..).I.J:§.;.a.
. :. .. .21

v

. ,
2: .5:
. . . t u t ,
t a. 3......“ r
. , v .
..
3......»

V {9

.r

I
1F
t

.1
‘ r .
I: anus...

‘9

3.47
‘ \

:81

L .

XI...
4(3)

I ..,...... Lamar...”
. .. x

 

.4.) l
1:. I... 1.3K.
mini-I:9’

“5.31:.

. . .5...
l... ‘y s:

 

 

. . ‘ ,3 o..ﬂ.‘.r.1.ﬂ.§.
5:. . “A" I V

 

This is to certify that the

dissertation entitled

VOTING RIGHTS, CORPORATE CONTROL,
AND FIRM PERFORMANCE

presented by

Yi Zhang
has been accepted towards fulﬁllment

of the requirements for

Ph. D. degree in Business Administration

W /6M¢A

 

 

 

Major plofcssor

Date ?/2/2002

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY
Michigan State
University

 

 

 

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.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:lCIRC/DateDuo.p65-p. 15

VOTING RIGHTS, CORPORATE CONTROL, AND FIRM PERFORMANCE
By

Yi Zhang

A DISSERA'ITION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Finance

2002

ABSTRACT
VOTING RIGHTS, COPORATE CONTROL, AND FIRM PERFORMANCE
By

Yi Zhang

This dissertation investigates the impact of corporate control on ﬁrm performance
of companies with dual class stock where voting rights are separated from cash flow
rights. I ﬁrst examine the issues concerning the decision to issue multiple classes of
common stock with different voting rights. Firms often issue these shares to maintain a
concentrated ownership structure, while at the same time raising new equity capital. I
identify all Compustat listed companies with dual/multiple class common stock during
the 1990-1999 time period using various databases including Compustat, CRSP, proxy
statements, and 10-K ﬁlings. This database should be the mOst comprehensive sample of
dual-class listed ﬁrms ever assembled. Using the collected data, I proceed to use logit
models to examine which factors affect the likelihood that a ﬁrm will issue dual-class
shares, and also to see if this relationship has changed over time. My principal ﬁndings
are as follows: (1) There are strong industry effects in the likelihood of having dual class
shares — printing, media, and motion picture ﬁrms are relatively more likely to have two
classes of common stock while utilities are relatively less likely to do so. (2) Company
age affects the likelihood of having a dual-class structure — younger ﬁrms are more likely
to have two classes outstanding. (3) Firm size does not have any signiﬁcant effect on the
likelihood of having dual—class shares. (4) Firms with high stock-retum volatility are

relatively less likely to have a dual class structure. I then examine issues concerning the

valuation and performance of dual class firms. I identify insider ownership in dual class
ﬁrms in 1995 using various databases including proxy statements and lO-K ﬁlings. This
database should be the most comprehensive ownership data sample of dual-class listed
ﬁrms ever assembled. I compare valuation and performance of dual class ﬁrms and single
class firms. I am able to investigate the effects of ownership of voting rights and
ownership of cash flow rights on ﬁrm value separately. My principal ﬁndings are as
follows: (1) overall dual class ﬁrms do not have a lower Tobin’s Q than single class
ﬁrms, (2) dual class ﬁrms that have both classes traded publicly have a lower Tobin’s Q
than single class ﬁrm and Tobin’s Q increases as insider ownership of cash ﬂow rights
rises and declines as insider ownership of voting rights rises, (3) dual class ﬁrms that
have one class of stock traded publicly do not have a lower Tobin’s Q than single class
ﬁrms, (4) within media industries, dual class ﬁrms do not have a lower Tobin’s Q than
single class ﬁrms while within non-media industries dual class ﬁrms have a lower
Tobin’s Q than single class ﬁrms, (5) there is no evidence that dual class ﬁrms
underperform single class peers on certain performance measures, such as the Market-to-
book equity ratio and the Price/Earnings ratio. There is some evidence that dual class
ﬁrms underperform single class peers on some other performance measures, such as
ROA, EBIT, and operating cash ﬂow. Of these measures, ﬁrm performance rises with
insider ownership of cash flow rights and decreases with insider ownership of voting
rights, (6) investment-cash flow sensitivities increase with insider ownership of cash flow
rights and decrease with insider ownership of voting rights. I explore the implications of

these results for our understanding of ownership, control and value.

To my past six years.

iv

ACKNOWLEDGENIENTS

I would like to express my thanks to the members of my dissertation committee —
Dr. Charles Hadlock (Chair), Dr. Edward Fee, Dr. Wei~Lin Liu, and Dr. Long Chen, for
their guidance and encouragement. I especially owe a great deal to Charlie Hadlock for
his help, support, patience and commitment of time. Without his guidance I would not
have been able to complete this dissertation. I am indebted to Ted Fee, who was always
ready to help me with literature, research direction, data analysis and stata. I thank Wei-
Lin Liu for consistent encouragement and insightful advice on career development. I
thank Long Chen for his valuable advice and critiques of my presentation and many other
aspects.

I thank Dr. James Wiggins, Dr. Kirt Butler, Dr. Mark Schroder and other faculty
members in the Finance department who provided help to me. I am also thankful to
Professor Jack Meyer, Professor Jeffrey Wooldridge and Professor Peter Schmidt in the
Economics department who taught me economics and econometrics.

I thank my fellow students in the ﬁnance program for their help and
companionship, especially Dr. Romana Sonti, Dr. Wei-Ling Song and Bei Zhou. I also
thank Chien-Ho Wang from the economics department for discussions on econometrics. I
also thank all of my friends at Michigan State University for supporting me and making

my life better.

LIST OF TABLES ............................................................................... viii
CHAPTER 1 DUAL CLASS STOCK AND THE BENEFITS OF CONTROL .......... 1
l . 1 Introduction ..................................................................... l
1.2 Existing Literature and Hypothesis Development ......................... 3
1.2.1 Existing Literature .................................................... 3
1.2.2 Relationship to Existing Literature ................................. 5
1.2.3 Hypothesis Development ............................................. 6
1.3 Data .............................................................................. 7
1.4 Analysis and results ............................................................ 10
1.4.1 The Logit Mode: Determinants of Dual-class
Common Equity and Their Measurements ........................ 10
1.4.2 Results of Logit Analysis ............................................ 13
1.5 Conclusion ..................................................................... 16
CHAPTER 2 DUAL CLASS STOCK, FIRM VALUE AND
PERFORMANCE .............................................................. 19
2. 1 Introduction ....................................... , .............................. 19
2.2 Existing Literature and Hypothesis Development .......................... 22
1.2.1 Existing Literature ..................................................... 22
1.2.2 Relationship to Existing Literature .................................. 23
1.2.3 Hypothesis Development ............................................. 24
2.3 Data .............................................................................. 26
2.4 Analysis and Results ............................................................ 29
2.4.1 The Separation of cash ﬂow rights and voting rights .............. 29
2.4.2 Matching of dual class ﬁrm with single class ﬁrm ................. 32
2.4.3 Cross Sectional Regression of Tobin’s Q ........................... 36
2.4.4 Other Performance Measures ......................................... 44
2.4.5 Dividend policy in dual class ﬁrms ................................. 46
2.5 Conclusions ...................................................................... 48
CHAPTER 3 DUAL CLASS STOCK, INVESTMENT, AND CASH FLOW...........51
3. 1 Introduction ...................................................................... 5 1
3.2 Hypothesis Development ...................................................... 52
3.3 Regression Specification ...................................................... 54
3.4 Estimation Results .............................................................. 55
3.5 Conclusion ....................................................................... 56
APPENDICES ..................................................................................... 59

TABLE OF CONTENTS

vi

Appendix A Tables for Chapter 1: Dual Class Stocks

and the Benefit of Control ........................................... 60
Appendix B Tables for Chapter 2: Dual Class Stocks
and Firm’s Value and Performance ................................. 74
Appendix C Tables for Chapter 3: Dual Class Stocks,
Investment and Cash Flow ........................................... 98
BIBLOGRAPHY .................................................................................. 100

vii

Table A1
Table A2
Table A3
Table A4
Table A5
Table A6
Table A7
Table A8
Table A9
Table B1
Table B2

Table 133

Table B4.A

Table B4.B

Table B4.C

Table B4.D

Table B4.E

Table B4.F

Table B5

Table B6.A

LIST OF TABLES

Number of US Firms With Dual Class Common Stock in 1990-1999 ....... 61

Description of the Variables ...................................................... 62
Summary Statistics for the Sample of Firms in 1990 .......................... 63
Summary Statistics for the Sample of Firms in 1999 ........................... 65

Distribution of Dual Class Firms in Certain Industries in 1990 and 1999.. .67

Logit Regressions for the Sample of Firms in 1990 ............................. 68
Logit Regression for the Sample of Firms in 1990-1994 ..................... 7O
Logit Regression for the Sample of Firms in 1995-1999 ..................... 71
Logit Regression for the Sample of Firms in 1999 ............................. 72
Description of the Variables ....................................................... 75
Summary Statistics for the Sample of Firms in 1995 ........................... 76
Distribution of Dual Class Firms in Certain Industries in 1995 ............... 78
Insider Ownership in The Dual Classes .......................................... 79
Insider Cash Flow Rights and Voting Rights in Dual Class Firms ........... 79
CEO Ownership in The Dual Classes ........................................... 80
Insider Cash Flow Rights and Voting Rights in Dual Class Firms

in Media Industries and Non-media Industries ................................. 80
Insider Cash Flow Rights and Voting Rights in Dual Class Firms ........... 81
The Largest Block Shareholder Ownership in The Dual Classes ............. 81

Summary Statistics for the Sample of Dual Class Firms and
the Matching Firms .................................................................. 82

Summary Statistics for the Sub sample of Dual Class Firms with
Both Classes Traded Publicly and the Matching Firms ........................ 83

viii

Table B6B
Table B7.A
Table B7.B

Table B8

Table B9
Table B 10

Table B1 1
Table B 12
Table B 13
Table B 14
Table B 15
Table B 16

Table C1

Summary Statistics for the Sub Sample of Dual Class Firms with

Only One Class Traded Publicly and the Matching Firms ..................... 84
Summary Statistics for the Sub Sample of Dual Class Firms in

Media Industries and the Matching Firms ....................................... 85
Summary Statistics for the Sub Sample of Dual Class Firms in

Non—media Industries and the Matching Firms ................................. 86
Cross Sectional Regressions of Q in 1995 ....................................... 87
Cross Sectional Regressions of Q in 1995: Sub Sample when Only

The Dual Class Firms Have Both Classes Traded Publicly .................... 89
Cross Sectional Regressions of Q in 1995: Sub Sample of Firms in

Media Industries and Firms in Non-media Industries ........................... 90
Regressions of Market-to-book Equity Ratio ................................... 91
Regressions of Price/Earnings Ratio ............................................. 92
Regressions of ROA ................................................................ 93
Regressions of EBIT ............................................................... 94
Regressions of Operating Cash Flow ............................................. 95
Cross Sectional Regressions of DIVIDEND in 1995 .......................... 97
Investment Regressions Results ................................................... 99

ix

CHAPTER 1

DUAL-CLASS STOCK AND THE BENEFITS OF CONTROL

1.1 Introduction

An important issue in the governance of modem corporations is the structure of
managerial incentives. Following the seminal work of Berle and Means (1932),
researchers have hypothesized that there are many situations where the preferences of
managers and shareholders may substantially diverge. In particular, managers and
shareholders may disagree on investment policy (e.g., Jensen (1986)), capital structure
policy (e. g., Berger, Ofek, and Yermack (1996)), corporate risk taking (e.g., Amihud and
Lev (1981)), and attitudes towards acquisition offers (e.g., Stulz (1988)).

One way to solve this agency problem is via the use of incentive compensation
and managerial ownership. Consistent with this hypothesis, Morck, Shleifer, and Vishny
(1988) demonstrate empirically that managerial ownership may raise ﬁrm value by
solving agency problems. However, their analysis indicates that there may be both costs
and beneﬁts to high managerial ownership. While high ownership may enhance a
manager’s incentive to act in shareholders interests, it may also entrench managers by
giving them the ability to insulate themselves from outside market forces (e. g., unwanted
acquisition attempts).

One interesting situation where increased ownership does not necessarily

increase both a manager’s degree of entrenchment and his economic incentive to raise

firm value arises in the context of dual-class share structures. For ﬁrms that adopt this
ownership form, cash flow rights and voting rights are partially separated. Thus, by
studying dual-class firms, we may be able to better understand some of the components
of the costs and beneﬁts to managerial ownership.

In this paper I conduct an analysis of ﬁrms that adopt a dual-class share structure.
While many existing studies have looked at the share price levels and behavior of dual-
class stock, no study has carefully examined the factors underlying the decision to adopt
a dual-class structure. Using a comprehensive sample of almost all publicly traded ﬁrms
in the US. over the 1990-1999 time period, I conduct a logit analysis which predicts
which types of ﬁrms choose to adopt the dual-class structure.

Following the work of Demsetz and Lehn (1985), I hypothesize that dual-class
structures will be more likely when a ﬁrm’s control potential (i.e., the potential
managerial beneﬁt from control) is high or when the costs Of maintaining managerial
control under a one-share one-vote system are high. As my proxy for control potential, I
use industry dummy variables. Following Demsetz and Lehn, I expect certain industries,
for example the media industry, to have greater control potential owing to managerial
preferences. Since it would be very costly for managers of large ﬁrms and ﬁrms with
high return volatility to have a large undiversified stake in their ﬁrm, I expect managers
of these ﬁrms to be relatively more likely to adOpt a dual-class structure as a means of
ensuring control while minimizing a manager’s economic exposure to the ﬁrm’s proﬁt
stream.

Consistent with my control potential hypothesis, I find that firms in the media

industry (e.g., publishing, broadcasting, motion pictures, etc.) are signiﬁcantly more

likely to have two classes of common stock. I also ﬁnd that younger ﬁrms are more likely
to have a dual-class structure. To the extent that managers of young ﬁrms are more likely
to be founders who care about maintaining control, this can be viewed as additional
evidence in support of the control motivation for selling two classes of stock. When I
turn to the ﬁrm size and volatility measures, my results may be puzzling to some. My
estimates suggest that the likelihood of having a dual-class structure is unrelated to ﬁrm
size and negatively related to volatility. One potential explanation for this last result is
that possible concerns about expropriation problems in highly volatile ﬁrms make it
difﬁcult to sell inferior voting common stock to outsiders.

The rest of the paper is organized as follows. In section 2 I discuss the related
literature and explain where my paper ﬁts in the context of these studies. In section 3 I
discuss my data and sample selection criteria. In section 4 I report my main results, while

section 5 concludes.

1.2 Existing Literature and Hypothesis Development

1.2.] Existing Literature

Several existing studies have examined the issue of dual-class stocks. The

majority of these studies examine the price differential between the superior voting shares

and the inferior voting shares (e.g. Lease, McConnell, and Mikkelson (1983, 1984)). In

general, the superior voting shares sell at a premium, and the time pattern and cross-

sectional variation in these premia are consistent with votes being worth more when the
likelihood of a control contest is high (Zingales (1994, 1995)).lc

A separate set of studies have examined the announcement returns when a ﬁrm
announces that it is adopting a dual class structure (e.g., Partch (1987), Jarrell and
Poulsen (1988), Comett and Vetsuypens (1989), Chang and Mayers (1992), and Shum,
Davidson and Glascock (1995)). These results are difﬁcult to interpret, since
announcement returns are likely to reﬂect both the value consequences of adopting a
dual-class structure and information concerning the likelihood that a firm is acquired.
Nevertheless, it is interesting that several of these studies indicate a positive market
reaction, which suggests that there may be some benefits to the dual-class structure.2

In addition to these studies of stock price behavior, other studies have looked at
the ownership composition of ﬁrms that have two classes of stock (e.g. DeAngelo and
DeAngelo (1985) and Partch (1987)). As one would expect, these studies indicate that
dual-class ﬁrms tend to have very concentrated inside ownership, and the percentage of
votes held by management is typically signiﬁcantly greater than the percentage of cash
flows held by management. Two recent studies by Smart and Zutter (2001) and Grullon
and Kanatas (2001) examine related issues by trying to understand the relationship

between dual-class stock and other ﬁrm decisions. 3

 

1 Other studies explore the value of voting rights when a single investor holds a large voting block. Barclay
and Holdemess (1989) ﬁnd that in private negotiations large blocks of stock trade at a premium to the
exchange price, and they argue that such premiums reﬂect the value of private beneﬁts of control.

2 Theoretical work demonstrates that dual-class structures can either beneﬁt or harm shareholders under
different conditions. See, for example, Grossman and Hart (1988), Harris and Raviv (1988), Ruback
(1988), Fischel (1987), Denis and Denis (1994) and Attari and Banerjee (1999).

3 See also Bohemer. Sanger and Varshney (1996) and Lehn, Netter and Poulson (1990).

Many studies have also looked at dual-class common stock in countries outside

the U84. In general, the results are similar to those of the US.5

1.2.2 Relationship to Existing Literature

While these existing papers present many interesting ﬁndings, they do not answer
the question as to why firms choose a dual-class structure in the ﬁrst place. There is a
common presumption that ﬁrms choose the dual-class structure to ensure managerial
control. While this is almost surely true, it begs the question as to why some managers
desire control and some do not. Presumably there are some economic tradeoffs that take
place in the decision to adopt the dual-class structure. In particular, I would expect
managers to weigh their desire for control with the penalties the market may assess to the
inferior voting shares. This is very analogous to the issue raised by Demsetz and Lehn
(1985) who ask whether patterns in managerial ownership are consistent with the various
costs and beneﬁts of control that ownership confers.

In this paper I try to answer this question by identifying which ﬁrms have dual-
class shares and which do not. I then run logit models that attempt to predict this
decision. In addition to understanding the cross-sectional variation in the decision to

adopt the dual-class structure, I am able to address two related and interesting issues.

 

4 Outside the US, dual-class ﬁrms are quite common in developed countries or regions such as Italy,
Sweden, Switzerland, Denmark, Finland, Netherlands, Belgium, Canada, Hong Kong, Japan, United
Kingdom and Israel and also emerging economies such as Argentina, Brazil, Mexico, South Korea,
Pakistan, Singapore and China.

5 See, for example, Jog and Riding (1986), Robinson and White (1990), and Foster and Porter (1993) on
Canadian ﬁrms; Ang and Megginson (1989) on British ﬁrms; Levy (1983) on Israel ﬁrms; Homer (1988),
Kunz and Angel (1996), and Caramanolis, Gibson and Tuchschmid (1996) on Swiss ﬁrms; BergstrOm and

First, I am able to assess for the ﬁrst time the overall level of dual-class activity in the
United States. This allows us to assess the economic importance of this type of corporate
organization in the modern US. business environment. Second, my sampling procedures
are designed to identify all instances of a dual-class structure — even cases where one of
the two classes is not publicly traded. This distinguishes my study from previous studies,
which generally restrict their attention to ﬁrms where both classes of common stock are
listed on CRSP (e.g., Lease, McConnell, and Mikkelson (1983)). Thus, my study is more
comprehensive and can for the ﬁrst time, uncover the propensity for ﬁrms to sell one

class of shares publicly and the other in the private markets.

1.2.3 Hypothesis Development

In explaining the cross-sectional variation in the choice to adopt a dual—class
structure, I must identify the potential costs and beneﬁts of this structure. One potential
beneﬁt, initially identiﬁed by Demsetz and Lehn (1985), is that managers may experience
a large utility payoff from being identiﬁed as the controlling party in certain high-proﬁle
ﬁrms, for example media ﬁrms and sports ﬁrms. This line of reasoning predicts that I
should generally observe strong industry trends in the use of dual-class shares, and
speciﬁcally I should observe many dual-class structures for ﬁrms in the media industry
(there are insufﬁcient sports ﬁrms in my sample to identify a sports ﬁrm effect).

One way managers can ensure control under a one-share one-vote system is to

purchase a majority of the shares. For large ﬁrms and ﬁrms with high volatility, the

 

Rydqvist (1990), and Liljeblom and Rydquist (1992) on Swedish ﬁrms; Zingales (1994) and Nicodano
( 1998) on Italian ﬁrms; Taylor and Whittred (1998) on Australian ﬁrms.

potential costs to the manager from pursuing such a policy in terms of the risk they bear
could be prohibitive. One way to control the ﬁrm while minimizing the manager’s
economic exposure is to adopt the dual-class structure. This suggests that, ceteris
paribus, dual-class structures should be more likely in large ﬁrms and in ﬁrms with high
levels of volatility.

A ﬁnal potential concern with dual-class shares is the valuation of the inferior
voting shares. If managerial control has only beneﬁts and no costs, the dual-class
structure should not result in a penalty on the inferior voting shares. However, if
managerial control results in potentially inefﬁcient managerial behavior and/or
expropriation of minority shareholders, this cost will enter into the decision to adopt the
dual-class structure. To the extent that large ﬁrms and ﬁrms with high volatility are more
subject to agency problems, these considerations would predict a negative relationship

between the likelihood of a dual-class structure and ﬁrm size and volatility.

