z . ..... kahuna
; 1:: :1. .
r...
u

.l. . 1‘15 ,

A <- .3‘; first, ..

r. 3 1.. .49.:
. . 3

3
‘1?th
w .w $1,.
I, I. N‘ I
ii. i V if .
nth
: ;
6...».
5 n

a
1...: was“.

r
. , :1. .?
(faxmwmwhl . .V} L... um.» t n
.1..;....w «a... k .

 

 

 

 

 

‘21}..31
.u r (is
1.3.1..

 

{r {fl 3 .V.
t ..)~x.r {2‘9 ‘

»

mv-vv—q

ymw-yu

_.u.~

3w

Aalv . .7 O ,,
1 .n L ‘9? it!»

u...

 

a.

2...: 3.2.».

This is to certify that the
dissertation entitled

The Role of Corporate Governance in Meeting or Beating
Analysts’ Expectations

presented by

Angela Banita Andrews

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Business Administration

 

myCin

\ Major Professor’sFignature

s/Iao losl

ate

MSU is an Affi/mative Action/Equal Opportunity Institution

 

 

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:/CIRC/DatoDue.p65-p.15

THE ROLE OF CORPORATE GOVERNANCE IN MEETING OR BEATIN G
ANALYSTS’ EXPECTATIONS

By

Angela Banita Andrews

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Accounting and Information Systems

2004

THE

tittxicsn :
expectant
anilxsts l

extent Ihd

Hie-chant:
the men;
fipectan
mechuni
Siiong ct

tipectu:

 

 

ABSTRACT

THE ROLE OF CORPORATE GOVERNANCE IN MEETING OR BEATING
ANALYSTS’ EXPECTATIONS

By

Angela Banita Andrews

This study provides empirical evidence on whether there is an association
between corporate governance mechanisms and meeting or beating analysts’
expectations. In the short run, the market may reward managers for meeting or beating
analysts’ expectations, but this may be detrimental to long run shareholder value to the
extent that meeting or beating analysts’ expectations is achieved purely through earnings
management. Consistent with the premise that firms with strong corporate governance
mechanisms emphasize long run shareholder value, I find a negative association between
the strength of corporate governance and the propensity of firms to meet or beat analysts’
expectations. In addition, I find that while firms with weak corporate governance
mechanisms use earnings management to meet or beat analysts’ expectations, firms with

strong corporate governance mechanisms achieve the same result by guiding analysts’

expectations.

lu'ould j
Felicia. l

 

 

 

I would like to dedicate this dissertation to my family: Walter, Elaine, Joyce, Patrick,
Felicia, Ishmael, Israel and India.

iii

traitor
p2oer 1
Jeffrey

like to

N

7‘.

I2.)

0‘

 

ACKNOWLEDGEMENTS

I would like to thank the members of my dissertation committee: Joseph
Anthony, Kimberly Ellis, Marilyn Johnson (chair) and Krishnamoorthy Ramesh. This
paper has also benefited from comments from Matthew Anderson, Christian Mastilak,
Jeffrey Wooldridge and seminar participants at Michigan State University. I would also

like to thank Freddy Coronado and Terence Pitre for their support.

iv

CHAPTEP

ll

4‘
-
4‘

IJIJIJ

,-l

CHAPTE

-‘ I

(J)

 

 

 

 

 

TABLE OF CONTENTS

LIST OF TABLES ........................................................................ vii

CHAPTER 1 INTRODUCTION ........................................................ 1

1.1 Motivation ................................................................ l

1.2 Overview of Hypotheses ................................................. 2

1.3 Overview of Research Design ........................................... 3

1.4 Overview of Results ...................................................... 4

1.5 Contribution to Literature .............................................. 4

1.6 Organization of the Dissertation ....................................... 5

CHAPTER 2 RELATED LITERATURE .............................................. 6
2.1 The Valuation Implications of Meeting or Beating

Earnings Benchmarks ................................................... 6

2.2 Mechanisms for Avoiding a Negative Earnings Surprise .. . . . 10

2.3 Corporate Governance and Earnings Management ................. 14

2.4 Corporate Governance and Expectations Management ............. 18

CHAPTER 3 HYPOTHESIS DEVELOPMENT ...................................... 23

3.1 Hypothesis One — Firms that fail to meet or beat analysis’

expectations have stronger corporate governance mechanisms

that firms that meet or beat analysts’ expectations (hereinafter

referred to as Hypothesis One) .......................................... 23
3.2 Hypothesis Two — Firms with strong corporate governance

mechanisms that meet or beat analysts’ expectations are more

apt to do so by managing expectations while firms with weak

corporate governance mechanisms that meet or be analysts’

expectations are more apt to do so my managing earnings

(hereinafter referred to as Hypothesis Two) ......................... 27

CHAPTER 4 RESEARCH DESIGN .................................................... 30
4.1 Sample Selection ......................................................... 30

4.2 Model Specification — Hypothesis One ............................... 31

4.3 Control Variables ......................................................... 34

4.4 Model Specification - Hypothesis Two .............................. 36
CHAPTER 5 RESULTS .................................................................. 43
5.1 Descriptive Statistics .................................................... 43

5.2 Regression Analysis — Hypothesis One .. . .- ........................... 45

5.3 Regression Analysis - Hypothesis Two .............................. 49

5.4 Summary .. . ._ .............................................................. 52

 

 

 

CHAPTER 6 CONCLUSION AND FUTURE RESEARCH

OPPORTUNITIES ........................................................................ 53
6.1 Conclusion ................................................................ 53
6.2 Future Research .......................................................... 54
6.3 Limitations ................................................................ 54
REFERENCES ............................................................................. 56
APPENDICES .............................................................................. 60
Appendix A - Descriptive Statistics ............................................ 61
Appendix B — Supplementary Analyses for Hypothesis One ............... 82
Appendix C - Descriptive Statistics for Hypothesis Two ................... 97
Appendix D — Supplementary Analyses for Hypothesis Two ............... 113
Appendix E — Additional Supplementary Analyses for Hypothesis Two.. 118
Appendix F — Valuation Supplementary Analysis ............................ 124

vi

abiel-

Table 2 -

Table 3 -

Table 4 -

Table 5 -

Table Al

Table A;

Table A,

Table A

Table A

Table A

 

 

LIST OF TABLES

Table 1 - Formation of Final Sample for Hypothesis Two ......................

Table 2 - Summary Statistics for Dependent Variable, Independent
Variables and Control Variables for 4,865 firm-year
observations for period 1997-2001 .....................................

Table 3 - Correlation Matrix for the Dependent Variable,
Independent Variables and Control Variables ........................

Table 4 - Logit Analysis of the Probability of Meeting or Beating
Analysts’ Expectations and Corporate Governance
Mechanisms ...............................................................
Table 5 - Logit Analysis of the Association between Corporate

Governance Mechanisms and the Probability of Managing
Expectations versus Managing Earnings ..............................

Table A] - Descriptive Statistics for the Dependent Variable, Independent
Variables and control Variables for firms that meet or beat
analysts’ expectations ....................................................

Table A2 - Descriptive Statistics for Dependent Variable, Independent
Variables and Control Variables for firms that fail to meet

or beat analysts’ expectations ...........................................

Table A3 - Descriptive Statistics for Dependent Variable, Independent
Variables and Control Variables .......................................

Table A4 - Descriptive Statistics for Dependent Variable, Independent
Variables and Control Variables by year ..............................

Table A5 - Descriptive Statistics for Dependent Variable, Independent
Variables and Control Variables for sample year 1997 ..............

Table A6 - Descriptive Statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 1998 ..............

Table A7 - Descriptive Statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 1999 ..............

Table A8 - Descriptive Statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 2000 ..............

vii

Table A9 -

Table Bl -'

1
Table B2 -

Table B3

Table B4

Table 8.5

TableC.l

Table C .j

Table C

Table C

 

 

Table A9 -

Table B.1 -

Table B.2 -

Table B.3 —

Table B.4 -

Table B.5 -

Table C.1 -

Table C.2 -

Table C.3 -

Table C.4 -

Table D.1 -

Descriptive Statistics for the Dependent Variable, Independent
Variable s and Control Variables for sample year 2001 .............

Logit Analysis of the Probability of Meeting or Beating Analysts’
Expectations and Corporate Governance Mechanisms ..............

Logit Analysis of the Probability of Meeting or Beating Analysts’
Expectations and Corporate Governance mechanisms (all
variables winsorized at 10% and 90% percentile) ....................

Logit Analysis of the Probability of Meeting or Beating Analysts’
Expectations and Corporate Governance Mechanisms by year. . ..

Multiple Regression Analysis between meeting or Beating
Analysts’ Expectations and Corporate Governance Mechanisms.

Multiple Regression Analysis between Meeting or Beating
Analysts’ Expectations and Corporate Governance Mechanisms
by year .....................................................................

Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for firms that managed earnings
upward (positive discretionary accruals) and managed
expectations downward (negative unexpected earnings forecast)..

Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for firms that failed to manage
earnings upward (negative discretionary accruals) and managed
expectations downward (negative unexpected earnings forecast)..

Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for firms that failed to manage
earnings upward (negative discretionary accruals) and failed to
manage expectations downward (positive unexpected earnings
forecast) ....................................................................

Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for firms that managed earnings
upward (positive discretionary accruals) and failed to manage
expectations downward (positive unexpected earnings forecast)..

Logit Analysis of the Association between Corporate Governance

Mechanisms and the Probability of Managing Expectations versus
Managing Earnings without Control Variables .......................

viii

80

83

84

85

90

92

98

102

106

110

113

Table D3 '

Table D3 -

Table 0-1

Table El

Table Fl

Table FL

Table F3

Table F-

 

Table D2 -

Table D3 -

Table D4 -

Table E.1 -

Table F.1 -

Table F.2 -

Table F.3 -

Table F.4 -

Logit Analysis of the Association between Corporate Governance
Mechanisms and the probability of Managing Expectations versus
Managing Earnings with Percentage of Institutional Holders as a

Control Variable ...........................................................

Logit Analysis of the Association between Corporate Governance
Mechanisms and the probability of Managing Expectations versus
Managing Earnings with all Individual Control Variables ..........

Logit Analysis of the Association between Corporate Governance
Mechanisms and the Probability of Managing Expectations versus
Managing Earnings with the Factor of Reliance on Implicit
Claims ......................................................................

Sequence of meting expectations using earnings management or
expectations management ................................................

Formation of the Final Sample for Hypothesis Two .................

Regression of Cumulative Abnormal Return on percentage of
shares held by the top five institutional investors and other
determinants of the cumulative abnormal return ......................

Regression of Cumulative Abnormal Return on percentage of
shares held by the top five institutional investors and other
determinants of the cumulative abnormal return by quintile of
level of earnings management as measured by the value of
discretionary accruals ....................................................

Regression of Cumulative Abnormal Return on percentage of
shares held by the top five institutional investors and other
determinants of the cumulative abnormal return by quintile of
level of expectations management as measured by the value of
unexpected earnings forecast ............................................

ix

114

115

117

119

131

133

134

135

Cl

011:
nenageme
control 1.“
ll'iCllefi D0
of manage

hate been

fraud iBe'.

Ibe discin
I
or beating

1.1 Nloti‘

 

 

Chapter 1 - Introduction

One role of corporate governance is to minimize incentive problems between
management and the providers of capital arising from the separation of ownership and
control (Watts and Zimmerman 1986). Corporate governance mechanisms, which
include boards of directors and the presence of institutional investors, mitigate the ability
of managers to behave in their own self interests. Corporate governance mechanisms
have been shown to mitigate behavior such as earnings management and accounting
fraud (Beasley 1996; Chtourou et al. 2001; Chung et al. 2002; Dechow et al. 1996; Farber
2003; Klein 2002; Peasnell et a1. 2000). The purpose of this paper is to examine whether
the disciplining effect of these governance mechanisms extends to the context of meeting
or beating analysts’ expectations (hereinafter referred to as MBE).

1.1 Motivation

Analysts’ expectations, as a performance benchmark, have taken on increased
significance in the last several years. Empirical evidence suggests that managers’
propensity to achieve particular thresholds has shifted over time. Currently avoiding
negative earnings surprises is the most important threshold (Brown and Caylor 2003).
However, the valuation literature provides mixed support for the shift in managers’
propensity to avoid negative earnings surprises. While short run market premiums and
increased earnings multiples accrue to firms that consistently MBE, these valuation
benefits subsequently disappear when firms that previously met expectations fail to do so
(Barth et a1. 1999; Bartov et a1. 2002). Related to this argument there is evidence of a
negative stock price response to a break in a string of earnings increases ( Ke and Petroni

2003; Ke et a1. 2003; Skinner and Sloan 2002). Additionally, managers and public

 

accountar
312311316111 1
n.3ang bu
anl Rotz I
1.2 0\ er\'
Sit
unclear M
that COrpo
run firm V
and “bath
shareholti

f”TITS 11",“ '

mechum,
351be Us;
perfigm‘

e’ll‘tTCte,

 

med-1m:

I
111 3.001
lIIC-‘dt

‘I,.
'l
u,

mechar

 

{IT-Ins \1‘

‘013111;

 

accountants have expressed concern that the “short term earnings game” is played to the
detriment of overall firm value. Short run targets, specifically MBE, could be met by
making business decisions that have a negative impact on a firm’s long run value (Favaro
and Rotz 2004; PricewaterhouseCoopers 2002).
1.2 Overview of Hypotheses

Since the short run benefits of MBE may result in long run costs to the firm, it is
unclear whether MBE is in the best interest of shareholders. The premise of this paper is
that corporate governance mechanisms provide incentives for managers to maximize long
run firm value. Examining the association between corporate governance mechanisms
and whether a firm MBE may shed light on whether MBE is truly in the interest of
shareholders. My first hypothesis predicts that concerns about long run costs motivate
firms with strong corporate governance mechanisms to avoid playing the MBE game.
Additionally, I examine earnings management and expectations management, two

mechanisms that are used to MBE (Matsumoto 2002). I define expectations management
as the use of voluntary disclosures to convey information about the underlying
performance of the firm such that the market participants adjust their forecasts to reflect
expected future performance. Prior literature suggests that corporate governance
mechanisms constrain the ability of managers to manage earnings upward (Chtourou et
a1. 2001; Chung et a1. 2002; Klein 2002; Peasnell et a1. 2000). The corporate governance
literature, however, has not established a relation between corporate governance
, mechanisms and expectations management. The literature, however, does suggest that
firms with strong corporate governance mechanisms are more likely to provide increased

voluntary disclosure through management earnings forecasts. In addition to increased

'al Vi." IV”

 

frequfn'
gaiemdi
mafia
expectat
2114113; Cl
a. 20;!)

1.3 Om

period 1‘
013 Illa-ll
intestor
of the be
directors
board :0

gOi'emat

 

 

 

frequency, these forecasts are less biased than those of firms with weak corporate
governance mechanisms (Ajinkya et al. 2003). Accordingly, my second hypothesis states
that firms with strong (weak) corporate governance mechanisms are predicted to use
expectations management and/or earnings management in order to MBE (Ajinkya et al.
2003; Chtourou et al. 2001; Chung et al. 2002; Klein 2002; Matsumoto 2002; Peasnell 'et
a1. 2000).
1.3 Overview of Research Design

I test these hypotheses using a sample of 4,865 firm year observations from the

period 1997-2001. Three measures of strong corporate governance are used: the presence
of a majority of outsiders on the board of directors, the presence of an institutional
investor with long term incentives/preferences and board size. The size and composition
of the board of directors are important governance mechanisms because boards of
directors have the right to hire, fire and compensate management. I hypothesize that a
board composed of a majority of independent directors is an indicator of strong corporate
governance. The corporate governance literature is inconclusive on whether larger or
smaller boards are better monitors (Chtourou et al. 2001; Core et al. 1999; Peasnell et al.
2001). As such, no directional hypothesis is set forth for board size. Institutional
investors are considered governance mechanisms because of their potential to influence
management through share ownership and indirectly by trading shares (Gillian and Starks
2003). Shleifer and Vishny (1986) suggest that only shareholders with significant
holdings have an incentive to monitor management. In order to allow monitoring

incentives to vary, I use the percentage of shares held by the top five institutional

._ -41

'- '4', ' rev". n..." -'

 

; 9‘ 1!
‘matk

"I if
(Gittlln ‘

1,4 Oi er

the top ft
consister
caneeme
increases
Thai as in

firmrsrr

New 512

1.5 Cont

ultimate
P026“ 5,
g0\eITt_1

short te:

 

3’15 pre‘

ln‘\ est?

 

investors as a proxy for institutional investors with long term incentives/preferences

(Gillian and Starks 2003).
1.4 Overview of Results

Supporting my first hypothesis, I find that as the concentration of shares held by
the top five institutional investors increases firms are less likely to MBE. This is
consistent with firms that have strong corporate governance mechanisms being more
concerned with long run costs such as those resulting from a break in a string of earnings
increases (Ke and Petroni 2003; Ke et al. 2003). Supporting my second hypothesis, I find
that as the percentage of shares held by the top five institutional investors increases the
firm is more likely to use expectations management and less likely to use earnings
management. The results are insignificant when corporate governance is measured using
board size and composition.
31.5 Contribution to Literature

This study contributes to the literature in several ways. First, this paper provides
support for managers’ and public accountants’ concerns that focusing on MBE may not
ultimately lead to increased shareholder value (Favaro and Rotz 2004;
PricewaterhouseCoopers 2002). The results suggest that firms with strong corporate
governance mechanisms are more concerned with long run consequences rather than
short term valuation increases, and thus are less likely to play the meet or beat game.

Second, this paper suggests that there are different incentives to monitor based on
the preferences/incentives of institutional investors. Using the percentage of institutional
investment as a proxy assumes that all institutions have the same incentives. By using

ownership concentration it is possible to differentially examine the incentives of

l'

Lfi'ynfl’ 4 "'

IVE-I" " 1".L— a M‘- f.“-

 

institutio

Ottnersht-p

institution
1.6 Orgal
Tl

btptathese

are act 101

erections

 

II‘\_?.-. .
\.A‘I‘u'l:

'3‘. ~
L’I irked.

 

institutional ownership. The results of this study suggest that concentrated institutional
ownership has a different monitoring effect than institutional ownership in general’. This
finding complements Gillian and Starks (2003) who find that concentration of
institutional ownership is associated with lower levels of executive compensation.
1.6 Organization of the Dissertation

The paper proceeds as follows: related literature is presented in section two;

hypotheses are formulated in section three; the research design and descriptive statistics
are set forth in section four; results are presented in section five; conclusions and

directions for future research are presented in section six.

 

lBushee and Noe (2000) suggest that improvement in the disclosure process is one way to attract
institutional investors. Additionally, tailored road show presentations specifically targeted to a certain type
of institutional investor is another way to attract institutional investment.

'quauanra

.4

q‘ .p on; 3W!"

. I'n \- WE‘J-Ya' rLLLI"

 

htptithes

beat em

focuses t'
2.1 The

model to
Jll’lcx
et al. (19
631an gs
firms Wt
Ail-ditto:
i~’v'1l‘liln g

with;

§

 

Penn

 

Chapter 2 - Related Literature

The valuation and corporate governance literature that are used to motivate the
hypotheses in section three will be discussed in detail. The review begins with a
discussion of the benefits of meeting or beating expectations, including higher returns
and market premiums. Secondly, literature relating to the mechanisms used to meet or
beat expectations will be discussed. Thirdly, the corporate governance literature that
focuses on mitigating the ability of firms to manage earnings will be discussed. Finally,
the literature that focuses on expectations management will be discussed.
2.1 The Valuation Implications of Meeting or Beating Earnings Benchmarks

The valuation literature begins with Barth et al. (1999) who use a price earnings
model to examine whether the earnings multiple for firms that had a pattern of increasing
earnings lasting five years or longer is higher than for firms with no such pattern. Barth
et al. (1999) find that the incremental earnings multiple for firms with increasing
earnings, compared to firms without this pattern of increases, is significantly larger for
firms with at least five years of increasing earnings after controlling for growth and risk.
Additional analysis examines the earnings multiple when a firm with a pattern of
earnings increases experiences an earnings decrease. The authors find that the
incremental earnings multiple decreases and after two years of decreased earnings the
earnings multiple is eliminated and insignificantly different from zero.

Bartov et al. (2002) use an event study methodology to examine the market

premium surrounding quarterly earnings after controlling for earnings forecast error for

 

 

Brim c
release

b} other
findtha
controll
premrur
1311):) 1
higher f
This su;
abnorm;
et al. [it
future p

lIlTOI'm a

 

 

firms that meet or beat expectations. Specifically, the authors examine whether the
premium is different for firms that meet rather than beat expectations. They also examine
whether the premium for beating is different from that for failing to meet expectations.
Bartov et al. (2002) focus on earnings forecasts for the current quarter made after the
release of the previous quarter’s earnings forecasts in an effort to eliminate contamination
by other news. After forming portfolios based on forecast error size, Bartov et al. (2002)
find that firms that beat expectations had the highest cumulative abnormal return after
controlling for the magnitude of the forecast error. This supports their contention that the
premium is higher for firms that beat rather than meet expectations. Further Bartov et al.
(2002) find that the average cumulative abnormal return, across all error groups, is 3.2%
higher for firms that beat expectations compared to those that fail to meet expectations.
This supports the authors’ hypothesis that the premium, in the form of the cumulative
abnormal return, for beating is different than that for failing to meet expectations. Bartov
et al. (2002) also examine the relationship between meeting or beating expectations and
future performance and find that the ability of a firm to meet or beat expectations is
informative with respect to future performance, after controlling for forecast error.
Kasznik and McNichols (2002) examine whether the market rewards firms that
meet expectations in the current period after controlling for the information in current
year earnings. This is tested by examining whether analysts’ predictions of subsequent
year earnings are affected by whether the firm met or beat expectations in prior years.
Kasznik and McNichols (2002) find that the abnormal annual returns are significantly

higher for firms that meet or beat expectations. The authors then attempt to ascertain

' L' ”01‘ error-'2]

 

theater t‘t‘
represent~ I
.\l;.\'icht§
assigned (I

tonsisters

thirty six

eredibtltt

tell as :I'

 

 

such. the
either be
ElbllOl'm;
met or t
Finns 1‘
’tl’the {1

results

whether the return is higher because of higher expected future earnings or whether it
represents a premium which is incremental to expected future earnings. Kasznik and
McNichols (2002) conclude that firms that meet or beat expectations have a higher value
assigned to them by the market and the market premium increases in value as the firm
consistently meets expectations. Further, the premium appears to be persistent as long as
thirty six months into the future.

Finally, Dopuch et al. (2003) extend the literature by examining whether there is a
credibility premium attached to the ability of a firm to beat both the analysts’ forecast as
well as the earnings of the comparable prior period, a time series based benchmark. As
such, there should be a premium for exceeding both benchmarks in addition to exceeding
either benchmark separately. Using quarterly data and examining the cumulative
abnormal returns surrounding the release of the forecasts, the authors find that firms that
met or beat both analysts’ and time series forecasts received an abnormal return of 4.5%.
Firms that only met or beat analysts’ forecasts received an abnormal return of 2.5% and
those that met or beat time series forecasts received an abnormal return of 3.3%. The
results suggest that time series forecasts provide incremental explanatory power to
meeting or beating analysts’ forecasts. In addition, the results suggest that the two
benchmarks are complements rather than substitutes. Time series forecasts are used to
assess the credibility of analysts’ forecasts and as such firms that exceed both forecasts
are rewarded with a credibility premium.

