 

 

 

l m .LIBRARY
1 o cg Michigan State
University

 

 

This is to certify that the
dissertation entitled

EARNINGS MANAGEMENT INCENTIVES OF TARGET FIRMS: AN
ANALYSIS OF SOLICITING AND UNSOLICITED TARGETS'
ACCRUAL ACCOUNTING CHOICES

presented by

Min Shen

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Department of Accounting and
Information Systems

 

 

K PK
Major Professor’s Signature

Masj I‘, 200;

 

 

Date

MSU is an Afﬁrmative Action/Equal Opportunity Institution

 

PLACE IN REIURN 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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 c'JCIRC/DctoDueJndd-Ms

 

 

 

EARNINGS MANAGEMENT INCENTIVES 0F TARGET FIRMS: AN
ANALYSIS OF SOLICITING AND UNSOLICITED TARGETS’ ACCRUAL
ACCOUNTING CHOICES
By

Min Shen

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

2005

ABSTRACT
EARNINGS MANAGEMENT INCENTIVES OF TARGET FIRMS: AN ANALYSIS
OF SOLICITING AND UNSOLICITED TARGETS’ ACCRUAL ACCOUNTING
CHOICES
By
Min Shen
Although prior literature suggests that bidders manage earnings upward to beneﬁt

from stock-ﬁnanced acquisitions, the results for target ﬁrms are mixed. Given that
targets are typically unaware of impending bids, they are unlikely to engage in
opportunistic earnings management. In this study I identify a setting in which targets not
only know that they are likely to be taken over, but they have relatively strong incentives
to consummate a takeover. Speciﬁcally I examine the pre-takeover accrual decisions of
240 target ﬁrms that publicly announced their intent to seek acquirers or merger partners
(soliciting targets) and compare them to the accrual decisions of 277 target ﬁrms that
received unsolicited takeover offers (unsolicited targets). In contrast to the suggestion in
the prior literature that targets have incentives to report positive discretionary accruals to
increase their takeover premium, I ﬁnd that neither soliciting targets nor unsolicited
targets appear to make positive accrual choices in the periods leading to the initial
takeover announcement. More importantly, the soliciting targets practice income-
decreasing earnings management in the year immediately preceding their solicitation
announcement. This ﬁnding is consistent with soliciting targets facing substantial costs
in the market for corporate control if opportunistic earnings management is detected.

Instead of endangering their chance of sale, soliciting targets choose to clean up their

balance sheets to credibly signal to the market their willingness to seek acquirers or

merger partners. Moreover, I ﬁnd that the managers of soliciting targets show declining
propensity to smooth earnings when approaching their solicitation decision, reﬂecting
managers’ lessened long-term economic and/or career incentives. My results highlight
the importance of controlling for target ﬁrms’ incentives when examining their accrual

accounting choices prior to takeover.

ACKNOWLEDGMENTS

I would like to express my gratitude to my dissertation advisor, Professor K.
Ramesh for his patience, inspiration and constant encouragement. His technical and
editorial advice was essential to the completion of this dissertation. My thanks also go to
other members of my dissertation committee, Kathy Petroni, Jun-Koo Kang, and Jeffrey
Wooldridge for reading drafts of this dissertation and providing guidance and insightful
comments on my dissertation.

I sincerely thank all faculty members and fellow doctoral students in the
Department of Accounting and Information Systems at Michigan State University who
gave me valuable comments during the workshop and practice presentation. Also thanks
are due to Deb North and other staff at Michigan Sate University for their generous help.

In particular, my parents, Yuexian Shen and Hongju Tang, my sister, Jie Shen,
and my husband, Xinran Wang, receive my deepest gratitude. Their dedication,

understanding and many years of support made this dissertation possible.

iv

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... vi
CHAPTER 1: INTRODUCTION .................................................................. 1
CHAPTER 2: LITERATURE REVIEW AND HYPOTHESES .............................. 7
2.1. Pre-takeover accrual choices by takeover targets ................................. 7
2.2. Income smoothing behavior of soliciting targets ................................. 10
CHAPTER 3: SAMPLE AND DESCRIPTIVE STATISTICS .............................. 14
3.1. Sample selection and description ................................................... 14
3.2. Descriptive statistics ................................................................. 18
CHAPTER 4: MEASURING DISCRETIONARY ACCRUALS ........................... 23
CHAPTER 5: ANALYSIS RESULTS .......................................................... 25
5.1. Accrual Analysis ..................................................................... 25
5.2. Income smoothing analysis ......................................................... 27
CHAPTER 6: OWNERSHIP STRUCTURE .......... ' ......................................... 35
CHAPTER 7: CONCLUDING REMARKS ................................................... 40

