.. 00:0 0.0 0. 000 ..0 0:.0> 05 000.0. 00.0 050.00., >88 0 0. 0.00.0.» 80000000 00.... .00 00000030 .5000 800 00.0000 800000 0.08 0.0....

00.50.50 0008090005. 0. 09:30 I 002000003— 0.00...—
0.m 0.00H

157

BIBLIOGRAPHY

158

BIBLIOGRAPHY

Bar, Michaela, Alexander Kempf, and Stefan Ruenzi, 2005, Team Management and

Mutual Funds, Working Paper, Centre for Financial Research (CFR) Cologne.

Barry, Christopher and Laura Starks, 1984, Investment Management and Risk Sharing
with Multiple Managers, Journal of Finance 39, 477-491.

Bliss, Richard T. and Mark E. Potter, 2002, Mutual Fund Managers: Does Gender

Matter?, Journal of Business and Economic Studies (forthcoming).

Brown, Keith C., W. V. Harlow and Laura T. Starks, 1996, Of Tournaments and

Temptations: An Analysis of Managerial Incentives in the Mutual Fund Industry, Journal
ofFinance 51, 85-110.

Carhart, Mark, 1997, On the Persistence of Mutual Fund Performance, The Journal of
Finance 52, 57-82.

Chen, Joseph, Harrison Hong, Ming Huang, and Jefﬁey D. Kubik, 2003, Does Fund Size
Erode Mutual Fund Perfrormance? The Role of Liquidity and Organization, Working
Paper.

Chevalier, Judith and Glenn Ellison, 1997, Risk Taking by Mutual Funds as a Response
to Incentives, Journal of Political Economy 105, 1167-1200.

Chevalier, Judith and Glenn Ellison, 1999, Career Concerns of Mutual Fund Managers,
Quarterly Journal of Economics 2, 389-432.

Chevalier, Judith and Glenn Ellison, 1999, Are Some Mutual Funds Managers Better
than Others? Cross-sectional Patterns in Behavior and Performance, The Journal of

Finance 54, 875-899.

159

 

Deli, Daniel N., 2002, Mutual Fund Advisory Contracts: An Empirical Investigation,
Journal of Finance 57, 103-133.

Gottesman, Aron A. and Matthew R. Morey, 2006, Manager Education and Mutual Fund

Performance, Journal of Empirical Finance 13, 145-1 82.

Golec, Joseph, 1996, The Effects of Mutual Fund Managers’ Characteristics on their

Portfolio Performance, Risk and Fees, Financial Services Review 5, 133-148.

Goya], Amit and Sunil Wahal, 2005, The Selection and Termination of Investment

Management Firms by Plan Sponsors, Working Paper, Emory University.

Grinblatt, Mark and Sheridan Titman, 1992, The Persistence of Mutual Fund
Performance, The Journal of Finance 47, 1977-1984.

Hendricks, Darryl, Jay Patel and Richard Zeckhauser, 1993, Hot Hands in Mutual Funds:
Short-run Persistence of Performance, 1974-1988, The Journal of Finance 48, 93-130.

Hill, G. W., 1982, Group vs Individual Performance: Are N+1 Heads Better than One?,
Psychological Bulletin 91, 517-539.

Herrenkohl, Roy C., 2004, Becoming a Team, South- Western, First Edition.

Hohnstrom, Bengt, 1982, Moral Hazard in Teams, The Bell Journal of Economics 13,
324-340.

Hollenbeck, John R., Daniel R. Ilgen, Jeffrey A. LePine, Jason A. Colquitt and Jennifer
Hedlund, 1998, Extending the Multilevel Theory of Team Decision Making: Effects of

Feedback and Experience in Hierarchical Teams, The Academy of Management Journal
41, 269-282.

160

Ibbotson, Roger and William Goetzmann, 1994, Do Winners Repeat? Patterns in Mutual
Fund Performance, Journal of Porﬂolio Management 20, 9-17.