The papers in this section provide evidence that there are valuation benefits that

accrue to firms that meet or beat expectations. These valuations benefits include an

.' .r.'J¥fl'

 

”"4

mtersaf
earning
expect:
ma'ltet
premiu:
r 4 4...

dUuLIOI

tests I]

a break

behat‘ie

 

 

earnings multiple that increases as a pattern of earnings increases continues. If there is a
reversal of fortune, the earning multiple is completely eliminated after two years of
earnings declines. Another valuation benefit accrues to firms that beat rather than meet
expectations in the form of a higher cumulative abnormal return. In addition, a higher
market premium exists for firms that meet or beat expectations. This higher market
premium continues to increase as the firm consistently meets expectations. Finally, in
addition to a premium for meeting or beating analysts’ expectations a credibility premium
exists for firms that meet both times series forecasts as well as analysts’ forecasts.

Another stream of literature that will be discussed focuses on the consequences of
a break in a string of consecutive earnings increases. Ke et al. (2003) examine the trading
behavior of insiders in the quarters prior to a break in a string of consecutive earnings
increases. Ke et al. (2003) first establish a negative relationship between the stock price
response and a break in a string of consecutive earnings surprises that becomes more
negative the longer the break. After this relationship is established, the authors examine
the behavior of insiders leading up to the break in earnings. Ke et al. (2003) find that in
the third through eighth quarters prior to the break, insiders make significantly more
stock sales. Ke and Petroni (2003) examine the trading behavior of institutional investors
in the quarters leading up to a break in a string of consecutive earnings increases. Ke and
Petroni (2003) find evidence that similar to insiders, transient institutional investors also
sell their stock prior to a break in earnings increases but the sales occur closer to the

break. Further, the sales are triggered by breaks that lead to the largest stock price

a «my: 6'. -

 

GEN

grass 11‘:
hi lon:
rim is :

CC‘Ili6S I

 

16an .
milling
a\Oid ‘.

[his aw,

Wat:

 

declines. The largest stock price declines are caused by breaks that are longer. Those of
growth firms show breaks with larger declines in earnings and breaks that are preceded
by longer strings of consecutive earnings increases. Both of these papers show that the
firm is affected in an adverse manner when a string of consecutive earnings increases
comes to an end.

Skinner and Sloan (2002) also address the consequences of a break in a string of
earnings increases. This paper focuses on the abnormal return surrounding earnings
surprises for growth stocks compared to all other stocks. The regression analysis allows
a differential response for good and bad news earnings. The results indicate that for
growth firms the average realized return is larger in magnitude for negative earnings
surprises than for positive earnings surprises. Skinner and Sloan (2002) conclude that
this asymmetric response is due to overly optimistic expectations for growth stock which,
when not met, lead to price declines.

The costs to the firm of an earnings decline include a sell off of shares by insiders
and institutions. Further, growth stocks appear to be affected by an asymmetric price
response when a firm reports bad earnings news. The smaller return for a negative
earnings surprise should lead firms to avoid such surprises. The mechanisms used to
avoid these negative earnings surprises will be discussed in the next section.

2.2 Mechanisms for Avoiding a Negative Earnings Surprise

The mechanisms used by firms to meet or beat expectations are also pertinent to

this analysis. Matsumoto (2002) examines management incentives to avoid a negative

earnings surprise. The author also examines the mechanisms firms use to avoid a

10

 

 

negative 6
upts'afd.
probabilit
o'tmerSl’ilj
nhere e‘n
prababrli
lr
rneet ore
conditiot
mOdlllfit
model is
the pros
probabt

and let.

 

and lost

 

negative earnings surprise including managing expectations downward and earnings
upward. Matsumoto (2002) finds that there is a positive relationship between the
probability that a firm meets or exceeds the analysts’ consensus forecast and institutional
ownership, reliance on implicit claims, growth prospects and membership in an industry
where earnings are more value relevant. There is also a negative relationship between the
probability that a firm meets or beats analysts’ consensus forecasts and loss firms.

In the second analysis, Matsumoto (2002) examines the mechanisms firms use to
meet or exceed expectations. The ability of firms to manage earnings upward,
conditional on meeting or beating expectations, is addressed first. The Jones and
modified Jones models are used to develop a proxy for discretionary accruals. A logit
model is then used to examine the relationship between positive abnormal accruals and
the proxies that represent managers’ incentives to avoid a negative earnings surprise. The
probability of positive abnormal accruals is positively related to institutional ownership
and long term growth. A negative relationship exists between positive abnormal accruals
and loss firms.

The issue of expectations management is then addressed. A method similar to the
Jones and modified Jones models is used to develop a measure of forecast guidance. The
method involves comparing the expected forecast, which is based on prior earnings
changes and returns during the period, to the consensus analyst forecast. Again, the
author uses a logit model to examine the relationship between the probability of a
unexpected negative forecast and managers’ incentives to avoid a negative earnings

surprise. The probability of an unexpected negative forecast is positively related to

11

 

lfiSlllu’liOl
negatiw‘.’
and loss 1
earnings
D
outcome
manager
became
gains. S
HCCTLILil.‘

O'ber f1:

 

discrett.

The an:

 

 

institutional ownership, long term growth and the firm’s reliance on implicit claims. A
negative relationship was found between the probability of a negative earnings surprise
and loss firms. Based on the analyses, Matsumoto (2002) concludes that firms manage
earnings upward and guide expectations downward to avoid negative earnings surprises.

Dechow et al. (2003) also examine the ability of firms to avoid a negative
outcome by focusing on the “kink” in the earnings distribution and whether earnings
management provides a complete explanation. The kink, or discontinuity at zero, occurs
because there are too few firms reporting small losses and too many firms reporting small
gains. Specifically, the authors test whether small profit firms have higher discretionary
accruals than small loss firms and the population at large by comparing them with all
other firms in the sample. Dechow et a1. (2003) find a significant difference between the
discretionary accruals of small profit firms and those of all other firms in the sample.
The authors, however, fail to find a significant difference between the discretionary
accruals of small profit and small loss firms. In the absence of a significant difference in
discretionary accruals between small profit and loss firms, Dechow et al. (2003) conclude
that the kink can not be explained completely by firms using earnings management to
avoid reporting a small loss. The kink can not be explained completely by earnings
management but this does not completely rule out earnings management as a mechanism
to avoid a negative earnings surprise.

Finally, Rowchowdhury (2003) examines the manipulation of real activities that
affect both accruals and cash flow from operations to meet or exceed certain thresholds.

Rowchowdhury (2003) examines deviations from expected cash flow from operations

12

 

pen 2
eal HC‘
unsust;
ecrea:
predu:
tdefine
‘irms a
firm-ye
exhibit
and um
conclue
sales ar
3\ bid lt‘

boost e

analyst

mans;

 

that if:
1’“) l “-
”UTE.

PIESejj

 

 

given a firm’s sales level for three types of manipulations of real activities. The three
real activities include: accelerating the timing of sales or generating additional
unsustainable sales through increased price discounts or more lenient credit terms,
decreasing discretionary expenses and reporting a lower cost of goods sold by increasing
production. In order to test for this deviation, firm-years in the interval just right of zero
(defined by Burgstahler and Dichev (1997)), are classified as suspect since in this interval
firms are suspected of managing earnings to report income marginally above zero. These
firm-years are then compared to the rest of the sample. Suspect firm year observations
exhibit unusually low cash flow from operations, unusually low discretionary expenses
and unusually high production costs for a given level of sales. Rowchowdhury (2003)
concludes that the results are consistent with managers offering price discounts to boost
sales and engaging in overproduction to reduce reported cost of goods sold in order to
avoid losses. The results also suggest that managers reduce discretionary expense to
boost earnings above zero.

The valuation literature enumerates several benefits for meeting or beating
analysts’ expectations. These benefits make the use of earnings and expectations
management as mechanisms to avoid negative earnings surprises more compelling. Now
that the mechanisms for avoiding negative earnings surprises have been discussed, the
corporate governance mechanisms that are used to mitigate earnings management will be

presented.

13

l'rw . 4“»

 

23C0|

comral

anali 515

and earr
manage!
tests the
indepen

board. 11

abnorm
tnuhix,
outside

Klein .

 

mans-5

board .
lltCte;
in Car
“hEn

meet,

 

2.3 Corporate Governance and Earnings Management
In addition to the literature related to meeting or exceeding expectations, the
corporate governance literature that focuses on earnings management is essential to the

analysis and will be discussed. Klein (2000) examines corporate governance mechanisms

and earnings management. The author predicts a negative relationship between earnings
management and board independence and the presence of a blockholder. The author also
tests the relationship between earnings management and CEO shareholdings. Board
independence is measured by the percentage of outsiders on the audit committee and the
board, indicator variables are used if the board or audit committee is dominated by
independent directors and the audit committee is either totally independent of
management or dominated by a majority of outside directors. The absolute value of
abnormal accruals is used to proxy for earnings management. The results of a
multivariate analysis show that boards and audit committees comprised of a majority of
outsiders are negatively related to the absolute value of adjusted abnormal accruals.
Klein’(2000) concludes that boards and audit committees structured to be independent of
management are best able to perform their independent oversight functions.

Another paper that examines the relationship between earnings management and
board monitoring is Peasnell et al. (2000). This paper, however, focuses on income
increasing earnings management around the threshold of zero earnings and zero changes
in earnings. Peasnell et al. (2000) also examine income decreasing earnings management
when earnings exceed zero and last period’s earnings by a large margin. The authors

predict a negative relationship between earnings management and the proportion of

14

rm“

 

.4 ~
I!“

 

outside (
L’tt exist
when m;
are used
earnings

below 2:

 

 

outside directors on the board. The authors also predict a negative relationship between
the existence of an audit committee and earnings management which is more pronounced
when managerial ownership is low. In order to test these relationships abnormal accruals
are used as a proxy for earnings management. A negative relationship is found between
earnings management and the proportion of outsiders on the board when earnings fall
below zero or prior year’s earnings. When earnings exceeded zero or last period’s
earnings by a large margin, no relationship was found between earnings management and
the proportion of outsiders on the board. There was no significant relationship between
earnings management and the existence of an audit committee when earnings either fell
below or exceeded a threshold. Based on the results, Peasnell et al. (2000) conclude that
the presence of outsiders seems to mitigate earnings management around certain
thresholds.

Chtourou et al. (2001) examine the relationship between earnings management
and board characteristics by using two sub-samples of firms, one with high levels of
discretionary accruals and one with low levels of discretionary accruals. This paper
differs from other papers in the cited literature because it not only focuses on board
independence but also examines the competence and activeness of the audit committee.
The hypotheses are based on recommendations from a variety of groups including the
Blue Ribbon Commission, SEC and stock exchanges. Chtourou et al. (2001) hypothesize
a negative relationship between earnings management and an independent audit
committee and board, a financially competent and active audit committee and outside

director ownership. N o prediction is given for the relationship between earnings

15

managéfl'.

measure
codified
compete
of outsit
negabxe
account
Abnorn
tin-alts:
financr
CFAti
IESpQr;
Clair

hi the

fianu:
BCCrUI
lncor
Sling

Cant.

 

management and board size. Board independence is measured by the inclusion of
independent directors on the board, separation of the roles of chair and CEO and the
presence of an independent nomination committee. Activity by the audit committee is
measured by the presence of a formal written charter where oversight responsibilities are
codified and the number of meetings of the audit committee in a given year. Financial
competence is measured by the average non-executive director’s tenure and the number
of outside directorships of non-executive directors. Peasnell et al. (2001) also predict a
negative relationship between earnings management and the presence of a certified public
accountant, CPA, or a certified financial accountant, CFA, on the audit committee.
Abnormal accruals are used to proxy for earnings management. The results of the
analysis indicate a negative relationship between earnings management and board size,
financial competence, percentage of independent directors, presence of either a CPA or
CFA on the audit committee, presence of a written charter that spells out the
responsibilities of the audit committee and meetings more than twice per year. Peasnell
et al. (2001) conclude that the results support the effectiveness of the reforms called for
by the SEC, stock exchanges and the Blue Ribbon Commission.

Chung et al. (2001) focus on the relationship between institutional investors and
earnings management and predict a negative relationship between the two. Discretionary
accruals are used to proxy for earnings management and the paper focuses on both
income increasing and decreasing discretionary accruals. The analysis focuses on two
situations where management has the incentive to manage earnings. The first situation

exists when firms have poor current performance but good future prospects. The second

16

a

.4 9" "c TE‘ICW.

 

{fl— "’

situatit
Each }‘
used tc

12001 t

intestc
earning
intestc
of insti
( 1999 l
locuse

IElEiIltft

 

 

 

 

situation exists when firms have good current performance but poor future prospects.
Each year the median of institutional investment is determined, and a dummy variable is
used to represent firms with institutional investment above the median. Chung et al.
(2001) predict that this group of investors has the incentive to monitor management. The
results indicate that large institutional shareholdings deter both income increasing and
decreasing discretionary accruals. The results provide evidence that institutional
investors do mitigate the ability of managers to behave opportunistically in the case of
earnings management.

Finally, Mohan et al. (1999) address the relationship between institutional
investors and earnings management. Using absolute discretionary accruals as a proxy for
earnings management, the authors find a negative relationship between institutional
investors and earnings management. Mohan et al. (1999) conclude that the sophistication
of institutional investors allows them to see through earnings management. Mohan et al.
(1999) build on this result by examining whether institutional investors are overly
focused on current earnings to the detriment of future earnings. In order to test this
relationship five quintiles are formed based on institutional holdings and a ratio is formed
that captures the extent to which prices lead accounting earnings. Mohan et al. (1999)
find that there is a positive relationship between the level of institutional holdings and the
extent to which prices lead earnings. Mohan et al. (1999) conclude that institutional
investors do not fixate on current period earnings and look beyond the current period to

assess firm value.

17

 

 

nitigatir
has been
on the be
hetti eer‘.
investor

the man

110“ c
author
Pmpt!‘
lnsttt;

amrz

 

lift] L. .

EOSTI;

Institutional investors as well as outsiders seem to have a role to play in
mitigating the ability of firms to manage earnings. Additionally, a negative relationship
has been established between earnings management and a higher proportion of outsiders
on the board of directors. Additionally, a negative relationship has also been established
between earnings management and the proportion of shareholdings of institutional
investors. Further, Mohan et al. (1999) provide empirical evidence that does not support
the managerial myopia hypothesis. Institutional investors appear not to fixate on current
earnings and take a long term view when assessing firm value. There is also evidence
that audit committees and board competence also reduce the ability of the firm to manage
earnings. The next section of the paper discusses expectations management.

2.4 Corporate Governance and Expectations Management

Expectations management is one of two mechanisms firms use to meet or beat
analysts’ expectations (Matsumoto 2002). Expectations management occurs when firms
communicate with analysts’ to dampen their forecasts and bring them more in line with
the company’s expectations (Boni and Womack 2003). Ajinkya et al. (2003) examine
how corporate governance mechanisms affect the disclosure process. Specifically the
authors examine the relationship between governance mechanisms (including the
proportion of the board consisting of outsiders and the percentage of shares held by
institutions) and properties of management forecasts. Ajinkya et al. (2003) hypothesize
a positive relationship between corporate governance mechanisms and the occurrence,
frequency, specificity and accuracy of the forecasts. The authors also hypothesize a
positive relationship between the conservativeness of the forecast and the presence of

corporate governance mechanisms. The results indicate a positive relationship between

18

 

 

tomorat
1018C :15 L‘

specific:

the boat
proi'idc
disclos _

then cu

 

comm 'u‘

nith if:

he six
autho:l
Caminl
manag

s

adéllh

 

5:21ij
6min
bl lss

ease;

I

SUI-TI

corporate governance mechanisms and occurrence, frequency and accuracy of the
forecasts. There is also a positive relationship between institutional investors and
specificity of the forecasts.

Finally, the presence of institutional investors and the proportion of outsiders on
the board results in the release of less optimistically biased forecasts. The analysis
provides support for the argument that corporate governance mechanisms improve the
disclosure process. Specifically, frequency and occurrence of forecasts are improved
when corporate governance mechanisms are in place. This suggests that firms are
communicating with the market more often and keeping the markets expectations in line
with those of management.

Kasznik and Lev (1995) examine management’s discretionary disclosures during
the sixty day interval leading up to’the announcement of a large earnings surprise. The
authors find that the likelihood of issuing a warning increases with the size of the
earnings surprise, the existence of an earlier prospective disclosure made by
management, a firm’s membership in a high technology industry and firm size. In
addition, firms with larger impending earnings surprises issued more quantitative and
eamings-related warnings. Kasznik and Lev (1995) conclude that the firms with larger
earnings surprises, particularly disappointments, attempt to narrow the expectations gap
by issuing a warning. Managers appear less concerned with narrowing positive
expectation gaps than with avoiding large earnings disappointments.

Skinner (1997) addresses discretionary disclosures but focuses on those

surrounding stockholder lawsuits. The author specifically focuses on whether managers

19

shortly
PIi‘flIITlt
expect;

preanr.

line:
5

title;

 

can reduce stockholder litigation costs by disclosing adverse earnings news early. Using
a sample of firms that were the subject of lawsuits attributable to earnings related
information releases and a matched sample of firms that were not the subject of lawsuits;
the author concludes that lower litigation costs in the form of smaller litigation
settlements result from timely disclosure of earnings problems.

Soffer et al. (2000) also examine the strategy of preannouncing tentative earnings
shortly before the formal release of earnings. The amount of information a manager
preannounces affects the degree of surprise and is a way of managing market
expectations. Soffer et al. (2000) find that firms with positive news at the
preannouncement date release only some of their news. On the other hand, firms with
negative news at the preannouncement date release all of their news. The results suggest
that managers strategically select preannouncements to avoid negative earnings surprises.
This conclusion is consistent with Kasznik and Lev (1995) who find that firms with
larger earnings surprises, particularly with disappointments, attempt to narrow the
expectations gap by making voluntary disclosures.

Richardson et al. (2001) address expectations management in the context of
equity issuance and insider trading incentives. The authors examine the process of
managers “walking down” analysts’ forecasts to beatable targets. The motivation for this
“walk down” is the possible issuance of additional company securities or the sale of
securities by insiders after the earnings announcement. The metric used to measure
whether expectations management occurred is whether between the beginning and end of

the forecast period analysts’ forecasts switch from optimism to pessimism. Richardson et

20

ffl'fl'llflt' "6'4 D '31:

 

 

al. 1 3'
comm
comm
folio“

analis

anal )5
attest;
analyst
eamin
quote

a beth

mana
and n
3171‘s.

l0 Sl'.

 

al. (2001) find that forecast pessimism prior to an earnings announcement is more
common for firms that are about to issue new equity. Forecast pessimism is also more
common when the insiders are net sellers of the firm’s stock in the period immediately
following an earnings announcement suggesting that managers opportunistically guide
analysts’ forecasts down.

Cotter et al. (2000) also investigate expectations management but focus on
analysts’ reaction to explicit management earnings guidance. Specifically, the paper
attempts to determine if analysts regurgitate information provided by management or if
analysts make their own assessment about the company based on the management
earnings forecast and other company information. Using a sample of firms that issued
quarterly management earnings forecasts Cotter et al. (2000) first address the question of
whether management attempts to lead analysts. Based on the directional bias in
management earnings forecasts the authors conclude that managers do indeed issue
pessimistic guidance. Based on this result, the authors investigate how analysts react to
management’s guidance. Cotter et al. (2000) find that analysts quickly react to
management’s forecast with revisions. Those revisions are independent of management
and not just a regurgitation of management’s forecast. However, although analysts
appear to behave independently, the end result is that their forecasts shift from optimism
to slight pessimism in response to bad news management guidance and from pessimism
to slight optimism in response to good news management guidance. The shift to
pessimism in the face of bad news (albeit less than management anticipated) may not

eliminate the firm’s chances to beat expectations. It is unclear whether the shift to

21

 

pessi:

enougl

oopofl

1156 m;

 

 

pessimism, in the face of bad news, although less than management anticipated is
enough for the firm to beat expectations.

There is evidence to suggest that firms use management earnings forecasts
opportunistically in the context of equity issuance or insider stock sales. Further, firms
use management guidance to bring analysts expectations in line with management
expectations. However, empirical evidence suggests that analysts may revise their

forecasts but fall short of revising to the extent suggested by management guidance.

22

 

 

3.1H)

farm's
10111111.
ireque
on the
tumor
The m
\Oltllll
disclor
to foe
Setting
capes

Value

 

 

 

 

Chapter 3 - Hypotheses Development

3.1 Hypothesis One

The existing literature documents a positive relationship between the strength of a
firm’s corporate governance mechanisms and both the quality and the quantity of its
voluntary disclosures. Ajinkya et al. (2003) find that the probability of occurrence and
frequency of management forecasts are positively related to the percentage of outsiders
on the board. Ajinkya et al. (2003) also find that forecasts issued by firms with strong
corporate governance mechanisms were more accurate and less optimistically biased.
The monitoring provided by corporate governance mechanisms is concluded to constrain
voluntary disclosure decisions. This monitoring effect should extrapolate to other
disclosure processes within the organization where there are incentives2 for management
to focus on objectives that are contrary to the maximization of shareholder value. In this
setting, this includes focusing on the valuation benefits of meeting or beating analysts’
expectations in the short run which may be to the detriment of long run shareholder
value.

A survey of chief financial officers conducted by Graham et al. (2004) supports
the contention that focusing on meeting or beating analysts’ expectations in the short run
may be to the detriment of long run shareholder value. Managers surveyed by Graham et

al. (2004) suggest that due to the severe market reaction caused by missing an earnings

 

2Core et a1 (1999) document a setting where management has an incentive to behave opportunistically but
the presence of corporate governance mechanisms mitigates this ability. The paper finds that CEOs earn
greater compensation when governance structures are less effective. Specifically, CEO compensation is
higher when the CEO is also the board chair, the board is larger and a greater percentage of the board is
composed of outside directors who are either appointed by the CEO or considered gray directors. CEO
compensation is also higher when outside directors are older and serve on more than three other boards.

23

3r.

 

 

target. 1
target.

mainte:
project
Coca C
at C oc;
strateg'
the lon

estima

prelim
111681 f
Cchc
PI€\‘j(
find '.
I0 me
prei j

331T;

 

 

 

target, firms are willing to sacrifice economic value in order to meet a short run earnings
target. Managers admit that they would take real economic action such as delaying
maintenance and advertising expenditure and even give up positive net present value
projects to meet earnings benchmarks. In response to this short run focus, firms such as
Coca Cola and Gillette have discontinued the issuance of earnings estimates. Managers
at Coca Cola felt that short term guidance prevented a more meaningful focus on the
strategic initiatives that the company was taking to build its business and succeed over
the long term. Gillette echoed this sentiment by suggesting that focusing on earnings
estimates distracted management from long term goals (Kueppers and Weaver 2003).