APPENDIX: INCOME SMOOTHING USING THE BACKING OUT APPROACH...43

BIBLIOGRAPHY ................................................................................... 62

Table 1:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table 7:

Table 8:

LIST OF TABLES

Sample Selection of 240 Soliciting Targets and 277 Unsolicited Targets
between 1990 and 2003 ................................................................. 49
Cited Reasons of Solicitation in News Releases and the Initial Announcement
Effects for Soliciting Targets ........................................................... 50
Panel A: Calendar Year of Initial Takeover Announcement for 240 Soliciting
and 277 Unsolicited Targets ............................................................ 51
Panel B: Two-Digit SIC Industry Classiﬁcation for 240 Solicited and 277
Unsolicited Targets ...................................................................... 52

Panel A: Descriptive Statistics on Selected Financial Characteristics (2-Digit
SIC Industry-Adjusted) of Soliciting and Unsolicited Targets One Year Prior to
the Initial Takeover Announcement ................................................... 53
Panel B: Probit Regressions of Soliciting versus Unsolicited Targets on Selected
Financial Characteristics (2-Digit SIC Industry-Adjusted) One Year Prior to the
Initial Takeover Announcement ....................................................... 55
Panel C: Comparison of Management Ownership and Institutional Ownership
(Industry-Unadjusted) One Year Prior to the Initial Takeover

Announcement ........................................................................... 56

Annual and Quarterly Discretionary Accruals of Soliciting and Unsolicited
Targets ..................................................................................... 58

Panel A: Annual OLS Regressions Of Discretionary Accruals on Pre-Managed
Earnings And Control Variables From Year —4 To Year 0, Where Year 0 Is The
Year Of Initial Announcement (Two-Tailed P Values In Parentheses) ............. 61
Panel B: Annual OLS Regressions Of Discretionary Accruals On Operating
Cash Flows And Control Variables From Year —4 To Year 0, Where Year 0 IS
The Year Of Initial Announcement (Two-Tailed P Values In Parentheses). ....62

Panel A: The Association between Management Ownership and Discretionary

Accruals in Year -1 ...................................................................... 64
Panel B: The Association between Institutional Ownership and Discretionary
Accruals in Year —1 ..................................................................... 65

OLS regressions of discretionary accruals on industry-unadjusted ownership
variables and control variables in year —1 (two-tailed p values in
parentheses) ............................................................................... 66

vi

Table A1: Randomization-adjusted income smoothing coefﬁcients of soliciting and
unsolicited targets from year —4 to year 0 (one-tailed p values in
parentheses) ............................................................................. 48

vii

CHAPTER 1: INTRODUCTION

This dissertation examines earnings management by target ﬁrms before takeover.
Conventional wisdom suggests that target ﬁrms have incentives to overstate reported
earnings using positive discretionary accruals to inﬂuence the short-term stock price prior
to takeover. However, attempts to document such evidence have been unsuccessful.
Erickson and Wang (1999) ﬁnd that target ﬁrms’ discretionary accruals are
insigniﬁcantly different from zero during periods preceding merger announcements. In
addition, Easterwood (1997/ 1 998) shows that only targets of hostile takeovers manage
earnings upward prior to a takeover announcement; for targets of friendly tender offers,
their discretionary accruals are not different from those of non-targets. These ﬁndings
suggest that opportunistic earnings management is not a widespread practice among
target ﬁrms in general. In contrast to the research on targets, prior literature documents
that acquiring ﬁrms use positive discretionary accruals to inﬂuence the exchange ratio
prior to stock-ﬁnanced acquisitions (Erickson and Wang, 1999; Louis, 2004).

This dissertation seeks to explore why some targets choose not to manage
earnings upward before takeover even though they could potentially beneﬁt. Prior
research argues that because targets are typically unaware of the impending bids, they are
unlikely to engage in upward earnings management as many bidders do (Erickson and
Wang, 1999). In this dissertation I identify a context in which targets not only know that
they are likely to receive bids, but they have incentives to adopt income-decreasing
accounting practices prior to any takeover attempts. Speciﬁcally, I examine the accrual

behavior of a group of target ﬁrms that publicly announce their intent to solicit takeover

offers (hereafter, “soliciting targets”). This group of takeover targets merits investigation
because they have the opportunity to manage earnings by timing their solicitation
announcement.