1.3 Data

My goal is to identify all dual-class firms chosen by a large set of publicly traded
ﬁrms. To accomplish this objective, I initially identify the set of all ﬁrms from 1990-
1999 listed on the Compustat tapes. After deleting limited partnerships, my sample
contains 12,069 ﬁrms representing 71,679 ﬁrm years over the 1990-1999 period.

For the resulting sample, I searched the company name ﬁeld in the Compustat
database and identiﬁed all ﬁrms with a share class indicator included in the name (e.g.,

AEL Industries -CL A.) This yields 503 US. ﬁrms representing 2,888 ﬁrm-years over

the 1990-1999 period. Next, I turn to the CRSP tapes and identify all cases where the
ﬁrm has two classes of common stock listed (e.g., AARON Rents Inc. has share classes
A and B.) This procedure yields 666 ﬁrms representing 3480 ﬁrm years over 1990-1999.
After merging the CRSP and Compustat databases (exact merging procedure available
from the author), I found that there was substantial overlap in shares identiﬁed by the two
procedures. After accounting for this duplication, my CRSP search procedure yields 183
new Compustat dual-class ﬁrms rcpresenting 868 ﬁrm-years.

The above procedure is incomplete in that it relies exclusively on the CRSP and
Compustat methods for reporting share classes and company names. For example, I
suspected that in many cases a ﬁrm may have one class of shares that is traded on a major
exchange and another that is not. It is not clear that the CRSP and Compustat procedure
outlined above will identify all of these cases. To identify other instances of dual-class
shares, I exploit the fact that CRSP typically reports a ﬁrm’s shares outstanding for a.
single-class of common stock, while Compustat reports the number of shares outstanding
for all classes of common stock added together. Thus, for cases where these two
numbers differ signiﬁcantly, it is likely that the ﬁrm has a dual-class structure.

Using this line of reasoning, I identify every instance where these ﬁgures differ by
more than 5%.6 I then searched through company ﬁnancial statements, proxy statements,
and the Dow Jones Interactive database to investigate whether the ﬁrm has multiple
classes of common stock. Using the ﬁrm’s IO-K statement, I looked up detailed

information on the share classes that are outstanding. Firms that are authorized to issue

 

6 My choice of a 5% rule is arbitrary, but probably rather conservative. As I report below, most ﬁrms 1
identify as dual-class by this procedure have a much larger difference in the shares outstanding ﬁgure
between Compustat and CRSP. Note that if I do miss a few ﬁrms, given that they were not ﬂagged by my

two classes but only have one outstanding are classiﬁed as single-class ﬁrms. Only ﬁrms
with outstanding shares with different voting rights are classiﬁed as dual-class ﬁrms.
This procedure identiﬁes 198 new dual-class ﬁrms representing 1022 ﬁrm years. The
fact that I found a substantial number of new dual-class ﬁrms via this procedure suggests
that procedures relying solely on CRSP and Compustat labeling are inadequate for
identifying a comprehensive list of dual—class ﬁrms. Note that the median difference
between the Compustat and CRSP share outstanding figures for the ﬁrms identiﬁed by
this last procedure is 31.6%.

Given my sampling procedures, I believe that I have the most comprehensive
database of dual-class listed ﬁrms ever assembled. The above procedures yield a total
sample of 884 dual class ﬁrms (4778 ﬁrm—years) out of 12069 ﬁrms (71679 ﬁrm-years).
Thus, the overall rate of adopting the dual-class structure appears to be approximately
4,778/71,679 = 6.67%. To compare my sample to previous samples in the literature, note
that Comett and Vetsuypens (1989) reported 271 dual class ﬁrms in 1962-1988 period
identiﬁed by the Investor Responsibility Research Center (IRRC) and Olrog and
Rickhamre (1992) estimated 4 percent of 7200 US. ﬁrms were dual class ﬁrms.

In Table 2 I report summary deﬁnitions for all of the variables used in the
subsequent analysis. In Tables 3 and 4 I present univariate comparisons of some of the
variables of interest for the single-class ﬁrms compared to the dual—class ﬁrms. I report
these results both for the ﬁrst year of the sample (1990) and the last year (1999). While
most of these variables are fairly standard from the literature and are derived in a

straightforward manner from the CRSP and Compustat tapes, a few variables are less

 

5% rule, it is almost surely the case that one of the classes has very few shares outstanding and, therefore. is
likely to be economically unimportant.

standard and should be discussed here. The variable Age is a proxy for the firm’s true
age and is measured by identifying the ﬁrst year that the firm’s stock has return data on
the CRSP tape. I derive ownership data from the Compact Disclosure CDs. According
to Anderson and Lee (1997), this is the most comprehensive database on inside
ownership that is widely available. The variable Inpct is deﬁned to equal the percentage
of all common stock held by insiders (i.e., ofﬁcers and directors).

As the t-tests in Tables 3 and 4 indicate, dual class listed ﬁrms appear to differ
signiﬁcantly from the single-class listed ﬁrms on many dimensions including working
capital, the market to book ratio, sales growth, liquidity, the P/E ratio, and some of the
volatility measures. As I report in Table 5, the percentage of dual-class ﬁrms is
particularly high in the publishing, media and motion picture industry, and is particularly

low for utility and ﬁnancial ﬁrms.

1.4 Analysis and results

1.4.] The Logit Model: Determinants of Dual-class Common Equity and their
measurements

I use a logit model to regress the binary dependent variable dual, which indicates
if a ﬁrm has dual-class stock, on predictors described below. Of the possible general
forces affecting issuing dual-class common equity, four are important enough to merit
investigation. One of these, size of the ﬁrm is not surprising. The second is the age of the
firm, which is related to the size of the firm. The third is the proﬁt potential from more

effective control of the ﬁrm. This is referred to as control potential. The fourth is

10

systematic regulation. Regulation imposes constraints on the scope and impact of
shareholder decisions. The amenity potential of ﬁrms should be also included. In
addition, ﬁnancial ratios representing ﬁnancial performance and market expectations
should be considered as well.
Size

The size of ﬁrms that fit in the various product and input markets varies within
and across industries. The larger the competitive viable size is, ceteris paribus, the larger
is the ﬁrm’s capital sources and the greater is the value of fractional ownership. The
higher price of a given fraction of the ﬁrm should reduce the degree to which ownership
is concentrated. This constitutes a legitimate reason for owners to issue dual class shares.
So it is logical to expect a nontrivial effect of ﬁrm size on the feasibility of issuing dual-
class shares. Risk aversion should reinforce the effect. On the other hand larger ﬁrms are
less likely to be a take—over targets so take-over defenses Such as dual-class common
stock are needed less for large ﬁrms.7 I use assets, equity and their logarithms to proxy
for size respectively.
Age

The age of the firm is closely related to firm size and ownership structure. A
young ﬁrm is more likely to have a concentrated ownership structure or to maintain
control by holding voting preferred classes of stock. As a ﬁrm grows, a larger amount of
capital is required and it is harder for managers to maintain control. The use of dual-class

stock as a defense against hostile takeovers becomes less important. Well-established

 

7 There were growing numbers of US. publicly traded ﬁrms issuing dual classes of common stock with
differential voting rights in 19805. The increased adoption of dual classes in 19805 parallels a takeover
wave and increased adoption of other antitakeover defense mechanism such as the issuance of ‘poison pills’
(Jarrell and Poulsen, 1988). This is also the case in Sweden (Liljeblom and Rydqvist, 1992).

11

ﬁrms would face investor pressure due to the issue of transparency and liquidity of the
stock if they keep dual-class shares. A reverse relationship between age and the
possibility to issue dual-class shares is expected.
Control Potential

Control potential is the gain achievable through more effective monitoring of
managerial performance by a ﬁrm’s owners. Demsetz and Lehn (1985) believe that
owners can inﬂuence the success of their ﬁrms and all outcomes are not completely
random. Consequently, they assert that a ﬁrm’s control potential is directly associated
with the noisiness of the environment in which it Operates. The noisier a firm’s
environment, the greater the payoff to owners in maintaining tighter control. Hence,
noisier environments should give rise to more a concentrated ownership structure. It is of
interest to see if the uncertainty also affects the decision to issue dual-class shares. The
three measures of instability examined here are (1) ﬁrm-speciﬁc risk, as measured by the
standard error of the estimate calculated from ﬁtting the market model, (2) the standard
deviation of monthly stock market rates of return, and (3) the standard deviation of
annual accounting proﬁt rates. I favor ﬁrm-speciﬁc risk as the factor most strongly
associated with the type of instability for which control is most useful. I include both a
ﬁrm-speciﬁc risk measure and a ﬁrm total risk measure.
Regulation

Systematic regulation restricts the options available to owners thus reduces the
control potential. Regulation also brings monitoring and disciplining of the management

of the regulated firms. Regulation should reduce the likelihood of issuing dual-class

l2

shares or maintaining a highly concentrated ownership structure. Thus ﬁnancial and
utility industries are expected to be less likely to issue dual class stocks.
Amenity Potential

There is nonpecuniary income associated with the provision of leadership and the
ability to deploy resources to suit one’s personal preference. Two industries are well
known examples for tighter control in order to indulge such personal preference. They are
professional sports and mass media ﬁrms. To reach the consumption goals arising from
the particular tastes of owners, it is necessary that owners are in a position to inﬂuence
managerial decisions. Previous literature has documented concentrated ownership in
these industries. The practice of dual-class shares might also be popular in these
industries.
Financial performance related to take-over possibilities

Palepu (1986), and Comment and Schwert (1995) use several accounting and
stock market performance measure to predict takeovers or the issuance of poison pills.
They are sales growth rate, liquidity, debt ratio, and market to book value and price to
earnings ratio. Since dual-class shares are a defense measure against takeovers, these

ratios will be helpful in explaining the issuance of dual-class shares.

1.4.2 Results of Logit Analysis

The logit regression is covers the years from 1990 to 1999. The basic results are

reported in Table 6-9.

13

Surprisingly, the size proxy (assets) generally is not statistically signiﬁcant when
assets, age, industry dummies and the standard deviation of monthly return are used as
predictors. When the ﬁnancial ratios such as sales growth, liquidity, debt ratio, market to
book ratio and P/E ratio are included in the regressions, assets becomes statistically
signiﬁcant and the sign is negative. However, this is mainly because many small ﬁrms
are dropped because of missing data on ﬁnancial ratios the year before or the average of 4
years before. The size of the ﬁrms after the small ﬁrms are dropped is comparable to
previous studies, and a similar result is generated: the assets factor is statistically
signiﬁcant and having large assets reduces the probability of having dual-class shares,
ceteris paribus. It does not affect the signiﬁcance of assets whether age is included in the
regression or not. The results are similar using the logarithm of assets

I also use equity, the market value of equity to replace assets in the regression.
Equity is statistically signiﬁcant and the sign is negative whether the ﬁnancial ratios are
included or not (the regressions are not reported here). The market value of equity might
be a better measure in the sense that it is directly related to the size of wealth to maintain
certain ownership levels.

Age of the ﬁrm is statistically signiﬁcant and the sign is negative. However, it
becomes only marginally signiﬁcant or not signiﬁcant when the ﬁve ﬁnancial ratios are
included in the regressions. The negative sign shows the young and immature ﬁrms are
more likely to have dual-class shares.

The industry effect is clear. The dummy publish is significant and positive in all
regressions, broadcast is statistically significant and positive in most regressions, mo-

picture is signiﬁcant and positive in all regressions. This is consistent with the notion that

14

tighter control is required to achieve the amenity potential. Another reason is that
personal preference is important in such industries in which it is hard to predict outcome
from managerial decisions. Managers are vulnerable and need tighter control and higher
ownership to protect them.

The utility dummy is statistically significant and negative in all regressions. The
ﬁnance dummy is statistically signiﬁcant and negative in most regressions. This indicates
the likelihood for systematically regulated industries to have dual-class shares is less than
that for other ﬁrms. It is likely that utility ﬁrms are affected more by regulation than are
ﬁnancial ﬁrms.

Firm-specific risk is statistically signiﬁcant in all the years except 1996 and
negatively related to issuing dual-class shares. The result is the same if another measure
of instability, total risk, is used in place of firm-speciﬁc risk. One would think this result
is different from what is expected from the literature on ownership concentration. Here,
however, I study the decision to have dual-class shares which is not equivalent to
ownership concentration. Investors usually would prefer transparency in managerial
decisions and the owner-management relationship. There is evidence that investors
regard dual-class ﬁrms as an indicator of low levels of transparency. Firms with high
instability would care more about transparency because investors require more
transparency for such volatile stocks. The investors also often believe dual-class ﬁrms
have much lower liquidity. Firms with high instability would also like to have high stock
liquidity because investors require these ﬁrms to have high liquidity. A firm with high
instability would prefer not to have dual-class shares although it tends to have a higher

ownership concentration on average. In addition to linearly estimating the probability to

15

have dual-class shares, I also estimate this relationship in nonlinear form by adding the
square of the instability measure. The squared value of the variables are positively related
to probability to have dual-class shares, indicating at lower values of these variables the
increase in the probability associated with given a decrease in instability diminishes.

The ﬁve ﬁnancial ratios, both value of the year before and the average of 4 years
before, are not statistically signiﬁcant in general. The merger and acquisition market in
the 19905 was very different from that in the 19805. Hence, dual-class shares may now be
less important as a defense against a takeover.

Insider ownership is closely related to dual-class shares. When insider ownership
is included in the regression, it is statistically signiﬁcant and positive as expected. High
insider ownership increases the probability of having dual-class shares. I also regress
insider ownership on dual, size, age, industry dummies and instability measures (not
reported here). As expected, dual is statistically signiﬁcant and positive, size is statistical
signiﬁcant and negative, age is statistical signiﬁcant and negative, the utility dummy is
statistically signiﬁcant and negative, m-picture dummy is statistically signiﬁcant and
positive. However, both the publish and broadcast dummies are not statistically
signiﬁcant. The instability measure is statistically signiﬁcant and positive which is

consistent with the control potential hypothesis.

1.5 Conclusion

In this paper I examine the propensity of ﬁrms to issue two classes of common

stock. I hypothesize that dual-class structures should be more common in situations

16

where a firm’s control potential is high. In addition, if the dual-class decision is driven
by managerial risk aversion and the fear of having a large undiversified economic stake
in a ﬁrm, I expect the dual—class structure to be more likely in large firms and ﬁrms with
high levels of volatility. Conversely, if agency problems are more severe in large and/or
risky ﬁrms, the dual-class structure may be relatively less likely in these ﬁrms owing to a
large penalty assessed by the outside market to the inferior voting shares.

To investigate these issues I identify the dual-class status of all Compustat listed
ﬁrms over the 1990-1999 time period. I ﬁnd that previous procedures to identify dual-
class ﬁrms are biased towards under-reporting the frequency of the dual-class structure,
and my sample reveals that approximately 6.67% of US. ﬁrms have adopted a dual class
structure. Consistent with my control potential hypothesis, in my logit models I ﬁnd that
firms in the media industry (e.g., publishing, broadcasting, motion pictures, etc.) are
signiﬁcantly more likely to have two classes of common stock. I also ﬁnd that younger
ﬁrms are more likely to have a dual-class structure. To the extent that managers of young
ﬁrms are more likely to be founders who care about maintaining control, this can be
viewed as additional evidence in support of the control motivation for selling two classes
of stock.

When I turn to the firm size and volatility measures my logit estimates suggest
that the likelihood of having a dual-class structure is unrelated to ﬁrm size and negatively
related to volatility. Thus, the diversiﬁcation motivation for adopting the dual-class
structure does not appear consistent with my data. However, the data do appear highly

consistent with the hypothesis that agency problems or concerns about expropriation in

17

highly volatile ﬁrms makes it difﬁcult to sell inferior voting common stock to outsiders in

these environments.

18

CHAPTER 2

Dual-Class Stock, Firm Value, and Performance

2.1 Introduction

One way to solve the agency problem is via the use of incentive compensation
and managerial ownership. Morck, Shleifer, and Vishny (1988) demonstrate empirically
that managerial ownership may raise ﬁrm value by solving agency problems. However,
their analysis indicates that there may be both costs and beneﬁts to high managerial
ownership. While high ownership may enhance a manager’s incentive to act in
shareholders interests, it may also entrench managers by giving them the ability to
insulate themselves from outSide market forces (e. g., unwanted acquisition attempts).

A dual class structure is an interesting way to keep control of the ﬁrm. There have
been mixed arguments and evidence in the ﬁnance literature on whether dual class stocks
are value enhancing or value decreasing. Harris and Raviv (1988), and Grossman and
Hart (1988) show that one share one vote is optimal although dual class stock can beneﬁt
the shareholder under certain conditions in takeovers.

On the other hand, Attari and Banerjee (1999) argue that if the ﬁrm requires
outside equity ﬁnancing to undertake the project, it will ﬁnd separation of the vote and
dividend claim optimal in cases. The separation increases the manager’s willingness to
undertake all positive NPV projects. Managers in single class ﬁrm will give up some

positive NPV projects requiring managers to issue equity to outsiders because they are

19

worried about losing control of the ﬁrm and being replace by outsiders. DeAngelo and
DeAngelo (1985) argue that a dual class structure can encourage managers to invest in
ﬁrm-speciﬁc human capital which adds value to the firm.

In this paper I conduct an analysis of the value of firms that choose to adopt a
dual-class share structure. While many existing studies have looked at the share price
levels and behavior of dual-class stocks, no study has convincingly examined the
relationship between dual class stock and ﬁrm value. No study has carefully examined
the effect of ownership of cashﬂow rights and ownership of voting rights separately.
Using a comprehensive sample of almost all publicly traded ﬁrms in the US. in 1995, I
conduct an analysis that describes whether dual class ﬁrms under perform and the effect
of dual class stock on ﬁrm value. My analysis also shows the relation between ﬁrm value
and ownership of cashﬂow rights and voting rights separately.

Following the work of Morck, Shleifer and Vishny (1988), I hypothesize that
managers in dual class ﬁrms can maintain control without worrying about being replaced
and having less cash ﬂow rights than voting rights, thus leading to sluggish management
and unwise investment. Dual class structure makes it more likely that an inefﬁcient
manager will keep control than in one share/one vote. One share/one vote aligns the
interests of managers more closely with shareholders. Thus I expect a lower Tobin’s Q
for a dual class firms compared with its industry peers.

On the other hand, according to the shareholder interest hypothesis, managers
may want some form of takeover defense for surplus-protection and surplus-extraction
which is in the interest of target shareholders. Perhaps a more common occurrence is that

a dual class structure creates stability that can be beneﬁcial. According to the under-

20

investment hypothesis developed by Attari and Banerjee (1999), the dual class structure
mitigates the under-investment problem when managers have to issue equity to ﬁnance
new projects. Thus, I expect a higher or similar Tobin’s Q for a dual class ﬁrm compared
with its industry peers if the shareholder interest hypothesis is true.

Overall, I ﬁnd that dual class ﬁrms do not have a lower Tobin’s Q than their peers
of sirrrilar size in the same industry. However dual class ﬁrms that have both classes
traded publicly have a lower Tobin’s Q than single class ﬁrms and Tobin’s Q increases as
insider ownership of cash ﬂow rights rises and declines as insider ownership of voting
rights rises in the sub sample of dual class ﬁrms that have both classes traded and single
class ﬁrms. On the other hand, dual class ﬁrms that have one class stock traded publicly
do not have a lower Tobin’s Q than single class ﬁrm.

The media industries are special in that the control potential is large as shown in
my ﬁrst essay. I ﬁnd within media industries, dual class ﬁrms do not have a lower
Tobin’s Q than single class ﬁrm while within non- media industries dual class ﬁrms have
a lower Tobin’s Q than single class ﬁrms. In media industries, dual class ﬁrms pay less
dividends than single class ﬁrms, the dividend payouts increases as insider ownership of
cash ﬂow rights rises and declines as insiders ownership of voting rights rises — there is
no such relation in the non-media industries.

There is no evidence that dual class firms underperform single class peers on
certain performance measures, such as Market-to-book equity ratio and Price/Earnings
ratio. There is some evidence that dual class ﬁrms underperform single class peers on

some other performance measures, such as ROA, EBIT, and operating cash ﬂow. On

these measures, ﬁrm performance rises with insider ownership of cash ﬂow rights and
decreases with insider ownership of voting rights.

The rest of the paper is organized as follows. In section 2 I discuss the related
literature and explain where my paper ﬁts in the context of these studies. In section 3 I
discuss my data and sample selection criteria. In section 4 I report my main results, while

section 5 concludes.

2.2 Existing Literature and Hypothesis Development

2.2.] Existing Literature

Many studies find that superior voting shares sell at a premium. They also
examine the time pattern and cross-sectional variation in price differential between the
superior voting shares and the inferior voting shares (e.g. Lease, McConnell, and
Mikkelson (1983, 1984)). In general this premium is consistent with votes being worth
more when the likelihood of a control contest is high (Zingales (1994, 1995))

Several existing studies have examined the issuance of dual-class stock. Many
authors study the wealth effect of dual class recapitalization in attempt to ﬁnd out the
impact of a dual class structure on ﬁrm value. Partch (1987) ﬁnds nonnegative abnormal
returns on the announcement of a recapitalization. Jarrel and Poulsen (1988) ﬁnd
signiﬁcant negative returns while Comett and Vetsuypen (1989) report positive abnormal
returnss. These results are inconclusive since announcement returns are likely to reﬂect

both the value consequences of adopting a dual-class structure and information

22

concerning the likelihood that a firm is acquired. Nevertheless, it is interesting that
several of these studies indicate a positive market reaction, which is suggestive that there
may be some beneﬁts to the dual-class structure.