Short run valuation benefits in the form of higher stock prices and market
premiums accrue to firms that MBE. Kasznik and McNichols (2002) find that firms that
meet expectations are rewarded with a higher stock price after controlling analysts’
expectations of future earnings. However, in the long run, the market penalizes firms that
previously met expectations but subsequently fail to do so. Similarly, Barth et al. (1999)
find that firms which meet expectations receive a higher return than their peers who fail
to meet expectations. However, the earnings multiple decreases when a firm that
previously exhibited a pattern of increasing earnings subsequently experiences an
earnings decrease. Finally, Bartov et al. (2002) find larger earnings multiples accrue to
firms exhibiting a pattern of increasing earnings. However, the future performance of
firms that used earnings or expectations management was inferior to that of firms that did
not use either of these mechanisms. Firms that managed earnings or expectations,

however, still fared better than firms that failed to meet or beat their earnings

24

2:}

“‘31:: .' I-Fl’"’£" P“) ' 1'5"" .

 

 

expectations. The valuation benefits suggest meeting or beating expectations is
advantageous to the firm in the short run but meeting or beatin g expectations may have
some long run negative ramifications when a firm that met or exceeded expectations in
the past subsequently fails to do so.

Additional empirical evidence that supports the long run costs associated with
failing to meet or beat analysts’ expectations is provided by Ke et al. (2003), who
establish a negative relationship between the stock price response and a break in a string
of consecutive earnings increases. Similarly, Skinner and Sloan (2002) find that negative
earnings surprises for growth firms are penalized more than positive earnings surprises
are rewarded. Collectively, these papers support the contention that there are costs
associated with an earnings break. Further, these papers are consistent with anecdotal
evidence in the financial press that detail the consequences associated with failing to
meet or beat analysts’ expectations3. Additionally, this research provides preliminary
support for the arguments of managers and public accountants that playing the earnings
game may erode long run firm value (Favaro and Rotz 2004; PricewaterhouseCoopers
2002).4 Consistent with practitioners’ arguments that meeting or beating analysts’

expectations forces managers to focus on short run performance, Roychowdhury (2003)

 

3 For example, on May 15 , 2003, when Target missed its first quarter earnings target by one cent, shares of
the stock fell in early morning trading (Reuters 2003). On December 22, 2002, when Tupperware warned
that its annual earnings would fall short, the stock lost 14% of its value (CNNMoney 2002).

4Favaro and Rotz (2004) suggest that a focus on short term targets, specifically MBE, stands in the way of
building the capabilities and making the investments required to increase a company’s longer term value.
Reducing the level of research and development activities, pulling back from marketing investments and
delaying projects with long term payoffs are moves to cut costs significantly and ratchet up earnings in the
near term. Over time they rob a business of growth drivers and can actually destroy value.
PricewaterhouseCoopers (2002, p. 76) argue that erosion in long run value occurs when companies meet or
beat expectations by making business decisions that provide a short run earnings boost at the expense of
lower future cash flows.

25

long rt
persor
oppor
Furthc
and K
nen l}

stocl;

 

5&1?

that
that

EXP;

 

 

finds that firms that report small annual profits exhibit unusually low cash flow from
operations and unusually high production costs. These suggest that managers seek to
avoid losses by offering price discounts to boost sales and engaging in overproduction to
reduce reported costs of goods sold.

Although the short term stock price run up may not result in an increase in the
long run value of the firm, it does provide managers with opportunities to enhance their
personal wealth at the expense of shareholders. Richardson et al. (2001) find firms that
opportunistically “walk analysts down” to beatable forecasts are net issuers of equity.
Further, the managers are net sellers of stock from their own personal accounts. Aboody
and Kasznik (2000) find that CEOs preannounce bad news to lower the strike price of
newly granted options. These opportunistic voluntary disclosers ultimately maximize
stock option compensation. Matsunaga and Park (2001) find that CEOs are penalized
with a smaller bonus which shrinks as their firms continue to miss earnings benchmarks.
Finally, CEO career concerns provide additional incentives to meet or beat expectations.
Puffer and Weintrop (1991) find that CEO turnover is more likely when reported annual
earnings per share fall short of expectations.

The premise of this paper is that corporate governance mechanisms exist to ensure
that long run value is maximized and may provide one way to differentiate between firms
that play the earnings game and those that do not. Meeting or beating analysts’
expectations does not appear to provide long run value, but the short run stock price run
up presents self interested managers with incentives to maximize personal wealth at the

expense of shareholders. This leads to my first hypothesis:

26

. .u Riff?! 1: cw"

 

ffr'

23‘
::

sine:
Eamn
althot
Fir. s
in a re
mama;
about

tells;

SUg:
size

at a‘

 

 

 

 

H1: Firms that fail to meet or beat analysts ’ expectations have stronger

corporate governance mechanisms than firms that meet or beat

analysts’ expectations
3.2 Hypothesis Two

Matsumoto (2002) presents evidence that firms meet or exceed analysts’
expectations by using both earnings management and expectations management.
Earnings management occurs when managers use their accounting discretion to MBE
although the underlying firm performance is not consistent with market expectations.
Firms manage earnings upward by increasing discretionary accruals to a level that results
in a reported earnings number that exceeds analysts’ forecasts. Expectations
management occurs when managers, through voluntary disclosures, convey information
about underlying firm performance such that market participants adjust their forecasts to
reflect managers’ predictions about expected firm performance.

Corporate governance mechanisms have been shown to constrain a firm’s ability to
manage earnings (Klein 2002; Peasnell et al. 2000). The corporate governance literature
suggests that the presence of a majority of outsiders on the board of directors, small board
size and the presence of institutional investors mitigate earnings management. Peasnell
et al. (2000) find that in a sample of United Kingdom firms the presence of outsiders
reduces the likelihood of income-increasing accounting choices. Klein (2002) also
examines this relationship and finds income increasing accounting choices vary
negatively with board independence, as measured by boards and/or audit committees

composed of a majority of independent directors.

27

C1110:
small;
at al. l
intest

Bane:
C

gown
make
foreca-
proces
manag
meOI
OCCUIT

“Elmo

 

‘Lhfi for

are mi

 

..
.._.
r n.

are u:

 

Chtourou et al. (2001) find that firms with low levels of earnings management had
smaller boards that were composed of experienced independent board members. Chung
et al. (2002) examine earnings management and find that the presence of institutional
investors discourages managers from behaving opportunistically. Bushee (1998) and
Bange and DeBondt (1998) examine the manipulation of research and development
expenditures and find that the presence of institutional investors with a long term
investment horizon reduces the level of earnings management.

Prior literature has also established an indirect relationship between corporate
governance mechanisms and expectations management. One way firms inform the
market of their revised estimates is through the release of a management earnings
forecast. These voluntary releases to the market are considered a part of the disclosure
process. Ajinkya et a1. (2003) examine the disclosure process, specifically focusing on
management earnings forecasts and conclude that institutional ownership and the
proportion of outside directors have a favorable effect on the likelihood of forecast
occurrence as well as the frequency of issuance. One would expect the positive
relationship that exists between corporate governance and disclosure to manifest itself in
the form of expectations management. This would suggest that firms MBE by keeping
analysts abreast of earnings news in a timely manner. As a result, analysts’ expectations
are more in line with the actual results of the firm.

Corporate governance mechanisms have been shown to mitigate the ability of a
firm to manage ea'mings. Therefore firms with strong corporate governance mechanisms

are unlikely to use earnings management to meet earnings targets. In contrast, firms with

28

strong corporate governance mechanisms that MBE are more likely to manage

expectations so that market expectations are consistent with underlying firm

performances. Based on incentives provided by corporate governance mechanisms the

following hypothesis is proposed:

H 2: Firms with strong corporate governance mechanisms that meet or beat analysts’
expectations are more apt to do so by managing expectations while firms with

weak corporate governance mechanisms that meet or beat analysts’ expectations
are more apt to do so by managing earnings

 

5 In addition, prior literature suggests benefits from a transparent disclosure policy. Botosan (1997)
provides direct evidence of a relationship between disclosure level and cost of equity capital. Lang and
Lundholm (1996) provide evidence that firms with more informative disclosure policies have a larger
analyst following, more accurate analyst earnings forecasts, less dispersion among individual analyst
forecasts and less volatility in forecast revisions. An exception is Bushee and Noe (2000) who find that
transient investors are attracted to increased disclosure, these investors exacerbate a firm’s stock return
volatility with their short term investment horizon and aggressive trading strategies.

29

I
filib'}

 

4.1 53:

datum
expect
al. 20"
allBE
cams.
lSlC t
indus'

from

 

 

Chapter 4 - Research Design
4.1 Sample Selection

The initial sample consists of 10,256 firm year observations from the I/B/E/S
database of analysts’ forecasts from 1997-2001. To ensure consistency, analysts’
expectations and actual earnings are both obtained from the I/B/E/S database (Bhojraj et
al. 2003). Firm year observations are classified as meeting or exceeding expectations
(MBEzl) if reported earnings met or exceeded the outstanding consensus forecast at the
earnings announcement. Banks, insurance companies and other financial services firms,
(SIC 6000-6999) are excluded, as are utilities (SIC 4800-4999) and other quasi regulated
industries (SIC 4000-4499 and 8000 and above). All other financial data are obtained
from the Compustat database.

The corporate governance and institutional holding data are obtained from the
Compact Disclosure database. This database takes financial data and text from
stockholders and Form 10-K reports, registration and proxy statements and other sources
for all current and new companies registered with the SEC and transfers it into Excel
format. Financial statements plus quarterly statements are provided with financial
footnotes, and the full text of the president's letter and management discussions. Also
included are lists of subsidiaries, major officers and board members and their salaries, 5%
individual and 10% institutional holders, ranges of earnings forecasts from advisory
services, and lists of additional SEC filings and exhibits. The database is updated
monthly and from this source 48, 035 firm year observations from 1997-2001 were

obtained. The final sample consists of 4,865 firm year observations. The dramatic drop

30

il

.IL’s‘l- ! H
'|.- '1” ‘. curativuv

 

in fart
inn)

4.2 .\l

 

a,

 

 

in firm-year observations is mainly due to analysts’ forecasts not being available for the
firm year observations from the I/B/E/S database.
4.2 Model Specification - Hypothesis One

In order to perform the analysis for hypothesis one, a logit regression, as shown in
equation one, is used to test the probability that a firm MBE6:

Prob(MBE= 1) = p0 + ,6, %o UT+ flZTOTAL+ amps + ,6, %INST+ ,BSUNCFE

(l)
+ ,66 LITRISK+ ,6,VREARN+ flsGROWTH+ ,69 SIZE + fllOICLAIM+ e

where:

Prob(MBE=l) = dummy variable, taking on the value 1 if the firm met or
beat expectations and 0 otherwise;

Corporate Governance Variables:

%OUT = percentage of outsiders on the board of directors;
TOTAL = total number of directors on the board;
TOPS = percentage of institutional shares held by top five

institutions;
Control Variables:
%IN ST = percentage of institutional holdings;

UNCFE absolute value of the difference between the first
consensus forecast and actual earnings scaled by price;

LITRISK = dummy variable, taking on the value 1 if the firm is in a
high risk industry and 0 otherwise;

VREARN = dummy variable, taking on the value of 1 if in prior year
had loss before extraordinary item;

GROWTH = market capitalization at beginning of the year divided

by beginning book value;

 

6 The analysis is conducted with time dummies to control for time period specific effects.

31

‘3

no
' .14"-
5" Iim
'W

 

90%
%Ol

man

 

pert
outs
0f
De,
Er.

De.

 

of _,

 

 

SIZE = logarithm of the market value of equity at the

end of the year;

ICLAIM = factor composed of the individual variables: DUR:
a dummy variable taking on the value of 1 if the
firm is in the durable goods industry. R&D: annual
research and development expenditures scaled by total
assets. LABOR is defined as one minus the ratio of
total gross property, plant and equipment to total assets;

E, = error term.

This model is an extension of that used in Matsumoto (2002), and incorporates
three corporate governance variables. First, I consider the size and composition of the
board of directors as proxies for board governance. Second, I consider the concentration
of a firm’s institutional holdings because Shleifer and Vishny (1986) suggest that only
shareholders with significant holdings have an incentive to monitor management.

I briefly discuss each of the three measures used to operationalize corporate
governance. The first measure is the percentage of outsiders on the board of directors,
%OUT. Ajinkya et al. (2003) examine the relationship between properties of
management earnings forecasts and corporate governance mechanisms including the
percentage of outside directors. Ajinkya et al. (2003) conclude that the percentage of
outside directors on the board is positively associated with the occurrence and accuracy
of forecasts and negatively associated with optimistic bias in the forecasts. The corporate
governance literature also suggests that outside directors mitigate the incidence of fraud.
Dechow et a1. (1996) use a sample of firms who were investigated by the Securities and
Exchange Commission for violations of Generally Accepted Accounting Principles.

Dechow et al. (1996) conclude that the firms investigated were less likely to have a board

of directors dominated by outsiders and less likely to have an external blockholder

32

 

 

monitc

im‘CSll

boart

of th-

H1617:

 

monitoring management. Beasley (1996) also uses a sample of firms that were
investigated by the Securities and Exchange Commission for financial statement fraud.
Beasley (1996) concludes that the fraud firms had boards with significantly lower
percentages of outside members than a matched sample of non-fraud firms. Finally,
Farber (2003) finds that fraud firms work to improve their corporate governance
mechanisms by increasing the outside director percentage and the market responds
positively when there are large improvements in their corporate governance mechanisms.
Overall, I expect a negative relationship between %OUT and MBE.

The second measure of corporate governance is the number of directors on the
board. Core et al. (1999) find that the board is less effective at monitoring when the size
of the board is larger. Using a sample of publicly traded firms they find that a one
member increase in the size of the board is associated with a $30,601 increase in total
CEO compensation. A smaller board may form a more cohesive group and not be
distracted by factions that may form if there were more board members. However,
Chtourou et al. (2001) find that a larger board is associated with less earnings
management, but only for income decreasing accounting choices. A larger board may
therefore provide a broader base of expertise since there are more directors present with
diverse backgrounds. No directional hypothesis between board size and MBE is set forth
since the literature supports both small and large board size as corporate governance
mechanisms.

Corporate governance is also operationalized with the presence of institutional

investors who have long term incentives/preferences. Two proxies for the presence of

33

 

instzt
uh]:
inc 1
until
moti'

COIN

rein:

 

 

 

 

institutional investors will be used in this analysis. The first is institutional concentration,
which is measured as the proportion of institutional ownership accounted for by the top
five institutional investors. The second is the Herfindahl index7, an alternative
institutional concentration measure. The two institutional concentration measures are
motivated by Hartzell and Starks (2003) who find that institutional ownership
concentration is negatively related to the level of executive compensation and positively
related to the performance sensitivity of managerial compensation. Overall, I expect a
negative relationship between institutional investors with long term
incentives/preferences and MBE.
4.3 Control Variables

The percentage of institutional holdings is used as a control variable (Hartzell
and Starks 2003). This allows the level of total institutional holdings to vary separately
from the percentage of shares held by the top five institutional investors. Additional
variables are used to control for factors that may be related to a firm’s ability to MBE
including uncertainty in the information environment, litigation risk, value relevance of
earnings, growth, size and reliance on implicit claims with stakeholders. Initial forecast
error, UNCFE, captures the uncertainty in the information environment. It is measured as
the absolute value of the first consensus forecast minus actual earnings scaled by price.
Matsumoto (2002) finds a negative relationship between the initial forecast error and the

probability that a firm MBE.

 

7 The Herfindahl-Hirschman index uses the share ownership of all institutions; the index is calculated as the
sum of the squared institutional ownership.

34

. .P...
".‘J
J WIH”

 

indust

lnclus

 

 

 

Shareholder litigation can be precipitated by a sharp price decline, and certain
industries are more volatile and are thus more vulnerable to the actions of shareholders.
Inclusion in a high risk industry, LIT RISK, is used to proxy for litigation risk. High risk
industries include: biotechnology (SIC 2833-2836), computers (SIC 3570—3577 & 7370-
7374), electronics (SIC 3600-3674) and retail (SIC 5200-5961). The delineation is
consistent with prior research (Ali and Kallapur 2001; Bolliger and Kast 2003; Francis et
al. 1994; Matsumoto 2002; Soffer et al. 2000). Matsumoto (2002) finds a positive
relationship between MBE and existence in a high risk industry.

Value relevance of earnings, VREARN, is represented by an indicator variable if
a loss occurred in the prior year (loss before extraordinary items). The inclusion of the
value relevance of earnings is based on Hayn (1995) who finds a significant difference
between the return earnings relation for loss firms. Matsumoto (2002) finds a negative
relationship between the probability of MBE and the value relevance of earnings

Long term growth, GROWTH, represents future opportunities available to the
firm. The average realized negative return surrounding a negative earnings response is
larger in magnitude than the average realized positive return surrounding a positive
earnings response (Skinner and Sloan 2002). High growth firms would therefore have a
greater incentive to MBE. Consistent with Bolliger and Kast (2003) market to book ratio
proxies for the growth prospects of the firm. It is measured as the market capitalization
of the firm divided by book value, both at the beginning of the year.

SIZE, is measured by the logarithmic value of the market value of equity.
Matsumoto (2002) finds a positive relationship between the probability of avoiding a

negative earnings surprise and size.

35

 

;
t’
L
i?
s

 

E?! ”:1?

impli.‘
Bone

induV

Eli?

 

Finally, as in both Matsumoto (2002) and Bolliger and Kast (2003), reliance on
implicit claims with stakeholders, ICLAIM, is measured with proxies developed by
Bowen et al. (1995): DUR, R&D and LABOR. DUR is membership in a durable goods
industry (SIC codes 150-179, 245, 250-259, 283, 301 and 324-399). R&D represents
research intensity and is annual research and development expenditures scaled by total
assets. Finally labor intensity, LABOR, is defined as one minus the ratio of total gross
property, plant and equipment to total assets. A positive relationship was found between
the probability of MBE and reliance on implicit claims (Matsumoto 2002).

4.4 Model Specification - Hypothesis Two

To test hypothesis two, I limit the sample to those firms that meet or exceed
analysts’ expectations. In addition, I need proxies for earnings management and
expectations management. The cross sectional Jones model (1991) is used to determine
discretionary accruals, which serve as a proxy for earnings management. A similar cross
sectional analysis is used to determine the expected change in earnings. This is used to
identify firms that engaged in expectations management (see below for details). Once
these two proxies are determined, a multinomial logit regression of the two mechanisms
on the percentage of outsiders, board size and the presence of institutional investors with
long term incentives/preferences is conducted.

Performance-matched discretionary accruals are used to proxy for the degree of
earnings management. Kothari et al. (2001) find that performance matching provides a
well-specified and powerful test. The test is considered well-specified because the null
hypothesis is rejected at the nominal significance level when it is true. The test is

considered powerful because a false null hypothesis is rejected with high probability.

36

 

 

1
1mm

That

there

TOT

5"
r»

sq,

11L

7.5?

 

The cross sectional Modified Jones model (1991) is used to determine total accruals. If
there are a minimum of eight firms in a particular industry, the following equation is

estimated for each two digit SIC year grouping:

TOTAL _ ACCRUALS = a + fl1(ASALEs — AREC) + ,62PPE + g,

(2)
where:

TOTAL_ACCRUALS = difference between operating cash flows and
income before extraordinary items;

A SALES = difference between current and prior year sales;

A REC = difference between current and prior year
accounts receivables;

PPE = end of year property, plant and equipment
balance.

Total Accruals, TOTAL_ACCRUALS, is measured as the difference between operating
cash flows and income before extraordinary items from the statement of cash flows.
Change in sales, A SALES, represents the difference in sales from the previous year.
Change in account receivables, A REC, represents the change in accounts receivable from
the prior year. Property, plant and equipment, PPE, is the end of the year balance in the
property, plant and equipment account. All variables are scaled by prior year total assets.
The fitted values from equation (2) are used to determine nondiscretionary accruals.
Finally, discretionary accruals are measured as the difference between total accruals from
equation (2) and the fitted values. In this analysis, only those firms that use discretionary
accruals to increase earnings are of interest. As such, firm year observations with

positive values of discretionary accruals are classified as earnings managers.

37

 

Etpc‘
shat:
CUES
L’- Ill:
man.

beat;

 

In .

 

 

After determining firm year observations with positive values of discretionary
accruals, the expected change in earnings is calculated and used to identify firms that
engaged in expectations management. The methodology used is based on Bolliger and
Kast (2003), who adapt the Matsumoto (2002) methodology for use with yearly data.
Expectations management is defined as the difference between the expected earnings per
share based on a model of prior earnings and stock price changes and the last analyst
consensus earnings forecast released by analysts prior to the earnings announcement date.
If the expected earning per share is smaller than the consensus forecast then expectations
management is assumed to have occurred and analysts may have been walked down to a
beatable forecast.

The first step is to determine the yearly change in earnings, which is modeled as a
function of changes in earnings and returns cumulated over the current year for each firm

in a given industry, j, during year t.

 

AEPSU, AEPSWl
= a], + ,6”, *————— + B.” * CUMRETU, + c” (3)
P. . P. “ J
(0-1 111-2
where:
AEPSUl = earnings per share for firm i in two digit SIC code j in
year t, less earnings per share from the prior year;
P”, 2 price per share for firm i in two digit SIC code j at the

end of year t;

CUMRETU-t = cumulative daily excess return for firm i in two digit
SIC code j during year t; returns are cumulated from
three days after year t-l earnings announcement to 20
days before year t earnings announcement.

This model is estimated for each industry using all firms in that year that have the same

two digit SIC code. In each year there must be a minimum of eight firms in a particular

38

 

 

 

mi".

in»

0b.

 

 

industry. The parameter estimates are then used to determine the expected change in
earnings per share, (E[AEPS]):

AEPS.. _ .
——£ + p,,,_, * CUMRET] * PUH (4)

ijt -2

E[AEPSm] = Ia).-. + 3W *

The expected change in earnings per share, (E[AEPS]), is added to the prior year’s
earnings to obtain an estimate of earnings expectations for the current year that

represents underlying firm performance based on past earnings changes, (E [FEPS]):

E[FEPS ]=EPS +E[AEPSU,] (5)

ijt ijt—l
Finally, the unexpected earnings forecast, UEF, is computed as the difference between
the last consensus forecast released by analysts prior to the earnings announcement date

(FEPS) and the expected analyst forecast from equation (5):

UEF. = FEPSU, — E[FEPSm] (6)

m

In this analysis, only those firms that use expectations management to meet or
beat analysts’ expectations are of interest. As such, firm year observations with negative
unexpected earnings forecasts are classified as expectations managers.

Equation (7) is a test of hypothesis two. This analysis focuses only on firms that
met or exceeded analysts’ expectations by managing expectations or managing earnings.
For completeness, however, firms that manage both earnings and expectations or abstain
from managing either are included in the analysis. These firms make up 907 firm year

observations. Table 1 provides a description of the sample selection procedure.