More importantly, as a subset of targets of friendly takeovers, soliciting targets
clearly face relatively higher costs and a stronger disincentive to manage earnings upward
than other target ﬁrms. Soliciting targets seek acquirers or merger partners to address
existing ﬁnancial problems or to exploit strategic growth opportunities. By explicitly
seeking buyers, these ﬁrms subject themselves to the close scrutiny of the entire market
for corporate control. Any opportunistic accounting ploys used by soliciting targets
could potentially endanger their ability to ﬁnd acquirers or merger partners at terms that
maximize their shareholder value. Consequently, companies seeking buyers are more
likely to “clean-up” their ﬁnancial statements to credibly signal their interest in
efficiently consummating a deal. Consistent with this line of reasoning, some merger and
acquisition experts believe that “honesty is the best policy when selling [a] company.” 1
Moreover, they recommend that companies seeking buyers proactively clean up their
ﬁnancial statements before putting themselves up for sale. Anecdotal evidence shows
that target ﬁrms follow this strategy to encourage or facilitate bids for their companies.
Therefore, I hypothesize that soliciting targets make income-decreasing accrual choices
prior to their solicitation announcement to credibly signal their interests in ﬁnding

strategic acquirers or merger partners.2

 

I See “What is your company worth”, by George Alexander, Black Enterprise, July 1, 2001 (vol. 31, issue
12).

2 Any observed income-decreasing accrual choices are subject to two interpretations. Soliciting targets
who have used income-increasing accrual choices in the past are using income-decreasing accrual choices
prior to solicitation to reverse the effects of past earnings management. Alternatively, in order to convince
potential bidders of their ﬁnancial reporting credibility, soliciting targets might report large negative

I also investigate the income smoothing behavior of mangers of soliciting targets
because the desire to seek a buyer may also be reﬂected in their weakened propensity to
smooth earnings. Previous studies ﬁnd that managers in general make discretionary
accounting choices to smooth earnings for compensation (Gaver, etal., 1995) or job
security (DeFond and Park, 1997). However, given their time horizon, managers of
soliciting targets may be less concerned about such longer-term beneﬁts of income
smoothing. Instead, large payouts from golden parachutes and other economic beneﬁts
from ﬁnding an immediate strategic partner might motivate the managers of soliciting
targets to take measures, such as balance Sheet cleanup, to welcome takeover attempts.3
Therefore, I hypothesize that the income smoothing incentives of managers of the
soliciting targets diminish as approaching the solicitation announcement.

From Securities Data Company (SDC) I identiﬁed 240 soliciting targets during
the period 1990-2003. On average the soliciting targets earn a signiﬁcantly positive 8%
abnormal stock return surrounding their solicitation announcement, suggesting that the
sale decision is perceived to enhance shareholders’ wealth. I further selected a group of
277 ﬁrms that received unsolicited takeover bids (hereafter, “unsolicited targets”) during
the same period. In contrast to soliciting targets, unsolicited targets may not foresee the
takeover bids, in which case their pre-takeover accrual choices are less likely to be
affected by the impending offers. Thus, these targets provide a powerful control for
earnings management not motivated by takeover.

My analysis reveals that neither soliciting targets nor unsolicited targets

systematically select accounting accruals to overstate reported earnings in the periods

 

discretionary accruals to present a conservative ﬁnancial position. This study does not attempt to
disentangle these two interpretations.
3 See “No wonder CEO’s love mergers”, by Gretchen Morgenson, New York Times, July 18, 2004.