In addition to these studies of stock price behavior, other studies have looked at
the ownership composition of firms that have two classes of stock (e.g. DeAngelo and
DeAngelo (1985), Partch (1987)). As one would expect, these studies indicate that dual-
class ﬁrms tend to have very concentrated inside ownership, and the percentage of votes
held by management is typically signiﬁcantly greater than the percentage of cash ﬂows

held by management.

2.2.2 Relationship to Existing Literature

While these existing papers present many interesting ﬁndings, they do not answer
the question as to whether a dual class structure destroys or enhances ﬁrm value and
performance. There is a well-known hypothesis that managerial control is inefﬁcient
since it leads to entrenchment. However, a ﬁrm’s decision to have dual class stock may
not be just to the interest of insiders. Presumably there are some economic tradeoffs that
take place in the decision to adopt the dual-class structure. In particular, I would expect
there to be a beneﬁt to outsiders and to the ﬁrm as a whole to have dual class stock, thus
firm value will not simply decline because of manageerial entrenchment.

In this paper I try to answer this question by identifying the separation of cash

ﬂow rights and voting rights of dual class ﬁrms. I then match dual class firms with their

industry peers to compare firm value and performance. I run regressions to examine the

 

8 Also see Chang and Mayers (1992) and Shum. Davidson and Glascock (1995).

23

cross sectional variation of firm value and performance with different degrees of
separation of cash ﬂow rights and voting rights. In addition to understanding the relation
between ﬁrm value and dual class stocks, I am able to address related interesting issues. I
am able to assess for the ﬁrst time the overall level of insider ownership in dual classes
and the degree of the separation of cash ﬂow rights and voting rights of dual class ﬁrms
in the United States. This allows me to assess the economic importance of this type of

corporate control in the modern US. business environment.

2.2.3 Hypothesis Development

In explaining the valuation and performance of the dual-class ﬁrms compared
with single class ﬁrms, I must identify the potential costs and beneﬁts of this structure. In
general there is trade off between the cost and the beneﬁts.

According to Morck, Shleifer, and Vishny (1988), the managerial entrenchment
hypothesis claims that managers in dual class ﬁrms can maintain control without
worrying about being replaced, thus leading to sluggish management and unwise
investment. The dual class structure protects inefﬁcient incumbent managers from a
control contest. A dual class structure makes it more likely that an inefﬁcient manager
will keep control than in one share/one vote. One share/one vote aligns the interest of
managers more closely with shareholders. If there is no beneﬁt of control, there is
substantial cost of dual class shares because outside investors would pay more for the
equity issued by a single-class ﬁrm than for the equity issued by a dual-class ﬁrm. Thus,

the managerial entrenchment hypothesis predicts a lower Tobin’s Q for a dual class ﬁrm

24

compared with its industry peers. Dual class firms could also be outperformed by single
class ﬁrms in other operating performance measures. The managerial entrenchment
hypothesis also predicts that Tobin’s Q declines with insider ownership of voting rights
while increases with insider cashﬂow rights.

On the other hand, according to the shareholder interest hypothesis, managers
may want some form of takeover defense for surplus-protection and surplus-extraction
which is in the interest of target shareholders. Perhaps more important, a dual class
structure creates stability which can be beneﬁcial. First, according to the
underinvestment hypothesis developed by Attari and Banerjee (1999), the dual class
structure mitigates the under-investment problem when managers have to issue equity to
ﬁnance new projects. Managers in a single class ﬁrm will give up the positive NPV
projects because the equity issue will dilute their equity ownership thus gradually
reducing their control. However, if a ﬁrm issues a restricted voting class of equity to
outside investors then managers can undertake all positive NPV projects without dilution
of voting control. Second, DeAngelo and DeAngelo (1985), Fischel (1987) and Denis
and Denis (1994) argue that the reduced threat of displacement provides managers with
incentives to invest in firm-speciﬁc human capital. Thus, the shareholder interest
hypothesis predicts a higher or similar Tobin’s Q for a dual class ﬁrm compared with its
industry peers. Dual class ﬁrms also should not be outperformed by single class ﬁrms in
other Operating performance measures. Unlike the managerial entrenchment hypothesis,
the shareholder interest hypothesis does not predict that Tobin’s Q declines with insider

ownership of voting rights while increases with insider cashﬂow rights.

25

2.3 Data

My goal is to identify all dual—class ﬁrms and their ownership chosen from a large

9. To accomplish this objective, I initially identify the set of

set of publicly traded ﬁrms
all ﬁrms from 1990-1999 listed on the Compustat tapes. For the resulting sample, I search
the company name ﬁeld in the Compustat database and identify all ﬁrms with a share
class indicator included in the name (e.g., AEL Industries —CL A.) Next, I turn to the
CRSP tapes and identify all cases where the ﬁrm has two classes of common stock listed
(e.g., AARON Rents Inc. has share classes of A and B.) The above procedure is
incomplete in that it relies exclusively on the CRSP and Compustat methods for reporting
share classes and company names. It is not clear that the CRSP and Compustat procedure
outlined above will identify all of these cases. To identify other instances of dual-class
shares, I exploit the fact that CRSP typically reports a ﬁrm’s shares outstanding for a
single-class of common stock, while Compustat reports the number of shares outstanding
for all classes of common stock added together. I identify every instance where these
shares ﬁgures differ by more than 5%.'0 I then search through company ﬁnancial
statements, proxy statements, and the Dow Jones Interactive database to investigate

whether the firm has multiple classes of common stock. The above procedures yield a

total sample of 391 dual class ﬁrms out of 8286 ﬁrms in year 1995.

 

9 See more details in the ﬁrst essay.

'0 My choice of a 5% rule is arbitrary, but probably rather conservative. A51 report below, most ﬁrms 1
identify as dual-class by this procedure have a much larger difference in the shares outstanding ﬁgure
between Compustat and CRSP. Note that if I do miss a few ﬁrms, given that they were not ﬂagged by my
5% rule, it is almost surely the case that one of the classes has very few shares outstanding and. therefore, is
likely to be economically unimportant.

26

Given my sampling procedure, I believe that I have the most comprehensive
database of dual-class listed ﬁrms ever assembled. I then search through the proxy
statements, 10K ﬁlings, annual reports and Dow Jones Interactive to identify the number
of votes per share for each class, number of shares outstanding for each class, the level of
insider ownership, and CEO ownership and block holdings in each class of dual class
ﬁrms in 1995. Insider ownership includes ownership of ﬁrm executives and board
directors. Block holdings refers to any block of ownership more than 5% of the class. I
use ownership data from Compact Disclosure for insider ownership of single class ﬁrm in
1995. According to Anderson and Lee (1997), this is the most comprehensive database
on inside ownership that is widely available. I also identify whether the dual class ﬁrm
trade both classes publicly or only trade one class publicly. The exchanges where the
shares traded are also recorded.

In Table 1 I report summary deﬁnitions of all the variables I use in the subsequent
analysis. In Table 2 I present univariate comparisons of some of the variables of interest
for the single-class ﬁrms compared to the dual-class ﬁrms. I report these results the year
1995. While most of these variables are fairly standard from the literature and are
derived in a straightforward manner from the CRSP and Compustat tapes, a few variables
are less standard and should be discussed here. Ownsuper is the insider ownership in the
superior voting class and Owninfer is the insider ownership in the inferior voting shares.
Sharesuper is the number of shares outstanding in superior voting class and Shareinfer is
the number of shares outstanding in the inferior voting class. Votesuper is the votes per
share for the superior voting class and Voteinfer is the votes per share for the inferior

voting class. The Voting Rights is insider ownership of voting rights of common stocks.

27

The insiders’ voting rights are measured as a percentage of the total number of
outstanding votes.

Ownsmger’“Sharesuper*Votesrmer+Owninfer*Shareinfer*Voteinfer

VotingRights =
Sharesuper*Votesuper+Shareinfer*Voteinfer

The Cashﬂow rights is insider ownership of common stocks based on monetary value and
I calculate the insiders’ implied common stock cash interest as the weighted average of
their percentage holdings in the superior and inferior voting classes:

CashﬂowRights = Ownsuper*Shagsuper+0wninfer*Shareinfer
Sharesuper+Shareinfer

The variable SEPARATION is the separation index of ownership of voting rights and
cashﬂow rights and SEPARATION = Voting Rights/Cashﬂow Rights. STDF is the
standard error of the monthly stock market rates of return in the previous four years
estimated from a market model in which the ﬁrm’s monthly return in the previous four
years is regressed on the average monthly return on a value-weighted market portfolio.
The variable Age is a proxy for the firm’s true age and is measured by identifying the ﬁrst
year that the ﬁrm’s stock has return data on the CRSP tape.

As the t-tests in Table 2 indicate, dual class listed ﬁrms appear to differ
signiﬁcantly from the single-class listed ﬁrms on many dimensions including the market-
to-book equity ratio, sales growth, liquidity, the P/E ratio, and some of the measures of
volatility. As I report in Table 3 and also in the ﬁrst essay, the percentage of dual-class
ﬁrms is particularly high in the publishing, broadcasting and motion picture industry, and

it is particularly low for utilities and ﬁnancial ﬁrms.

28

2.4 Analysis and Results

2.4.] The Separation of cash ﬂow rights and voting rights

Table 4A document insider ownership in each of the two classes. ( The table does
not include the ﬁrms with more than two classes of common stocks.) Table 4B reports
the insiders’ voting rights and cash ﬂow rights from common stocks. Insiders include
executive ofﬁcers and board directors. Ownership includes shares owned by ofﬁcers and
directors, trusts for their benefactors and foundations or corporations they control. The
level of insider ownership in each class is hand collected from proxy statements, 10K
ﬁlings and annual reports.

Insiders in dual class ﬁrms hold a substantially higher percentage of the stock
with more votes than of the stocks with fewer votes. In Table 4A the mean insider
ownership in the superior voting class is 63.73% and the mean insider ownership in the
inferior voting class is 21.65%. The t-statistic for equal means is 20.81. Insider ownership
in the superior voting shares is substantially greater than that of inferior voting shares.
The insiders hold a median of 71.23% in the superior voting class while they hold a
median of 14.60% in the inferior class. The Wilcox sign-rank test for equal medians is
13.51. Both the mean insider ownership-and median insider ownership are signiﬁcantly
higher in the superior voting class than in the inferior voting class. In Table 4.B., the
insiders hold a mean of 34.65% of cash ﬂow rights from common stock and a mean of
54.80% of voting rights. The t-statistic for equal means is 17.92. The insiders hold a

median of 32.25% of cash ﬂow rights from common stock and a median of 60.58% of

29

voting rights. The Wilcox sign-rank test for equal medians is 12.59. Both the mean and
median insider cash ﬂow rights are significantly lower than those of voting rights. The
mean SEPARATION Index is 1.875 and the median SEPARATION Index is 1.500.

Table 4C shows the holdings of CEOs in the two classes. The mean CEO
ownership in the superior voting class is 34.75% and the mean CEOs’ ownership in the
inferior voting class is 9.94%. The t-statistic is 13.53. The CEOs hold a median of
25.05% in the superior voting class while they only hold a median of 2.2% in the inferior
class. The Wilcox sign-rank test is 11.28. Both mean CEO ownership and median CEO
ownership are signiﬁcantly higher in the superior voting class than in the inferior voting
class. Table 4F shows the holdings of the largest block shareholder in the two classes.
The mean largest blockholder’s ownership in the superior voting class is 60.13% and the
mean largest blockholder’s ownership in the inferior voting class is 27.25%. The largest
blockholder holds a median of 56.67% in the superior voting class while they only hold a
median of 19.29% in the inferior class.

The voting rights and cash rights of insiders in firms in the media industry are
compared with those in non-media industries in Table 4D. Insiders hold similar mean and
median ownership of voting rights and cash ﬂow rights in media industries and non-
media industries. There is no signiﬁcant difference found. The separation of voting rights
and cash ﬂow rights are similar across industries with different likelihood to issue dual
class stock (or different control potential.)

Table 4B reports the voting rights and cash ﬂow rights in ﬁrms that have both
classes publicly traded and have only one class traded publicly. In both sub-samples the

insiders in dual class firms hold substantially higher levels of ownership of voting rights

30

compared to cashﬂow rights. The insiders in ﬁrms with one class traded publicly have a
mean ownership of voting rights of 57.0%, signiﬁcantly higher than their mean
ownership of cashﬂow rights of 34.3%. The insiders in ﬁrms with one class traded
publicly have a median ownership of voting rights of 63.8%, signiﬁcantly higher than
their median ownership of cashﬂow rights of 31.4%. The insiders in ﬁrms with both
classes publicly traded have a mean ownership of voting rights of 49.2%, signiﬁcantly
higher than their mean ownership of cashﬂow rights of 33.0%. The insiders in ﬁrms with
both classes traded publicly have a median ownership of voting rights of 52.2%,
signiﬁcantly higher than their median ownership of cashﬂow rights of 32.6%.

A more interesting issue is whether dual class ﬁrms with both classes traded
publicly have less separation of voting rights and cash ﬂow rights since the voting
preferred stocks can be purchased in the market by outsiders. The insiders in ﬁrms with
one class traded publicly have a mean ownership of voting rights of 57.0%, signiﬁcantly
higher than a mean of 49.2% for firms with both classes traded publicly, t-statistic =
1.963. The insiders in ﬁrms with one class publicly traded have a median ownership of
voting rights of 63.8%, signiﬁcantly higher than a median of 52.2% for ﬁrms with both
classes traded publicly, Wilcox sign-rank test = 1.963. Insiders in the two types of dual
class ﬁrms have about same mean and median ownership of cash ﬂow rights. The
insiders in ﬁrms with one class traded publicly have a mean ownership of cashﬂow rights
of 34.3% and a median of 31.4%. The insiders in ﬁrms with both classes traded publicly
have a mean ownership of cashﬂow rights of 33.0% and a median of 32.6%. By letting
both classes publicly trade the dual class ﬁrms do have a little less control. However,

these dual class ﬁrms still hold a median of 52.2% voting rights compared to a median of

31

32.58% of cash ﬂow rights. Letting the superior voting shares trade publicly does not
signiﬁcantly reduce insider control of the ﬁrm. Insiders still hold a large amount of those

superior voting shares, although they appear to sell out a little portion.

2.4.2 Matching of dual class ﬁrms with single class ﬁrms

My major goal is to compare the value and performance of dual class firms with
single class ﬁrms. It is well established that corporate governance affects ﬁrm value.
Gompers, Ishii and Metrick (2001) show corporate governance is highly correlated with
ﬁrm value and stock returns. The authors excluded dual class ﬁrms from their sample. In
this section I use a matching approach to study whether the variation in a ﬁrm’s decision
to issue dual class stock is related to cross sectional differences in ﬁrm value and
performance. As is well known, industry and size are the majOr factors in ﬁrm value and
performance. I match each dual class ﬁrm with a ﬁrm closest in size (total assets) in the
same industry. I use the two-digit sic code as the proxy for industry. Table 5 documents
the dual class ﬁrms’ value and performance measure and those of matching ﬁrms.

My valuation measure is Tobin’s Q, which has been used routinely for this
purpose in corporate governance studies. I follow Kaplan and Zingales (1997) and
compute Q as the market value of assets divided by the book value of assets, where the
market value of assets is computed as book value of assets plus the market value of
common stock less the sum of the book value of common stock and balance sheet
deferred taxes. Perfect and Wiles (1994) and Chung and Pruitt (1994) show that the

improvements obtained from a more involved computation of Q are fairly limited.

32

I also include market-to-book equity ratio in the comparison. Operating
performance measures include the Price/Earning ratio, Return on assets, cash ﬂow
(Operating Cash Flow/Total Assets), EBIT (EBIT/Total Assets), and the Dividend Ratio
(Cash Dividends/Total Equity).

Table 5 shows that overall there is no strong signiﬁcant difference between dual
class ﬁrms’ value and performance and those of matching single class ﬁrms, although
dual class ﬁrm have a lower median Tobin’s Q in the next year (1996) and a lower
Market-to-Book equity ratio than single class matching ﬁrms. The dual class ﬁrms
perform as well as single class ﬁrms. This is not consistent with the managerial
entrenchment hypothesis. Insiders do not use a dual class structure to entrench inefﬁcient
management.

Table 6A reports the performance of the sub-sample of dual class ﬁrms that have
both classes publicly traded compared to the matching ﬁrms. Table 6B reports the
performance of the sub-sample of dual class ﬁrms that have only one class traded
publicly compared with the matching ﬁrms. Surprisingly, I ﬁnd that dual class ﬁrms with
both classes traded publicly actually under-perform their single-class peers and dual class
ﬁrms that have only one class traded perform as well as their single-class peers. Dual
class ﬁrms with both classes traded have a mean Tobin’s Q of 1.454. This is signiﬁcantly
lower than the matching firms’ mean of 1.813 (t statistics: -1.650). Dual class ﬁrms with
both classes traded have a median Tobin’s Q of 1.146 significant lower than matching
ﬁrms’ median of 1.396 (statistically signiﬁcant at 1% level). Dual class ﬁrms with both
classes traded publicly also have a signiﬁcantly lower median Tobin’s q in the next year

( 1996). Besides Tobin’s q, dual class ﬁrms with both classes traded publicly also have a

33

signiﬁcantly lower median Market-to-Book equity ratio, a signiﬁcantly lower median
investment level and a significantly lower mean dividend payout ratio.

As shown in the previous section, dual class ﬁrms with both classes traded
publicly have a lower degree of separation of voting rights and cash ﬂow rights than dual
class ﬁrms that have only one class traded publicly. If the managerial entrenchment
hypothesis were true, we would expect dual class ﬁrms that have both classes traded
publicly to perform better than dual—class ﬁrms that have only one class traded relative to
their matching single class ﬁrms. My results show that the higher degree of separation of
ownership of voting rights and cashﬂow rights does not destroy ﬁrm value. This is not
consistent with the managerial entrenchment hypothesis or there is something more than
entrenchment.

One potential benefit, initially identiﬁed by Demsetz and Lehn (1985), is that
managers may experience a large utility payoff from controlling firms with high control
potential or amenity potential and/or for being identiﬁed as the controlling party in
certain high-proﬁle firms, for example media ﬁrms and sports ﬁrms. This line of
reasoning predicts that I should generally observe strong industry trends in the use of
dual-class shares, and specifically I should observe many dual-class structures for ﬁrms
in the media industry. My findings in the ﬁrst essay are consistent with these predictions.
If control is also beneﬁcial to outside shareholders, a further prediction is that dual class
ﬁrms in media industries should not have a lower value than single class ﬁrms while dual
class ﬁrms in other industries may have a lower value. Table 7A reports the performance
of the sub-sample of dual class ﬁrms in media industries, compared with the matching

firms. Table 7B reports the performance of the sub-sample of dual class ﬁrms in non-

34

media industries compared with the matching ﬁrms. Media industries includes SIC code
from 2700 to 2799 (publishing), from 4800 to 4899 (broadcasting) and 7800 to 7899
(motion pictures). All other industries (SIC code from 0100 to 9999) are defined as non-
media industries. 1 ﬁnd that dual class ﬁrms in media industries actually do not under-
perform their single-class peers. There is some evidence that dual class ﬁrms in non-
media industries do not perform as well as their single-class peers. Dual class ﬁrms in
non-media industries have a next-year median Tobin’s Q of 1.356. This is statistically
signiﬁcantly lower than matching ﬁrms’ median of 1.447. Dual class ﬁrms in non-media
industries do not have Price-Earning ratios signiﬁcantly different from their single class
peers. Dual class ﬁrms in the media industries have a median Price-Eaming ratio of
17.75. This is signiﬁcantly higher than the matching ﬁrms’ median of 0 (statistically
signiﬁcant at the 5% level). Dual class ﬁrms in the media industries have a mean Price-
Eamings ratio of 38.06. This is signiﬁcantly higher than matching ﬁrms’ mean of 13.83
(t-statistic = 1.819). Dual class ﬁrms in media industries also have a signiﬁcantly higher
mean and median a-year-forward Price-Eaming ratio than matching ﬁrms while non-
media dual class ﬁrms do not. Dual class firms in media industries have a signiﬁcantly
higher mean EBIT/Total Assets ratio than matching firms while dual class ﬁrms in non-
media industries do not have an EBIT/Total Assets ratio signiﬁcantly different from the
matching ﬁrms. Dual class ﬁrms in media industries also have a signiﬁcantly higher
mean Operating Cashﬂow/Assets ratio in the following year (1996) than matching ﬁrms
while dual class ﬁrms in non-media industries do not have 3 Operating Cashﬂow/Assets

ratio signiﬁcantly different from the matching ﬁrms.