39

LlllI

 

 

Pa

0f

 

or;

r.
fl.‘

Table 1
Formation of Final Sample for Hypothesis Two

Panel A: Description of sample selection procedure
Firms used to test

H1 3,185
Firms excluded

because no

discretionary

accrual data (997)
Firms excluded

because no

unexpected

earnings forecasts (750)
Firm year

observations 1 ,43 8

Firms that failed to
meet or beat
expectations (531)

907

Panel B: Composition of sample based on mechanism used to meet
or beat analysts’ expectations

Earnings Management
Positive Negative

Expectations Positive 283 226
Management Negative 254 144
907

 

A multinomial logit analysis that takes into consideration the four potential
outcomes is performed. Firms with negative unexpected earnings forecasts are classified
as expectations managers and the firm year observation is coded: MANAGE=1. Firms

with positive discretionary accruals are classified as earnings managers and the firm year

40

 

 

else:

'16" '

t . ‘

l -V
»

llllil‘. ._

REST.

 

erpc
prob

The

. ~13

 

observation is coded: MAN AGE=2. Firms with both positive discretionary accruals and
negative unexpected earnings forecasts are classified as both earnings and expectations
managers and the firm year is coded: MAN AGE=3. Firm with negative discretionary
accruals and positive unexpected earnings forecast are classified as neither earnings nor
expectations managers and the firm year observation is coded: MANAGE=0. The
probability of each outcome is calculated against the base case of managing earnings.
The following logit analysis is estimated for 907 firm year observations:

P(MANAGE= 1) = p, + p, %o UT+ flZTOTAL+ amps + ,3, %INST+ ,BSUNCFE

7
+ fl61CLAIM+ B, LITRISK+ ,6,VREARN+ ,6, GROWTH+ @0512)“ e ( )

where:

Prob(MANAGE=1) = categorical variable, taking on the value 0 if the firm
manages neither earnings nor expectations, 1 if the firm
manages expectations, 2 if the firm manages earnings
and 3 if the firm manages both earnings and
expectations;

Corporate Governance Variables:

%OUT = percentage of outsiders on the board of directors;

TOTAL = total number of directors on the board;

TOP5 = percentage of institutional shares held by top five
institutions;

Control Variables:

 

%INST = percentage of institutional holdings;

UNCFE = absolute value of the difference between the first
consensus forecast and actual earnings scaled by price;

LITRISK = dummy variable, taking on the value 1 if the firm is in a
high risk industry and 0 otherwise;

VREARN = dummy variable, taking on the value of 1 if in prior year

had loss before extraordinary item;

41

GROWTH

SIZE

ICLAIM

2 market capitalization at beginning of the year divided

beginning book value;

logarithm of the market value of equity at the

end of the year;

factor composed of the individual variables: DUR:

a dummy variable taking on the value of 1 if the

firm is in the durable goods industry. R&D: annual
research and development expenditures scaled by total
assets. LABOR is defined as one minus the ratio of total
gross property, plant and equipment to total assets;

= error term.

42

art

of r

are

to

me

 

 

Chapter 5 — Results

5.1 Descriptive Statistics

Table 2 reports. descriptive statistics for the total sample. Firms met or exceeded
expectations in approximately 60% of the firm years in the sample compared to 65% '
reported by Matsumoto (2002) and Bolliger and Kast (2003). Firms in the durable goods
industry represented 42% of the firm years in the sample which is the same as that
reported by Bolliger and Kast (2003) and the 45% reported by Matsumoto (2002).
Highly litigious industries represented 39% of the firm years in the sample compared to
37% reported by Matsumoto (2002) and 35% reported by Bolliger and Kast (2003).
Sample firms had a mean market value of equity of $3.96 billion and in 25% of the firm
years there was a loss in the prior year. Additionally, institutions owned on average 43%
of the stock outstanding compared to the 45% reported by Matsumoto (2002). On
average, boards in the sample are composed of eight members which is the same as that
reported by Peasnell et al. (2000). The average board is composed of 72 % outside board

members compared to the 71.3% reported by Ajinkya et al. (2003).

43

 

 

 

 

Table 2

Summary Statistics for Dependent Variable, Independent Variables and Control
Variables for 4,865 firm year observations for period 1997-2001

Variable“ b

MBE
%OUT
TOTAL

TOPS
%INST

UNCFE
LITRISK
VREARN
GROWTH

SIZEc
DUR
R&D

LAB OR

ICLAIM

a Prob(MBEzl)

%OUT
TOPS
TOTAL
%INST
UNCFE

LITRISK
VREARN
GROWTH

SIZE
ICLAIM

 

Standard 3rd
E Mean Deviation 1 st uartile Mag uartile

4,865 0.60 0.49 0.00 1.00 1.00
4,717 0.72 0.16 0.62 0.75 0.83
4,865 8.00 3.00 6.00 8.00 10.00
4,865 0.58 0.25 0.37 0.54 0.82
4,820 0.44 0.27 0.20 0.44 0.66
4,750 0.16 2.13 0.00 0.01 0.05
4,865 0.39 0.49 0.00 0.00 1.00
4,865 0.25 0.43 0.00 0.00 1.00
4,865 4.20 26.52 1.11 2.07 4.04
4,858 3,706 16,836 82 305 1,268
4,865 0.42 0.49 0.00 0.00 1.00
3,367 0.09 0.18 0.01 0.05 0.12
4,802 0.52 0.35 0.33 0.60 0.78
3,363 0.00 1.00 -0.54 0.09 0.56

dummy variable, taking on the value 1 if the firm met or beat expectations
and 0 otherwise;

percentage of outsiders on the board of directors;

percentage of institutional shares held by top five institutions;

total number of directors on the board;

percentage of institutional holdings;

absolute value of the difference between the first consensus forecast and
actual earnings scaled by price;

dummy variable, taking on the value 1 if the firm is in a high risk industry
and 0 otherwise;

dummy variable, taking on the value of 1 if in prior year had loss before
extraordinary item;

market capitalization at beginning of the year divided by beginning book
value;

= logarithm of the market value of equity;

factor composed of the individual variables: DUR: a dummy variable
taking on the value of 1 if the firm is in the durable goods industry. R&D:
annual research and development expenditures scaled by total assets.
LABOR is defined as one minus the ratio of total gross property, plant and
equipment to total assets;

bStatistics in this table are presented as actual values while logarithmic values are used in the

analysis.

cSize is presented in millions.

tang

 

 

ana

 

Table 3 reports Pearson (Spearman) correlations of MBE and the independent
variables in equation 1. The correlations between MBE and the three governance
variables are significant. Consistent with hypothesis 1, the correlation between MBE and
the proportion of institutional shares held by the top five institutions is negative and
significant. Also consistent with hypothesis 1, the correlation between MBE and the
percentage of outsiders on the board is negative and significant. However, the correlation
between MBE and board size is positive and significant. With respect to the control
variables, RD, DUR, LABOR, are also positively and significantly correlated. Factor
analysis is used to combine these three variables into one factor, ICLAIM. This factor is
used in the analysis singly and the individual variables are used collectively in separate
logit analysesg.

5.2 Regression Analysis - Hypothesis One

The results of the logit analysis used to test hypothesis One appear in Table 4.
Columns 2-3 present the results when the individual implicit claim variables are used.
The analysis is repeated in Columns 4-5 where the construct implicit claim, ICLAIM,
replaces the three individual variables. In the first analysis, TOPS (p<.05) and TOTAL
(p<.05) are significantly negative. This result suggests that as the shares held by the top
five institutional investors increase the firm is less likely to MBE, which is consistent

with hypothesis one. The results also suggest that as board size increases the firm is less

 

8 Principal components factor analysis is used to reduce the three variables to a single construct, ICLAIM.
One construct results from retaining factors with an eigenvalue greater than one. Since all eigenvalues
were greater than one this indicates that DUR, RD and Labor proxy for a single construct. The final
communality estimates for DUR, Labor and RD are .32. .47 and .40 for the single construct ICLAIM.

45

likely to MBE. There is mixed evidence in the corporate governance literature about the
influence of board size. The result is consistent with the literature that suggests larger
boards provide a broader base of expertise since there are more directors present with
diverse backgrounds (Chtourou et al. 2001). The coefficient on percentage of outside
directors is insignificant. This may be due to the lack of variability throughout the
sample years. The control variables in the analysis have the same signs although varying
levels of significance relative to those reported by Matsumoto (2002) and Bolliger and

Kast (2003).

46

1.9—3.3-..31 _:.:==.v uuC-a 1.u—a_-;..:> ci..-i.lu.~1i- ill-:1- .I t
It 5:23.?

 
 

**

**

**

**

**

**

N. _ d
\- _ d
w _ d-
m _ d-
odd
cod
.1 d
9:

Ed-
Nod-
Ed-
m _ d-
mod
Ed

MmEC-H

**

*

**

**

**

**

**

**

m _ d
cod-
2 d
mod
odd-
Ed-
Omd
Ed-
do;

Ed-
Edd
w _ d-
Ed
Ed-

$575

in

**

«1*

as."

**

**

in

**
**

:25me 323 323.880 553% can 25% 8032280 5835

N _ d-
Ed-
m _ d-
mod
mmd
Ed-
Nad-
mod-
Ed.
2:

dod-
d. d-
omd-
dmd

.5215

in

3

**

**

**

**

**

**

**

S d
Ed
m _ d
Odd
mod-
Ed
vmd
Ed-
Sud
dud-
do;
Ed-
dmd
mmd-
EOE

**
In
.I,
.1”
I”

**

**

**

**

**

vmd-
w _ d-
3 d-
ood-
dad
cod
w _ d-
S d-
vod-
_ .d
mad-
dd.—
dad-
mod
dad-OH

**

“In

**

**

**

*

Ed
add
add-
mod-
m _ d
Ed-
Ed-
Ed
mod
Ed-
Ed
dmd-
dd.—
Ed-
H3005

.1,

**

**

i...

**

sdva a aacewar

2:8 3 285:»; ..
8d- 2242
8o «35
5o- 9.
:3. man
43 Ba
:3 5395
_ 3- zm<ma>
8.0 625.:
one- Buzz
N8 521s
3o. 28
8o 45.8
:3. Sea
2: mm:
mm:

83325 .2280 ES 83323 33531:— .ofiatg Eugenon 3.. £552 zeta—2.30

m 2:5.

47

haulj‘ ‘II-Ii‘l till I I

 

 

. my. :25» p. 612:.

 

F§.-l.r_iv4 .f r... E. . .. r .1
Fig,“ - L

**

**

*4.

I

**

**

*1.

“.3."

**

**

dd.—

Ed
Nbd
\- _ d
n _ d-
Ed-
mmd
3 d
ddd
N _ d-
N _ d

mmd-
Ed-
Ed-

2353

**

**

**

**

**

Nwd
dd.—
add
add
Ed-
Edd
Ed
0 _ d
Ed
Odd
Ed
vmd-
mod
Ed
«Oaxa—

**

**

**

,1"

**

**

**

**

**

**

Z-d
de
dd.—

Ed
5 _ d-
Ed-
mmd
m _ d
_ .d
\- Z d-
o. d
dmd
Ed-
Ed-
QM

**

**

**

.2836 323 223256 5:53am use 30% Ease—6:8 cemaom

\- _ d
8.0
m _ d
E;
Ed
Nod
mod-
m _ d-
Ed
8d
Ed
Ed-
mod-
Ed-
~50

**

**

**

“I,

**

**

.I.

**

**

**

**

odd-
Nod
N“ d-
Ed-
do;
Ed
\- _ d-
odd
odd-
Rd
ovd-
mmd
Ed-
2 d
mNE

Ed-
Nod-
Ed-
Edd
Ed
9:
odd
Nod
ddd
Odd
Ed-
mod
ddd
mod

EOMO

**

“1*

**

**

**

**

.1"

**

**

**

:58 a 285mm...-
modva a 58.58% -
woo E52
25 «83

93 9.
85. mag
Rd. mg
:3. F565
9: 52m”;
Ed amen-E
:3 E62:
m2- 5sz
3c Boy
25. 45.8
8d- .Sos
o3- mm:
52mg

$.aata> .8280 ES 823...; Emu—33...: 633:3 Emu—anon he in: 5:22.80
3.288 m «Ea-m.

48

 

 

V
‘Sl'

c
31‘
a)

the i
sin:
bit:
as 1

mo

 

5.3 Regression Analysis - Hypothesis Two

The results of the multinomial logit analysis used to test hypothesis Two appear in
Table 5. Columns 2-3 present the results when the firm manages earnings compared to
the base case of managing earnings. The coefficient on the concentration of the top five
shareholders variable, TOPS (p<.05) is positive and significant. The coefficient on the
board size variable, TOTAL (p<.05) is negative and significant. The results suggest that
as the number of shares held by the top five institutional investors increases the firm is
more likely to manage expectations downward rather than manage earnings upward.
Further, firms that manage expectations have smaller boards than firms in the sample that
manage earnings. This result is consistent with the stream of literature that finds smaller

boards are more cohesive and more effective at monitoring (Core et al. 1999).

49

 

..-IL

(3’

rL,

Table 4
Logit Analysis of the Probability of Meeting or Beating Analysts’ Expectations
and Corporate Governance Mechanisms
(all variables winsorized at .01 and .99 percentile)

Prob(MBE= 1) = )3, + p, %0UT+ p2T0TAL+ ,B3T0P5 + p, %INST+ ,BSUNCFE+ flOLITRISK
+ fl7VREARN+ BSGROWTH+ ,6,SIZE+ pm ICLAIM+ 8

Independent Predicted Coefficient p-value Coefficient p- value Coefficient p-value
Variablea Sign
Intercept -0.21 0.72 -0. 18 0.74 -0.33 0.47
Corporate
Governance
Mechanisms
%OUT - —0. 1 1 0.67 -0.08 0.75 0.02 0.91
TOTAL ? -0.05 0.01 -0.04 0.02 -0.04 0.01
TOPS - -l.08 0.00 -1. 10 0.00 -0.86 0.00
Control
Variables
%INST + 0.50 0.04 0.48 0.05 0.62 0.00
UNCFE - -l.l4 0.00 -1.17 0.00 -1.12 0.00
LITRISK + 0.14 0.14 0.21 0.02 0.19 0.01
VREARN - -0. 12 0.28 -0.1 l 0.31 -0.07 0.39
GROWTH + 0.01 0.08 0.01 0.09 0.01 0.03
SIZE + 0.25 0.00 0.25 0.00 0.23 0.00
DUR + .022 0.02
RD + 0.07 0.86
LABOR + 0.30 0.05
ICLAIM + 0.06 0.30
N = 3185 3185 4569
Log -l937 -l94l.7 -2818
Likelihood=
p-value= <01 <01 <01

3 Refer to Table l for variable definitions.

50

 

 

L1

 

Table 5
Logit Analysis of the Association between Corporate Governance Mechanisms and
the Probability of Managing Expectations versus Managing Earnings

P(MANAGE =1)“ = p, + p, %OUT + ,BZTOTAL + 5310105 + ,6, %INST + flSUNCFE
+ flézmelsx + B7VREARN + ,6, GROWTH + e, SIZE + 5,0 ICLAIM + 5

Manage Expectations Manage Neither Manage Both
Independent Coefficient p-value Coefficient p-value Coefficient p-value
Variableb
INTERCEPT -1.26 .22 -l.87 .10 -0.87 .39
Corporate
Governance
Mechanisms
%OUT -0.57 .35 -.59 .39 -0.49 .39
TOTAL —0. 13 .00 -0.07 . 13 -0.08 .05
TOPS 2.21 .00 1.55 .06 -0.71 .06
Control
Variables
%INST -0. 12 .82 0.17 .77 -0.77 .10
UNCFE 10.70 .00 8.70 .01 9.69 .00
LITRISK 0.56 .01 0.66 .00 -0.11 .55
VREARN 0.31 .21 0.62 .02 -0.46 .08
GROWTH 0.02 .16 0.02 .17 0.02 . l 1
SIZE 0.13 .15 -0.11 .10 0.33 .00
N: 907
Log -1142.03
Likelihood:
p-value= <05

3 MANAGE is a categorical variable equal to 0 if a firm manages neither earnings nor
expectations, 1 if a firm manages expectations, 2 if a firm manages earnings and 3 if a
firm manages both.

b Refer to Table 1 for variable definitions.

51

 

 

all

103

CO.“

I .11.

it"

fin

301'

 

 

5.4 Summary
As discussed above, the results suggest that the holdings of the top five

institutional investors limit the ability of the firm to manage earnings upward which is
consistent with prior corporate governance literature (Chtourou et al. 2001; Chung et al.
2001; Klein 2002; Peasnell et al. 2000). The results also suggest that the holdings of the
top five institutional investors are associated with a firm that manages expectations.
This result is consistent with the expectations management literature, which finds that
firms walk analysts down to beatable forecasts (Cotter et al. 2000; Richardson et al.

2001).

52

 

 

3ft.

dt'

6.1

 

r9
I

=4

 

 

 

Chapter 6 - Conclusions and Future Research Opportunities

This chapter presents conclusions related to the results of the multivariate
analyses of Chapter 6. The conclusions are followed by suggestions for further
developing this study. Finally, limitations are discussed.
6.1 Conclusion

This paper examines the relationship between corporate governance mechanisms
and MBE. I document that firms with strong corporate governance mechanisms are less
likely to MBE. This suggests that firms with strong corporate governance are more
concerned about long run costs such as those resulting from a break in a string of
earnings increases. This also suggests that firms with strong corporate governance
mechanisms may judge corporate managers on the basis of long term reported earnings
and thus have less of a need to play the earnings game (Wahal and McConnell 1999).

For firms that choose to MBE, this paper also examines the relationship between
corporate governance mechanisms, earnings management and expectations management.
These are two mechanisms firms use to MBE (Matsumoto 2002). I find that corporate
governance mechanisms rniti gate the ability of firms to manage earnings upward. In this
setting, the concentration of shares held by the top five institutions is the mechanism that
appears to have the most impact. This result is consistent with prior corporate
governance literature which finds a negative relationship between earnings management
and corporate governance mechanisms. Finally, it appears that expectations management
is used by firms with strong corporate governance mechanisms. These firms keep the

market informed and this ultimately leads to alignment between management’s

53

 

01.
d1 5

€\
1
61

 

-C.
.tu -.k-
w“ . F“ .
6
nu. .

 

 

expectations and those of the market at the end of the reporting period (Cotter et al.
2000).
6.2 Future Research

Additional research that I plan to conduct includes examining the valuation
differences associated with the level of a firm’s corporate governance mechanisms. In
addition, examining the magnitude of long run costs such as a break in a string of
earnings increases would provide evidence that firms with strong corporate governance
mechanisms are more interested in managing the costs associated with such a break
instead of focusing on MBE in the short run.
6.3 Limitations
There are several limitations to this study. First is the quality of the corporate

governance data. This data originated from the Compact Disclosure database which
transfers into Excel format financial data and text from stockholders and Form 10-K
reports, registration and proxy statements and other sources for all current and new
companies registered with the SEC. Also included are lists of subsidiaries, major officers
and board members and their salaries, 5% individual and 10% institutional holders.
Although the database lists the names of board members and their ages it does not
provide additional information about those board members. Information that could have
provided richer analyses includes whether the board member is a financial expert, what
board committees he/she serves on or whether he/she serves on additional boards. The
present limited information allowed only the basic governance variables to be used in the

analyses.

54

 

 

 

 

 

An additional limitation is the measurement error that is a result of using the
Jones model to obtain a proxy for earnings management (in this case the level of
discretionary accruals). The use of performance matched discretionary accruals in the
analysis should have mitigated the level of measurement error. The same limitation holds
for the use of a J ones—like model to obtain a proxy for expectations management, which

in this case is the level of unexpected portion of the earnings forecast.

55

References

Aboody, D. and Kasznik, R. (2000). CEO stock option awards and the tinting of
corporate voluntary disclosures. Journal of Accounting and Economics, 29, 73-
100.

Ajinkya, B., Bhojraj, S. and Sengupta, P. (2003 ). The governance effect of institutional
investors and outsider directors on the properties of management earnings
forecasts. Working Paper, University of Florida.

Ali, A. and Kallapur, S. (2001). Securities price consequences of the private securities
litigation reform act of 1995 and related events. The Accounting Review, 76, 431-
460.

Bange, M. and DeBondt, W. (1998). R&D budgets and corporate earnings targets.
Journal of Corporate Finance, 4, 153-184.

Barth, M., Elliott, J. and Firm, M. (1999). Market rewards associated with patterns of
increasing earnings. Journal of Accounting Research, 37, 387-413.

Bartov, E., Givoly, D. and Hayn, C. (2002). The rewards to meeting or beating earnings
expectations. Journal of Accounting and Economics, 33, 173-204.

Beasley, M. (1996). An empirical analysis of the relation between the board of director
composition and financial statement fraud. The Accounting Review, 71, 443-465.

Bhojraj, S., Hribar, P. and Picconi, M. (2003). Making sense of cents: An examination
of firms that marginally miss or beat analyst forecasts. Working Paper, Cornell
University.

Bolliger, G. and Kast, M. (2003). Executive compensation and analyst guidance: The
' link between CEO compensation and expectations management. Working Paper,

University of Néuchatel.

Botosan, C. (1997). Disclosure Level and the Cost of Equity Capital. The Accounting
Review, 72, 323-349.

Bowen, R., DuCharme, L. and Shores, D. (1995). Stakeholders’ implicit claims and
accounting method choice. Journal of Accounting and Economics, 20, 255-295.

Brown, L. and Caylor, M. (2003). A temporal analysis of thresholds: Propensities and
valuation consequences. Working Paper, Georgia State University.

Bushee, B. (1998). The influence of institutional investors on myopic R&D investment
behavior. The Accounting Review, 73, 305-333.

56

Bushee, B. and Noe, C. (2000). Corporate Disclosure Practices, Institutional Investors
and Stock Return Volatility. Journal of Accounting Research, 38, 171-202.

Chtourou, S., Bedard, J. and Courteau, L. (2001). Corporate governance and earnings
management. Working Paper, Universite Laval.

Chung, R., Firth, M. and Kim, J.B. (2002). Institutional monitoring and opportunistic
earnings management. Journal of Corporate Finance, 8, 29-48.

Core, J ., Holthausen, R. and Larcker, D. (1999). Corporate governance, chief executive
officer compensation and firm performance. Journal of Financial Economics, 51,
371-406.

Cotter, J ., Tuna, A. and Wysocki, P. (2000). The expectations management game: Do
analysts act independently of explicit management earnings guidance? Working
Paper, University of Southern Queensland.

Dechow, P., Sloan, R. and Sweeney, A. (1996). Causes and consequences of earnings
manipulation: An analysis of firms subject to enforcement actions by the SEC.
Contemporary Accounting Research, 13, 1-36.

Dopuch, N., Seethamraju, C. and Xu, W. (2003). An empirical assessment of the
credibility premium associated with meeting or beating both time series
expectations and analysts’ forecasts. Working Paper, Washington University.

Farber, D. (2003). Restoring trust after fraud: Does corporate governance matter?
Working Paper, Michigan State University.

Favaro, P. and Rotz, G. (2004). The perils of playing to quarterly expectations.
Financial Executives International, 1-5.

Francis, J ., Philbrick, D. and Schipper, K. (1994). Shareholder litigation and corporate
disclosures. Journal of Accounting Research, 32, 137-164.

Friday’s losers. (2002, December 20). CNN Money, Retrieved October 15, 2003,
from http://wwwcnnmoneLcom.

Gillian, S. and Starks, L. (2003). Corporate governance, corporate ownership, and
the role of institutional investors: A global perspective. Journal of Applied
Finance, Fall/Winter, 4-22.