leading to a takeover. This result is consistent with the intuition that the unsolicited
targets have insufﬁcient time to respond to the unexpected takeover bids, and that the
potential costs associated with opportunistic earnings management are signiﬁcant enough
to hinder the soliciting targets from inﬂating reported earnings. Consistent with my ﬁrst
hypothesis, I ﬁnd that soliciting targets report signiﬁcantly negative discretionary
accruals in the year before and in three out of four quarters immediately preceding their
initial solicitation announcement. In addition, the level of soliciting targets’ discretionary
accruals is signiﬁcantly lower than that of the unsolicited targets. Moreover, the negative
discretionary accruals are not performance-driven nor do they reverse in the year
(quarter) following the solicitation announcement. These ﬁndings suggest that soliciting
targets clean up their balance sheets to make them credible and worthy targets.
Alternatively stated, soliciting targets’ income-decreasing accrual choices before
solicitation primarily reﬂect their desire to consummate a sale. To lend support to the
assumption that soliciting targets do not frustrate takeover, I Show that none of the
soliciting targets have any poison pill defenses in place, whereas 12.6% of unsolicited
targets have at least one poison pill defense to resist potential takeover attempts.
Consistent with my second hypothesis, I ﬁnd that the negative association
between discretionary accruals and pre-managed earnings for the soliciting targets
declines over the period leading up to the solicitation decision. This result suggests that
the managers of soliciting targets have less proclivity to smooth earnings before takeover.
In contrast, I ﬁnd that the earnings smoothing behavior of unsolicited targets is unabated

by the impending takeover bids.

To further understand the income-decreasing accounting practices adopted by
soliciting targets, I investigate the impact of management ownership and institutional
ownership on their accrual choices one year prior to their solicitation. While
management ownership better aligns the interests of managers and investors, it could also
result in entrenched management when an inefﬁcient manager controls a substantial
fraction of stock equity (Morck et al., 1988). 1 ﬁnd that soliciting targets with relatively
low (high) management ownership report signiﬁcantly negative (positive) discretionary
accruals. Because the interest of soliciting target managers who hold a relatively small
portion of their ﬁrms’ equity converges to that of investors, these managers make
income-decreasing accrual choices to facilitate the sale of their companies, while
entrenched managers make income-increasing accrual choices to make takeover bids
costly. However, this association between discretionary accruals and management
ownership is not present in unsolicited targets.

Another aspect of ownership structure, institutional ownership, is assumed to play
an important role in monitoring and disciplining managers’ opportunistic behavior
(Shleifer and Vishny, 1986, 1997; McConnell and Servaes, 1990; Denis et al., 1997).
Consistent with this argument, I ﬁnd a negative association between the level of
institutional ownership and discretionary accruals, but this association is not signiﬁcantly
different between soliciting and unsolicited targets. It appears that the governance role of
institutional investors compelling managers to make more conservative (less aggressive)
accrual choices is not contingent on the ﬁrms’ solicitation decision.

This dissertation contributes to the existing literature on earnings management in

takeover targets. My result that soliciting targets indeed clean up their balance sheet

prior to takeover suggests that, in the population of takeover targets, earnings
management through positive discretionary accruals may only account for a fraction of
earnings managing attempts by targets to maximize shareholder value. Other target ﬁrms
with incentives similar to soliciting targets may not manage earnings upward or may even
choose to manage earnings downward, especially when they are pressured to solve
ﬁnancing problems or satisfy their growth needs through a merger and acquisition
transaction. This study also enhances our understanding of the role of ownership
structure in corporate control actions by showing that, when managers’ and investors’
interests are aligned, managers of soliciting targets make negative accrual choices to
facilitate takeover bids. Finally, I ﬁnd that managers of soliciting targets are less likely to
be involved in income smoothing due to the expected sale of their companies. Thus this
study adds to the literature on income smoothing by providing a setting in which
managers’ incentives for income smoothing are weaker.

The remainder of the dissertation is organized as follows. Chapter 2 provides a
review of related literature and develops the rationale for my hypotheses. The sample
selection criteria of soliciting targets and unsolicited targets are presented in Chapter 3,
along with descriptive statistics. Chapter 4 discusses the methodology of estimating
discretionary accruals, and Chapter 5 details the empirical tests and results of the
managers’ accrual choices and income-smoothing behavior in the two groups of target
ﬁrms. Chapter 6 reports analysis results of associations between ownership structure and
discretionary accruals. Concluding remarks are provided in Chapter 7. The appendix
includes a discussion of the backing out method used in testing the income smoothing

hypothesis.