35

If the managerial entrenchment hypothesis were true, we would expect that dual
class ﬁrms in both media industries and non-media industries under-perform their
matching single class counterparts. My results show that the dual class ﬁrms in media
industries do not under—perform the matching single class firms while the dual class ﬁrms
in non-media industries somewhat under-perform their peers. This is not consistent with
the managerial entrenchment hypothesis. The control obtained by the dual class structure

is not harmful to firm value in industries where the control potential is high.

2.4.3 Cross sectional regression of Tobin ’s Q

2.4.3.1 Empirical regression models

I use several regression models to study the effect of a dual class structure and

separation of voting rights and cash ﬂow rights on ﬁrm value. I ﬁrst estimate
Q=a+bDUAL+cW+e (1)

where DUAL is a dummy variable equal to 1 for dual class ﬁrms and 0 for single class
ﬁrms. W, is a vector of ﬁrm characteristics. As elements of Wi, I follow Gompers, et al.,
(2001) and Shin and Stulz (2000) and use the log of the book value of assets, the log of
ﬁrm age and a dummy variable for each of the four~digit SIC codes. The book value of
assets, age, and two-digit SIC code are also used as a robustness check. It is obvious that
industry, size, and age are critical control variables in regression of Q. My empirical

results below show that stock volatility is also an important control variable.

36

Regression with the dummy variable DUAL can give an estimate of the effect of a
dual class structure on ﬁrm value. More speciﬁcally I can estimate the effect of

separation of voting rights and cash flow rights on Q. So I estimate

Q=a+bDUAL+cCASH+dW+e (2)
and

Q=a+bVOTE+cCASH+dW+e (3)
and

Q=a+bSEPARATION+CW+e (4)

Where CASH is the cash ﬂow rights of insiders and VOTE is the voting rights of
insiders. Separation index equals VOTE divided by CASH.
A piecewise regression may be desirable as in Morck, Shelifer and Vishny (1988).
Q may not be a monotonic function of insider ownership of voting rights and cash ﬂow
rights, but a piece wise function of insider ownership. I use the following variable to
estimate and report my piece wise linear regressions:
Vote.0tolO = insider ownership of voting rights if insiders ownership < 0.10
=O.lO if insiders ownership of voting rights if insiders ownership 2 O. 10
Vote.0VERlO= 0 if insider ownership < 0.10
= insider ownership of voting rights minus .10 if insider ownership of
voting rights 2 0.10
Cash.0to75 = insider ownership of cash ﬂow rights if insider ownership of cash ﬂow
rights < 0.75
= 0.75 if insider ownership of cash flow rights if insider ownership of cash

ﬂow rights 2 0.75

37

Cash.OVER75= 0 if insider ownership of cash ﬂow rights < 0.75
= insider ownership of cash ﬂow rights minus .75 if insider ownership of
cash ﬂow rights 2 0.75

The theoretical justiﬁcation for the piecewise regressions and these particular
turning points is not very strong. Morck, Shleifer and Vishny (1988) use 5% and 25% as
turning points for board ownership which combines ownership of voting rights and cash
ﬂow rights. My analysis is different from Morck, Shleifer and Vishny (1988) in that a
dual class structure allows me to separate the insider ownership of voting rights from the
insider ownership of cash ﬂow rights. It is not well justiﬁed that there is a 5% and 25%
turning point for insider ownership of cash ﬂow rights. I use a turning point of 75% for
ownership of cash ﬂow rights in attempt to describe the diminishing incentives from the
ownership of cash flow rights when the ownership is high enough. I use 10% as a turning
point for ownership of voting rights since an ownership of voting rights less than 10%
does not confer signiﬁcant power while an ownership of voting rights higher than 10%
would give insiders signiﬁcant power. For regressions using other turning points (not
reported here) I deﬁne the variables analogously. The results are either similar or hard to
justify.

I estimate cross sections of (1) - (4) using data of 1995. I also use this procedure
when studying other performance measures in the next a couple of sections. 1 hand
collect the 1995 ownership data of dual class ﬁrms from proxy statements, 10 K ﬁlings,
and annual reports. The single class ﬁrms’ ownership data is from Compact Disclosure

which is very reliable for single class ﬁrms according to Anderson and Scott (1997).

38

2.4.3.2 Regression of ﬁll] sample

Table 7 summarizes the regression results based on the above regression models
for the full sample. All but one regression reports no statistically signiﬁcant relationship
between DUAL dummy (equals one if a dual class ﬁrm, zero otherwise) and Q, between
insider ownership of voting rights and Q or between voting rights-cashﬂow rights
SEPERATION Index and Q. Hence, there is no convincing evidence that a dual class
structure diminishes ﬁrm value. This is not consistent with the managerial entrenchment
hypothesis.

I also do a robustness check. To deal with the possibility that a variety of factors
can jointly affect board ownership and Q, and thus induce a spurious correlation between
them, I control for additional variables in the regression. Besides size, age, and industry
effect, R&D expenditure and advertising expenditure may be measures of intangible
assets that affect Q. Consequently, I include RD/ASSETS (R&D Expenses divided by
Total Assets) and ADV/ASSETS (Advertising Expenses divided by Total Assets) in the
regression. The results are similar if I exclude ﬁnancial ﬁrms with SIC code in the range
of 6000-6999 and utility ﬁrms with SIC code in the range of 4900-4999, which is the
practice in many corporate ﬁnance studies.

I regress on a subsample of the larger firms, the results, not reported here, are
similar to Morck, Shleifer and Vishny (1988). Their sample is composed of Standard &
Poor 500 ﬁrms which are the largest ﬁrms. So my full sample results are different from

their results because my sample contains all the smaller ﬁrms in Compustat.

39

2.4.3.3 Dual class ﬁrms with both classes traded vs. dual class ﬁrms with only one class

traded

Table 9 summarizes the results for regressions of the subsample with dual class
ﬁrms that have both classes traded publicly and the dual class ﬁrms that have one class
traded publicly. I also include a stock volatility measure, RD/ASSETS and/or
ADV/ASSETS in the regression. The results are similar if I exclude ﬁnancial ﬁrms with
SIC code in the range of 6000-6999 and utility ﬁrms with SIC code in the range of 4900-
4999. The ﬁrst couple of columns show that dual class ﬁrms with both classes traded
publicly have a lower q value. In the third and fourth column, the DUAL dummy (1 if
dual class ﬁrms, 0 otherwise) is signiﬁcantly negative. More striking, given the
regression equation (3), the sixth column show that the insider ownership of voting rights
is negatively related to Q, while the insider ownership of cashﬂow rights is positively
related to Q. For a dual—class ﬁrm that has both classes traded publicly, everything else
held constant, a 10 percentage point increase in inside ownership of cash ﬂow rights will
increase Tobin’s Q by 0.73, while a 10 percentage point increase in inside ownership of
voting rights will decrease Tobin’s Q by .58. The coefﬁcients are statistically and
economically signiﬁcant.

Table 9 also shows that dual class firms with one class traded publicly do not
have a lower Tobin’s Q. There is an interesting question as to why dual class ﬁrms with
both classes traded publicly are valued lower. The difference between the two types of
dual class ﬁrms is that the ﬁrms that have both classes traded publicly, insiders can sell

their superior voting shares in the market. Insiders in the ﬁrms with one class traded

40

publicly cannot sell their shares easily. As these firms’ proxy statements show, the shares
of the privately held class usually are not transferable or will be converted to shares of a
publicly traded class upon transfer, or are transferable upon permission of the board.
Since they may not be able to bail out by selling shares if the ﬁrms are in ﬁnancial
trouble, the insiders in these ﬁrms have a commitment to the success of the ﬁrm. So there
is a signal of interest alignment by holding a class of stocks that cannot be traded
publicly.

As is well known, there is a difference between prices of the two classes of stock
when both are traded. When dual class ﬁrms have both classes traded publicly, the shares
with more votes usually are traded at a higher price than the shares with fewer votes. The
relative premium investors are willing to pay for additional votes is called a voting
premium. As documented by Zingales (1995), the voting premium on the voting-
preferred class of stock on average is 10.5% (the median is 3%). However, there are other
differences between the two classes that affect the relative price of shares between the
two classes, such as difference in dividends, conversion rights and liquidity. Following
Zingales (1995), the nominal voting premium is given by:

Wu = Bo + 5101mm +BzDIVrr + BsCONVrr + B4SIZEir + BSLIQUIDITYR + 8n (5)
where VP is the nominal voting premium. ((1)/1t) proxies for the voting power and B]
represents the real voting premium. DIV is a dummy variable equals one for those ﬁrms
in which inferior voting shares received a larger dividend. CONV is a dummy variable
that takes a value of one for ﬁrms in which the superior voting shares are convertible into
inferior voting shares at the holder’s will. SIZE is the total market capitalization of stock.

LIQUIDITY is the relative stock liquidity difference of the two classes. If equation (6)

41

were used to price the superior voting shares not traded publicly, I would expect a
negative or non-positive nominal premium. Because there is a large liquidity premium for
voting inferior shares over voting superior shares if the liquidity of shares that cannot be
transferred easily is assumed to be close to zero. Therefore, insiders in dual class ﬁrms
with one class traded publicly do sacriﬁce the value of shares they hold for keep the
voting control.

This also leads me to suspect that the motivation for dual class firms to let the
superior voting shares trade publicly is to increase the liquidity of the shares insiders
previously held privately thus increasing the value of their holdings and diversifying the

risk they face.

2.4.3.4 Dual class ﬁrms in media industries vs. dual class ﬁrm in non-media industries

Table 10 summarizes the results for the sub-sample of ﬁrms in media industries
and the sub-sample of ﬁrms in non-media industries. The ﬁrst few columns show that
dual class ﬁrms in media industries do not have lower q values than their peers. All
regressions report no statistically signiﬁcant relationship between the DUAL dummy and
Q, between insider ownership of voting rights and Q, or between voting rights-cashﬂow
rights SEPERATION Index and Q. Hence, there is no convincible evidence that a dual
class structure harms a media ﬁrm’s value. Next, the regressions of the sub-sample of
dual class ﬁrms in the non-media industries show that the non-media dual class ﬁrms
have lower Tobin’s Q values than their single class peers. All else equal, a dual class ﬁrm

in a non-media industry has a Tobin’s Q which is 0.633 lower than that of a single class

42

firm. However, no statistically signiﬁcant relationship between Tobin’s Q and insider
ownership of voting rights is reported. The results are similar if I exclude ﬁnancial ﬁrm
with SIC code in the range of 6000-6999 and utility ﬁrms with SIC codes in the range of
4900.

One potential beneﬁt, initially identiﬁed by Demsetz and Lehn (1985), is that
managers may experience a large utility payoff from being identiﬁed as the controlling
party in certain high-proﬁle ﬁrms, for example media ﬁrms and sports ﬁrms. This line of
reasoning predicts that I should generally observe strong industry trends in the use of
dual-class shares, and speciﬁcally I should observe many dual-class structures for ﬁrms
in the media industry. My findings in the ﬁrst essay are consistent with these predictions.
The results in this section show that the dual class ﬁrms in the industries with the greatest
control potential actually have the same value as single class ﬁrms in the same industries.
One possibility is that there is not a traditional managerial entrenchment problem in
media industries. However the results also show that there is a managerial entrenchment
problem in non-media ﬁrms. Therefore, a more plausible explanation is that the control
by insiders has certain beneﬁts to the ﬁrm as a whole. In media industries, the product,
business, and management are highly personalized. The business decisions frequently are
determined by the personal tastes of management. For example, ﬁlmmaking, book and
newspaper publishing and broadcast networks, depend more on personal style instead of
mechanical operation. Personal creativity is key to success. Since it is hard to evaluate
such stylized management, business and managerial decisions are challenged more
frequently than in other industries. Without certain control, personalized or stylized

business and management will not be able to be executed by insiders. The media ﬁrms

43

need to have constant stylized management and business strategy to survive. The dual
class structure or high ownership of voting rights provides a relatively “stabilized” space
for the managers to deliver and talented artists, ﬁlmmakers, and broadcasters to perform.
This effect can be called managerial stabilization. The overall effect of management
entrenchment and managerial stabilization on a media ﬁrm’s value resulting from dual

class stock can be insigniﬁcant because of a counteracting effect.

2.4.4 Other Performance Measures

Following similar reasoning, I also run regressions to determine if dual class ﬁrms
underperform single class ﬁrms on other performance measures. The empirical
speciﬁcation is similar to equations (1) — (4).

An alternative measure of performance is market-to-book equity ratio, market
value of common stock divided by book value of common stock. The market-to book
ratio is not as informative as Tobin’s Q. However, it still may be a useful measure of
ﬁrm performance. Table 11 summarizes the regression results. There is no evidence in
the regression of the full sample that dual class firms underperform their single class
peers on market-to book equity ratio. The regression on the sub-sample of dual class
firms that have two classes traded publicly and sub sample of ﬁrms in media industries
yield same results. Overall, the results are inconsistent with the managerial entrenchment

hypothesis.

Table 12 summarizes the regression results of Price/Earning ratio. The regression
results in all of the samples do not provide any evidence of underperforrnance of dual
class ﬁrms. Note the R2 is very low for all of the regressions.

Table 13 documents results of the regression of return on assets (ROA) in the full
sample and sub-sample of dual class firms only. There is some evidence that dual class
ﬁrms underperform the single class peers as measured by ROA. The coefficients of
insider ownership of cash ﬂow rights are positive and significant. ROA increases with
insider ownership of cash ﬂow rights. There is some evidence that ROA decreases with
insider ownership of voting rights. In the regression with a piecewise speciﬁcation for
insider ownership of voting rights, ROA rises sharply as insider ownership increase from
zero to 5%, RCA then decreases rapidly as insider ownership increases from 5% to 25%.
There is some evidence that ROA decreases less rapidly as insider ownership of voting
rights increases above 25%. This is consistent with some studies in that a low insider
ownership of voting control of 0—5% is beneﬁcial to ﬁrm performance.

Table 14 demonstrates the results of the regression of EBIT/ Assets in the full
sample and the sub-sample of dual class ﬁrms only. There is signiﬁcant evidence that
dual class ﬁrms underperform their single class peers on the aspect of EBIT/Assets. The
coefﬁcients of insider ownership of cash ﬂow rights are positive and significant.
EBIT/Assets increases with insider ownership of cash ﬂow rights. There is also evidence
that EBIT/Assets decreases with insider ownership of voting rights. In the piecewise
regression for insider ownership of voting rights, EBIT/Assets rises rapidly as insider

ownership increases from zero to 5%, EBIT/Assets then decreases less rapidly as insider

45

ownership increases from 5% to 25%. There is some evidence that EBIT/assets decreases
less rapidly as insider ownership of voting rights increases above 25%.

Table 15 demonstrates the results from the regression of Operating Cash Flow/
Assets in the full sample and the sub-sample of dual class firms only. There is signiﬁcant
evidence that dual class firms underperform their single class peers on the aspect of Cash
Flow/Assets. The coefﬁcient of insider ownership of cash ﬂow rights is positive and
signiﬁcant. Cash Flow/Assets increases with insider ownership of cash flow rights. There
is also evidence that Cash Flow/Assets decreases with insider ownership of voting rights.
In the regression with a piece wise speciﬁcation for insider ownership of voting rights,
Cash Flow/Assets rises rapidly as insider ownership increases from zero to 5%, Cash
ﬂow/Assets then decreases less rapidly as insider ownership increases from 5% to 25%.
There is some evidence that Cash Flow/assets decreases less rapidly as insider ownership
of voting rights increases above 25%.

In summary, there is no evidence that dual class ﬁrms underperform single class
peers on certain performance measures, such as Market-to-book equity ratio and
Price/Earnings ratio. There some evidence that dual class ﬁrms underperform their single
class peers on some other performance measures, such as ROA, EBIT, and operating cash
ﬂow. Of these measures, ﬁrm performance rises with insider ownership of cash ﬂow

rights and decreases with insider ownership of voting rights.

2.4.5 Dividend policy in dual class ﬁrms

46

Table 7A shows that dual class ﬁrms in media industries pay less dividends than
the matching single class ﬁrms while dual class ﬁrms in non-media industries do not pay
less dividends. I use various regression models to study the factors that affect ﬁrm
dividend payouts. I ﬁrst estimate

DIVIDEND/CAPITAL = a + b DUAL +c W + e (6)

where DUAL is a dummy variable that equals 1 for dual class ﬁrms and 0 for single class
ﬁrms. Wi is a vector of ﬁrm characteristics. As elements of W, I use the log of the book
value of assets, the log of ﬁrm age, Operating Cashﬂow/Assets, Tobin’s Q and dummy
variables for each of the two-digit SIC codes. EBIT/Assets, Net Income/Assets,
Liquidity, Debt ratio, the book value of assets, and age are also used as a robustness
check. More specifically, I can estimate the effect of the separation of voting rights and
cash ﬂow rights on dividend payout. I estimate

DIVIDEND / CAPITAL = a + b DUAL + c CASH + d W + e (7)
and

DIVIDEND/CAPITAL=a+bVOTE+cCASH+dW+e (8)
where CASH is the ownership of cash ﬂow rights of insiders and VOTE is the ownership
of voting rights of insiders.

Table 17 reports the regression results. Concerning firms in the media industries,
dual class ﬁrms pay signiﬁcantly less dividends (Dividends/Capital ratio) than single
class ﬁrm, as the coefﬁcient for dummy Dual is -0.028 (t statistic = -3.846). In non-media
industries, dual class firms do not have a lower dividend payout than single class ﬁrms.
More interestingly, in a media ﬁrm, insider ownership of voting rights is positively

related to Dividends/Capital ratio while insider ownership of cashﬂow rights is positively

47

related to Dividends/Capital ratio. For a media ﬁrm, a 10 percentage point increase in
insider ownership of voting rights decreases the Dividend/Capital ratio by .0073, while a
10 percentage increase in insiders ownership of cashﬂow rights decreases the
Dividend/Capital ratio by 0.0062. This also proves that control potential is high in media
industries in that the insiders can manipulate the dividend payout without penalty on firm
valuation — as shown earlier in media industries dual class ﬁrms do not have a different
Tobin’s Q from single class ﬁrm. The managers will pay out more dividends if they have
a larger stake in dividend cash ﬂow rights. The insiders will pay out less dividends if they
have more control in voting rights so they can have more cash ﬂow to spend. From this
angle, the dual class structure does provide the necessary condition for entrenchment.
However, dual class ﬁrms in media industries do not have lower values.

The results are similar if I exclude ﬁnancial ﬁrms with SIC codes in the range of

6000-6999 and utility ﬁrms with SIC codes in the range of 4900.

2.5 Conclusion

In this chapter I examined the relationship between ﬁrm value and dual class
common stock. If the dual class structure protects inefﬁcient incumbent managers from
control contest as the managerial entrenchment hypothesis states, a prediction is a lower
Tobin’s Q for a dual class ﬁrm compared with its industry peers. On the other hand,
according to the shareholder interest hypothesis, a dual class structure creates a level of

stability that can be beneﬁcial by mitigates the under-investment problem or encouraging

48

human capital investment. If this were true, the prediction is that dual class ﬁrms will not
have a lower Tobin’s Q.

To investigate these issues I identify the dual-class status and insider ownership of
voting rights and cashﬂow rights of all Compustat listed ﬁrms in 1995. Consistent with
the shareholder interest hypothesis, I ﬁnd that overall that dual class ﬁrms do not have a
lower Tobin’s Q than single class ﬁrms. However, dual class firms that have both classes
traded publicly do have a lower Tobin’s Q than single class ﬁrms and Tobin’s Q
increases as insider ownership of cash ﬂow rights rises and declines as insider ownership
of voting rights rises — only valid in this sub-sample. I also ﬁnd that dual class ﬁrms that
have one class of stock traded publicly do not have lower a Tobin’s Q than single class
ﬁrms. Consistent with the shareholder interest hypothesis, I ﬁnd within media industries,
dual class ﬁrms do not have a lower Tobin’s Q than single class ﬁrm. But within non-
media industries, dual class ﬁrms have a lower Tobin’s Q than single class ﬁrms. Overall
the ﬁndings suggest that dual class stock is beneﬁcial if there is strong managerial
comrrritment to the ﬁrm or there if there is a strong need for managerial stabilization.
There is managerial entrenchment problem when the dual class ﬁrms have both classes
traded publicly or when non-media ﬁrms issue dual class stock.

Besides Tobin’s Q, there is no evidence that dual class ﬁrms underperform their
single class peers on certain performance measures, such as Market-to-book equity ratio
and Price/Earnings ratio. There is some evidence that dual class ﬁrms underperform their
single class peers on some other performance measures, such as ROA, EBIT, and
operating cash flow. Of these measures, ﬁrm performance rises with insider ownership of

cash ﬂow rights and decreases with insider ownership of voting rights.

49

In media industries, dual class ﬁrms pay less dividends than single class ﬁrms, the
dividend payout increases as insider ownership of cash ﬂow rights rises and declines as
insider ownership of voting rights rises. The fact that managers can manipulate dividend

payouts demonstrates the large control potential in media industries.