Graham, J., Harvey, C. and Raj gopal, S. (2004). The economic implications of corporate
financial reporting. Working Paper, Duke University.

Hartzell, J. and Starks, L. (2003). Institutional investors and executive compensation.
The Journal of Finance, 58, 2351-2374.

57

Hayn, C. (1995). The information content of losses. Journal of Accounting and
Economics, 20, 125-153.

Jones, J. (1991). Earnings management during import relief investigations. Journal of
Accounting Research, 29, 193-228.

Kasznik, R. and Lev, B. (1995). To warn or not to warn: management disclosures in the
face of an earnings surprise. The Accounting Review, 70, 113-134.

Kasznik, R. and McNichols, M. (2002). Does meeting earnings expectations matter?
Evidence from analyst forecast revisions and share prices. Journal of Accounting
Research, 40, 727-759.

Ke, B., Huddart, S. and Petroni, K. (2003). What insiders know about future earnings

and how they use it: Evidence from insider trades. Journal of Accounting and
Economics, 35, 315-346.

Ke, B. and Petroni, K. (2003). How informed are institutional investors? Evidence
from their trading behavior before a break in a string of consecutive earnings
increases. Working Paper, Pennsylvania State University.

Klein, A. (2002). Audit committee, board of director characteristics and earnings
management. Journal of Accounting and Economics, 33, 375-400.

Kothari, S., Leone, A. and Wasley, C. (2001). Performance matched discretionary
accrual measures. Working Paper, Massachusetts Institute of Technology.

Kueppers, R. and Weaver, G. (2003). Meeting the street: a discussion of earnings and
other guidance provided to investors. Financial Executives Research Foundation.

Lang, M. and Lundholm, R. (1996). Corporate disclosure policy and analyst behavior.
The Accounting Review, 71, 467-492.

Matsumoto, D. (2002). Management’s incentives to avoid negative earnings surprises.
The Accounting Review, 77, 483-514.

Matsunaga, S. and Park, C. (2001). The effect of missing a quarterly earnings benchmark
on the CEO’s annual bonus. The Accounting Review, 76, 313-332.

Peasnell, K., Pope, P. and Young, S. (2000). Board monitoring and earnings
management: do outside directors’ influence abnormal accruals? Working Paper,

Lancaster University.

PricewaterhouseCoopers. (2002). Value reporting review 2003: Transparency in
corporate reporting, United Kingdom, Authored by PricewaterhouseCoopers.

58

Puffer, S. and Weintrop, J. (1991). Corporate performance and CEO turnover: The role
of performance expectations. Administrative Science Quarterly, 36, 1-19.

Richardson, 8., Teoh, S. and Wysocki, P. (2001). The walkdown to beatable analyst
forecasts: The role of equity issuance and insider trading incentives. Working

Paper, Ohio State University.

Roychowdhury, S. (2003). Management of earnings through the manipulation of real
activities that affect cash flow from operations. Working Paper, Massachusetts

Institute of Technology.
Shleifer, A. and Vishny, R. (1986). Large shareholders and corporate control. Journal of
Political Economy, 94, 461-488.

Skinner, D. (1997). Earnings disclosures and stockholder lawsuits. Journal of
Accounting and Economics, 23, 249-282.

Skinner, D. and Sloan, R. (2002). Earnings surprises, growth expectations and stock
returns or don’t let an earnings torpedo sink your portfolio. Review of Accounting

Studies, 7, 289-312.
Soffer, L., Thiagarajan, S. and Walther, B. (2000). Earnings preannouncements. Review

of Accounting Studies, 5, 5-26.
Target misses target. (2003, May 15). CNN Money, Retrieved October 15, 2003, from
http://www.cnnmonev.com.

Wahal, S. and McConnell, J. (1999). Do institutional investors exacerbate managerial
myopia? Working Paper, Emory University.

Watts, R. and Zimmerman, J. (1986). Positive accounting theory. New Jersey:

Prentice-Hall.

59

APPENDICES

60

Appendix A - Descriptive Statistics

This appendix presents a detailed analysis of the descriptive statistics presented in
Table 1 of Chapter 5. Table A.l examines the descriptive statistics for firms that meet or
beat analysts’ expectations. Table A2 examines the descriptive statistics for firms that
fail to meet or beat analysts’ expectations. Table A3 examines differences between the
sets of statistics in Tables A.l and A2. From Table A3 there appears to be very little
variation in the percentage of outsiders on the board. The two groups of firms also
appear to have the same total number of individuals on the board of directors. However,
significance tests show that the difference in means is statistically significant at less than
the .05 level. The final corporate governance variable, percentage of shares held by the
top five institutional investors appears to be larger in the sample of firms that fail to meet
or beat analysts’ expectations. As shown in Table A3, the difference in means is
statistically significant at less than the .01 level. The percentage of shares held by
institutions appears to be larger for the sample of firms that meet or beat analysts’
expectations. The means are statistically significant at less than the .01 level as shown in
Table A3. The means of uncertainty of forecast error as well as value relevance of
earnings appear to be larger for firms that fail to meet or beat analysts’ expectations. The
differences in means are statistically significant. Litigation risk, growth and size appear
to be smaller for firms that fail to meet or beat analysts’ expectations. Again, the
differences in means are statistically significant. Firms that fail to meet or beat analysts’
expectations have larger research and development expenditures and are less labor
intensive. Both of these variables are statistically significant at less than the .01 level.

Finally, total assets are not statistically significant.

61

Table A]

Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for firms that meet or beat analysts’

 

expectations
Standard

Variable N Mean Deviation Median
MBEa 2,925 1.00 0.00 1.00
%OUT 2,849 0.73 0.16 0.75
TOTAL 2,925 8.19 2.95 8.00
TOPS 2,925 0.53 0.24 0.47
%INST 2,908 0.49 0.26 0.51
UNCFE 2,919 0.05 0.46 0.01
LITRISK 2,925 0.41 0.49 0.00
VREARN 2,925 0.22 0.41 0.00
GROWTH 2,925 5.02 32.00 2.10
SIZEb 2,922 635.00 610.72 476.00
DUR 2,925 0.40 0.49 0.00
R&D 2,045 0.09 0.14 0.05
LABOR 2,042 0.53 0.35 0.61
ICLAIM 2,042 (0.04) 0.91 0.07
TOTAL ASSETS 2,912 2,163.96 7,902.88 345.68

aMBE is a dichotomous variable equal to 1 if a firm’s actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis.

bSize is presented in millions.

62

Table A2
Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for firms that fail to meet or beat
analysts’ expectations

 

Standard

Variable N Mean Deviation Median
MBEa 1,940 0.00 0.00 0.00
%OUT 1,868 0.72 0.16 0.75
TOTAL 1,940 8.00 3.16 7.00
TOPS 1,940 0.67 0.25 0.68
%INST 1,912 0.36 0.26 0.34
UNCFE 1,831 0.35 3.38 0.03
LITRISK 1,940 0.37 0.48 0.00
VREARN 1,940 0.31 0.46 0.00
GROWTH 1,940 3.08 13.90 2.02
SIZEb 1,936 428.59 539.43 262.50
DUR 1,940 0.42 0.49 0.00
R&D 1,322 0.12 0.23 0.05
LABOR 1,909 0.50 0.36 0.59
ICLAIM 1,321 0.03 1.12 0.13

TOTAL ASSETS 1,916 1,856.88 6,990.58 179.13

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis.

bSize is presented in millions.

63

Table A3

Descriptive statistics for the Dependent Variable, Independent

Firms that meet or

beat analysts’

Variables and Control Variables

Firms that fail to meet

or beat analysts’

 

expectations expectations

Variable N Mean N Mean
MBE“ 2,925 1.00 1,940 0.00
%OUT 2,849 0.73 1,868 0.72
TOTAL” 2,925 8.19 1,940 8.00
TOPS“ 2,925 0.53 1,940 0.67
%INST“ 2,908 0.49 1,912 0.36
UNCFE“ 2,919 0.05 1,831 0.35
LITRISK“ 2,925 0.41 1,940 0.37
VREARN“ 2,925 0.22 1,940 0.31
GROWTH“ 2,925 5.02 1,940 3.08
SIZE“ 2,922 635.00 1,936 428.59
DUR 2,925 0.40 1,940 0.42
R&D“ 2,045 0.09 1,322 . 0.12
LABOR” 2,042 0.53 1,909 0.50
ICLAIM 2,042 (0.04) 1,321 0.03
TOTAL ASSETS 2,912 2,163.96 1,916 1,856.88

aMBE is a dichotomous variable equal to 1 if a firm’s actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis. Size is presented in millions.

bStatistically significant at less than the .05 level
cStatistically significant at less than the .01 level

64

Table A4 provides descriptive statistics for the sample by year. In panels A-E,
there appears to be very little variation in the percentage of outsiders on the board. The
mean percentage ranges from 71% in 1997 to 74% in 1998. In the remaining three years
the percentage remains Stable at 72%. This lack of variation may explain the lack of
significance in the analyses. In addition there appears to be very little variation in the
board size variable. The mean variable ranges from 7.64 in 1997 to 8.43 in 1998. In the
remaining three years the variable remains fairly stable at 8.22 in 1999, 8.16 in 2000 and
8.14 in 2001. Unlike the other two corporate governance variables there is significant
variation by year in mean percentage of shares held by the top five institutional investors.
The mean ranges from 50% in 1999 to 63% in 1997. The mean percentage of shares held
by institutional investors also varies throughout the years in the sample. The mean
ranges from 42% in 1997 to 51% in 2001. In the remaining years of the sample the mean
remains fairly stable at 43% in 1998 and 1999 and 44% in 2000. The mean of
uncertainty of forecast error shows variation when the sample is presented by year. The
mean ranges from .07 in 1999 to .27 in 1997. Litigation risk also varies throughout the
years in the sample. The mean ranges from .36 in 1999 to .42 in 2000. The mean value
relevance of earnings varies from a low of .20 in 1998 to .29 in 2001. The mean of
growth ranges from a low of 3.97 in 1998 to a high of 4.98 in 1997. The mean size of
firms in the sample varies from a low of 499.11 in 2001 to 678.61 in 1999. The mean
number of firms in a durable goods industry remains fairly stable throughout the sample.
The mean ranges from .39 in 1999 to .43 in 2001. The mean level of research and

development expenditures ranges from .08 in 1999 and 2000 to .12 in 1997. Mean labor

65

intensity .50 in 1998 to .53 in 2001. Finally, mean total assets vary from 1,551.86 in

1997 to 2439.06 in 2000.

Table A4
Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables by year

Panel A: Annual period 1997

 

Standard

Variable N Mean Deviation Median
MBE"l 1,064 0.59 0.49 1.00
%OUT 1,033 0.71 0.16 0.75
TOTAL 1,064 7.64 3.12 7.00
TOPS 1,064 0.63 0.23 0.59
%INST 1,055 0.42 0.26 0.41
UNCFE 1,051 0.27 4.03 0.01
LITRISK 1,064 0.39 0.49 0.00
VREARN 1,064 0.25 0.43 0.00
GROWTH 1,064 4.98 48.00 2.07
SIZEb 1,063 512.79 471.44 381.25
DUR 1,064 0.42 0.49 0.00
R&D 763 0.12 0.19 0.06
LABOR 1,051 0.54 0.32 0.61
ICLAIM 761 0.07 1.01 0.08
TOTAL ASSETS 1,056 1,551.86 9,832.04 161.76

z"MBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis. .

bSize is presented in millions.

66

Table A.4 (cont’d).

Panel B: Annual period 1998

 

Standard

Variable N Mean Deviation Median
MBE" 1,046 0.57 0.50 1,00
%OUT 1,021 0.74 0.16 0.78
TOTAL 1,046 8.43 3.18 8.00
TOPS 1,046 0.62 0.24 0.59
%INST 1,037 0.43 0.26 0.44
UNCFE 1,026 0.14 1.12 0.01
LITRISK 1,046 0.40 0.49 0.00
VREARN 1,046 0.20 0.40 0.00
GROWTH 1,046 3.97 18.65 2.17

SIZEb 1,045 515.22 553.29 357.81
DUR 1,046 0.42 0.49 0.00
R&D 716 0.12 0.27 0.05
LABOR 1,032 0.50 0.36 0.58
ICLAIM 716 0.02 1.20 0.06

TOTAL ASSETS 1,038 1,714.50 5,680.67 231.94

alMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

67

Table A.4 (cont’d).

Panel C: Annual period 1999

 

Standard

Variable N Mean Deviation Median
MBE"1 1,254 0.65 0.48 1.00
%OUT 1,217 0.72 0.16 0.75
TOTAL 1,254 8.22 3.00 8.00
TOPS 1,254 0.50 0.27 0.42
%IN ST 1,242 0.43 0.27 0.44
UNCFE 1,221 0.07 0.43 0.01
LIT RISK 1,254 0.36 0.48 0.00
VREARN 1,254 0.26 0.44 0.00
GROWTH 1,254 3.84 15.93 2.00
SIZEb 1,250 678.61 798.80 450.78
DUR 1,254 0.39 0.49 0.00
R&D 825 0.08 0.13 0.04
LABOR 1,236 0.51 0.37 0.60
ICLAIM 824 (0.09) 0.91 0.02
TOTAL ASSETS 1,245 2,304.68 7,400.33 351.94

a'MBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

68

Table A.4 (cont’d).
Panel D: Annual period 2000

 

Standard

Variable N Mean Deviation Median
MBEa 852 0.60 0.49 1.00
%OUT 826 0.72 0.16 0.75
TOTAL 852 8.16 3.04 8.00
TOPS 852 0.62 0.25 0.58
%INST 843 0.44 0.27 0.44
UNCFE 821 0.23 1.80 0.02
LITRISK 852 0.42 0.49 0.00
VREARN 852 0.27 0.45 0.00
GROWTH 852 4.46 10.98 2.01
SIZEb 851 505.07 494.99 346.88
DUR 852 0.40 0.49 0.00
R&D 600 0.08 0.10 0.05
LABOR 842 0.53 0.34 0.62
ICLAIM 600 (0.05) 0.89 0.13
TOTAL ASSETS 845 2,439.06 6,953.33 350.80

a’MBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

69

Table A.4 (cont’d).

Panel E: Annual period 2001

 

Standard

Variable N Mean Deviation Median
MBE" 649 0.58 0.49 1.00
%OUT 620 0.72 0.18 0.75
TOTAL 649 8.14 2.57 8.00
TOPS 649 0.56 0.24 0.49
%IN ST 643 0.51 0.27 0.54
UNCFE 631 0.16 0.78 0.02
LIT RISK 649 0.40 0.49 0.00
VREARN 649 0.29 0.45 0.00
GROWTH 649 4.02 16.97 2.21
SIZEb 649 499.11 420.45 392.25
DUR 649 0.43 0.50 0.00
R&D 463 0.09 0.12 0.06
LABOR 641 0.52 0.35 0.61
ICLAIM 462 0.01 0.88 0.20
TOTAL ASSETS 644 2,345.48 6,884.08 370.98

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet exceed the analysts’ consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

70

Tables A.5 through A.9 provide descriptive statistics by year separately for firms
that meet or beat analysts’ expectations and those firms that fail to do so. These tables
provide for a more in-depth analysis of the individual variables. Consistent with Table
A.3, there appears to be very little variation in the mean percentage of outsiders on the
board when the sample is broken down by sample year. In 1999, there is the same mean
percentage of outside directors on boards that met or beat analysts’ expectations and
boards that failed to do so. The mean board size variable shows a small amount of
variation based on whether the firm meets or beats analysts’ expectations or fails to do so.
For firms that met or beat analysts’ expectations the mean ranges from a low of 7.72 in
1997 to a high of 8.39 in 2001. For firms that failed to meet or beat analysts’
expectations the mean ranges from a low of 7.79 in 2001 to a high of 8.47 in 1998.

The mean percentage of shares held by the top five institutional investors shows
variation based on whether the firm meets or beats analysts’ expectations or fails to do so.
For firms that met or beat analysts’ expectations the mean ranges from a low of 45% in
1999 to a high of 58% in both 1997 and 1998. For firms that failed to meet or beat
analysts’ expectations the mean ranges from a low of 60% in 1999 to a high of 72% in
2000. There appears to be no overlap in the ranges based on whether a firm met or beat
analysts’ expectations or failed to do so. The mean percentage of shares held by
institutional investors also shows variation based on whether the firm meets or beats
analysts’ expectations or fails to do so. For firms that met or beat analysts’ expectations
the mean ranges from a low of 46% in 1997 to a high of 56% in 2001. For firms that
failed to meet or beat analysts’ expectations the mean ranges from a low of 35% in the

first four years of the sample to a high of 45% in 2001.

71

Table A5
Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 1997

Panel A: Firms that met or beat analysts’ expectations

 

Standard

Variable N Mean Deviation Median
MBEa 626 1.00 0.00 1.00
%OUT 609 0.71 0.16 0.75
TOTAL 626 7.72 3.06 7.00
TOPS 626 0.58 0.22 0.54
%INST 623 0.46 0.25 0.48
UNCFE 623 0.03 0.10 0.01
LITRISK 626 0.39 0.49 0.00
VREARN 626 0.20 0.40 0.00
GROWTH 626 6.19 62.36 2.12
SIZE” 626 572.28 429.41 475.00
DUR 626 0.41 0.49 0.00
R&D 448 0.10 0.16 0.05
LABOR 619 0.55 0.32 0.62
ICLAIM 446 0.01 0.93 0.06

TOTAL ASSETS 624 1,674.97 11,845.99 195.20

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet exceed the analysts’ consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis.

bSize is presented in millions.

72

Table A5 (cont’d).
Panel B: Firms that fail to meet or beat analysts' expectations

 

Standard

Variable N Mean Deviation Median
MBEa 438 0.00 0.00 0.00
%OUT 424 0.72 0.15 0.75
TOTAL 438 7.54 3.20 7.00
TOPS 438 0.70 0.23 0.74
%INST 432 0.35 0.25 0.33
UNCFE 428 0.61 6.30 0.03
LIT RISK 438 0.39 0.49 0.00
VREARN 438 0.33 0.47 0.00
GROWTH 438 3.25 6.04 1.98
SIZE” 437 427.57 514.47 254.68
DUR 438 0.43 0.50 0.00
R&D 315 0.14 0.22 0.07
LABOR 432 0.52 0.33 0.58
ICLAIM 315 0.15 1.12 0.15
TOTAL ASSETS 432 1,374.04 5,807.98 109.31

a‘MBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

73

Table A6
Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 1998

Panel A: Firms that met or beat analysts' expectations

 

Standard

Variable N Mean Deviation Median
MBE" 596 1.00 0.00 1.00
%OUT 589 0.75 0.16 0.78
TOTAL 596 8.39 3.05 8.00
TOPS 596 0.58 0.23 0.53
%INST 592 0.48 0.26 0.50
UNCFE 596 0.07 0.46 0.01
LITRISK 596 0.45 0.50 0.00
VREARN 596 0.18 0.39 0.00
GROWTH 596 4.86 20.99 2.17
SIZE” 596 594.79 614.80 433.01
DUR . 596 0.39 0.49 0.00
R&D 415 0.10 0.19 0.05
LABOR 588 0.52 0.35 0.58
ICLAIM 415 (0.03) 0.96 (0.00)

TOTAL ASSETS 592 1,639.32 5,061.88 267.64

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet exceed the analysts’ consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis.

bSize is presented in millions.

74

Table A6 (cont’d).
Panel B: Firms that fail to meet or beat analysts' expectations

 

Standard

Variable N Mean Deviation Median
MBEa 450 0.00 0.00 0.00
%OUT 432 0.74 0.15 0.78
TOTAL 450 8.47 3.35 8.00
TOPS 450 0.68 0.24 0.68
%INST 445 0.36 0.25 0.33
UNCFE 430 0.23 1.63 0.02
LITRISK 450 0.35 0.48 0.00
VREARN 450 0.23 0.42 0.00
GROWTH 450 2.79 14.95 2.16
SIZE” 449 409.61 437.94 270.31
DUR 450 0.46 0.50 0.00
R&D 301 0.14 0.36 0.04
LABOR 444 0.47 0.38 0.57
ICLAIM 301 0.08 1.48 0.11
TOTAL ASSETS 446 1,814.29 6,415.08 158.95

"MBE is a dichotomous variable equal to 1 if a firm’s actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

75

Table A7
Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 1999

Panel A: Firms that meet or beat analysts’ expectations

 

Standard

Variable N Mean Deviation Median
MBEa 815 1.00 0.00 1.00
%OUT 792 0.72 0.16 0.75
TOTAL 815 8.24 2.78 8.00
TOPS 815 0.45 0.25 0.38
%INST 809 0.48 0.26 0.50
UNCFE 813 0.03 0.10 0.01
LITRISK 815 0.39 0.49 0.00
VREARN 815 0.24 0.43 0.00
GROWTH 815 4.54 15.19 2.02
SIZE” 813 754.32 799.13 534.38
DUR“ 815 0.39 0.49 0.00
R&D 547 0.08 0.13 0.04
LABOR 805 0.51 0.36 0.60
ICLAIM 546 (0.10) 0.91 0.02

TOTAL ASSETS 811 2,146.78 5,676.23 450.12

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet exceed the analysts’ consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis.

bSize is presented in millions.

76

Table A.7 (cont’d).
Panel B: Firms that fail to meet or beat analysts' expectations

 

Standard

Variable N Mean Deviation Median
MBEa 439 0.00 0.00 0.00
%OUT 425 0.72 0.17 0.75
TOTAL 439 8.17 3.38 8.00
TOPS 439 0.60 0.28 0.56
%INST 433 0.35 0.26 0.30
UNCFE 408 0.15 0.72 0.03
LITRISK 439 0.31 0.46 0.00
VREARN 439 0.29 0.45 0.00
GROWTH 439 2.53 17.17 1.97
SIZE” 437 8,190.63 779.75 306.25
DUR 439 0.39 0.49 0.00
R&D 278 0.09 0.13 0.04
LABOR 431 0.50 0.39 0.62
ICLAIM 278 (0.07) 0.91 0.05
TOTAL ASSETS 434 2,599.73 9,845.41 213.98

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

77

Table A8
Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 2000

Panel A: Firms that meet or beat analysts’ expectations

 

Standard

Variable N Mean Deviation Median
MBEa 509 1.00 0.00 1.00
%OUT 497 0.73 0.16 0.75
TOTAL 509 8.29 3.10 8.00
TOPS 509 0.55 0.23 0.50
%INST 506 0.49 0.26 0.53
UNCFE 508 0.09 0.93 0.01
LIT RISK 509 0.43 0.50 0.00
VREARN 509 0.21 0.41 0.00
GROWTH 509 4.37 12.18 2.04
SIZE” 508 610.55 525.28 464.06
DUR 509 0.43 0.50 0.00
R&D 360 0.07 0.09 0.05
LABOR 505 0.55 0.34 0.64
ICLAIM 360 (0.03) 0.86 0.17

TOTAL ASSETS 507 2,909.93 7,804.31 495.68

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet exceed the analysts’ consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis.

bSize is presented in millions.