CHAPTER 2: LITERATURE REVIEW AND HYPOTHESES

2.1. Pre-takeover accrual choices by takeover targets

The importance of accounting earnings in equity valuation creates an incentive for
earnings management (Healy and Wahlen, 1999). This incentive also manifests itself
when companies are engaged in corporate control and capital structure transactions. For
instance, extant literature ﬁnds that managers systematically select accounting procedures
and/or make accrual choices in an attempt to inﬂuence short-term stock performance
prior to management buyouts (MBO) (DeAngelo, 1986; Perry and Williams, 1994),
equity offerings (Teoh, Welch, and Wong, 1998a, b; Tech, Wong, and Rao, 1998;
Shivakumar, 2000), and in acquiring ﬁrms prior to stock-ﬁnanced acquisitions (Erickson
and Wang, 1999; Louis, 2004).

Similarly, the managers of target ﬁrms may be motivated to overstate reported
earnings prior to takeover to increase their negotiating power, and consequently, the
takeover price. However, the evidence on income-increasing earnings management by
target ﬁrms has not been convincingly documented. Easterwood (1997/ 1998) ﬁnds
higher discretionary accruals for targets of hostile tender offers than a control group of
non-targets during the quarter immediately preceding the initiation of takeover attempts,
but not for targets of friendly tender offers. Easterwood argues that only managers faced
with hostile takeovers choose income-increasing discretionary accruals as a takeover
defense. Erickson and Wang (1999) investigate the accrual choices of 55 target ﬁrms
involved in stock-for-stock mergers and ﬁnd that the target ﬁrms’ discretionary accruals

during pre-merger periods are positive but not signiﬁcantly different from zero.

Furthermore, Eddey and Taylor (1999) Show that for 43 Australian Stock Exchange
target ﬁrms, there is no systematic evidence of income-increasing earnings management
during the takeover. The mixed ﬁndings of prior research suggest that not all target ﬁrms
may have incentives to manage earnings upward.4

One obstacle to opportunistic earnings management by targets is that they are
subject to acquirers’ scrutiny during due diligence. Managers of acquiring ﬁrms have
strong incentives to ensure that the target ﬁrms’ ﬁnancial statements are free of material
ﬁnancial reporting management. Acquiring ﬁrms’ managers and boards of directors are
subject to shareholder litigation if they do not perform their ﬁduciary duties on behalf of
acquiring ﬁrms’ shareholders.5 In the takeover market, it is typical that acquiring ﬁrms
hire an investment banker and other advisers to evaluate the potential targets’ ﬁnancial
statements before sending letters of intent. The cost of detection of any earnings
management or fraud during due diligence could be signiﬁcant to the target ﬁrms because
they could lose their credibility in the market for corporate control, resulting in a lower
takeover price or even broken merger talks.6

In the setting I have chosen, the cost of such detection is likely to be prohibitively
high. Soliciting targets deem the sale of the whole company as an effective way to

address depressed ﬁrm value or ﬁnancing constraints that they face (more details to

 

4 There is also mixed evidence in studies focusing on a single industry. For example, Chamberlain (1991)
ﬁnds “no anticipatory earnings management prior to the merger” in 192 bank mergers, whereas Bettinghaus
(2001) documents negative unexpected loan loss provision for 641 bank/year observations in a period
surrounding the merger.
5 On April 15, I998, Cendant Corp., which was created by the merger of CUC lntemational Inc. and HFS
Inc. in December 1997, restated its 1997 earnings downward by more than $100 million due to improper
and fraudulent accounting in the former CUC business. Investors who purchased Cendant common stocks
ﬁled lawsuits against Cendant officers and directors, including former CEOs, Chairmen, CFOS and
directors of CUC and HFS]. The defendants reached a settlement of $2.8 billion with the class action
plaintiffs on December 7, 1999.

Tyson Foods broke its merger talks with [HP once Tyson Foods learned of the accounting problems at
IBP. In addition, Tyson Foods sued lBP for economic damages caused by the termination of the merger
agreement. As a result, lBP’s stock price fell sharply (Dow Jones Newswires, March 29, 2001).

follow in Chapter 3). To the extent that the takeover market readily sees through the
aggressive accounting techniques, and that the detection of opportunistic earnings
management will adversely impact the takeover negotiations, soliciting targets have little
incentive to overstate earnings even though they have sufﬁcient time to do so. Instead,
the soliciting targets are likely to take discretionary earnings charges prior to their
solicitation announcements. DeAngelo et a1. (1994) ﬁnd that ﬁnancially troubled ﬁrms
exhibit large negative accruals around violation of debt covenants. They argue that
managers of these ﬁrms make negative accrual choices to signal to the lenders that they
are willing to acknowledge and deal with their ﬁrms’ problems. Likewise, to increase the
conﬁdence of interested bidders, the soliciting targets have an incentive to take
discretionary write-offs to signal to the takeover market that they are credible and worthy
targets.