50

CHAPTER 3

Dual-class Stock, Investment and Cash Flow

3.1 Introduction

Many studies show that internal capital is an important determinant of
investments. Following Fazzari, Hubbard, and Peterson (1988), several studies
demonstrate that cash ﬂow is a more important determinant of investment for ﬁrms that
are most likely to be constrained by internal capital when they invest.11 Lamont (1997)
and Shin and Stulz (1998) provide evidence from another angle. Investments of segments
of a diversiﬁed ﬁrm are dependent on cash ﬂow from other segments. In general this
literature provides evidence that when ﬁrms invest there is a signiﬁcant difference
between the cost of internal capital and external capital. However, the source of the
difference in the cost of capital has not been fully explained.

Myers and Majluf (1984) and Greenwald, Stiglitz, and Weiss (1984) show that
there is a premium on external funds arising from contracting and information problems.
According to this hypothesis, the relationship between liquidity and investment is
typically a symptom of underinvestment. Firms pass up some positive NPV projects
because of the cost of external capital. An alternative hypothesis is agency problem
oriented. The free cash ﬂow theory developed by Jensen (1986) and Stulz (1990) shows

that managers overspend internal capital on unproﬁtable projects. The reality is not that

 

” For empirical evidence on the two explanations see Vogt (1994), Carpenter (1995), Lopez-de-Silanes
and Shleifer (1994) and Jung. Kim. and Stulz (1996).

51

external capital is too expensive but is that internal capital is too inexpensive. The
relationship between cashﬂow and investment is a symptom of overspending.

It is an important issue to distinguish the two explanations for empirical
relevance. Many previous empirical tests do not yield conclusive results.12 Hadlock
(1998) examines how managerial ownership, or more generally the alignment of the
interests between managers and shareholders, affects the sensitivity of investment to cash
ﬂow. Hadlock (1998) ﬁnds a nonlinear relationship between insider shareholdings and
the sensitivity of a ﬁrm’s investment to its cash ﬂow. As insider holdings increase from
zero, investment-cash ﬂow sensitivities rise sharply. This relationship weakens at high
levels of insider ownership. Investment-cash flow sensitivities decrease slowly with
insider holdings after a certain point. In my dual class firm sample insider ownership of
voting rights and cash ﬂow rights is separated. This separation provides an approach to

study the entrenchment effect on investment-cash ﬂow problem.
3.2 Hypothesis Development

The free cash ﬂow hypothesis advanced by Jensen argues that managers have a
propensity to overinvest internal cash. Managers are not the owners of the ﬁrm and the
interests of mangers and shareholders are not perfectly aligned. Managers may reap
personal beneﬁt from overinvesting without fully internalizing the costs of the investment
decision borne by outsider shareholders. If a firm has free cash ﬂow, managers will spend

a large fraction of the ﬁrm’s internal funds. If a ﬁrm does not have free cash flow

 

'2 See Hoshi, Kashyap, and Scharfstein (1991), Oliner and Rudebusch (1992), Whited (1992), Fazzari and
Peterson (1993), Himmelberg and Peterson (1994) and Calomiris and Hubbard (1995).

52

managers will not invest as much since the external ﬁnancing through capital markets
disciplines managers and managers do not want be monitored. This investment behavior
will generate a positive relationship between investment and cash ﬂow. The free
cashﬂow problem arises because insider ownership of cash ﬂow rights is too low to align
the interests of insiders and shareholders. Thus, managers overinvest. Therefore, ﬁrms
with levels of low inside ownership of cash ﬂow rights will exhibit a substantial
sensitivity of investment to cash ﬂow. For a ﬁrm with free cash ﬂow but a high insider
ownership of cash ﬂow rights, managers' interests are well aligned with shareholders and
thus we expect that they will make investment decisions that maximize the return to
shareholders. The managers will invest at the efﬁcient level and ignore the amount of free
cash ﬂow in making investments. The investment level of a ﬁrm does not depend
substantially on the level of cash ﬂow. So ﬁrms with higher insider ownership of cash
ﬂow rights will exhibit a relatively low sensitivity of investment to cash ﬂow.

On the other hand, when a ﬁrm has a high level of insider ownership of voting
rights, managers will feel free to overspend without the presence of effective monitoring.
If a ﬁrm has free cash ﬂow then managers will spend the free cash flow without much
discretion. Hence, ﬁrms with high insider ownership of voting rights will exhibit a
substantial sensitivity of investment to cash flow. For a ﬁrm with free cash ﬂow but a low
level of insider ownership of voting rights, managers have to be cautious with investment
decisions that do not maximize the return to shareholders. The managers will invest at a
more efﬁcient level and the amount of free cash ﬂow is less in making investments. In

this situation, the investment level of a ﬁrm depends substantially less on level of cash

53

ﬂow. The reasoning implies that ﬁrms that have free cash ﬂow with low ownership of
voting rights will exhibit relatively low investment-cash ﬂow sensitivities.

Myers and Majluf (1984) demonstrate how a ﬁrm underinvests in the presence of
asymmetric information problems in the capital markets. Myers and Majluf (1984)
assume that mangers act in the interests of shareholders when making the investment and
financing decisions, which needs a more essential assumption that ﬁrms typically have a
high ownership of cash ﬂow rights. Dybvig and Zender (1991) show that if managerial
ownership is low, the underinvestment problem in alleviated. Hadlock (1998) uses a
model of investment under asymmetric information where investment is a function of
both cash ﬂow and managerial ownership. Hadlock (1998) has shown that investment-
cash ﬂow sensitivity is increasing in managerial ownership. Based on these models, I
infer that the underinvestment problem worsens as insider ownership of cash ﬂow rights
increases, which should imply that investment-cash ﬂow sensitivity is increasing in
insider ownership of cash ﬂow rights. On the other hand based on empirical studies on
entrenchment, I infer that the underinvestment problem is alliviated as insider ownership
of voting rights increases, which should imply that the investment-cash ﬂow sensitivity is

increasing in insider ownership of cash ﬂow rights.

3.3 Regression Speciﬁcation

Following the model in Hadlock (1998) and the regression speciﬁcation in Shin

and Stulz (1999), I use the following regression model:

I/A = BOQ + BlF/A + Bz(F/A)*Voteright + B3(F/A)*Cashright +Y + s (6)

54

Where I is investment, F is cash flow, A is total assets, Q is a measure of Tobin’s q,
Voteright is the insider ownership of voting rights, Cashright is insider ownership of cash
ﬂow rights and Y is controls. If ﬁrms face asymmetric information problems, I expect B3
to be positive and, if entrenchment effects are important, [32 to be negative. If ﬁrms face
free cash ﬂow problems, I expect (3;; to be negative and, if entrenchment effects are
important, 32 to be positive. If asymmetric information problems and free cash ﬂow

problems do not affect ﬁrms investment decisions, I expect B3 and [32 to equal to zero.

3.4 Estimation Results

The first a couple of columns in Table 16 are the regression results from the full
sample of ﬁrms including single class ﬁrms and dual class ﬁrms. Column 1 reports
estimates from a standard investment-cash ﬂow regression fOr the entire sample of 1995.
As in previous studies, the coefﬁcient on Q is positive and signiﬁcant. However, lagged
sales is not signiﬁcant. The estimated coefﬁcient on cash ﬂow of .015 is highly
signiﬁcant but smaller in magnitude to the cash-ﬂow coefﬁcient estimates reported by
Fazzari, Hubbard and Peterson (1988) and Hadlock (1998). My full sample is much
larger than that in previous studies and contains a large number of small companies. That
may be part of the reason for the difference. Column 2 includes terms interacting cash
ﬂow with insider ownership of cash ﬂow rights and voting rights. The regression on the
full sample does not show that insider ownership affects investment-cash ﬂow sensitivity.

Again this may be due to the large number of small firms in the sample.

55

Column 3 reports estimates from a standard investment cash ﬂow regression for
the dual class ﬁrm sample in 1995. As in previous studies, the coefﬁcient on Q is positive
and signiﬁcant. However, lagged sales is not signiﬁcant. The estimated coefﬁcient on
cash ﬂow is positive and not signiﬁcant. At least part of the reason is that the dual class
ﬁrm sample is very small. Column 4 includes terms interacting cash ﬂow with insider
ownership of cash ﬂow rights and voting rights. The estimated coefﬁcient on cash ﬂow
of -.O28 is small and insigniﬁcantly different form zero. This implies that for a ﬁrm with
no insider ownership in either cash ﬂow rights or voting rights, investment is not
sensitive to cash ﬂow. The estimated coefﬁcient on cash ﬂow interacted with insider
ownership of cash ﬂow rights is highly signiﬁcant. The sensitivity of investment to cash
ﬂow increases rapidly with insider ownership of cash ﬂow rights.

The coefficients on cash flow interacted with insider ownership of voting rights is
signiﬁcant at the 1% level, implying that the sensitivity of investment to cash ﬂow
decreases rapidly with insider ownership of voting rights. Note that the R2 is as high as
0.9146.

These estimates support the asymmetric information theories. The results are
inconsistent with the free cash ﬂow theories. The conclusion is similar to that of Hadlock
(1998). As managers care more about shareholder value the asymmetric information

problem becomes more severe.

3.5 Conclusion

56

An intriguing issue in the investment literature on ﬁnancial constraints is the
question of why ﬁrms behave as if there is a difference between internal and external
capital. A large body of literature argues that capital market imperfections result in a
premium on external capital. Hoshi, Kashyap, Sharfstein (1991) and Hubbard, Kashyap
and Whited (1995) provide some evidence that free cash ﬂow problems do not explain
the observed relationship between cash ﬂow and investment. However, their evidence is
indirect and skeptics remain. Hadlock (1998) provides evidence that investment becomes
more sensitive to cash ﬂow as insider holdings increase from an initial point of zero.
However, the entrenchment effect on investment is less conclusive as investment-cash
ﬂow sensitivity stops increasing in ownership after ownership reaches an unclear level.

The separation of voting rights and cash ﬂow rights in dual class ﬁrms provides a
good ground to study the investment-cash ﬂow sensitivity. In this chapter I have
presented direct evidence that is inconsistent with the free cash ﬂow problems driving the
observed sensitivity of investment to cash ﬂow. If the sensitivity of investment to cash
ﬂow was caused by free cash ﬂow problems, the sensitivity of investment to cash ﬂow
should decrease as managers care more about shareholder value. However I ﬁnd that
there is evidence that investment-cash ﬂow sensitivity increases with insider ownership
of cash ﬂow rights and investment-cash ﬂow sensitivity decreases with insider ownership
of voting rights. I am able to explicitly demonstrate the managerial entrenchment effect
on investment-cash ﬂow relation.

The ﬁnding presented above is consistent with the asymmetric information
theories on explanation of investment —cash ﬂow sensitivity. The asymmetric information

problems become more severe as managers care more about shareholder value. My

57

ﬁndings add to the literature suggesting that capital market imperfections result in a

premium on external capital.

58

APPENDICES

59

APPENDIX A

TABLES FOR CHAPTER 1: DUAL-CLASS STOCK AND THE BENEFITS OF
CONTROL

60

Table A1. Number of US Firms With Dual Class Common Stock in 1990-1999

 

Year Number of ﬁrms Number of ﬁrms with dual

 

in Total Class common stocks
1990 5320 379
1991 5392 396
1992 5633 417
1993 6579 455
1994 6921 494
1995 8286 391
1996 8138 563
1997 8902 573
1998 9310 566
1999 7198 544

 

Note - The full sample is composed of all ﬁrms listed on Compustat from 1990 to
1999. Foreign ﬁrms (ADR) and limited partnerships are excluded. The dual class
ﬁrm sample includes ﬁrms with multiple classes of stocks.

61

Table A2. Description of the Variables

 

Dual
Inpct
Inpct_1
Assets_l
Assets-4
Lgasset

Equity_1
Age
Salegrow_l
Salegrow-4
Liquid_l
Liquid-4
DE_1

DE-4
MB_1

MB-4
PE_1
PE-4
Publish
Broadcast
Fin
Insure
Estate
Utility
mpicture

STDf_l

STDt_l

dummy variable, equals one for dual class ﬁrms, zero otherwise.
percent of common stock hold by insiders.
precious year’s percentage of common stock hold by insiders.
previous year’s total assets.
average of previous four year’s total assets
logarithm of previous year’s total assets
pevious year’s market value of equity
proxy for a ﬁrm’s actual age. Deﬁned as current year — ﬁrst year
listed on CRSP
sales growth in previous year
average sales growth in previous four years
liquidity in previous year
average liquidity in previous four years
debt to equity ration in previous year, Compustat (data5[t-1]
+data9[t-1])/data216[t-1]
average debt to equity ratio in previous four years
market to book ratio in previous year, Compustat data25[t-l]
*data199[t-1]/data60[t-1]
average market to book ratio in previous four years
P/E ratio in the previous year
average P/E ratio in the previous four years
dummy, equals one if SIC code is between 2700 and 2800, zero
otherwise
dummy, equals one if SIC code is between 4800 and 4900,
zero otherwise
dummy, equals one if SIC is between 6000 and 6300 or SIC is
between 6700 and 6800, zero otherwise
dummy, equals one if SIC is between 6300 and 6500, zero
otherwise
dummy, equals one if SIC is between 6500 and 6600, zero
otherwise
dummy, equals one if SIC is between 4900 and 5000, zero
otherwise
dummy, equals one if SIC is between 7800 and 7900, zero
otherwise
standard error of estimate from the market model in which ﬁrm’s
monthly return in previous four years is regressed on the
average monthly return of the value-weighted market
portfolio
standard deviation of monthly stock market rates of return in the
previous four year

 

62

 

 

 

 

80¢; 3.9.2 mwm. memo N.mm0~ Wan _ S moom o 33. 0.80— 0360 v-8mm<
hmwmc 3mm: 3 o. Sema— n.wm§ w _ m.m0_ mgomm o M02. 5.00m _ :.wm .1336.
2.5
Saw; 0322 BO. 302 mdm: 9&3 0303 o chow 0.002 02.00 88.02.. .58.
0mg.“ 3 .wm mi m: mg... 2%. 5.2. 053m QUE..- 304. vam. @09. Yaocwuﬁm
mﬂm. fawn T 034. 0:3. awe. wvwwm m0m..- wmw0 3mm: memo. .IBEonm
***ow0m.m $5.0 _- 0mg. 29. 02.0. 00.9:V 00d- 36 03%. $00. 5380 8.5m
33.- aﬁwmm o minom oaﬁm 20; 3.0? 0 mad. me.m vow; v-92
gem; QWNS o mms 29m 804 5.33 o 0.9» 0o:V mam; _Im2
40000.. 2.: o 2.0 aid 3 2 000 o mméa 2.3. EN; 2:: m2
2:: 356m
00mm. imam 09m- mvﬁ v.08 www.0v mmwmv :Vmwv- moi madam 30.2 03:5 xoom
Dunn. Sham C03”.— 03. 3.000 3.02 5.8000 02 N. wbvm w.mm0 02.0 «Axiom
2.0—. oommm Pow. wénﬂ Néwm NS 320 o mm0m w.m©0 03.3. .rbicm
EE 0 095cm
ammo. E mom ammo. o0: $.me £30 50000 mummoo. Comm 8.03 03.3 2:? Exam—2
002. mm _ mm Nmmm- mg. 8.3m moose «mm? 0.02.? vME 0.8m 2.0.x— EEV .933 .28.
$2 8 e- 2.2 $6 m 8 a- 3.2 3.2 A a?
9004. 39 23 ﬁﬂ owe £2 32 mmo. v.2 $2 one mmmu
:o :6» LEE
.2. _ mmvN- 2o: Sm- 5:: 3&0 mtgm 0002 $00- @me E .0v mmoé EEC
6:96 waste?
90.. imma— o owns 0.009 $.02 0nmm§ omwé- N0? Némw 3.0 3:: 33m
58—2 £2 Om :32 52022 x32 :22 am :32 5:022
33:. Race . 202:3
.50 3.2;. 295% SE 3.20 .25 2&an =5

 

3.: 5 was... .s 23% as 8:535 555.5 .2 2.5

63

as. .03 as s 28:290....
.92 .03 as a 28:50....
.22 0.“ as a sscema...

 

 

 

 

 

 

:26. 003 2: a E80E05... - 202
000 0000 3:0 0o 50:52
.1 _00._ N00. 0000. 0000. 000.. .3 00. 0000. _ 000. 0000. 00_ 0. 00000. 1.500
30000.0 0000. 00000. 000. 00_ _. N000. _ 000. 0000. 0000. 000_. 000_. 00.00
000 0. _ 0000. 00000. 0000. 000.. .00_. 0000. 0000. N. 00. 002. 00. _. 15-00
.0000.— 0000. 0 _ 000. 00000. 000 _. 0N_ _. 0000. 0000. 0000. 000.. 000_. 5.00
$0000. 000. 00000. 0000. 0000. 0000. 0000. 00-0000 000. 0000. 050. 169:
._. $00000- 0000. 00000. 0000. 00_ 0. 0000. 00000. 00-00. .0 0000. 000. 000_. 80:—
0000. _ - 00.000 0 .000 00.00 00.0_ 00.000 0 00.00 :40 _0.0_ 0-mm
00_ 0.- 000. 0 .000— 0000 00.: 0000 0 00. _ 0 00.00 000.0 7mm
2me
30000.0 0000— 0 00.00 00.0_ 000.0 0000 0 00.00 2.00 000.0 0:_:.am\8_.£
0000.7 0000— 0 _0.0_ 000.0 00: 0.000 0 000.0 000.0 000.. 005
0_0N._- 00.000 0 00.00 0000.0 0000. 0.000. 0 00.00 0000.0 020. _Imn_
0000.- 0.000 0 00.0_ 00_ .0 000. .00 0 00.0_ 00. .0 000. 260 30G
1.0000. _ - 00 .0. 000. _ - 000. 0000. 0000. 0000.0 000_ - 00.0. 0000.- 0000. 0-0505
$00000- 0000. 000.0- 0000. 0.00. 0_ 00. 000.00 0007 00.00 _000.- 0000. _IE:05
1000—0- 0000. 0.- 0000. 0000. 0000. _ 000 _- 0.0_ 0000.- 000_. 2:: 2:05
532 ES Gm 532 5622 x32 0:2 00 :82 .8602
$82: 3:00 035:3
50 3.2-.0 20:80 SE 3.03 :25 29:00 :2”—

 

33 5 2:3 E 29.50 as 800.50 0.25% €35.59 .3. ass-0

64

 

 

 

 

 

SS.- 888 80.. 80: 28 .08 888 88. 88. 88 ~32 0.3....
8.0.- $28 8... 80: K: 02.8 2.88 o 8% Sn. .08 .43....
88..- 888 8... 208 08% 0.8 080: o 8.8 0.3.8 9%. .._e:.aw_...so.-
0.8.- 8.02 .8...- ms... 80... 9.... 8.8“ 2%..- 08... 88. 82. 0-32028
omn- 808 83.- 2.2 $8. 8... 8.0% 8.....- 83 :8. .8... .13o.w....m
8:.- 38m 8....- QN. 28.0 82. 380 88..- 2.9. $0.. 38.. 53203.8
$.85... 8m: 8 :2. 82 08.. o8. 0 9.0.0 9.0.. 2.0 0-..:
8%.. .08 o .80. 200 80. 88 o 8...... :3 8.0. .02
1.8.0 2.: o ....m 9.3 w... 88 o 8.8 2.3. 20. 82......
2.5 0.5.0.
1.08.0- 08.: 38¢ 3.8.. 38 0...“. 28.: 8.2...- 028 3% 0.8....“ 22> .00..
88.- 2.2... 88. 05 NS. N8 .820 88. 8.: an. 8.2. 1.3.8
8.0.- 88m. 0%... .88 08. .8 Q82 0 38. a: 0.8 1.5.5
260.95%.
88m..- 85$ 8:. 58. 3.08 8.8.. 88% 288. to... 2: 8.0.. U.....>§.._.§.
.55
:23- 89: 0.58- 29. 2.8 8.0... Re: 058- $8 0.5 00% 053.28
.80 E o :0. am... 0 i o 000. 8... 0 cm...
ammo
08:.- 8 8 2.: 3... 8 8 8 .02 .5 8 8.8.2....
.55
~82- 8% 89... m.... 8.8. at... 080. 8.”...- 03 9.2: 3.3. 2.0.5808...
28..- 0.2: o 3.8. 22 0.80 23. 8.0.- 8% 0...... 0.3.. 95:28
mcmvEEzcu 08.). ES. Om :32 5:00.). .83. ES. Om 0.3.). 5:002
5:02-...
0.0800 6.... 80.0 .25 0.0600 :2". 0.0.00.0}

 

0000 :. 95.0 00 20:80 .3 8.6.3.0 0.85:50 .04. 0.00..-

65

.26. .0.... 6... a 280.85....-
.6.6. .00... 65 .6 58.0.8.0 .3...-
..66. .00 2.. .0 28020.0 .3.
.26. .00. 2.. a 2.30.8.0 .-

 

 

 

 

66

 

 

 

000 00.0 06.... 0o ..00::.Z
1......0. 00.0 00.0. .000. . 000. .0... 0000. 0000.. .0000. 00000. 000.. 000.. .0000
1.2000 0 .0 0000. 00000. 00000. .00.. 000. . .000 . . 00000. 00000. 000. . . 0. .. 1.0.00
0.00.- 0. 00. 00000.0 00. 0. 0. 00. 000. . 0000. 00-0000 0000. 0000. 000. . .1800.
.0000..- 000. 0 00.00 ...00 00.0. 000. 0 00.00 00.00 00.0. 0-0.0
.0000..- 0000 0 0.00. 00.00 00.0. 0000 0 00.. 0 00.00 000.0 .19.
005.
0000. 0.00. 0 00. .0 000.0. 000.0 00.0000 0 0.00. 00.00 000.0 00.0-50.0005
00...- 00.00 0 000.0 000.0 000.. 0.000 0 00.0. 000.0 0000. 0-0.0.
000. . 0.000 0 0. .0. 000.0 00 . 0. 0000 0 00.00 .000 0000. .Imn.
000 .. 00.00 0 000 .0 000.0 0000. 00.000. 0 00.00 000.0 0000. 2.3. .000.
000..- 0000. 0000.0- 00. 0. 0.0.. 0000. . 00.00- .00.. 0. 0.. 0000. 0.0505
0000..- 0000. 00.0- 00.0. ..0.. 0.0.. . 000- 0.0.0 00.000. 00 . 0. .0505
$0. ..0- 0000. 000.0- 0000. 000.. 000.. . 000.- 00.0. 0000.- 00. 0. 0...: 08.05
0.80:. .0300 .83. 52 Q0 5.0.). 00.00.). .83. ES. Q0 5.0.). 00.00.).
.00 .00..-.- 0.0.00.5
0.0..-.00 EE 8.0.0 .020. 20:80 :3".