78

Table A8 (cont’d).
Panel B: Firms that fail to meet or beat analysts' expectations

 

Standard

Variable N Mean Deviation Median
MBE“| 343 0.00 0.00 0.00
%OUT 329 0.72 0.16 0.75
TOTAL 343 7.97 2.96 8.00
TOPS 343 0.72 0.23 0.73
%INST 337 0.35 0.26 0.31
UNCFE 313 0.47 2.66 0.04
LITRISK 343 0.42 0.49 0.00
VREARN 343 0.37 0.48 0.00
GROWTH 343 4.59 8.93 1.99
SIZEb 343 348.85 398.64 200.00
DUR 343 0.36 0.48 0.00
R&D 240 0.09 0.12 0.05
LABOR 337 0.50 0.35 0.61
ICLAIM 240 (0.09) 0.94 0.09
TOTAL ASSETS 338 1,732.75 5,366.76 240.06

aMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the '
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

79

Table A9
Descriptive statistics for the Dependent Variable, Independent
Variables and Control Variables for sample year 2001

Panel A: Firms that meet or beat analysts’ expectations

 

Standard

Variable N Mean Deviation Median
MBE" 379 1.00 0.00 1.00
%OUT 362 0.71 0.17 0.75
TOTAL 379 8.39 2.62 8.00
TOPS 379 0.51 0.22 0.44
%INST 378 0.56 0.26 0.59
UNCFE 379 0.08 0.34 0.02
LITRISK 379 0.41 0.49 0.00
VREARN 379 0.25 0.43 0.00
GROWTH 379 5.25 14.90 2.25
SIZE” 379 579.41 443.69 450.00
DUR 379 0.40 0.49 0.00
R&D 275 0.08 0.10 0.05
LABOR 376 0.53 0.36 0.62
ICLAIM 275 (0.02) 0.85 0.19

TOTAL ASSETS 378 2,829.15 7,562.67 537.00

“MBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet exceed the analysts’ consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the
analysis.

bSize is presented in millions.

80

Table A9 (cont’d)
Panel B: Firms that fail to meet or beat analysts' expectations

 

Standard

Variable N Mean Deviation Median
MBE" 270 0.00 0.00 0.00
%OUT 258 0.72 0.17 0.78
TOTAL 270 7.79 2.47 7.00
TOPS 270 0.64 0.25 0.61
%INST 265 0.45 0.28 0.44
UNCFE 252 0.29 1.15 0.03
LIT RISK 270 0.39 0.49 0.00
VREARN 270 0.35 0.48 0.00
GROWTH 270 2.29 19.40 2.09
SIZE” 270 386.39 356.88 284.75
DUR 270 0.48 0.50 0.00
R&D 188 0.11 0.15 0.06
LABOR 265 0.52 0.34 0.59
ICLAIM 187 0.04 0.94 0.21
TOTAL ASSETS 266 1,658.14 5,728.41 216.44

alMBE is a dichotomous variable equal to 1 if a firm's actual earnings
meet or exceed the analysts' consensus forecast outstanding at the
earnings announcement and 0 otherwise. Statistics in this table are
presented as actual values while logarithmic values are used in the

analysis.

bSize is presented in millions.

81

Appendix B - Supplementary Analyses for Hypothesis One

This appendix presents supplementary analyses for the results presented in Table
4 of Chapter 5. Tables B.1 provides results without winsorized variables. The results in
Table 4 Show continuous variables winsorized at the .01 percentile. Table B.2 provides
results when continuous variables are winsorized at the .10 percentile. The results are
consistent with those presented in Table 4 of Chapter 5. Both TOPS (p<.05) and
TOTAL (p<.05) are significantly negative. This suggests that as the shares held by the
top five institutional investors increase the firm is less likely to MBE, which is consistent
with Hypothesis One. The results also suggest that as board size increases the firm is less
likely to MBE. There is mixed evidence in the corporate governance literature about the
influence of board size. The result is consistent with the literature that suggests larger
boards provide a broader base of expertise. The percentage of outsiders on the average
board remains insignificant. The lack of results may be due to the lack of variability

throughout the sample years as Shown in appendix A.

82

Table B.1
Logit Analysis of the Probability of Meeting or Beating Analysts' Expectations

and Corporate Governance Mechanisms

Pr ob(MBE = 1) = 60 + ,6, %OUT + ,BZTOTAL + ,63T0P 5 + ,6, %INST + ,BSUNCFE
+ ,66LITRISK + B7VREARN + flgGROWIW + ,6, SIZE + ,6“) ICLAIM + 5

Independent Variable

Intercept

Corporate Governance

Mechanisms
%OUT
TOTAL
TOPS

Control Variables
%INST
UNCFE

LIT RISK
VREARN
GROWTH

SIZE
DUR
RD
LABOR
ICLAIM

N:
Log Likelihood:
p-value=

Coefficient

-0.21

-0.11
-0.05
-l.08

0.50
-1.14
0.14
-0. 12
0.01
0.25
-0.22
0.07
0.30

p-value

0.72

0.67
0.01
0.00

0.04
0.00
0.14
0.28
0.08
0.00
0.02
0.86
0.05

3185
-1937.87
<01

83

Coefficient

-O.18

-0.08
-0.04
— 1 . 10

0.48
-1.17
0.21
-0.11
0.01
0.25

0.06

p-value

0.74

0.75
0.02
0.00

0.05
0.00
0.02
0.31
0.09
0.00

0.30

3185
-l941
<.01

Coefficient

-0.33

0.02
-0.04
-0.86

0.62
-l. 12
0.19
-0.07
0.01
0.23

p-value

0.47

0.91
0.01
0.00

0.00
. 0.00
0.01
0.39
0.03
0.00

4569
-2818
<.01

Table B.2
Logit Analysis of the Probability of Meeting or Beating Analysts' Expectations
and Corporate Governance Mechanisms (all variables winsorized at 10% and 90%
percentiles)

Pr ob(MBE = 1) = fl- + 61%0UT + flzTOTAL + fl-TOP 5 + fl- %INST + flSUNCFE
+ fléurmsx + B7VREARN + pgokowm + [3, SIZE + ,BmICLAIM + 5

Independent Variable

Intercept

Corporate Governance
Mechanisms
%OUT
TOTAL
TOPS

Control Variables
%INST
UNCFE

LITRISK
VREARN
GROWTH

SIZE
DUR
RD
LABOR
ICLAIM

N:
Log Likelihood:
p-value=

Coefficient

0.52

0.08
-0.04
-O.88

0.59
-9.25
0.13
-0.08
0.04
0.11
-0.20
0.78
0.22

p-value

0.45

0.80
0.08
0.00

0.02
0.00
0.15
0.49
0.03
0.18
0.03
0.27
0.23

3185

-l914.55

<.01

Coefficient

0.60

0.11
-0.04
-0.91

0.56
-9.32
0.21
-0.04
0.04
0.10

0.07

p- value

0.37

0.73
0.08
0.00

0.03
0.00
0.01
0.71
0.03
0.21

0.29

3185
-l917
<.01

Coefficient

0.53

0.20
-0.05
-0.69

0.69
-9. 18
0.19
-0.01
0.03
0.09

p-value

0.35

0.45
0.01
0.01

0.00
0.00
0.01
0.95
0.06
0. 19

4569
-2782
<.01

Table B.3 provides the regression analysis by year for unwinsorized variables.

The corporate governance variables except 2001 are mostly insignificant. As shown in

panel B, in 2001, the percentage of shares held by the top five investors is significantly

negative. Again, this supports the original contention that as these investors increase

84

their stake in the company the firm has a smaller probability of meeting or beating
analysts’ expectationsg.

Table B.3
Logit Analysis of the Probability of Meeting or Beating Analysts' Expectations
and Corporate Governance Mechanisms by year

Panel A: Annual period 1997

Pr ob(MBE = 1) = 60 + 13- %OUT + 6,T0TAL + fl3T0P 5 + ,6, %INST + ,BSUNCFE
+ 66LITRISK + B,VREARN + flSGROWI‘H + ,6, SIZE + 6,,ICLAIM + 5

Independent Variable Coefficient p-value Coefficient p- value Coefficient p—value
Intercept -0. 10 0.93 -0.22 0.84 0.46 0.63
Corporate Governance
Mechanisms
%OUT -0.45 0.38 -0.39 0.44 -0.37 0.38
TOTAL -0.02 0.55 -0.02 0.45 -0.02 0.41
TOPS -0.75 0.20 -0.68 0.25 -0.85 0.08
Control Variables

%INST 0.39 0.40 0.42 0.35 0.34 0.38
UNCFE -1.40 0.05 -1.45 0.04 -2.42 0.01
LIT RISK -0. 13 0.43 -0. 12 0.47 -0.09 0.49
VREARN 0.00 0.99 -0.05 0.78 -0.19 0.25
GROWTH 0.01 0.42 0.01 0.45 0.01 0.35
SIZE 0.23 0.10 0.25 0.08 0.16 0.18

DUR -0.09 0.58

RD 062 0.21

LABOR 0.27 0.32

ICLAIM -0.01 0.88

N: 726 726 1013
pseudo R2: 0.0749 0.0726 0.0818

 

9 The results remain unchanged when continuous variables are winsorized at the .01 and .10 level.

85

Table B.3 (cont’d).
Panel B: Annual period 1998

Independent Variable Coefficient
Intercept - l . 13
Corporate
Governance
Mechanisms
%OUT 0.07
TOTAL 007
TOPS -0. 17
Control Variables
%INST l . l7
UNCFE - l . l4
LIT RISK 0.29
VREARN 0.33
GROWTH 0.01
SIZE 0.26
DUR -0.41
RD 007
LABOR 0.38
ICLAIM
N:
pseudo R2:

p-value

0.32

0.89
0.03
0.77

0.01
0.16
0.10
0.16
0.20
0.05
0.02
0.89
0.21

685
0.0908

86

Coefficient

-1.05

0.05
-0.07
-0.27

1.20
—1.30
0.42
0.37
0.01
0.24

0.00

p- value

0.35

0.91
0.03
0.64

0.01
O. 14
0.01
0.11
0.21
0.07

0.97

685
0.084

Coefficient

-1.57

-0.08
-0.07
0. 10

1.30
-0.10
0.36
0.20
0.01
0.30

p-value

0.07

0.85
0.01
0.83

0.00
0.43
0.01
0.29
0.29
0.00

992
0.0588

Table B.3 (cont’d).
Panel C: Annual period 1999

Independent Variable Coefficient
Intercept 048
Corporate
Governance
Mechanisms
%OUT 0.10
TOTAL 005
TOPS 077
Control Variables
%IN ST 0.83
UNCFE -l.65
LIT RISK 0.29
VREARN -0.34
GROWTH 0.01
SIZE 0.26
DUR 0.02
RD 1.06
LABOR -0.24
ICLAIM
N:
pseudo R2:

p-value

0.64

0.84
0.14
0.10

0.06
0.23
0.1 1
0.08
0.19
0.03
0.91
0.20
0.38

781
0.0868

87

Coefficient

-0.55

0.07
-0.04
-0.74

0.86
-l.50
0.31
-0.28
0.01
0.26

0.00

p- value

0.59

0.89

0.17
0.11

0.05
0.22
0.06
0.14
0.22
0.03

0.98

781
0.0845

Coefficient

-0.44

0.12
-0.02
-0.55

0.90
-2. 12
0.29
-0.05
0.01
0.18

p-value

0.59

0.77
0.41
0.13

0.01
0.09
0.04
0.74
0.13
0.09

1175
0.0732

Table B.3 (cont’d).
Panel D: Annual period 2000

Independent Variable Coefficient
Intercept 072
Corporate
Governance
Mechanisms
%OUT -0.06
TOTAL -0.04
TOPS -1.04
Control Variables
%INST -0.03
UNCFE -0.95
LIT RISK 0.05
VREARN -0.47
GROWTH 0.00
SIZE 0.35
DUR -0.1 1
RD -0. 2 1
LABOR 0. 85
ICLAIM
N:
pseudo R2:

p-value

0.57

0.92
0.26
0.12

0.95
0.17
0.82
0.06
0.88
0.02
0.58
0.87
0.01

563
0.1192

88

Coefficient

-0.27

0.01
-0.03
-l.07

-0.04
-0.99
0.1 1
-0.50
0.00
0.34

0.25

p- value

0.83

0.99
0.31
0.10

0.93
0.15
0.58
0.04
0.82
0.02

0.05

563
0.1162

Coefficient

-0.54

0.44
-0.04
- l .06

0.37
-0.04
0.11
-0.41
0.00
0.30

p-value

0.59

0.37
0.08
0.06

0.40
0.58
0.47
0.03
0.53
0.01

789
0.083

Table B.3 (cont’d).
Panel E: Annual period 2001

Independent Variable Coefficient p-value Coefficient p- value Coefficient p-value

Intercept 1.94 0.18 1.39 0.32 0.70 0.55
Corporate
Governance
Mechanisms
%OUT -0.98 0.1 1 -0.89 0.15 -0.24 0.62
TOTAL 0.01 0.79 0.03 0.56 0.01 0.73
TOPS -2. 12 0.00 -1.97 0.01 -1.70 0.01
Control Variables
%INST -0.35 0.53 -0.29 0.58 -0.22 0.64
UNCFE -0. 15 0.59 —0. 14 0.62 -0.27 0.45
LITRISK 0.10 0.69 0.29 0.17 0.24 0.19
VREARN -0.25 0.34 -0. 16 0.53 -0.08 0.69
GROWTH 0.01 0.04 0.01 0.08 0.02 0.03
SIZE 0.07 0.66 0.10 0.57 0.13 0.38
DUR -0.60 0.01
RD 088 0.47
LABOR 0.72 0.05
ICLAIM 0.06 0.64
N = 430 430 600
pseudo R”: 0.0833 0.0658 0.0602

Table B.4 provides the regression analysis when the absolute value of the amount
by which a firm meets or beats analysts’ forecasts or fails to do so is used as the
dependent variable. The absolute difference is larger for firms with a larger percentage
of shares held by the top five investors. This supports the original contention that as
these investors increase their stake in the company the firm has a smaller probability of

meeting or beating analysts’ expectations").

 

’0 The results remain unchanged when continuous variables are winsorized at the .01 and .10 level.

89

Table B.4

Multiple Regression Analysis between Meeting or Beating Analysts' Expectations
and Corporate Governance Mechanisms

ABS (MBE ) = 60 + ,6,%0UT + 62TOTAL + ,6’3T0P 5 + fl.- %INST + ,BSUNCFE
+ fl6LITRISK + B7VREARN + ,BSGROWI‘H + 6, SIZE + 6,,ICLAIM + 5

Independent Variable Coefficient p-value Coefficient p- value Coefficient p-value

Intercept 0.17 0.23 0.13 0.34 -2.29 0.00

Corporate Governance

Mechanisms
%OUT 0.06 0.32 0.06 0.31 0.40 0.07
TOTAL 0.00 0.73 0.00 0.64 -0.01 0.02
TOPS 0.01 0.93 0.00 0.97 0.72 0.00
Control Variables

%INST -0.22 0.00 -0.22 0.00 -0.23 0.02
UNCFE 0.42 0.00 0.42 0.00 1.39 0.00
LIT RISK -0.01 0.82 0.00 0.97 0.02 0.80
VREARN 0.10 0.00 0.10 0.00 -0.11 0.33
GROWTH 0.00 0.53 0.00 0.52 0.00 0.48
SIZE 0.01 0.71 0.01 0.57 0.32 0.00

DUR -0.03 0.31

RD -0. 16 0.03

LABOR 0.00 0.99

ICLAIM -0.03 0.06

N: 3185 3 185 4569
R2: 0.1923 0.193 0.6342

90

Table B.5 provides the regression analysis when the absolute value of the
amount by which a firm meets or beats analysts’ forecasts or fails to do so is used as the
dependent variable. This table presents the analysis by year. In Panel B for 1998 the
percentage of shares held by the top five investors is positive and significant. This
supports the original contention that as these investors increase their stake in the
company, the firm has a smaller probability of meeting or beating analysts’ expectations.

In the remaining four years the corporate governance mechanisms are insignificant”.

 

H The results remain unchanged when continuous variables are winsorized at the .01 and .10 level.

91

Table B.5
Multiple Regression Analysis between Meeting or Beating Analysts' Expectations and
Corporate Governance Mechanisms by year

Panel A: Annual period 1997

ABS (MBE ) = ,60 + 6,%0UT + 6,T0TAL + 6,701) 5 + fl.- %INST + ,BSUNCFE
+ 66LITRISK + B7VREARN + flSGROWI‘H + 6, SIZE + 6,,ICLAIM + e

92

Independent Variable Coefficient p-value Coefficient p- value Coefficient p-value
Intercept 0.68 0.20 0.68 0.20 -4.04 0.06
Corporate Governance
Mechanisms
%OUT -0.23 0.45 -0.23 0.45 0.55 0.35
TOTAL 0.00 0.52 0.00 0.55 -0.02 0.09
TOPS 0.03 0.78 0.04 0.72 1.07 0.03
Control Variables

%IN ST -0. l 3 0.06 -0. 12 0.07 -0.11 0.50
UNCFE 0.15 0.01 0.16 0.01 1.29 0.00
LITRISK -0.02 0.62 -0.02 0.53 0.04 0.52
VREARN 0.1 l 0.03 0.10 0.02 0.13 0.25
GROWTH 0.00 0.22 0.00 0.45 0.00 0.26
SIZE -0.07 0.17 -0.07 0.16 0.54 0.03

DUR 0.03 0.51

RD -0.05 0.69

LABOR 0.02 0.61

ICLAIM 0.00 0.88

N: 726 726 1013
R2: 0.2621 0.262 0.8857

Table B.S (cont’d).

Panel B: Annual period 1998

Independent Variable

Intercept

Corporate Governance
Mechanisms

%OUT

TOTAL

TOPS

Control Variables
%INST
UNCFE
LIT RISK
VREARN
GROWTH
SIZE

DUR

RD
LABOR
ICLAIM

Coefficient

-0.19

-0.07
0.00
0.24

-0.1 l
0.46
-0.01
0.08
0.00
0.03
0.02
-0.21
0.02

p-value

0.24

0.23
0.21
0.01

0.07
0.00
0.69
0.01
0.97
0.15
0.52
0.04
0.66

685

0.3198

93

Coefficient

-0. 19

-0.09
0.00
0.26

-0.1 l
0.42
-0.02
0.07
0.00
0.04

-0.03

p- value

0.23

0.15
0.45
0.00

0.08
0.00
0.25
0.02
0.80
0.1 1

0.07

685
0.3077

Coefficient

-5.75

0.05
-0.04
1.65

0.03
4.58
-0.06
-0.19
0.00
0.84

p-value

0.00

0.86
0.02
0.04

0.91
0.00
0.60
0.18
0.27
0.00

992
0.8805

Table B.S (cont’d).
Panel C: Annual period 1999

Independent Variable Coefficient
Intercept -0.7 1
Corporate Governance
Mechanisms
%OUT 0.14
TOTAL 0.00
TOPS 0.09
Control Variables

%INST -0. l7
UNCFE 1.90
LIT RISK -0.04
VREARN 0.14
GROWTH 0.00
SIZE 0.12
DUR -0.09
RD 056
.LABOR 0.07

ICLAIM

N:
R3:

p-value

0.02

0.45
0.65
0.57

0.16
0.04
0.53
0.04
0.10
0.01
0.10
0.21
0.44

781
0.2281

94

Coefficient

-0.78

0.16
0.00
0.07

-0.20
1.79
-0.02
0.1 l
0.00
0. 12

-0.03

p- value

0.01

0.41
0.74
0.68

0.15
0.04
0.70
0.02
0.12
0.00

0.07

781
0.2153

Coefficient

-l.03

0.19
-0.01
0.22

-0.32
1.81
—0.09
0.06
0.00
0.18

p-value

0.01

0.34
0.34
0.19

0. 16
0.00
0.16
0.45
0.42
0.03

1175
0.3291

Table B.5 (cont’d).
Panel D: Annual period 2000

Independent Variable Coefficient
Intercept -0.59
Corporate Governance
Mechanisms
%OUT 0.26
TOTAL 0.01
TOPS 0.05
Control Variables

%INST -0.34
UNCFE 0.62
LIT RISK -0.03
VREARN 0.01
GROWTH 0.00
SIZE 0.09
DUR -0.02
RD 2.05
LABOR -0. l9

ICLAIM

N:
R2:

p-value

0.01

0.1 1
0.43
0.75

0.02
0.00
0.45
0.92
0.32
0.01
0.70
0.33
0.09

563
0.757

95

Coefficient

-0.51

0.21
0.01
0.05

-0.38
0.63
-0.01
0.10
0.00
0.09

0.04

p- value

0.02

0.11
0.49
0.78

0.05
0.00
0.70
0.08
0.34
0.01

0.61

563
0.7437

Coefficient

-0.78

0.16
0.00
0.25

-0.34
0.55
0.02
0.10
0.00
0.12

p-value

0.00

0.11
0.92
0.15

0.03
0.00
0.60
0. 15
0.28
0.00

789
0.6743

Table B.S (cont’d).
Panel E: Annual period 2001

Independent Variable Coefficient
Intercept -0.27
Corporate Governance
Mechanisms
%OUT 0.] 7
TOTAL 001
TOPS 0. l 2
Control Variables
%INST -0. l 7
UNCFE 0.57
LIT RISK 0.01
VREARN 0.1 1
GROWTH 0.00
SIZE 0.06
DUR 00]
RD 053
LABOR 0.00
ICLAIM
N:
R“-

p-value

0.36

0.14
0.21
0.59

0.30
0.07
0.79
0.03
0.42
0.28
0.79
0.21
0.99

430

0.3183

96

Coefficient

-0.39

0.17
-0.01
0.15

-0. l 4
0.57
0.00
0.09
0.00
0.07

-0.03

p- value

0.24

0.15
0.18
0.46

0.34
0.07
0.93
0.04
0.43
0.27

0.15

430
0.3142

Coefficient

-0.46

0.13
-0.01
0.23

-0.06
0.54
-0.03
0.14
0.00
0.06

p-value

0.11

0.21
0.41
0.13

0.63
0.06
0.30
0.00
0.37
0.21

0.2942

Appendix C - Descriptive Statistics for Hypothesis Two

This appendix presents descriptive statistics for the data used in Table 5 of
Chapter 5. Table C.1 provides descriptive statistics for firms that managed earnings
upward and managed expectations downward. Panel A presents the total sample while
Panels B and C divide the sample into firms that met or beat analysts’ expectations and
those that failed to do so. From Panel B, the mean percentage of Shares held by the top
five institutional investors for firms that met or beat analysts’ expectations is .48. From
Panel C, the mean percentage of shares held by the top five institutional investors for
firms that failed to meet or beat analysts’ expectations is .57. The difference between the
two is statistically significant. Further, the only other variable that is statistically

different between the two samples is total assets.

97

Table C.1
Descriptive statistics for the Dependent Variable, Independent Variables
and Control Variables for firms that managed earnings upward
(positive discretionary accruals) and managed
expectations downward (negative unexpected earnings forecast)

Panel A: Total Sample

 

Standard

Variablea N Mean Deviation Median
%OUT 428 0.73 0.15 0.75
TOTAL 428 8.14 2.37 8.00
TOPS 428 0.51 0.21 0.45
%INST 428 0.52 0.23 0.55
UNCFE 427 0.04 0.10 0.01
LITRISK 428 0.36 0.48 0.00
VREARN 428 0.11 0.32 0.00
GROWTH 428 6.79 76.13 2.16
SIZE” 428 599.52 409.27 510.94
DUR 428 0.54 0.50 1.00
R&D 428 0.05 0.07 0.02
LABOR 428 0.51 0.29 0.57
ICLAIM 337 (0.31) 0.75 (0.38)
TOTAL ASSETS 428 1,782.24 4,798.34 356.27

aStatistics in this table are presented as actual values while logarithmic values are
used in the analysis.
bSize is presented in millions.