Ippolito, Richard A., 1989, Efficiency with Costly Information: A study of Mutual Fund
Performance, 1965-1984, Quarterly Journal of Economics 104, 1-23.

Jensen, Michael C., 1968, The Performance of Mutual Funds in the period 1945-1964,
The Journal of Finance 23, 389-416.

Janis, I. L., 1982, Groupthink: A Psychological Study of Policy Decisions and Fiascoes,
Boston, Houghton Mifﬂin Company.

Krishnan, Hema A., Alex Miller, and William Q. Judge, 1997, Deversiﬁcation and Top
Management Team Complementarity: Is Performance Improved by Merging Similar or

Dissimilar Teams?, Strategic Management Journal 18, 361-374.

Khorana, Ajay, Peter Tufano and Lei Wedge, 2005, Board Structure, Mergers and
Shareholder Wealth: A Study of the Mutual Fund Industry, Journal of Financial

Economics, forthcoming.

LePine, Jefﬁey A., Hollenbeck, John R., Ilgen, Daniel R., Colquitt, Jason A, and
Aleksander Ellis, 2002, Gender Composition, Situational Strength, and Team Decision-
Making Accuracy: Acriterion Decompsition Approach, Organizational Behavior and

Human Decision Processes 88, 445-475.

Ley, Eduardo and Mark F. J. Steel, 1995, A Model of Management Teams, Working
Paper.

Malkiel, Burton G., 1995, Returns from Investing in Equity Mutual Funds 1971 to 1991,
Journal of Finance 50, 549-572.

161

 

Massa, Massino and Rajdeep Patgiri, 2005, Compensation and Managerial Herding:
Evidence from the Mutual fund Industry, Working Paper, INSEAD.

Massa, Massimo, Reuter, Jonathan, and Eric Zitzewitz, 2006, The Rise of Anonymous
Teams in Fund Management, Working Paper, INSEAD.

Nanda, Vikram K., Jay Z. Wang and Lu Zheng, 2005, The ABCs of Mutual Funds: A
Natural Experiment on Fund Flows and Performance, AF A 2005 Philadelphia Meetings.

Pichler, Pegaret, 2004, Optimal Contracts for Teams of Money Managers, Working
Paper.

Prather, Larry J. and Karen L. Middleton, 2002, Are N+1 heads better than one? The

Case of Mutual Fund Managers, Journal of Economic Behavior & Organization 47, 103-
120.

Rasmusen, Eric, 1987, Moral Hazard in Risk—Averse Teams, The RAND Journal of
Economics 18, 428-435.

Sah, Raj Kumar and Joseph E. Stiglitz, 1988, Committees, Hierarcies and Polyarchies,
The Economic Journal 98, 451-470.

Schmidt, Jeffrey B., Mitzi M. Montoya-Weiss, and Anne P. Massey, 2001, New Product
Development Decision-Making Effectiveness: Comparing Individuals, Face-to-Face

Teams, and Birtual Teams, Decision Sciences 32, 575-600.
Simons, Tony, Lisa Hope Pelled, and Ken A. Smith, 1999, Making Use of Difference:

Diversity, Debate, and Decision Comprehensiveness is Top Management Teams,

Academy of Management Journal 42, 662-673.

162

Sharpe, W.F., 1981, Decentralized Investment Management, Journal of Finance 36, 217-
234.

Slezak, Steve L. and Naveen Khanna, 2000, The Effect of Organizational Form on
Information Flow and Decision Quality: Informational Cascades in Group Decision

Making, Journal of Economics & Management Strategy 9, 115-156.

Thompson, Leigh L., 2004, Making the Team: A guide for managers, Pearson-Prentice

Hall, second edition.

Tufano, Peter and M. Sevick, 1997, Board Structure and Fee-setting in the US. Mutual
Fund Industry, Journal of Financial Economics 46, 321-355.

Ding, Bill amd Russ R. Wermers, 2005, Mutual Fund Performance and Governance

Structure: The Role of Portfolio Managers and Boards of Directors, AFA 2006 Boston
Meetings Paper.

163