 

000— :. 0:5,.— ..0 0.0500 .80 8.00.050 0.85.5.0 2.0.5.3000 .0< 0.00.0

Table A5. Distribution of Dual Class Firms in Certain Industries in 1990 and 1999

 

 

 

 

1990 1999
Industry Dummy Full Sample Dual class Full sample Dual class
ﬁrm sample ﬁrm sample

Publish 83 24 84 24
Broadcast 122 26 204 53
Finance 5 89 27 1 l 13 55
Insurance 157 19 209 19
Real Estate 62 3 69 5
Utility 260 8 194 5
Motion picture 62 ll 51 7
Number of ﬁrms 5320 379 7198 544

in total

 

Note- Publishing industry includes ﬁrms with SIC codes between 2700 and 2800. Broadcasting
industry includes ﬁrms with SIC codes between 4800 and 4900. Financial industry includes
ﬁrms with SIC codes between 6000 and 6300 or SIC between 6700 and 6800. Insurance
industry includes ﬁrms with SIC code between 6300 and 6500. Real estate industry includes
ﬁrms with SIC between 6500 and 6600. Utility industry includes ﬁrms with SIC between 4900
and 5000. Motion picture industry includes ﬁrms with SIC between 7800 and 7900.

67

Table A6. Logit Regressions of Sample of Firms in 1990

 

 

Independent
variables and 1 2 3
summary statistics
Assets 4.8e-06 6.7e-06 -.000204**
(.791) (1.112) (-2.297)
Age -.00844 -.001570
(-1.344) (-.207)
Salegrow -.004821
(~. 144)
Liquid -. 1520
(-.420)
Debt/Equity .008968
(1.558)
Market lbook -.0071 19
(-0.790)
price/eaming -.004175
(-l.334)
Publisher(dummy) l.687***** 1.653***** 1.668*****
(5.417) (5.292) (5.108)
Broadcast(dummy) .7181* .677* .6669
(1.813) (1.711) (1.306)
Utility (dummy) -2.845**** -2.802*** -2.738***
(-2.8 19) (-2.775) (-2.692)
Finance (dummy) -.676l** -.7615*** -.2290
(-2.316) (-2.590) (-.312)
Insurance (dummy) .4894 .4385 .5751
(1.416) (1.295) (.757)
Estate (dummy) -l.008 -1.005 dropped in
(-.0991) (-.989) reg
Mpicture (dummy) 1.408** 1.385M l.723****
(2.462) (2.420) (2.873)
STDf -3.530*** -4.250**** -6.32*****
(-2.621) (-2.889) (3.572)
Intercept —2.256 ~2.019 -1.663
(-12.62) (8.09) (-5.209)
Log-likelihood -722.83 -721.88 -583.30
Number of 3 164 3 164 2522
Observations

 

Note - Logit regression of the probability of issuing dual class common stock on
accounting and stock return data. The dependent variable is a dummy that equals one if a
ﬁrm has a dual common stock and 0 otherwise. Publish, broadcast, utility, ﬁnance.
insurance, estate and mpicture industry dummies. t-statistics are in parentheses.

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

68

Table A6. (Continued) Logit Regressions of Sample of Firms in 1990

 

 

Independent
variables and 4 5 6
summary statistics
Insider ownership 1.343*****
(3.352)
Assets 0000152“
(2.422)
assets-4 -.00027**
(-2.418)
Equity -.64e-05*
(- l .730)
Age -.006281 -.000824
-.818 (-. 106)
Salegrow-4 .001132
(.054)
Liquid-4 -.08372
(-. 197)
debt/equity-4 .009198
(.542)
Market/book-4 .009642
(1.328)
price/earning -4 -.002947
(-.735)
Publisher (dummy) 1.662***** 2.062***** 1.53*****
(5.335) (6.330) (4.596)
Broadcast(dummy) .9760** .9422** .6414
(2.407) (2.047) (1.266)
Utility (dummy) -2.83**** dropped -2.690***
(~2.803) in reg (-2.636)
Finance(dummy) -.6770** -.4890 -. 1894
(-2.397) (-1.214) (-.257)
Insurance (dummy) .5268 .8530** .6212
(1.572) (2.334) (.800)
Estate (dummy) -1.053 -.5462 dropped
(- 1.036) (-.532) in reg
Mpicture (dummy) 1.439** l.552** 1.776****
(2.513) (2.345) (2.910)
STDf 4.54“” -3.990** -6.13*****
(-3. 155) (~2.065) (-3.349)
Intercept -2.083 -2.719 -1.736
(-10.86) (-8.410) (5.046)
Log-likelihood -719.70 -524.45 -574.62
Number of 3141 2309 2408
observations

 

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

69

Table A7. Logit Regression of Sample of Firms in 1990-1994
Logit estimate of the dependence of dual/multiple class common stock structure probability on accounting
and stock return data. The dependent variable is a dummy that equals one if a ﬁrm has a dual/multiple
common stcok structure.

 

Independent variables

 

. and summary statistics 1990 1991 1992 1993 1994
total assets-4 3.76e-06 4.00e-06 4.04e-06 2.90e-06 1.80e-06
(.450) (.547) (.595) (.416) (.273)
Age -.0116** -.0109* -.01334** -.0164*** -.0145**
(-1.796) (- l .789) (-2.290) (-2.772) (~2.526)
Publisher (dummy) 1.582***** 1.529***** 1.517***** l.71***** l.72*****
(5.03) (5.035) (4.998) (5.78) (5.945)
Broadcast .5879 1.018**** .9966***** l.20***** 1.09*****
(dummy) (1.47) (2.996) (3.179) (4.12) (3.581)
Utility -2.947**** -2.32***** -1.94***** -1.123*** -1.166
(dummy) (-2.917) (-3.218) (-3.282) (-2.623) ( 1.3975)
Finance -.8419*** -.8297*** -.930***** -.843**** -.8067****
(dummy) (~2.671) (~2.762) (~3.346) (—3. 124) (2979)
Insurance .4220 .5999** .24009 .04772 .006675
(dummy) (1.181) (2.015) (.801) (.144) (.984)
Estate -1.021 -.9819 -1.024 -.7331 -.6914
(dummy) -l.003 (—.965) (-1.006) (-.718) (-.676)
Mpicture (dummy) 1.431** 1.262** 1.1836** 1.131** l.398****
(2.491) (2.488) (2.546) (2.233) (2.935)
STDf _10.4***** -l0.0***** _6.82***** 4.3***** .4'56*****
(-4.431) (-4.710) (-4.287) (-3.66) (-3.831)
S'I'Df2 13.44***** 12.95***** 6.411***** 1.212** 1.1959“
(3.956) (4.311) (3.756) (2.684) (1.984)
Intercept -l .472 -1.494 -1.593 -l.720 -1.683
(-5.013) (-5.393) (-6.740) (7.972) (—7.622)
Log-likelihood -703.50 -78l.96 -884.73 -896.60 -896.61
number of 3081 3345 3596 3579 3547
observations

 

Note - Logit regression of the probability of issuing dual class common stock on accounting and stock
return data. The dependent variable is a dummy that equals one if a ﬁrm has a dual common stock and 0
otherwise. Publish, broadcast, utility, ﬁnance, insurance, estate and mpicture are indusU'y dummies. t-
statistics are in parentheses.

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

70

Table A8. Logit Regression of Sample of Firms in 1995-1999:

Logit estimate of the dependence of dual/multiple class common stock structure probability on accounting
and stock return data.

 

Independent variables

 

, and summary statistics 1995 1996 1997 1998 1999
total assets-4 1.54e-06 3.62e-07 3.42e-07 l .32e-06 8.62e-06
(.251) (-062) (.946) (.320) (.228)
Age -.0144** -.0211***** -.01395*** -.008225* -.00903*
(-2.554) (-3.652) (-2.672) (-1.706) (-1.906)
Publisher (dummy) 1.73***** 1.737***** l.693***** l.75***** l.7l6*****
(5.969) (6.207) (6.080) (6.157) (5.842)
Broadcast 1.25***** 1.025***** 1.356***** 1.435***** 1.432*****
(dummy) (4.008) (3.225) (4.700) (5.219) (5.039)
Utility -1.173*** -1.245 -l.469**** -l.339**** -1.548****
(dummy) (~2.742) (-2.877) (-3. 138) (-2.866) (-2.981)
Finance -.847***** -.6485*** -l.01***** -.925***** -1.09*****
(dummy) (-3.043) (-2.653) (-4.298) (-4. 125) (.4830)
Insurance .1326 -.3088 -.06688 -.0212 -. 1578
(dummy) (.414) (-.896) (-.231) (-.076) (-.531)
Estate -.6719 .1372 -.1174 -.07635 -.0299
(dummy) (—.657) (.184) (-. 159) (-. 103) (-.040)
Mpicture (dummy) 1.508***** 1.192** 1.118** l.062** -1.142**
(3.353) (2.525) (2.202) (2.111) (2.226)
STDf ~4.620***** -6.480* -8. 194** -7.10***** -7.78*****
(-3.820) (-1.633) (-2.551) (-3.517) (-4.602)
STD;2 l. 196* -2.189 3.382 4.7000 5.298“
(1.832) (-.182) (.369) (1.150) (2.310)
Intercept -l.675 -1.245 -1.306 -1.556 -1.358
(-7.631) (-3.632) (-4.363) (-6.333) (-5.722)
Log-likelihood -898.55 -945.81 998.28 -1055.85 - 1021.71
number of observations 3518 3593 3931 4104 3866

 

Note - The dependent variable is a dummy that equals one if a ﬁrm has a dual common stock and 0
otherwise. Publish, broadcast, utility, ﬁnance, insurance. estate and mpicture are industry dummies. t-
statistics are in parentheses.

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

71

Table A9. Logit Regression of Sample of Firms in 1999

 

Independent variables

 

and summary statistics 1 2 3 4 5
Insider ownership 1.034****
(3.046)
Total Assets -1.03e-06 4.25e-07 -.0000231
(-.310) (.089) (-1.415)
total assets-4 -.0000244
(-1.287)
Equity -.0000132*
(-l.910)
Age -.005599 -.00909* -.00943*
(-1.029) (-1.666) (-1.723)
Salegrow -.42455*
(-1.908)
Salegrow-4 -.4079*
(-1.941)
Liquid .1401
(.451)
Liquid-4 .09635
(.288)
Debt/Equity .02132
(2.168)
debt/equity-4 .01464*
(1.833)
market lbook -.04414**
(-2.248)
Market/book-4 -.02902
(-1.558)
price/earning .0002664
(.677)
price/earning -4 .001336*
(1.749)
Publisher (dummy) 1.747***** 1,7209***** 1.886***** 1.762***** l.76*****
(5.963) (5.833) (5.797) (5.747) (5.701)
Broadcast 1.482***** 1.6086***** 1.561 ***** 1.672***** 1.48*****
(dummy) (5.234) (5.552) (4.665) (5.021) (4.406)
Utility -1.591**** -1.6237**** -l.289* -1.439*** -l.454***
(dummy) (-3.085) (-3. 139) (-2.155) (-2.738) (~2.760)
Finance -.9358***** -l.014***** -1.063 -.9909 -l.0205
(dummy) (-4.356) (-4.684) (-3.660) (-.971) (-1.000)
Insurance -.05683 -. 1046 .0635***** .8691 * .8298
(dummy) (-.l98) (-.355) (.201) (1.709) (1.642)
Estate -.0633 -. 1051 -.5128 dropped in dropped in
(dummy) (-.085) (-.141) (-.498) reg reg
Mpicture (dummy) 1.167*** 1.205** 1.193** 1.235* 1.3906“
(2.279) (2.348) (2.012) (1.847) (2.340)
STDf _4.860***** _7.252***** _6.07***** _6.67***** _7.8*****
(-4.506) (-4.534) (—3. 188) (-4.702) (-4.283)
STD.2 4.920** 3951* 5.688**
(2.139) (1.601) (2.662)
Log-likelihood - 1024.98 -1020.32 -808.10 -795.01 -803.53
number of observations 3877 3860 2967 2907 2882

 

72

Note - The dependent variable is a dummy that equals one if a ﬁrm has a dual common stock and 0
otherwise. Publish, broadcast, utility, ﬁnance, insurance, estate and mpicture are industry dummies. t-
statistics are in parentheses.

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

73

APPENDIX B

TABLES FOR CHAPTER 2: DUAL-CLASS STOCK, FIRM
VALUE, AND PERFORMANCE

74

Table Bl. Description of the Variables

 

DUAL
Ownsuper
Owninfer
Sharesuper
Shareinfer
Votesuper
Voteinfer
Separation

Vote rights
Cash rights
Size
Size_l
Lgsize
Age

Lgage
Salegrow_l
Liquid

DE

M/B equity
PE

Stdf

Stdt

R&D
Advertising
Investment
Cashflow
EBIT

ROA
Dividend

dummy variable, equals one for dual class ﬁrms, zero otherwise.
the insider ownership in the superior voting class
the insider ownership in the inferior voting shares
the number of shares outstanding in superior voting class
the number of shares outstanding in inferior voting class
the votes per share for the superior voting class
the votes per share for the inferior voting class
Insider ownership of voting rights/ Insiders ownership of
cashﬂow rights
Insider ownership of voting rights
Insider ownership of cashflow rights
total ﬁrm assets
previous year’s total assets.
logarithm of total assets
proxy for a ﬁrrn’s actual age. Deﬁned as current year - first year
listed on CRSP
logarithm of age
sales growth in previous year
Assets liquidity
debt to equity ratio, Compustat (data + data9)/data216
market-to-book equity ratio, Compustat data25*data199/data60
Price/Earnings ratio
standard error of monthly stock market rates of return in previous
four years estimated from a market model in which ﬁrm’s
monthly return in previous four years is regressed on the
average monthly return on value-weighted market portfolio
standard deviation of monthly stock market rates of return in
previous four years
Research and development expense/total assets
Advertising expense/ total assets
Capital expenditure/ total assets
operating cashﬂow/ total asstes
Earnings before interest and taxes/ total assets
Net Income/ total assets
Cash dividend! total capital

 

75

 

23m

 

 

 

8m. 0 - Q _ N..- 000: o 52: 0N0~ 3.9 N. g m0m o 0m.0o w0.- m.~_ wEEmmBotm
iiimd- 85.. dem 0 Sim ooo.~ ﬂaw. No: 0 Omgm $0; :00. 2:: Boa
ammo.
***wwhd- «NE .- 9.5. 54.0. CM memo. $3.. 0 $9- 000 - Rom. 0:5 2:34
0
***ixwd- 300.. $83 $0. 00m: 0.5m”: QwE memmnm o ammo ._ :1 0.5.. 7083‘
0 2::
33.5.2- 800.. mwmmvm 3.0. bvmmm 002 imam 030: o @002 .502 wN.:. mgomm< :23.
0:. wdwh. wm.m_ T 3 : :3. m: _. mm». 8.7 00.0 3.00. 23. 530.5 83m
ommg- wa0. mdm 0 Sam wand 0w0._ 3.3. o v.00 __w.v «(00.— oE: 5:50 92
met was?
.8 958,56
*****m.0_- iiium- _ o :5. 00mm. 3.00. mac. 0 w0om. mmmm. 3.3. 8055
30w: 25: 5.8
0c 9:20:30
****i.md- $33.0- 030. o aim. 5.9%. mowm. whoa. o w0om. mmmm. 3.3. 505:—
EEV bS—om
***iwwd- 000w: :mvm Emmw- «of mem vWNw 099. 042 _- «02 04mm 8.5 32> xcom
*****m.w- omoﬁ .- 0_ mwm ommg .2m :00 wdE 30$ 0 2.— m v0.0 0m.mm 1.3.23.0
BE 0.92.5
1.3.1.80- Rum: monm 58. mmmm ©wa Wm: owoo. _ 3:00. 009» 00: 91.0 62.3 8x82
09... 0vmm. o\. v- w. 2 0.2 w E. v- NE M2: N. ow<
oov. 03m; age 32 w.: $9 $9 08— mme N.m_ Swo— wwo— ammo
:c EPA 5.5
$33.36 50m. vwwm 35. mm.m v00; mmm._ Sow mmvm. 5m: Sand 30.. 0o 0 $508.
$3.336 3%. 0.3: oo0m. Q06 mmm.m 05w— 0§w 0?. 000— omvé 3.0.. 0 £505.
.502: :53 SSE 08—2 52 Om :32 22022 582 ES Om can—2 5:022
.50 an. 6.5: :33 6.03:3
-cwﬁ 080:3 5.. 52.. 29:3 EE mam—o :25 29:3 :2: $20 o_w:_m

 

ma: 5 2:3 a. 295% .8 82.5w 3555 .2 2...;

76

ES. :0: .0 on: :0 anEcwﬁ 40:00:: E5025 5823:...
.05. 000.0 20 S Ewe-2:06 0300:: 20:00:: 582::
$3. :0: of 3 2:89:06. 0300:: 33025 532...:
_o>o_ :00 on: E Enucﬁwi 40:00:: 23025 502......

 

65. 000— 0:: E Ewe-:06 00:00:: A5025 :32... - 202
2:5
:00 0005 .6 508:2
3890
.310: .N- 00N0.- 00: ._ 0 000 _. .000. >000. 000.N 0 00: N. E 00. 0N00. \0:_m::o>0<
*****N_0.0 v00N. 00.N 0 movN. 0000. 00.0. 000 0 N50: 00mm. 0500. 23.35000
$30000 3:000...“ 0000. NN00. 000. 000:. 0000. 000.0 0 : N0. 0000. :0N:. 000.. 000.0
130030 0.31000 300. NN00. 0000. 30:. 000:. N30 0: N0. 0000. 0N0 _. :N: _. no.“
00 :3 o
*****000.v- NO0N. 03 ._ 00N: .- N08. 0: N0. 0 0N.0m 0N0.N- N000. 0NNO. 0 \m0:M0_>__:Q
:3 m
it... : N06- 0NNm. N3.— 000N.- N000. 0 _ N0. 0 N: .0 N000- 000:. 00N0. 0 \m0co0_>5
00 £o$<
0:0. N0:0.- 0N: 000.0- 050. NON0. 0050. 00.0 00:- 000.N N90.- Kno. 350030
83?.
000.. N0 0 0 .- 5.0— 00.5 - 000.: >000. 00:0. 0000. 00_ 7 N00: 0NON.- _ 0 :0. 300080
00:. 0NNN; 0:0.0 00.N_- 030. 0000.- 0000. NNNN :0: - 00.m_ NO0N. 0N8. <00
03 . _ - 0:00.- 0N.0N 00.N_- 000.: 030. 0000. 000.0 0800- 00.0 000.- 2 >0. 332 2.5m
00 aomm<
:0. 0000. 000.: 0 002. 0: 00. 02:0. 0N0 0 0NNN. :000. 0000. 22:39::
23.3:
N05- 000. _ - 00.3 0 0000. 003. 0000. 0.00 0 NON0. 30:. 0000. 05:58.»:—
00 00:: mi
**00_.N- 11000.0- 0N; 0 0.0:.N 00: 00.0: 0.N:0N 0 00.00 0N.0N 0N:
x32 52 D0 :32 5022 :82 52 D0 :32 50022
5200:: 0300 33:: 3:06
:8 :8: 0:0: :8 :8:- o_0:::0 E5 320 0:5 20an :5 o_02._:>

:05 50:3

 

 

 

m9: 5 22:: .5 «Beam 8: 8:235 racism €35.50: .8 2.5

77

Table B3. Distribution of Dual Class Firms in Certain Industries in 1995

 

 

Industry Dummy Full Sample Dual class
ﬁrms

Publishing 96 22
Broadcasting 191 39
Finance 1269 19
Insurance 275 23
Real Estate 68 2
Utility 264 5
Motion picture 61 8
A gric ulture-Crops 15 2
Oil and gas extraction 216 4
Number of ﬁrms in total 8286 391

 

Note- publishing industry includes ﬁrms with SIC codes between 2700 and 2800.
Broadcasting industry includes firms with SIC codes between 4800 and 4900.
Financial industry includes ﬁrms with SIC codes between 6000 and 6300 or SIC
between 6700 and 6800. Insurance industry includes ﬁrms with SIC code
between 6300 and 6500. Real estate industry includes ﬁrms with SIC between
6500 and 6600. Utility industry includes ﬁrms with SIC between 4900 and
5000. Motion picture industry includes ﬁrms with SIC between 7800 and 7900.
Agriculture-Creps industry includes ﬁrms with SIC between 0100 and 70200.
Oil and extraction industry includes ﬁrms with SIC between 1300 and 1400.