98

Table C.1 (cont’d).

Panel B: Firms that met or exceeded anal ysts' expectations

 

Standard

Variablea N Mean Deviation Median
%OUT 283 0.73 0.15 0.75
TOTAL 283 8.29 2.43 8.00
TOPS 283 0.48 0.19 0.43
%INST 283 0.54 0.23 0.56
UNCFE 283 0.03 0.06 0.01
LITRISK 283 0.39 0.49 0.00
VREARN 283 0.12 0.32 0.00
GROWTH 283 9.36 92.57 2.13
SIZE” 283 650.73 422.10 571.88
DUR 283 0.51 0.50 1.00
R&D 283 0.05 0.07 0.02
LABOR 283 0.51 0.30 0.58
ICLAIM 224 (0.27) 0.74 (0.31)
TOTAL ASSETS 283 2,190.43 5,524.68 432.68
DISCRETIONARY 283 0.05 0.06 0.03

ACCRUALS
UNEXPECTED 283 (0.89) 1.29 (0.51)
EARNINGS FORECAST

aStatistics in this table are presented as actual values while logarithmic values are

used in the analysis.
bSize is presented in millions.

99

Table C.1 (cont’d).
Panel C: Firms that failed to meet or beat analysts' expectations

 

 

Standard
Variablea N Mean Deviation Median
%OUT 145 0.73 0.15 0.75
TOTAL 145 7.86 2.23 8.00
TOPS 145 0.57 0.22 0.54
%INST 145 0.50 0.24 0.52 "'
UNCFE 144 0.06 0.15 0.02
LITRISK 145 0.32 0.47 0.00
VREARN 145 0.11 0.31 0.00
GROWTH 145 1.77 19.16 2.23
SIZE” 145 499.58 364.10 385.93
DUR 145 0.59 0.49 1.00
R&D 145 0.05 0.07 0.03
LABOR 145 0.50 0.26 0.55
ICLAIM 1 13 (0.04) 0.77 (0.46)
TOTAL ASSETS 145 985.56 2,744.77 301.40
DISCRETIONARY 145 0.05 0.05 0.04
ACCRUALS
UNEXPECTED 145 (0.94) 1.69 (0.59)
EARNINGS FORECAST

2’Statistics in this table are presented as actual values while logarithmic values are
used in the analysis.
bSize is presented in millions.

100

 

Table C.2 provides descriptive statistics for firms that failed to manage earnings
upward but managed expectations downward. Panel A presents the total sample while
Panels B and C divide the sample into firms that met or beat analysts’ expectations and
those that failed to do so. From Panel B, the mean percentage of Shares held by the top
five institutional investors for firms that met or beat analysts’ expectations is .61. From
Panel C, the mean percentage of shares held by the top five institutional investors for
firms that failed to meet or beat analysts’ expectations is .67. The difference between the
two is statistically Si gnificant. Further, the only other variable that is statistically

different between the two samples is total assets.

101

 

A

Table C.2
Descriptive statistics for the Dependent Variable, Independent Variables and
Control Variables for firms that failed to manage earnings upward
(negative discretionary accruals) and managed expectations downward
(negative unexpected earnings forecast)

Panel A: Total Sample

 

Standard

Variablea N Mean Deviation Median
%OUT 373 0.72 0.16 0.75
TOTAL 373 7.23 2.76 7.00
TOPS 373 0.63 0.23 0.61
%IN ST 372 0.43 0.25 0.42
UNCFE 371 0.09 0.18 0.02
LITRISK 373 0.48 0.50 0.00
VREARN 373 0.34 0.47 0.00
GROWTH 373 3.40 16.30 1.94
SIZE” 373 404.22 414.59 268.75
DUR 373 0.50 0.50 0.00
R&D 373 0.14 0.24 0.07
LABOR 373 0.59 0.34 0.69
ICLAIM 300 0.21 1.12 0.01
TOTAL ASSETS 373 922.32 4,226.61 109.65

2’Statistics in this table are presented as actual values while logarithmic values are
used in the analysis.
bSize is presented in millions.

102

Table C.2 (cont’d).
Panel B: Firms that met or beat analysts' expectations

 

Standard

Variablea N Mean Deviation Median
%OUT 226 0.72 0.16 0.75
TOTAL 226 7.16 2.60 7.00
TOPS 226 0.61 0.23 0.57
%INST 225 0.46 0.26 0.44
UNCFE 224 0.07 0.14 0.02
LIT RISK 226 0.51 0.50 1.00
VREARN 226 0.32 0.47 0.00
GROWTH - 226 4,144.00 6.25 2.00
SIZE” 226 456.13 463.72 289.84
DUR 226 0.48 0.50 0.00
R&D 226 0.13 0.21 0.08
LABOR 226 0.61 0.33 0.71
ICLAIM 184 0.23 1.09 0.06
TOTAL ASSETS 226 1,267.75 5,347.83 121.93
DISCRETIONARY 226 (0.1 1) 0.28 (0.04)

ACCRUALS
UNEXPECTED 226 (0.91) 1.41 (0.47)
EARNINGS FORECAST

”Statistics in this table are presented as actual values while logarithmic values are
used in the analysis.
bSize is presented in millions.

103

Table C.2 (cont’d).

Panel C: Firms that failed to meet or beat analysts' expectations

 

Standard

Variable“ N Mean Deviation Median
%OUT 147 0.73 0.16 0.78
TOTAL 147 7.33 2.99 7.00
TOPS 147 0.67 0.23 0.67
%INST 147 0.40 0.23 0.38
UNCFE 147 0.12 0.23 0.05
LITRISK 147 0.43 0.50 0.00
VREARN 147 0.37 0.48 0.00
GROWTH 147 2.25 24.79 1.85
SIZE” 147 324.41 309.79 206.25
DUR 147 0.52 0.50 1.00
R&D 147 0.15 0.27 0.06
LABOR 147 0.57 0.35 0.67
ICLAIM 116 0.17 1.18 (0.11)
TOTAL ASSETS 147 391.26 986.95 90.93
DISCRETIONARY 147 (0.12) 0.32 (0.05)

ACCRUALS
UNEXPECTED 147 (0.80) 1.66 (0.34)
EARNINGS FORECAST

3 Statistics in this table are presented as actual values while logarithmic values are

used in the analysis.
bSize is presented in millions.

104

Table C.3 provides descriptive statistics for firms that failed to manage earnings
upward and failed to manage expectations downward. Panel A presents the total sample
while Panels B and C divide the sample into firms that met or beat analysts’ expectations
and those that failed to do 80. From Panel B, the mean percentage of shares held by the
top five institutional investors for firms that met or beat analysts’ expectations is .55.
From Panel C, the mean percentage of shares held by the top five institutional investors .
for firms that failed to meet or beat analysts’ expectations is .71. The difference between
the two is statistically significant. Further, the only other variable that is statistically

different between the two samples is total assets.

 

I

105

Table C.3
Descriptive statistics for the Dependent Variable, Independent Variables
and Control Variables for firms that failed to manage earnings upward
(negative discretionary accruals) and failed to manage expectations
downward (positive unexpected earnings forecast)

Panel A: Total Sample

 

Standard

Variable“ N Mean Deviation Median
%OUT 267 0.71 0.16 0.75
TOTAL 267 7.40 2.40 7.00
TOPS 267 0.63 0.24 0.62
%INST 267 0.45 0.28 0.45
UNCFE 266 0.18 0.66 0.02
LIT RISK 267 0.54 0.50 1.00
VREARN 267 0.43 0.50 0.00
GROWTH 267 3.87 10.77 2.01
SIZE” 267 455.04 459.40 334.75
DUR 267 0.47 0.50 0.00
R&D 267 0.15 0.33 0.08
LABOR 267 0.64 0.27 0.71
ICLAIM 225 0.33 1.18 0.27
TOTAL ASSETS 267 439.56 1,087.21 119.59

“Statistics in this table are presented as actual values while logarithmic values are
used in the analysis.
bSize is presented in millions.

106

 

_l

Table C.3 (cont’d).
Panel B: Firms that met or exceeded analysts' expectations

 

Standard

Variable“ N Mean Deviation Median
%OUT 144 0.72 0.16 0.75
TOTAL 144 7.62 2.57 7.00
TOPS 144 0.55 0.22 0.50
%INST 144 0.51 0.27 0.53
UNCFE 143 0.04 0.12 0.01
LITRISK 144 0.54 0.50 1.00
VREARN 144 0.33 0.47 0.00
GROWTH 144 4.18 13.00 2.20
SIZE” 144 589.16 522.79 478.13
DUR 144 0.44 0.50 0.00
R&D 144 0.11 0.12 0.07
LABOR 144 0.67 0.25 0.75
ICLAIM 124 0.30 0.83 0.27
TOTAL ASSETS 144 569.03 1,366.27 188.21
DISCRETIONARY 144 (0.08) 0.11 (0.04)

ACCRUALS
UNEXPECTED 144 0.78 1.28 0.40
EARNINGS FORECAST

“Statistics in this table are presented as actual values while logarithmic values are
used in the analysis.
bSize is presented in millions.

107

 

Table C.3 (cont’d).

Panel C: Firms that fail to meet or exceed analysts' expectations

 

Standard

Variable“ N Mean Deviation Median
%OUT 123 0.70 0.16 0.71
TOTAL 123 7.16 2.16 7.00
TOPS 123 0.71 0.23 0.78
%INST 123 0.37 0.27 0.31
UNCFE 123 0.34 0.94 0.05
LITRISK 123 0.53 0.50 1.00
VREARN 123 0.55 0.50 1.00
GROWTH 123 3.50 7.40 1.86
SIZE” 123 298.03 306.11 176.55
DUR 123 0.50 0.50 1.00
R&D 123 0.19 0.46 0.08
LABOR 123 0.60 0.28 0.67
ICLAIM 101 0.37 1.51 0.27
TOTAL ASSETS 123 287.98 588.43 74.69
DISCRETIONARY 123 (0.14) 0.54 (0.06)

ACCRUALS
UNEXPECTED 123 1.10 4.44 0.42
EARNINGS FORECAST

“Statistics in this table are presented as actual values while logarithmic values are

used in the analysis.
bSize is presented in millions.

108

 

Table C.4 provides descriptive statistics for firms that managed earnings upward
and failed to manage expectations downward. Panel A presents the total sample while
Panels B and C divide the sample into firms that met or beat analysts’ expectations and
those that failed to do so. From Panel B, the mean percentage of shares held by the top
five institutional investors for firms that met or beat analysts’ expectations is .48. From
Panel C, the mean percentage of shares held by the top five institutional investors for r:
firms that failed to meet or beat analysts’ expectations is .62. The difference between the
two is statistically Significant. Further, the differences between total assets and

discretionary accruals between the two samples is statistically significant.

 

109 P

Table C.4
Descriptive statistics for the Dependent Variable, Independent Variables
and Control Variables for firms that managed earnings upward
(positive discretionary accruals) and failed to manage expectations
downward (positive unexpected earnings forecast)

Panel A: Total Sample

 

Standard

Variable“ N Mean Deviation Median
%OUT 370 0.74 0.15 0.75
TOTAL 370 8.41 3.38 8.00
TOPS 370 0.53 0.23 0.46
%IN ST 370 0.52 0.24 0.56
UNCFE 370 0.06 0.33 0.01
LITRISK 370 0.35 0.48 0.00
VREARN 370 0.17 0.38 0.00
GROWTH 370 3.58 6.32 2.04
SIZE” 370 619.16 475.62 516.40
DUR 370 0.34 0.47 0.00
R&D 370 0.04 0.07 0.00
LABOR 370 0.46 0.32 0.51
ICLAIM 215 (0.21) 0.89 (0.23)
TOTAL ASSETS 370 1,275.64 3,832.55 287.76

“Statistics in this table are presented as actual values while logarithmic values are
used in the analysis.
“Size is presented in millions.

110

 

Table C.4 (cont’d).

Panel B: Firms that met or beat analysts' expectations

 

Standard
Variable“ N Mean Deviation Median
%OUT 254 0.74 O. 15 0.77
TOTAL 254 8.43 3.40 8.00
TOPS 254 0.48 0.21 0.42
%INST 254 0.56 0.24 0.59
UNCFE 254 0.01 0.03 0.00
LITRISK 254 0.37 0.48 0.00
VREARN 254 0.16 0.36 0.00
GROWTH 254 3.18 4.83 1.95
SIZEb 254 699.22 484.33 614.06
DUR 254 0.37 0.48 0.00
R&D 254 0.04 0.08 0.00
LABOR 254 0.47 0.31 0.52
ICLAIM 144 (0.22) 0.95 (0.18)
TOTAL ASSETS 254 1,186.63 3,758.24 321.88
DISCRETIONARY 254 0.06 0.05 0.04
ACCRUALS
UNEXPECTED 254 0.60 0.79 0.32
EARNINGS FORECAST

“Statistics in this table are presented as actual values while logarithmic values are

used in the analysis.
“Size is presented in millions.

111

 

Table C.4 (cont’d).

Panel C: Firms that failed to meet or beat analysts' expectations

 

Standard

Variable“ Mean Deviation Median
%OUT 116 0.73 0.14 0.75
TOTAL 116 8.37 3.35 8.00
TOPS l 16 0.62 0.24 0.59
%INST l 16 0.44 0.23 0.46
UNCFE 116 0.17 0.58 0.02
LITRISK l 16 0.32 0.47 0.00
VREARN 1 16 0.20 0.40 0.00
GROWTH 116 4.44 8.71 2.16
SIZE” 1 16 443.86 405.74 306.25
DUR 1 16 0.28 0.45 0.00
R&D 1 16 0.03 0.06 0.00
LABOR l 16 0.44 0.33 0.49
ICLAIM 71 0.44 0.33 0.49
TOTAL ASSETS 116 1,470.54 4,000.16 178.59
DISCRETIONARY 1 16 0.05 0.05 0.03

ACCRUALS
UNEXPECT ED 1 16 0.64 1.27 0.30
EARNINGS FORECAST

“Statistics in this table are presented as actual values while logarithmic values are

used in the analysis.
“Size is presented in millions.

112

 

I]. .
I... 4. . . Iii/«Hill! pi..hr.’.bn.w

Appendix D - Supplementary Analyses for Hypothesis Two
This appendix presents supplementary analyses for the results presented in Table
5 of Chapter 5. Table D.l provides results for a regression that includes the corporate
governance variables without control variables. The coefficients are consistent with
those in Table 5 of Chapter 5. The percentage of outsiders on the board continues to
remain insignificant”.
Table D.1
Logit Analysis of the Association between Corporate Governance Mechanisms and the

Probability of Managing Expectations versus Managing Earnings without Control
Variables

 

 

1"..V'-a.§.‘" I'A. .2. I -'-

P(MANAGE=1)“ = ,60 + ,6l %OUT + ,6,T0TAL+ ,6,T0P5 + ,6, %INST + ,BSUNCFE
+ ,66LITRISK + B,VREARN + ,68 GROWTH + 6, SIZE + 6,,ICLAIM + 6

Manage Expectations Manage Neither Manage Both
Independent Variable“ Coefficient p-value Coefficient p-value Coefficient p-value
INTERCEPT 0.18 0.76 -0. 16 0.81 0.75 0.16
Corporate Governance
Mechanisms
%OUT -0.66 0.27 -0.55 0.41 -0.57 0.30
TOTAL -0. 14 0.00 -0.08 0.04 -0.01 0.61
TOPS 2.27 0.00 1.26 0.01 -0.20 0.63
N: 907.00
Log Likelihood: -1,l92.7
p-value= <.05

“ MANAGE is a categorical variable equal to 0 if a firm manages neither earnings nor
expectations, 1 if a firm manages expectations, 2 if a firm manages earnings and 3 if a firm
manages both.

“ Refer to Table 1 for variable definitions.

 

12 . . . . .
The results remarn unchanged when continuous variables are wrnsorized at the .01 and .10 level.

113

5.1—I

Table D2 provides results for a regression that includes the corporate
governance variables with only the control variable for percentage of institutional
holders. The coefficients are consistent with those in Table 5 of Chapter 5. The
percentage of outsiders on the board continues to remain insignificant”.
Table D2
Logit Analysis of the Association between Corporate Governance Mechanisms and the

Probability of Managing Expectations versus Managing Earnings with Percentage of m.
Institutional Holders as Control Variable

P(MANAGE =1)“ = 50 + 61%0UT + ,6,T0TAL + fl3T0P5 + 6,, %INST + ,BSUNCFE
+ 66LITRISK + B7VREARN + A GROWTH + 6, SIZE + 6,0 ICLAIM + e

 

Manage Expectations Manage Neither Manage Both
Independent Variable“ Coefficient p-value Coefficient p-value Coefficient p-value
INTERCEPT 0.59 0.46 -0.07 0.93 1.53 0.03
Corporate Governance
Mechanisms
%OUT -0.60 0.32 -0.56 0.41 -0.57 0.30
TOTAL -0. 14 0.00 -0.08 0.04 -0.02 0.43
TOPS 1.92 0.00 1.21 0.07 -0.79 0.15
Control Variables
%INST -0.47 0.35 -0.08 0.88 -0.79 0.09
N: 906.00
Log Likelihood: -1190
p-value= <.05

“ MANAGE is a categorical variable equal to 0 if a firm manages neither earnings nor
expectations, 1 if a firm manages expectations, 2 if a firm manages earnings and 3 if a firm
manages both.

“ Refer to Table 1 for variable definitions.

Table D3 provides results for a regression that includes the corporate governance

variables with all of the control variables including the three individual variables for

 

13 . . . . .
The results remain unchanged when contlnuous varIables are wmsorrzed at the .01 and .10 level.

114 @

 

reliance on implicit claims. The coefficients are consistent with those in Table 5 of
Chapter 5. The percentage of outsiders on the board continues to remain insignificant. In
addition, the coefficient on board Size becomes insignificant”.

Table D3
Logit Analysis of the Association between Corporate Governance Mechanisms and
the Probability of Managing Expectations versus Managing Earnings with all
Individual Control Variables

P(MANAGE =1)” = ,60 + ,61 %OUT + 62T0TAL + ,6,T0P5 + fl.- %INST + ,BSUNCFE
+ fl6LITRISK + B,VREARN + ,68 GROWTH + 6, SIZE + ,BmlClAIM + e

 

Manage Expectations Manage Neither Manage Both
Independent Variable“ Coefficient p-value Coefficient p-value Coefficient p-value
INT ERCEPT -2.80 0.01 -3.84 0.00 - l .42 0.12
Corporate Governance
Mechanisms
%OUT -0.73 0.25 -0.78 0.27 —0.49 0.39
TOTAL -0.06 0.18 0.01 0.88 -0.07 0.06
TOPS 2.47 0.00 1.75 0.04 0.57 0.40
Control Variables

%INST 0.29 0.60 0.50 0.41 -0.68 0.16
UNCFE 10.90 0.00 9.18 0.00 10.03 0.00
LITRISK 0.42 0.05 0.48 0.05 -0.08 0.70
VREARN -0. 12 0.66 0.21 0.47 -0.56 0.04
GROWTH 0.02 0.08 0.02 0.09 0.02 0.06
SIZE 0.09 0.35 0.07 0.52 0.31 0.00
RD 5.39 0.00 4.72 0.00 1.48 0.32
DUR 0.48 0.02 0.33 0.15 0.50 0.01
LABOR 1.06 0.00 2.10 0.00 0.23 0.42

N = 903.00

Log Likelihood: -1 102.5

p-value= 0.00

“ MANAGE is a categorical variable equal to 0 if a firm manages neither earnings nor
expectations, 1 if a firm manages expectations, 2 if a firm manages earnings and 3 if a firm
manages both.

“ Refer to Table 1 for variable definitions.

 

14 . . . . .
The results remain unchanged when continuous variables are wrnsorrzed at the .01 and .10 level.

115

firm-

Table D.4 provides results for a regression that includes the corporate
governance variables with all of the control variables including reliance on implicit
claims. The coefficients are consistent with those in Table 5 of Chapter 5. The

percentage of outsiders on the board continues to remain insi gnificant15 .

 

 

15 . . . . .
The results remam unchanged when continuous variables are wrnsonzed at the .01 and .10 level.

116

Table D.4
Logit Analysis of the Association between Corporate Governance Mechanisms and
the Probability of Managing Expectations versus Managing Earnings with the Factor
Reliance on Implicit Claims

P(MANAGE =1)” = ,6, + ,6l %OUT + 6,T0TAL + 6,70P5 + ,6, %INST + ,BSUNCFE
+ ,6,L1TRISK + B,VREARN + 6, GROWTH + ,6, SIZE + 6,,ICLAIM + 6

Manage Expectations Manage Neither Manage Both
Independent Variable“ Coefficient p—value Coefficient p-value Coefficient p—value
INTERCEPT 0.06 0.95 -0.37 0.78 1.63 0.13
Corporate Governance
Mechanisms
%OUT -1.19 0.13 -l.48 0.07 -1.32 0.07
TOTAL -0.11 0.03 -0.03 0.57 -0.08 0.05
TOPS 2.19 0.01 1.52 0.12 -0.71 0.40
Control Variables
%INST -0.21 0.74 0.38 0.58 -0.98 0.08
UNCFE 7.81 0.01 5.81 0.08 6.54 0.05
LIT RISK 0.04 0.87 0.18 0.50 -0.26 0.27
VREARN 0.08 0.78 0.30 0.31 -0.89 0.01
GROWTH 0.02 0.33 0.01 0.75 0.02 0.27
SIZE 0.09 0.42 0.03 0.78 0.21 0.02
RD
DUR
LABOR
ICLAIM 0.42 0.00 0.58 0.00 -0.01 0.95
N: 673.00
Log Likelihood: -830.4
p-value= <.05

“ MANAGE is a categorical variable equal to 0 if a firm manages neither earnings nor
expectations, 1 if a firm manages expectations, 2 if a firm manages earnings and 3 if a firm
manages both.

“ Refer to Table 1 for variable definitions.

117

 

I'_l

Appendix E - Additional Supplementary Analyses for Hypothesis Two

This appendix presents supplementary analyses regarding whether a firm has a
strategy of managing earnings and/or expectations. Each table represents the number of
years a firm is in the sample. For each firm, y, represents whether a firm had a negative

value for the unexpected earnings forecast for the expectations management variable and

 

thus would be classified as a firm that managed expectations. Y also represents a firm I.
with positive discretionary accruals and as such is classified as a firm that managed
earnings.

Panel A of Table E.1 examines firms that appear in the sample two years. The E

table shows that a significant number of firms manage expectations rather than earnings.
Panel A also shows that 11% of the firms that are in the sample twice appear to manage

both earnings and expectations.