78

Table B4.A. Insider Ownership in The Dual Classes

Insiders are deﬁned as executive ofﬁcers and board directors. The insider
ownership includes shares owned by ofﬁcers and directors, trusts for their
beneﬁts and foundations or corporations they control.

 

Super-voting Inferior-voting

 

 

class class
Mean insider ownership 6373* 2165*
Medium insider ownership .7123** .1460**
Total number of ﬁrms 289

 

* t=20.81 for hypothesis that mean insider ownership in super class is
large than mean insider ownership in inferior class

** 2:13.51 for the Wilcox sign-rank test of the equal median insider
ownership.

Table B4.B. Insider Cash Flow Rights and Voting Rights in Dual Class Firms

Cash ﬂow rights and voting rights are based on common stock ownership.
SEPARATION Index is deﬁned as insider ownership of voting rights divided by
ownership of cash ﬂow rights.

 

 

 

Cash flow rights Voting
rights
Mean Insider rights 3465* 5480*
Medium Insider rights .3225** .6058**
Mean SEPARATION 1.87
Medium SEPARATION 1.50
Total number of ﬁrms 242

 

* t: 17.92 for hypothesis that mean insider voting rights in dual class ﬁrms is
large than mean insider cashflow tights
** 2: 12.59 for the Wilcox sign—rank test for equal median.

79

 

Table B4.C. CEO Ownership in The Dual Classes

 

 

 

Super-voting Inferior-voting
class class
Mean CEO ownership 3475* 0994*
Medium CEO ownership .2505** .022**
Total number of ﬁrms 290

 

* t=l3.53 for hypothesis that mean CEO ownership is larger in Super-voting class than
mean CEO ownership in inferior-voting class.

** z =11.28 for Wilcox sign-rank test of the hypothesis that the median CEO ownership
is larger in Super-voting class than median CEO ownership in inferior-voting class.

Table B4.D. Insider Cash Flow Rights and Voting Rights in Dual Class Firms in
Media Industries and Non-media Industries
Cash ﬂow rights and voting rights are based on common stock ownership.

 

 

 

 

 

 

Firms in media industries Firms in non-media industries
Cash ﬂow Voting Cash ﬂow Voting
Mean Insider rights .3318 .5340 .3413 .5508
Medium Insider rights .2612 .6249 .3202 .6021
Mean SEPARATION 1.793 1.891
Median SEPARATION 1.460 1.523
Total number of ﬁrms 44 202

 

80

Table B4.E. Insider Cash Flow Rights and Voting Rights in Dual Class Firms

Cash ﬂow rights and voting rights are based on common stock ownership.

 

 

 

 

 

Both classes traded Only one class traded
Jublicly publicly

Cash ﬂow Voting Cash ﬂow Voting
Mean Insider rights .3302 .4916* .3425 .5702*
Medium Insider rights .3258 .522** .3141 .6378**
Mean SEPARATION l.65*** l.96***
Median 1.41 1.52
SEPARATION
Total number of ﬁrms 69 201

 

Table B4.F. The Largest Block Shareholder Ownership in The Dual Classes

The block shareholder is deﬁned as any person, corporation or institution that
owns more than 5% of the class. Note there are no block shareholders in the
inferior voting class in some ﬁrms.

 

 

Super-voting Inferior-voting
class class
Mean largest block ownership .6013 .2725
Medium 1argest block .5667 .1929
ownership
Total number of ﬁrms 292 255

 

81

Table B5. Summary Statistics for the Sample of Dual Class Firms and the Matching

Firms

Dual class ﬁrms are matched with peer single ﬁrms with the closest book assets in the
same industry (same 2-di git SIC code).

 

 

 

 

Dual class firms Matching single class ﬁrms Number
Variable of
Median Mean SD Median Mean SD observations
Tobin’s Q 1.318 1.844 1.646 1.445 1.933 1.834 288
Q 96 1.388** 2.100 3.617 l.479** 2.083 2.645 315
PB 12.62 25.48 106.0 13.49 21.88 51.50 320
PE 96 14.01 36.34 177.1 13.07 33.58 180.7 350
Market/Book 1.651** 2.645** 3.260 2.094** 3.395** 4.803 326
equity
Age 5** 9.148 11.77 7** 10.42 12.76 406
Debt ratio .889 2.223 5.449 .942 4.314 28.18 375
Investment .0556 .1539 .8429 .0652 .0957 .1076 336
Investment96 .0497 .1052 .2452 .0610 .1021 .2034 363
Cash flow .0845 .0881 1.251 .0876 .0224 .8009 363
Cash flow 96 .07884 -.00577 .5397 .0905 .00793 .4318 402
EBIT .0883 .1134 1.447 .0796 .0409 .6328 372
ROA .0373 -.0281 .6708 .0306 -.0377 .7912 372
Dividend 0 .0153 .1362 0 -.0039 .8055 429
Note - * different mean or median significant at the 10% level

** different mean or median signiﬁcant at the 5% level

*** different mean or median signiﬁcant at the 1% level
**** different mean or median significant at the 0.5% level
***** different mean or median signiﬁcant at the 0.1% level

 

Table B6.A. Summary Statistics for the Subsample of Dual Class Firms with Both
Classes Traded Publicly and the Matching Firms

Dual class ﬁrms are matched with peer single ﬁrms with the closest book assets in the
same industry (same 2-digit SIC code).

 

 

 

Dual class ﬁrms with both Matching single class ﬁrms Number of
Variable classes traded observations
Median Mean SD Median ' Mean SD

Tobin’s Q 1.146*** 1454* 1.009 1.396*** 1.813* 1.567 68
Q 96 1.178** 1.458 .898 1.340** 1.809 1.785 64
PE 12.50 20.13 4.335 13.37 26.02 5.686 72
PE 96 12.62 23.61 48.44 14.81 41.18 140.2 71
Market/Book 1.613 2.494 3.086 2.187 3.165 3.642 74
equity

Age 11 13.09 10.45 9 12.69 13.31 75
Debt ratio 1.110 1.776 2.566 1.018 1.771 1.929 70
Investment .0509 .0874 .1317 .0668 .091 1 .0849 74
Investment96 .0428 * * .0764 .1254 0648* * .0923 .1300 72
Cash flow .1007 .0623 .3480 .0948 .0927 .1 114 77
Cash ﬂow 96 0935* .0684 .2686 .1165* .1012 .1363 76
EBIT .0904 .0641 .314 .0893 .101 .123 79
ROA .0439 .00903 .336 .0402 .0351 .102 79
Dividend .00549 .0240 .0448 .0124 .0508 .144 83

 

Note - * different mean or median signiﬁcant at the 10% level
** different mean or median significant at the 5% level
*** different mean or median signiﬁcant at the 1% level
**** different mean or median signiﬁcant at the 0.5% level
***** different mean or median signiﬁcant at the 0.1% level

83

Table B6.B. Summary Statistics for the Subsample of Dual Class Firms with Only
One Class Traded Publicly and the Matching Firms

Dual class ﬁrms are matched with peer single ﬁrms with the closest book assts in the
same industry (same 2-di git SIC code).

 

Dual class firms with only

 

Matching single class firms

 

 

Variable one class traded Number of
observations
Median Mean SD Median Mean SD
Tobin’s Q 1.357 1.892 1.579 1.443 1.941 1.821 197
Q 96 ~ 1.414 2.262 4.194 1.476 2.072 2.583 226
PB 12.95 28.926 125.4 13.99 21.81 55.16 222
PB 96 14.73 30.85 118.6 12.64 33.73 200.8 249
Market/Book 1667* 2.637** 3.173 2.059* 3.395** 4.962 224
equity

Age 4** 8.857 12.30 6** 10.15 12.79 286
Debt ratio .8274 2.392 6.267 .8759 3.920 24.94 264
Investment .0561 .1740 .9991 .0645 .0972 .1 156 231
Investment96 .0547 .1167 .2801 .0595 .1062 .2302 263
Cash flow .0841 .1483 1.240 .0854 .0031 .9534 251
Cash ﬂow 96 .0801 -0.0058 .5454 .0841 -.00830 .4820 282
EBIT .0891 .1743 1.647 .0769 .02455 .7498 257
ROA .0365 .03812 .417 .0294 -0.0590 .9425 257
Dividend 0 .0123 .160 0 -0.229 .960 300

 

Note - * different mean or median signiﬁcant at the 10% level
** different mean or median significant at the 5% level
*** different mean or median signiﬁcant at the 1% level

**** different mean or median signiﬁcant at the 0.5% level
***** different mean or median signiﬁcant at the 0.1% level

84

 

Table B7.A. Summary Statistics for the Subsample of Dual Class Firms in Media

Industries and the Matching Firms

Media Industries includes SIC code from 2700 to 2799 (publishing), from 4800 to 4900
(broadcasting) and 7800 to 7900 (motion pictures). Dual class ﬁrms are matched with
peer single firms with the closest book assets in the same industry (same 2-digit SIC

 

 

 

 

 

code)
Dual class ﬁrms in media Matching single class ﬁrms
Variable industries Number of
observations
Median Mean SD Median Mean SD

Tobin’s Q 1.616 1.761 .7262 1.762 1.789 .5736 46
Q 96 1.461 1.998 2.07 1.523 1.667 .776 50
PE 17.75** 3806* 86.88 0** 13.83* 25.13 49
PE 96 15.73** 107.5* 425.8 0** 13.97* 25.99 55
Market/Book 2.26** 3.2**** 4.33 3.74** 5.7**** 6.66 49
equity

Age 3 7.09* 10.38 5 1044* 13.43 66
Debt ratio 124* 2.66 4.83 1.70* 9.14 46.98 61
Investment .0620 .170 .438 .1085 .136 .122 52
Investment 96 .0776* .197 .487 .0914* .207 .444 59
Cash flow .0925 .140 .396 .112 .0856 .151 59
Cash flow 96 .0807 .0430* .459 .1056 -.0162* .662 71
EBIT .0921 .1305* .4229 .0622 .0507* .1455 66
ROA .0348 -.0726 .7033 .0293 -.0108 .1558 66
Dividend 0* .0199** .0483 0* .0429** .0751 74

Note - * different mean or median significant at the 10% level

** different mean or median significant at the 5% level

*** different mean or median signiﬁcant at the 1% level
**** different mean or median signiﬁcant at the 0.5% level
***** different mean or median signiﬁcant at the 0.1% level

85

 

Table B7.B. Summary Statistics for the Subsample of Dual Class Firms in Non-
media Industries and the Matching Firms

Media Industries includes SIC code from 2700 to 2799 (publishing), from 4800 to 4900
(broadcasting) and 7800 to 7900 (motion pictures). All other industries (SIC code from
0100 to 9999) are defined as non-media industries. Dual class ﬁrms are matched with
peer single ﬁrms with the closest book assets in the same industry (same 2-digit SIC
code)

 

Dual class ﬁrms in non- Matching single class firms

 

 

 

Variable media industries Number of
observations

Median Mean SD Median Mean SD

Tobin’s Q 1.222 1.859 1.768 1.372 1.960 1.985 242

Q 96 .136** 2.119 3.84 l.447** 2.162 2.86 265

PB 12.15 23.21 109.1 13.80 23.15 54.87 271

PE 96 13.95 23.08 53.04 13.63 37.26 196.3 295

Market/Book 1.520 2.54* 3.03 1.87 2.99* 4.29

equity

Age 5 9.55 12.00 7 10.42 12.65 340

Debt ratio .8071 2.138 5.564 .8938 3.377 22.82 314

Investment .0551 .1510 .8979 .578 .0883 .1033 284

Investment96 .0482 .0882 .161 1 .0545 .0826 .0823 317

Cash flow .0798 .0766 1.260 .0848 .0084 .8823 297

Cash flow 96 .0779 -.0162 .5555 .0850 .0131 .3651 331

EBIT .0880 .1097 1.584 .0815 .0378 .6947 306

ROA .0387 -.0185 .664 .0314 -.0435 .8696 306

Dividend 0 .0144 .148 0 -.0136 .885 355

 

Note - * different mean or median signiﬁcant at the 10% level
** different mean or median signiﬁcant at the 5% level
*** different mean or median signiﬁcant at the 1% level
**** different mean or median signiﬁcant at the 0.5% level
***** different mean or median signiﬁcant at the 0.1% level

86

.EazquU E 2.6 3383 Odom w:_mm_E 95 3.6 9:883; wEmﬂE 2: 9 26 8:6: 2 mcozaioms: mo 52:5: 2: -202 2

 

 

mcocetomno

 

87

3mm moi 23¢ a2: 33 053 Owen 2 mm m0 89:52
3:. 3:. emom. avZ. 0&2. 2.2. :2. .39 «m
83.8 95va 638.8 Amwmﬁv 83.8 2358 Avoodv $3.9
C m.m wom.m god mood ovmd mm: .m EON 2 .wm 385::
3:556
ovco x326:—
mo> wo> mo> mot» mo> m»; mm; mm; :wﬁé
3089 momcoaxm
8%. @5223.
22E 28.0 8225
*****mo.v *****mo._ Odom
38.8 23.: SON: $5.8 Goody ammdv
**voo._ mac; two.— ****O©o._ **moo._ ****wmv._ Emu
Song; 600.: 8&3; $60.7v Amwm._-v §VN4¢ amorv 929$
83 .- *ommmr 1.33.- ... _ mmmr o3: .- wag: Soon N38 Aowﬁwoq
$00.4; Amnog¢ 33¢ Qrch¢ Amoo.m-v Scowd; Ao— Zl Suwv
*****Nm_.- **wOZr 2N6: **m_N_.- ***...xbgr NON—r *****m—_.- *****w\r.o- Aoﬁmvwoi—
2 :31 :33 ENE 33 2.2: 33.1
*****wm._- *****Ov._- mg.— 0500. 350.. *****ON._- 2m<U
agrv C184; :wwrv
Sm. _ - mew. _ - mnmor m,_.0>
Am _ wrv
ammo... ZOE.<mmmmm
:2 .: Angrv 605—; 33::
owmvr 32.. 3.3%.. omov. 1223
85:53
faEEzm
was 832:3
.cowcogow:_

 

maa— E 0 he m:o_mmo._wo~_ 733.com $20 .wm 035,—.

 

.22 e3 2: a .=8E=w_m..._.:._.
35. e8 2: a swatcymiz
.26. as. 2: 3 285299.:
3.6. gem 2.. a Escewa t.
.25. so. 2: a 285ch ._.

.womoﬁcoba 5 8m 85:83 .353? 85:8 ecu mec 25: 58
new 8%: was? .EoEmE no 5?: cosﬁaom .2828.» >883 $20 :26 so a FEDS. Co .5693on 225:8» $8? 802

88

Table B9. Cross Sectional Regressions of Q in 1995: Subsample with Only The Dual
Class Firms Have Both Classes Traded Publicly

Note —cross sectional regression of Tobin’s q on dual class dummy variable, Separation
index or insider voting rights and cash ﬂow rights and control variables. t-statistics are in

parentheses.

 

Independent
variables and
summary statistics

Subsample with only dual class ﬁrms
have both classes traded publicly

 

DUAL 2.653 -4.112** -.8473*
(0.362) (- l .773) (1.677)
VOTE -4.112** -7.290**
(-2. 167) (-2.366)
CASH -. 1829 2.699 5.847*
(-.380) (1.384) (1.836)
VOTE.0tolO —6.642
(-1.411)
VOTE.over10 -7.295**
(-2.359)
CASH.Oto75 5754*
(1.791)
CASH.over75 7.327
(.924)
Log(size) -6.75***** -.111***** -0211 -.13***** -.1193* -.1175*
(-7.7) (-4.07) (-.397) (-3.81) (-1.939) (-1.773)
Log(age) 7.412 -. 198 —.269* -.148 -.3155 -.311
(2.018) (-1.246) (-1.837) (-l.268) (- l .586) (-1.543)
Fstd l.43**** 9753* 1.406 1.380
(2.913) (1.837) (1.109) (.983)
R&D 1.94***** 4.20***** 4.20*****
(3.60) (7.63) (7.60)
Advertising .3239
(.323)
4-digit Industry Yes Yes Yes Yes Yes Yes
code dummies
Intercept 28.13 2.628 2.906 3.295 3.603 3.552
(5.998) (11.85) (6.334) (8.916) (5.226) (4.288)
R2 .0681 .2503 .5100 .1326 .1311 .1295
Number of 5715 3480 474 2443 1406 1406
observations

'~I|.- g -_l\_‘- r. shill-$5211,
5.

 

 

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Significant at the 1% level
****Significant at the 0.5% level
*****Signiﬁcant at the 0.1% level

89

Table BIO. Cross Sectional Regressions of Q in 1995: Subsample of Firms in Media
Industries and Firms in Non-media Industries

Note —cross sectional regression of Tobin’s q on the dual class dummy variable,
Separation index or insider voting rights and cash ﬂow rights and control variables. t-
statistics are in parentheses.

 

 

 

Independent Subsample 0f Subsample of Subsample of Subsample of Subsample of
variables and ﬁrms in media ﬁrms in media ﬁrms in media ﬁrms in non- ﬁrms in non-
summary industries industries industries media media industries
statistics industries
DUAL -.234 -.217 -.360* -.6334* E
(-.873) (-731) (-1.824) (-1.635) T
VOTE -.488 E
(-.459) 5;
CASH -l.05* -.659 -.0368
(-l.778) (-500) (-.O78) ;
Log(size) -.0092 -.0065 -.0575 -.123***** -.00137 i
(-. 145) (-.093) (-.825) (4255) (-.027)
Log(age) —.262 -.424* -.456 -.0824 -.2403*
(-1.158) (-l.736) (-1.822) (-.8l6) (-l.647)
Fstd .498 .300 -.513 1.437****
(.258) (.149) (-.251) (2.855)
R&D Expense l.945*****
(3.637)
Advertising .372
Expense (.383)
4-digit Industry Yes Yes Yes Yes Yes
code dummies
Intercept 2.52 3.249 3.781 2.712 2.722*****
(3.172) (3.869) (4.277) (8.754) (6.036)
R2 .0729 .1319 .0162 .1527 .5050
Number of 156 126 123 4125 492”
observations

 

* Signiﬁcant at the 10% level
** Significant at the 5% level
***Significant at the 1% level

****Signiﬁcant at the 0.5% level
*****Significant at the 0.1% level

 

'4 Note- the number of the observations is limited due to the missing R&D Expense and Advertising
Expense data in Compustat.

90

Table B11. Regressions of Market-to-book Equity Ratio

Note —cross sectional regression of market-to-book equity ratio on the dual class dummy
variable, Separation index or insider voting rights and cash ﬂow rights and control
variables. t-statistics are in parentheses.

 

Independent variables Full Subsample of Subsample Subsample of
and summary sample firms that have of ﬁrms in dual class
statistics both classes media ﬁrms only

traded publicly industries

 

 

l' ;
VOTE . -l .825 -.2221 -4.332 .9589 .i
- (-.506) (-025) (-.630) (.448) '
CASH .5059 -1.066 1.466 - l .782
(.127) (-. 1 16) (.186) (-.614) ~.
Log(size) -.0624 -.05600 -.4997 .1653 ‘
(-.403) (—.344) (— l .530) (.543)
Log(age) -.7016 -.7855 -2.421** .00607
(-1 .323) {-1.412) (-2.229) (.010)
Fstd 5.464** 5.316
(2.148) (2.033)
4-digit Industry code Yes Yes Yes Yes
dummies
Intercept 5.213 5.448 ’ 13.73 1.598
(3.117) (3.131) (4.582) (.645)
R2 .1852 .2969 .0715 -.3081
Number of 2955 2836 235 235
observations

 

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

91

 

Table B12. Regressions of Price/Earnings Ratio

Note —cross sectional regression of Price/Earnings ratio on the dual class dummy
variable, Separation index or insider voting rights and cash flow rights and control
variables. t-statistics are in parentheses.

 

 

Independent Full Full sample Subsample of Subsample Subsample
variables and sample ﬁrm that of ﬁrms in of dual
summary have both media class ﬁrms
statistics classes traded industries only
publicly

VOTE -24.14 -47. 10* -87.59 -174.1 4.741
(—1.281) (-l.687) (-1.227) (-.873) (.191)

CASH 26.25 60.10** 104.23 120.3 -.6276
(1.261) (2.020) (1.443) (.511) (-.018)

Log(size) 1.690** 4.853***** 5.298***** 7.874 -4.533
(2.097) (5.109) (5.389) (.774) (- 1 .222)

Log(age) 5031* -4.324* -4.757* -24.96 -3.309
(1.823) (-l.689) (-1.800) (-.766) (-.452)

Fstd -2.310
(-. 174)

Tobin’s Q .4477 ** .4532** 2.919 -l.664

(2.075) (2.069) (.623) (-.537)

4-di git Industry Yes Yes Yes Yes Yes

code dummies

Intercept -l.903 9.692 8.010 79.79 54.47
(-.219) (1.212) (.970) (.853) (1.750)

R2 —.0137 0 .0051 -.0557 -4572

Number of 5369 4623 4471 224 217

observations

 

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Significant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

 

Table B13. Regressions of ROA

Note —cross sectional regression of ROA (return on assets) on the dual class
dummy variable, Separation index or insider voting rights and cash flow rights

and control variables. t-statistics are in parentheses.

 

 

Independent Full sample Full sample Full sample Subsample of Subsample of
variables and dual class ﬁrms dual class ﬁrms
summary statistics only only
Dual class dummy -.0375
(-1.083) E
VOTE -.1544 -.1130 1:
(-.l 17) (-1.546) .
VOTE.0t05 2.043*** 11.19* E.
(2.590) (1.889)
VOTE.5to25 -.O701 -2.536* "_
(-.358) (-2.017) ‘3
VOTE.over25 -.2085 -.0816
(-1.483) (-.3l6)
CASH .0818 .2374 .2108 .2346** .5193
(2.007) (1.598) (1.403) (2.37) (1.546)
Log(size) .0428***** .0429***** .0484***** .0239* .0694
(9.615) (9.582) (10.167) (1.799) (1.616)
Log(age) .0048 .0053 .00815 .0413 .136*
(.380) ( .420) (.640) (1.371) (1.909)
Fstd -1.521*****
(3.946)
Lagged Tobin’s q -.0259***** -.0258 -.0255***** .0086 .0393
(-7.076) (-7.039) {-6.967) (.482) (.0942)
R&D Expense -1.340***** - 1 .400 -l.400***** -l.494**
(-48.6) (-7.039) (-48.56) (-2.481)
4-digit Industry Yes Yes Yes Yes Yes
code dummies
Intercept -.0703 -.0726 -.203 -.0985 -.953
(-1.791) (-l.83 1) (-3.583) (-3.946) (~2.343)
R2 .7491 .7048 .7062 .3538 .2781
Number of 2303 2292 2292 156 102
observations

 

 

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Significant at the 0.5% level
*****Signiﬁcant at the 0.1% level

93

Table B14. Regressions of EBIT

Note —cross sectional regression of EBIT/Assets on the dual class dummy variable,
Separation index or insider voting rights and cash ﬂow rights and control variables. t-
statistics are in parentheses.

 

 

Independent Full sample Full sample Full sample Subsample Subsample Subsample
variables and of dual of dual of dual
summary class ﬁrms class ﬁrms class ﬁrms
statistics only only only
Dual class -.0598
dummy (-1.709)
VOTE -.2199 -.139 -.l66
(-l.576) (-1.408) (-.755)
VOTEOIOS 2.237**** 10.34*
(2.811) (1.809)
VOTE.5t025 -.l70 -2.26*
(-.082) (-1.865)
VOTE.over25 -.271* -.102
(-l.909) (-.410)
CASH .0592 .278* .257* .221 .423 .438
(1.440) (1.859) (1.698) (1.615) (1.259) (1.353)
Log(size) .0557 .0556***** .0614***** .0523***** .0622* .0594
(12.40) (12.32) (12.80) (3.527) (1.725) (1.434)
Log(age) -.0134 -.0129 -.0098 .0425 .0773 .0885
(-l.048) (-1.002) (-.761) (1.333) (1.140) (1.289)
Lagged -.0289***** -.029***** -.028***** .0444“ .0732* .0408
Tobin’s q (-7.825) (-7.781) (-7.698) (2.144) (1.955) (1.015)
R&D Expense -1.347 -l.35***** -l.35***** -1.834**** -1.756***
(-46.381) (-46.26) (-46.34) (-3.134) (-3.022)
4-digit Yes Yes Yes Yes Yes Yes
Industry code
dummies
Intercept —.0153 -.Ol70 -. 159 -.401 -.572 -.691
(-.387) (-.425) (-2.78) (-3.003) (-l.579) (-l .760)
R2 .6978 .6976 .6992 .2168 .3058 .3553
Number of 2303 2292 2292 204 102 102
observations

 

* Signiﬁcant at the 10% level
** Signiﬁcant at the 5% level
***Significant at the 1% level

 

****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

94

Table B15. Regressions of Operating Cash Flow

Note —cross sectional regression of Operating cash Flow/Assets on the
dual class dummy variable, Separation index or insider voting rights and
cash flow rights and control variables. t-statistics are in parentheses.

 

 

Independent Full sample Full sample Full sample
variables and
summary statistics
Dual class dummy -.0452
(-1.314)
VOTE -. 171
(-1.248)
VOTE.0t05 2.10***
(2.690)
VOTE.51025 -. 142
(-.730)
VOTE.over25 -.214
(-l.539)
CASH .0945** 266* .249*
(2.341) (1.809) (1.675)
Log(SIZC) _044***** .044***** .O49*****
(9.995) (9.952) (10.47)
Log(age) .00305 .0036 .0065
(.243) (.288) (.515)
Fstd
Lagged Tobin's q -.026***** -.026***** -.025*****
(-7.075) (-7.04) (-6.951)
R&D Expense -1.38***** -1.38***** -l.38*****
(-48.35) (48.24) (-48.30)
4-di git Industry code Yes Yes Yes
dummies
Intercept -.0226 -.0247 -. 154
(-.05 8) (-.629) (-2.751)
R2 .7024 .7022 .7483
Number of 2298 2288 2288
observations

 

* Significant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

95

Table B15. (Continued) Regressions of Operating Cash Flow

Note —cross sectional regression of Operating cash Flow/Assets on the dual class dummy
variable, Separation index or insider voting rights and cash flow rights and control
variables. t-statistics are in parentheses.

 

 

Independent Subsample Subsample Subsample Subsample of
variables and of dual class of dual class of dual class dual class
summary ﬁrms only ﬁrms only ﬁrms only ﬁrms only
statistics
VOTE -.151 -.184 -.l33* F‘
(-1.496) (-.808) (-1.699) 3 .f
VOTE.0t05 11.01 * .9
(1.878)
VOTE.5t025 -2.549*
(-2.050)
VOTE.over25 -.0916
(-.359) . ‘
CASH .221 .473 .251** .488 i
(1.592) (1.358) (2.331) (1.469)
Log(size) .044**** .0700* 0258* .0629
(2.92) (1.872) (1.808) (1.480)
Log(age) .0518 .121 .027 .136*
(1.624) (1.715) (.835) (1.937)
Fstd -1.37*****
(3.312)
Lagged Tobin’s q .0296 0733* .0073 .390
(1.406) (1.890) (.382) ( .946)
R&D Expense -1.68** -1.618**
(-2.77) (—2.715)
4-digit Industry Yes Yes Yes Yes
code dummies
Intercept -.350 -.740 -.0207 -.840
(~2.570) (- l .97) (-. 135) (-2.086)
R2 .2539 .2651 .4264 .3322
Number of 203 102 156 102
observations

 

 

* Significant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

96

 

Table B16. Cross Sectional Regressions of DIVIDEND in 1995

 

 

Independent Sub Sub Sub Sub Subsample Sub Subsample
variables -sample of -sample of -sample of -sample of of ﬁrms in -sample of of ﬁrms in
and ﬁrms in ﬁrms in ﬁrms in ﬁrms in media ﬁrms in non-media
summary media media media media industries non-media industries
statistics industries industries industries industries industries
DUAL -.028***** -.0007
dummy (-3.846) (-.068)
VOTE -.070**** -.O70**** -.070**** -.073**** -.00186
(-2.981) (2969) (-2.98) (-2.664) (-.057)
CASH .05 34* .0530* 0533* 0620* -.00505
(1.925) (1.909) (1.923) (1.913) (-.l47)
Log(SlZC) _007***** .OO7***** “m7***** .009***** .009***** .00283101‘ .003*****
(5.701) (5.660) (5.77) (6.49) (7.14) (2.521) (-4.03)
Log(age) .020***** .020***** .020***** .021***** .016***** .Ol4***** __009*****
(5.020) (5.014) (5.059) (4.695) (3.93) (4.271) (-3.34)
EBIT m.00271
(-.515)
CSHFLO -.00268 -.0033 -.00028 .00002 .000382
(-.6l7) (-.666) (-.057) (.138) (.491)
NET -.00389
INCOME (-.851)
Tobin Q .00188**
(2.109)
PE ratio -3.6e-06
(-.384)
Liquid .Ol 12
(1.205)
2-digit SIC Yes Yes Yes Yes Yes Yes Yes
code
Intercept -.0570 -.0570 -.0583 -.0713 -.0542 -.0260 -.Ol70
(-5.070) (-5.036) (-5.093) (-5.089) (-5.276) (-2.928) (-1.958)
R2 .3390 .3387 .3400 .4047 .2283 .026 .04
Number of 229 229 229 176 345 6355 4578
observation

 

Note -cross sectional regression of DIVENDEND on the dual class dummy variable,
Separation index or insider voting rights and cash flow rights and control variables.
DIVIDEND is deﬁned as Cash Dividend/Total Equity Capital. t-statistics are in

parentheses.

* Signiﬁcant at the 10% level

** Signiﬁcant at the 5% level
***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level
*****Signiﬁcant at the 0.1% level

97

my

 

 

APPENDIX C

APPENDIX FOR CHAPTER 3: DUAL-CLASS STOCK,
INVESTNIENT AND CASH FLOW

_9OI.'H'I 1:!
. r v i ‘
S

I!
: i

 

 

 

98

 

Table C1. Investment Regression Results

 

 

Independent variables Full sample Full sample Subsample Subsample of
and summary statistics of dual dual class
class ﬁrms ﬁrms
Sales growth_l .000337 .00024 .000894 .0141
(.851) (.590) (.303) (1.214)
Tobin’s q_1 .0067***** .0108***** .00767 0135*
(7.542) (9.493) (.901) (1.742)
Cash flow .0149***** .02805***** .06169 -.0284
(3.552) (3.322) (.744) (-.374)
Cashflow*Voteright .1215 -.5297***
(.792) (-2.822)
Cashflow*Cash —.l698 l.291*****
(-1.091) (6.364)
4-di git SIC code Yes Yes Yes Yes
Intercept .0685 .0604 .0705 .0498
(27.717) (20.399) (4.730) (4.507)
R2 .2245 .2504 .6006 .9146
Number of observation 5766 4494 300 196

 

Note —cross sectional regression of Investment/Asset on cash ﬂow, interaction term of
cash ﬂow and insider ownership of voting rights and interaction term of cash flow and
insider ownership of cash ﬂow rights, and control variables. t-statistics are in parentheses.

* Signiﬁcant at the 10% level
** Significant at the 5% level

***Signiﬁcant at the 1% level
****Signiﬁcant at the 0.5% level

*****Signiﬁcant at the 0.1% level or lower

99

 

BIBLOGRAPHY

100

 

BIBLOGRAPHY

Amihud, Yakov and Baruch Lev, 1981, Risk Reduction as a Managerial Motive for
Conglomerate Mergers, Bell Journal of Economics, 12, 2, 605-617

Anderson, Ronald C., and D. Scott Lee, 1997, Ownership Studies: The data Source Does
matter, Journal of ﬁnancial and Quantitative Analysis, 32, 3, 311-329

Ang, J .S. and Megginson, W.L.,1989, Restricted Voting Shares, Ownership Structure,
and the Market Value of Dual-class Firms, Journal of Financial Research, 12, 301-318

Attari M., Bannerjee S., 1999, The Investment Decision and Dual Class Shares, working
paper, University of Iowa

Barclay, M. J., and CG. Holdemess, 1989, Private Beneﬁts from Control of Public
Corporations, Journal of Financial Economics, 25, 1989, 371-95

Berger, Philip (3., Eli Ofek, and David L. Yermack, 1997, Managerial Entrenchment and
Capital Structure Decisions, Journal of Finance, 52, 4, 1411-1438

Bergstrém, C., Rydqvist, K., 1990, Ownership of Equity in Dual-class Firms, Journal of
Banking and Finance, 14, 225-69

Berle, Adolf A., and Means, Gardiner C, The Modern Corporation and Private Property.
New York: Macmillan, 1932

Bohemer E., Sanger G., Varshney S., 1996, The Effect of consolidated control on ﬁrm
performance: the Case of Dual-class IPOs. Empirical Issue in Raising Capital,
Elservier.

Caramanolis, Birgul, Rajina Gibson and Nils S. Tuchschmid, 1996, Dual Class Shares
Firms and Seasoned Equity Offerings: Empirical Evidence form the Swiss Stock
Mar. :1. Empirical Issue in Raising Capital, Elservier

Chang, S. and D. Mayers, 1992, Managerial Vote Ownership and Shareholder Wealth,
Journal of Financial Economics, 32, 103-131

Comment Robert, G. William Schwert, 1995, Poison or placebo? Evidence on the
deterrence and wealth effects of modern antitakover measures, Journal of Financial
Economics, 39, 3-43

Comett Marcia Million, Michael R. Vetsuypens, 1989, Voting Rights and Shareholder

Wealth: The issuance of limited Voting common stock, Managerial and Decision
Economics, Vol. 10, 175-188

101

 

DeAngelo, H. and DeAngelo, L., 1985, Managerial Ownership of Voting Rights - A
Study of Public Corporations with Dual Classes of Common Stock, Journal of
Financial Economics, 14, 33-69

Demsetz, Harold, Kenneth Lehn, 1985,The Structure of Coporate Ownership: Causes and
consequences, Journal of Political Economy, 93,1 155-1 177

Denis D., Denis D., 1994, Majority Owner-managers and Organizational Efﬁciency,
Journal of Corporate Finance, 1, 91-118

Dybvig, PH. and Zender, J .F., 1991, Capital Structure and Dividend Irrelevance with
Asymmetric Information, Review of Financial Studies, 4, 201-219

Fazzari, S.M., Hubbard, R.G., and Peterson, BC, 1988, Financiang Constraints and
Corporate Investment, Brookings Paper on Economic Activity, 141-195

Fischel D., 1987, Organized Exchange and the Regulation of Dual Class Common Stock,
Chicago Law Review, 54, 119-52

Foster, Stephen R., David C. Porter, 1993, Dual Class Shares: Are There Returns
Differences?, Journal of Business Finance & Accounting, 20(6), November, 893—903

Gompers, Paul A., Joy L. Ishii and Andrew Metrick, 2001, Corporate Governance and
Equity ‘

Prices, working paper, Harvard university, University of Pennsylvania and NBER

Greenwald, B., Stiglitz, IE, and Weiss, A., 1984, Information Imperfections in the
Capital Market and Macroeconomic Fluctuations, American Economic Review, 74,
194-199

Grossman, Sanford J. and Oliver D. Hart, 1988, One Share / One Vote and the Market for
Corporate Control, Journal of Financial Economics, 20, 175-202

Grullon, Gustavo, and George Kantas, 2001, Managerial Incentives, Capital Structure,
and Firm Value: Evidence from Dual-Class Stocks, working paper, Rice University

Hadlock, Charles J., 1998, Ownership, Liquidity, and Investment, RAND Journal of
Economics, 29, No. 3, 487-508

Harris, Milton and Artur Raviv, 1988, Corporate Governance: Voting Rights and
Majority Rules, Journal of Financial Economics, 20, 203-235

Homer, M.R., (1988), The Value of the Corporate Voting Right - Evidence from
Switzerland, Journal of Banking and Finance, 12, 69-83

 

J arrell, Gregg A., Annette B. Poulsen, 1988, Dual-class Recapitalizations as Antitakeover
Mechanisms — the Recent Evidence, Journal of Financial Economics 20, 129-152

Jensen, Michael C., 1986, Agency Costs of Free Cash Flow, Corporate Finance, and
Takeovers, American Economic Review, 76, 323-329

Jog, V. M., Riding, A.L., (1986), Price Effects of Dual-class Shares, Financial Analysts
Journal, 86 (Jan/Feb), 58-66

Kaplan, Steven N., and Luigi Zingales, 1997, Do Investment-cash Flow Sensitivities
Provide Useful Measures of Financing Constraints? Quarterly Journal of Economics,
112, 169-216

Lamont, Owen, 1997, Cash Flow and Investment: Evidence From Internal Capital
Markets, Joumal of Finance, 52, 83-110

Kunz, Roger, M., James J. Angel, 1996, Factors Affecting the Value of the Stock Voting
Right: Evidence from the Swiss Equity Market, Financial Management, Vol. 25, 3,
autumn, 7-20

Lease, Ronald C., John J. McConnell and Wayne H. Mikkelson, 1983, The Market Value
of Control in Publicly Traded Corporations, Journal of Financial Economics, April,
439-472

Lease, Ronald C., John J. McConnell and Wayne H. Mikkelson, 1984, The Market Value
of Differential Voting Rights in Closely Held Corporations, Journal of Business,
October, 443-

467

Lehn, Kenneth, Jeffry Netter and Annette Poulson, 1990, Consolidating Corporate
Control: Dual-class recapitalizations versus Leveraged Buyouts, Journal of Financial
Economics, 27, 557-80

Levy, H., (1983), Economic Evaluation of Voting Power of Common Stock, Joumal of
Finance, 38, 79-93

Liljeblom, E., and Rydqvist, K., 1992, Wealth Effects of Dual-class Shares - An
Empirical Study on Swedish Data, unpublished manuscript, Stockholm School of
Economics

Myers, S.C., and Majluf, NS, 1984, Corporate Financing and Investment Decisions
When Firms Have Information That Investors Do Not have, Journal of Financial
Economics, 13, 187-221

Morck, R., Shleifer, A. and R. Vishny, 1988, Management Ownership and Market
Valuation: An Empirical Analysis, Journal of Financial Economics, 20, 293-315

103

 

 

Nicodano, Giovanna, 1998, Corporate Groups, Dual-class Shares and the Value of
Voting Rights, Journal of Banking & Finance 22, 1117-37

Olrog, C., and Rickhamre, KG, 1992, F éretagskép idetframtidaEuropa, SNS
Publishing Company, Stockholm

Palepu, Krishna G., 1986, Predicting Takeover Targets: a Methodological and Empirical
Analysis, Journal of Accounting and Economics, 8, 3-35

Partch, M. Megan, 1987, The Creation of a class of Limited Voting Common Stock and
Shareholder Wealth, Journal of Financial Economics, 18, 313-39

Perfect, Steven B., and Kenneth W. Wiles, 1994, Alternative Construction of Tobin’s q:
An Empirical Comparison, Journal of Empirical Finance, 1, 313-341

Robinson, C., White, A., 1990, The Value of a Vote in the Market for Corporate Control,
unpublished manuscript, York University

Rydqvist, Kristian, 1993, Dual-class Shares: A Review, Oxford Review of Economic
Policy, Vol. 8, No. 3, 45-57

Ruback R., 1988, Coercive dual-class exchange offers, Journal of Financial Economics,
20, 153 -173

Shin, Hyun-Han, and Rene M. Stulz, 1998, Are Internal Capital Markets Efﬁcient?,
Quarterly Journal of Economics, May, 531-552

Shin, Hyun-Han, and Rene M. Stulz, 2000, Firm Value, Risk, and growth opportunities,
NBER working paper #7808

Smart, Scott 3., Chad J. Zutter, 2000,Control As a motivation for Underpricing: A
Comparison of Dual- and Single- Class IPOs, August, Working paper, Indiana
University

Shum, Connie M., Wallace N. Davidson and John L. Glascock, 1995, Voting Rights and
Market Reaction to Dual Class Common Stock Issues. Financial Review 30, 275-87

Stulz, Rene M., 1988, Managerial Control of Voting Rights: Financing Policies and the
Market for Corporate Control, Journal of Financial Economics, J anuary/March, 25-54

, 1990, Managerial Discretions and Optimal Financing Policies, Journal of
Financial Economics, 26, 3-27

 

Taylor S., Whittred G., 1998, Security Design and the Allocation of Voting Rights:
Evidence From the Australian IPO Market, Journal of Corporate Finance 4, 107-31

104

 

Wooldridge, Jeffrey, M., 1999. Introductory Econometrics: A Modern Approach, South-
Westem College Publishing

Zingales, L., 1994, The Value of the Voting Rights: A study of the Milan Stock
Exchange Experience, Review of Financial Studies, VII (1994), 125—48

, 1995, What determines the Value of Corporate Votes?, Quarterly Journal of
Economics, November, 1047-73

105

 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIII

llnllﬂll11111111Mllllzlllﬂlﬂlﬁlll