118 g

Table E.l
Sequence of meeting expectations using earnings
management or expectations management

Panel A: Firms appear in the sample two years

 

 

Expectations Earnings Percentage
N Management Management Of Total

14 nn nn 6.7
l l nn yn 5.3
19 nn yy 9.6
14 ny nn 6.7
4 my ny 1.4
9 ny yn 4.3
23 ny yy 1 Ll
l7 yn nn 8.2
yn ny 3.8

yn yn 3.8

12 yn yy 5.8
28 yy nn 13.5
9 yy ny 4.3
10 yy yn 4.8
23 W W l 1-1

 

Panel B of Table E.l examines firms that appear in the sample three years. With
these firms it is more difficult to see a discernible pattern in the use of expectations
management and/or earnings management. Only two firms in the sample managed
expectations in all three years while not managing earnings. Five firms in the sample

managed earnings in all three years while not managing expectations.

119

Table E.l (cont’d)
Panel B: Firms appear in the sample three years

 

120

Expectations Earnings Percentage

N Management Management Of Total

1 nnn nnn 1.6
1 nnn nyn 1.6
5 nnn yyy 8. 1
l nny nny 1.6
l nny yny 1.6
5 nny yyy 8. l
l nyn nny 1.6
l nyn nyy 1.6
1 nyn yyn 1.6
1 nyn yyy 1.6
1 nyy nyn 1.6
2 nyy ynn 3.2
3 nyy yyy 4-8
l ynn nnn 1.6
3 ynn nny 4.8
l ynn nyn 1.6
3 ynn nyy 4.8
1 ynn yyy 1.6
2 yny nnn 3.2
1 yny nyn 1.6
2 yny yyy 3-2
2 yyn nnn 3.2
l yyn nny 1.6
1 yyn nyy 1.6
3 yyn yyn 4.8
l yyn yyy 1-6
2 yyy nnn 3.2
1 yyy nyn 1.6
2 yyy nyy 3-2
2 yyy ynn 3.2
2 yyy yny 3-2
2 yyy yyn 3-2
5 yyy yyy 8-1

Panel C of Table E.l examines firms that appear in the sample four years. There
appears to be no pattern in terms of firms that consistently use one method versus
another. With these firms it is more difficult to see a discernible pattern in the use of
expectations management and/or earnings management. Only one firm in the sample
managed expectations in all four years while not managing earnings. Two firms in the
sample managed expectations and earnings in all four years they appeared in the sample.

Table E.l (cont’d)
Panel C: Firms in the sample four years

 

 

 

Expectations . Earnings Percentage of

N Management Management Total

1 nnnn ynyy 2.3
l nnny nnnn 2.3
l nnny nyyn 2.3
1 nnny yynn 2.3
1 nnny yyyy 2.3
1 nnyn . yyyn 2.3
1 nnyy yyny 2.3
1 nynn nnnn 2.3
1 nynn ynyy 2.3
l nyny nyyn 2.3
l nyny yyyn 2-3
l nyyn ynyn 2.3
1 nyyn ynyy 2.3
l nyyn yynn 2.3
l nyyn yyyy 2-3
l nyyy ynny 2.3
2 ynnn nnnn 4.7
1 ynny nnyn 2.3
1 ynny nyny 2.3
2 ynny nyyn 4.7
1 ynny yynn 2.3
2 ynny yyny 4.7
1 ynny yyyn 2.3

p

Table E.l (cont’d)

Panel C: (cont’d) Firms in the sample four years

Nu—au—u—Np—.y—.~_-

Panel D of Table E.l examines firms that appear in the sample five years. There
appears to be no pattern in terms of firms that consistently use one method versus
another. Because of the length of the string it is difficult to see a pattern in the use of
expectations management and/or earnings management. Only one firm in the sample
managed expectations in all five years while not managing earnings. One firm in the

sample managed expectations and earnings in all five years they appeared in the sample.

ynyy
ynyy
ynyy
ynyy
yynn
yynn
yyyn
YYYY
YYYY
yyyy
YYYY

Table E.1 (cont’d)

Panel D: Firms appear in the sample five years

Expectations

nnnnn
nnnny
nnnny
nnnny
nnnyn
nnnyn
nnnyy

nnnn
nnyn
yynn
yyyy
nnyy
nyyn
yyyn
nnnn
ynnn
ynny
yyyy

Earnings
Management Management

yyyyy
nyynn
ynyyy

YYYYY
nnynn

nyynn

nynny

122

Total

2L3
213
2L3
2L3
2L3
2L3
437
2L3
2L3
2L3
437

Percentage of

2:3
437
2:3
2L3
2L3
2L3
2L3

 

 

 

Table E.l (cont’d)

Panel D: (cont’d) Firms in the sample five years

~I—sh—‘F—‘Nt—ih—h—ID—it—‘U—it—‘F—F—I—Hh—H—‘D—Hh—IF—r—I—‘D—DF—t—‘F—i—h—‘b—D—H

nnnyy
nnnyy
nnnyy
nnnyy
nnynn
nnyny
nnyny
nnyyy
nynny
nyynn
nyyny
nyyny

nyyyn
ynnnn

ynnnn
ynnny
ynnyy
ynnyy
ynnyy
ynyyn
ynyyn
yynny
yynyn
yyynn
yyynn
yyyny
yyyny
yyyny
yyyny
yyyyn
yyyyn
yyyyn
yyyyn
YYYYY

nyyny
ynyyy
yyynn
yyyyy
yyyny
nnynn
yyyyy
yyyyy
yyyny
ynnnn
ynnyy
yynnn
ynnyy
yyynn
yyyyy
yyyyn
nynnn
yyynn
yyyyy
ynynn
yyyyy
ynyyn
yyynn
yynnn
yyyny
ynnny
ynyyy
yyynn
yyyny
ynnyn
ynyyn
yyyyn
yyyyy
yyyyy

123

253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
253
457
253
253
253
253

 

 

 

Appendix F — Valuation Supplementary Analyses

This appendix presents supplementary analyses concerning the market reaction
associated with a strategy of meeting or beating analysts’ expectations.
Introduction

Meeting or beating analysts’ expectations has taken on increased significance in
the last several years. Empirical evidence suggests that managers’ propensity to achieve
particular thresholds has shifted over time and currently, avoiding negative earnings
surprises is the most important threshold (Brown and Caylor 2003). However, the
valuation literature provides mixed support for the shift in managers’ propensity to avoid
negative earnings surprises. While short run market premiums and increased earnings
multiples accrue to firms that consistently meet or beat analysts’ expectations, these
valuation benefits subsequently disappear when firms that previously met expectations
fail to do so (Barth et al. 1999; Bartov et al. 2002). Related to this argument there is
evidence of a negative stock price response to a break in a string of earnings increases
(Ke and Petroni 2003; Ke et al. 2003; Skinner and Sloan 2002).

Additionally, managers and public accountants have expressed concern that the
“short term earnings game” is played to the detriment of overall firm value. Short run
targets, specifically meeting or beating analysts’ expectations could be met by making
business decisions that have a negative impact on a firm’s long run value (Favaro and
Rotz 2004; PricewaterhouseCoopers 2002; Rowchowdhury 2003). Nortel is an example
of a company that destroyed value by making acquisitions and over investing in an effort

to satisfy analysts’ growth expectations. As a result of the acquisitions and

124

 

 

overinvestment, there was a significant reduction in firm value. According to Fuller and
Jensen (2001) this kind of damage can be stopped if managers can just say no to the
pressure to fulfill unrealistic market expectations.
Since the benefits of playing the “short term earnings game” may result in long
run costs to the firm, it is unclear whether this phenomenon is in the best interest of

shareholders. The premise of this paper is that corporate governance mechanisms

provide incentives for managers to maximize long run firm value and as such forego the

market premium associated with meeting or beating analysts’ expectations in the short

run. Examining the association between corporate governance mechanisms and the

market premium surrounding a firm that either meets or beats analysts’ expectations or r"

 

fails to do so may shed light on whether corporate governance mechanisms influence a
firm’s decision to forego playing the meet or beat game in the short run. Jensen (2004)
suggests that corporate governance in the form of good control systems and monitoring
are necessary to stop the destruction of part or all of the core value of a firm due to
managers promising results they may not be able to deliver. The purpose of this paper is
to determine if investors are more interested in long run shareholder value and less
focused on market premiums that accrue in the short run. Raj gopal et al. (1999) find that
institutional investors look beyond current earnings compared to individual investors, in
this context, this would suggest a smaller market premium for firms with a sophisticated
investor base.

I test these hypotheses using a sample of 1,729 firm year observations from the
period 1997-2001. I use one measure of the shareholder base, the presence of an

institutional investor with long term incentives/preferences. Raj gopal et al. (1999) find

B

that the absolute value of discretionary accruals declines with institutional ownership.
This result is attributed to institutional investors being better informed than individual
investors. In order to allow monitoring incentives to vary, I use the percentage of shares
held by the top five institutional investors as a proxy for institutional investors with long
term incentives/preferences (Gillian and Starks 2003).
Hypothesis Development

There has been a move by many firms to change the type of guidance provided to
the market. Coca-Cola and Gillette are just two examples of companies that have
stopped providing earnings estimates to the market. The change was a result of too much
focus on short term guidance which distracted management from long term goals. In
February 2003, Mattel joined the list of firms that decided against providing earnings
estimates in the hopes that the change would refocus management on long term
shareholder value rather than meeting short term expectations (Kueppers and Weaver
2003). A survey by Deloitte and Touche found that 22 percent of partners had clients
that reduced the amount, or changed the type of information provided to the public. An
additional 20 percent of partners had clients that were debating the issue (Kueppers and
Weaver 2003). Although this survey focuses on clients of only one public accounting
firm the sentiment is echoed by managers and public accountants who suggest that
focusing on playing the earnings game may erode long run firm value (Favaro and Rotz

2004; PricewaterhouseCoopers 2002).16

 

"’Favaro and Rotz (2004) suggest that a focus on short term targets, specifically MBE, stands in the way of
building the capabilities and making the investments required to increase a company’s longer term value.
Reducing the level of research and development activities, pulling back from marketing investments and
delaying projects with long term payoffs are moves to cut costs significantly and ratchet up earnings in the

126

 

l

 

There is empirical evidence to suggest that playing the earnings game has real
costs to the firm. Consistent with the sentiments expressed by managers and public
accountants that meeting or beating analysts’ expectations forces managers to focus on
short run performance, Roychowdhury (2003) finds that firms that report small annual
profits exhibit unusually low cash flow from operations and unusually high production
costs. This suggests that managers may avoid losses by offering price discounts to boost
sales and engaging in overproduction to reduce reported costs of goods sold.

Additional empirical evidence that supports the long run costs associated with
failing to meet or beat analysts’ expectations is provided by Ke et al. (2003), who
establish a negative relationship between the stock- price response and a break in a string
of consecutive earnings increases. Similarly, Skinner and Sloan (2002) find that negative
earnings surprises for growth firms are penalized more than positive earnings surprises.
Collectively, these papers support the contention that there are costs associated with an
earnings break. Further, these papers are consistent with anecdotal evidence in the
financial press that detail the consequences associated with failing to meet or beat
analysts’ expectations”. The costs that accrue to the firm in the form of negative market
responses and real costs associated with directing resources away from long run projects
provide support for the reasons given by firms who decide to move away from providing

earning guidance to the market.

 

near term. Over time they rob a business of grth drivers and can actually destroy value in the long run.
PricewaterhouseCoopers (2002, p. 76) argue that erosion in long run value occurs when companies meet or
beat expectations by making business decisions that provide a short run earnings boost at the expense of
lower future cash flows.

'7 For example, on May 15, 2003, when Target missed its first quarter earnings target by one cent, shares of
the stock fell in early morning trading (Reuters 2003). On December 22, 2002, when Tupperware warned
that its annual earnings would fall short, the stock lost 14% of its value (CNNMoney 2002).

127

 

 

 

Although there are costs associated with meeting or beating analysts’
expectations this does not explain why the majority of firms continue to focus on meeting
analysts’ expectations in the short run. However, empirical research provides evidence of
short run valuation benefits in the form of higher stock prices and market premiums that
accrue to firms that meet or beat analysts’ expectations. Kasznik and McNichols (2002)

find that firms that meet expectations are rewarded with a higher stock price after

'7'"
controlling for analysts’ expectations of future earnings. However, in the long run, the
market penalizes firms that previously met expectations and subsequently fail to do so.

Similarly, Barth et al. (1999) find that firms which meet expectations receive a higher
return than their peers who fail to meet expectations. However, the earnings multiple "3

 

decreases when a firm that previously exhibited a pattern of increasing earnings
subsequently experiences an earnings decrease. Finally, Bartov et al. (2002) find larger
earnings multiples accrue to firms exhibiting a pattern of increasing earnings. However,
the future performance of firms that used earnings or expectations management was
inferior to that of firms that did not use either of these mechanisms. Firms that managed
earnings or expectations, however, still fared better than firms that failed to meet or beat
their earnings expectations. The valuation benefits suggest meeting or beating
expectations is advantageous to the firm in the short run but meeting or beating
expectations may have some long run negative ramifications when a firm that met or
exceeded expectations in the past subsequently fails to do so.

Raj gopal et al. (1999) examine the relationship between institutional ownership
and discretionary accruals. The paper finds that institutional investors are sophisticated

investors who are less likely to be fooled by earnings management. Because of a

D

sophisticated investor base, there is less incentive for this type of behavior and a strong

negative relationship exists between the absolute value of discretionary accruals and
institutional ownership. The paper also examines the stock market response to earnings.
Specifically, the focus is on whether institutional investors are overly focused on current
earnings. The paper finds that stock prices reflect a relatively greater proportion of the
information in future period earnings when institutional ownership is high. This is
consistent with institutional investors as sophisticated investors who look beyond current
earnings in assessing firm value. In the context of meeting or beating analysts’
expectations, I would expect sophisticated investors to understand the meet or beat game
and assign a valuation premium accordingly. Specifically, if a premium exists, I would
expect it to be smaller for firms with strong corporate governance mechanisms. This
leads to the following hypothesis:

[1,: If a premium exists, the premium to meeting or beating analysts’

expectations is smaller for firms with strong corporate governance

mechanisms than for firms with weak corporate governance mechanisms
Sample Selection

The initial sample consists of 4,865 firm year observations from the I/B/E/S

database of analysts’ forecasts from 1997-2001. To ensure consistency, analysts’
expectations and actual earnings are both obtained from the I/B/E/S database (Bhojraj et
al. 2003). Firm year observations are classified as meeting or exceeding expectations
(MBEzl) if reported earnings meet or exceed the outstanding consensus forecast at the

earnings announcement in both the current and previous year. Banks, insurance

companies and other financial services firms, (SIC 6000-6999) are excluded, as are

129

 

 

 

 

 

utilities (SIC 4800-4999) and other quasi regulated industries (SIC 4000-4499 and 8000
and above). All other financial data is obtained from the Compustat database.
Three criteria are set forth for the analysts’ forecasts to ensure that each firm year

observation is based on an initial analyst forecast and a revision that occurs later (Bartov

et al. 2002):
(1) There are at least two individual earnings forecasts for the year, which
are at least twenty trading days apart.
(2) The release date of the earliest forecast occurs at least three trading
days after the release of the previous year’s earnings.
(3) The release of the latest forecast precedes the earnings release by at

least three days.

The corporate governance and institutional holding data is obtained from the
Compact Disclosure database. This database takes financial data and text from
stockholders and Form 10—K reports, registration and proxy statements and other sources
for all current and new companies registered with the SEC and transfers it into Excel
format. Financial statements plus quarterly statements are provided with financial
footnotes, and the full text of the president's letter and management discussion. Also
included are lists of subsidiaries, major officers and board members and their salaries, 5%
individual and 10% institutional holders, ranges of earnings forecasts from advisory
services, and lists of additional SEC filings and exhibits.

The final sample after the above criteria are implemented includes 1, 729 firm

year observations. The impact of the three criteria on the sample size is set forth in Table

F.l.

130

 

Table F.1
Formation of the Final Sample for Hypothesis Two

Initial Sample 4,865

Firms excluded because there are not
two individual forecasts twenty days
apart (1,000)

Firms excluded because the release

date of the earliest forecast doesn’t

occur at least three days after the

release of the previous year’s earnings (1,825)
Firms excluded because the release

date of the latest forecast doesn’t

precede the earnings release by at least

three days (250)

Firms that appeared in sample one year (200)
1,590

Model Specification
To test whether the corporate governance mechanisms influence the premium to

meet or beat expectations the following regression is estimated:

CAR = 30 + ,6, ERROR+ ,BZSURP+ fl3MBE+ ,64T0P5 + flSMBE’“ S URP
+ ,B6TOP5 * S URP

CAR is the cumulative abnormal return over the period beginning two days following the
date of the earliest analyst forecast and ending one day after the release of the year’s 10K.
ERROR is the forecast error for the year and is the difference between actual earnings
and the earliest analyst forecast, deflated by the beginning of year stock price. SURP is

the difference between actual earnings and the latest analyst forecast, deflated by the

131

, ' .a ."u .n'
f
A ~

 

 

 

beginning of year stock price. By including both the surprise and error, the incremental
abnormal return is measured. MBE is an indicator variable taking on the value 1 if the
firm met or beat expectations in the current and previous year and 0 otherwise.
Additionally, TOPS represents the percentage of shares held by the top five institutions.
Hypothesis one states that the market premium, if it exists, will be smaller for firms with
a larger percentage of shares held by the top five institutional investors. This suggests
that [34 and Bo will be negative and significant which is consistent with a smaller market 5;

premium.

 

Results

 

The results of hypothesis one are presented in Table F.2. Terms [34 and [36 are both
negative. Term Bo is significant while [34 is marginally significant. This is partial support
for Hypothesis One. The market appears not to reward firms that manage to meet or beat
expectations with an additional premium when corporate governance mechanisms are
strong. This is consistent with the results of Raj gopal et a1. (1999) who find that
institutional investors are sophisticated investors who look beyond current earnings in

assessing firm value.

 

132 '

Table F .2
Regression of Cumulative Abnormal Return on percentage of shares held by the top five
institutional investors and other determinants of the cumulative abnormal return

CAR = )9, + )3, ERROR + ,BzSURP + ,63MBE + ,64T0P5 + ,BSMBE * SURP

 

 

 

 

+ ,B6T0P5 * SURP
(p-values are in parentheses)
B0 B1 B2 B3 B4 B5 B6 R2(%)
0.125 0.309 0.929 0.074 -0. 191 -0.302 -l.10 1.23
h
(0.06) (0.03) (0.16) (0.06) (0.04) (0.04) (0.13) .
CAR is the cumulative abnormal return over the period beginning two days following the
date of the earliest analyst forecast and ending one day after the release of the year’s l
results; '

ERROR is the forecast error for the year and is the difference between actual earnings
and the earliest analyst forecast deflated by the beginning of year stock price;
SURPRISE is the difference between actual earnings and the latest analyst forecast
deflated by beginning of year stock price;

MBE is an indicator variable taking on the value 1 if the firm met or beat expectations in
the current and previous year and 0 otherwise;

TOPS represents the percentage of shares held by the top five institutional investors.

Table F.3 presents the results of a regression of cumulative abnormal returns on
the percentage of shares held by the top five institutional investors and other determinants
of the cumulative abnormal return. The results are presented based on quintiles of the

level of earnings management. The coefficients of interest are B4 and [36. From Table F.3

neither coefficient is significant in any of the quintiles.

 

133 I

Table F.3
Regression of Cumulative Abnormal Return on percentage of shares held by the
top five institutional investors and other determinants of the cumulative abnormal '
return by quintile of level of earnings management as measured by the value of
discretionary accruals

CAR = ,60 + ,6, ERROR + ,BZSURP + ,B3MBE + BJOPS + ,BSMBE * SURP

 

 

 

 

 

 

+ fl6T0P5 * SURP
(p-values are in parentheses)
Quintile B0 B1 B3; B3 B4 B5 B6
0.010 0.802 5.95 0.366 -O.108 -10.33 -5.57
One (.74) (.05) (.65) (.07) (.78) (.05) (.68)
0.011 2.59 8.69 0.027 0.112 -33.82 -10.22
Two (.96) (.01) (.57) (.10) (.78) (.11) (.50)
0.220 -0.320 1.04 -0. 132 -0.487 9.14 -6.49
Three (.37) (.63) (.96) (.45) (.21) (.59) (.78)
0.123 -1.11 20.80 -0.304 0.010 19.11 -25.23
Four (.46) (.23) (.11) (.03) (.97) (.21) (.46)
0.037 1.69 5.01 -0.096 0.209 13.77 8.43
Five (.84) (.34) (.86) (.45) (.49) (.39) (.80)

 

CAR is the cumulative abnormal return over the period beginning two days following the

date of the earliest analyst forecast and ending one day after the release of the year’s

results;

ERROR is the forecast error for the year and is the difference between actual earnings

and the earliest analyst forecast deflated by the beginning of year stock price;

SURPRISE is the difference between actual earnings and the latest analyst forecast
deflated by beginning of year stock price;

MBE is an indicator variable taking on the value 1 if the firm met or beat expectations in

the current and previous year and 0 otherwise;

TOPS represents the percentage of shares held by the top five institutional investors.

134

 

 

Table F.4 presents the results of a regression of cumulative abnormal returns on
the percentage of shares held by the top five institutional investors and other determinants
of the cumulative abnormal return. The results are presented based on quintiles of the
level of expectations management. The coefficients of interest are [34 and [36. From Table

F4, [34 is significant in the fourth quintile and 13¢, is significant in both the first and fourth

Table F .4 q

Regression of Cumulative Abnormal Return on percentage of shares held
by the top five institutional investors and other determinants of the
cumulative abnormal return by quintile of level of expectations
management as measured by the value of unexpected earnings forecast

quintiles.

 

 

CAR = p, + ,6, ERROR + ,BZSURP + ,6,MBE + amps + ,BSMBE * SURP

 

 

 

 

 

 

 

+ ,86T0P5 * S URP
(p-values are in parentheses)
QUll’ltllC B0 B1 Bz B3 B4 B5 B6
0100 -1.51 78.9 0.158 0.106 -20.38 -84.10
One (.68) (.18) (.00) (.45) (.75) (.12) (.01)
-0.1 15 0.648 5.94 -0.186 0.397 25.4 -7.78
Two (.60) (.16) (.66) (.34) (.22) (.20) (.64)
0173 0.166 -0.158 0.171 0.219 -0.153 17.58
Three (.40) (.88) (.99) (.29) (.47) (.29) (.99)
0.502 0.839 -28.42 -0.063 -0.784 20.97 28.74
Four (.01) (.11) (.02) (.63) (.01) (.12) (.02)
-0.010 0.373 21.06 0.084 0.233 -8.83 -20.03
Five (.71) (.48) (.16) (.60) (.58) (.08) (.19)

 

CAR is the cumulative abnormal return over the period beginning two days following the
date of the earliest analyst forecast and ending one day after the release of the year’s
results;

ERROR is the forecast error for the year and is the difference between actual earnings
and the earliest analyst forecast deflated by the beginning of year stock price;

SURPRISE is the difference between actual earnings and the latest analyst forecast
deflated by beginning of year stock price;

MBE is an indicator variable taking on the value 1 if the firm met or beat expectations in
the current and previous year and 0 otherwise;

TOPS represents the percentage of shares held by the top five institutional investors.

 

135