Anecdotal evidence shows that, just like preparing a house for sale, soliciting
targets make efforts to “clean house” to put their ﬁnancial statements in the best light. A
senior vice president of Geostar Corp, for example, announced the intent to “clean
themselves up and search for a new corporate partner” (the Washington Post, February
28, 1991). European American Bancorp wrote off hundreds of millions of bad loans and
added $110.1 million to its loan loss reserve before it began searching for buyers (Crains
New York Business, April 13, 1987). Similarly, Beneﬁcial Corp decided to add reserves
for potential liabilities in preparation for selling the company (Wall Street Journal,
September 2, 1986). Apparently the managers of these companies believed that adopting
conservative accounting practices would make them more favorable acquisition targets.

Furthermore, accounting practitioners recommend that soliciting targets proactively clean

up their ﬁnancial statements before putting them on the block to mitigate unnecessary
frictions during takeover negotiations (Alexander, 2001).

In light of the above reasoning and anecdotal evidence, my ﬁrst hypothesis is as
follows:

HYPOTHESIS 1. Compared to unsolicited targets, soliciting targets report more

negative (less positive) discretionary accruals before takeover.

In this study I use unsolicited targets as a benchmark against which to compare
soliciting targets’ accrual choices. When compared to the soliciting targets, the
unsolicited targets are more likely to receive the takeover offers as a surprise. This
control group serves to reduce the likelihood that the accrual choices of soliciting targets

are attributable to incentives unrelated to the takeover.

2.2. Income smoothing behavior of soliciting targets

This dissertation also investigates the soliciting target managers’ income
smoothing behavior. Prior literature argues that managers make accounting choices to
reduce ﬂuctuations in reported earnings around some target level (Ronen and Sadan,
1981). Speciﬁcally, managers manage discretionary accruals or a particular component
of earnings to bring the reported earnings closer to the earnings target. Therefore, an
inverse relation between the smoothing component of earnings and the pre-managed
earnings is interpreted as evidence of income smoothing.

This inverse relation has been found in studies that use discretionary accruals as a
proxy for the smoothing component. For example, Gaver et a1. (1995), based on a

sample of 837 ﬁrm-year observations between 1980-1990, show that managers manage

10

earnings upward when pre~managed earnings fall below the lower bound of the bonus,
and vice versa. In addition, DeFond and Park (1997) predict and ﬁnd consistent results
that income-increasing (-decreasing) discretionary accruals are associated with ﬁrms that
have relatively poor (good) current performance and relatively good (poor) expected
future performance. They argue that managers smooth earnings to reduce the risk of job
dismissal. Other studies document evidence of earnings smoothing via a particular
component of income, including provision for bad debts (McNichols and Wilson, 1988),
income from debt equity swaps (Hand, 1989), write-offs (Francis et al., 1996), special
items (Kinney and Trezevant, 1997), R&D (Perry and Grinakcr, 1994; Bushee, 1998),
restructuring charges (Moehrle, 2002), income from LIFO liquidation (Dhaliwal et al.,
1994; Kinney and Wempe, 2004), value of employee stock options (Matsunaga, 1995),
gain/loss from asset sales (Bartov, 1993; Schrand and Walther, 2000), and realized
gain/loss on investment securities (Moyer, 1990; Scholes, et al., 1990; Beatty et al., 1995;
Collins et al., 1995; Hunton et al., 2004).

The above studies implicitly assume that the ultimate objective of income
smoothing is for the managers to maximize their own welfare in the long run. The
current study, however, provides a setting in which target managers’ incentives
associated with their ﬁrms’ long-term performance are not as strong as those of managers
in general. As an empirical matter, most senior managers of soliciting targets leave
around the sale of the company,7 and usually they receive a lump sum payment when
terminating their employment. Thus, long-term compensation and/or job security provide

lower incentives, and consequently, maintaining reported earnings around some pre-

 

7 I randomly collected management turnover data for 53 soliciting targets and found CEOS of 32 (60%)
ﬁrms resigned or retired around the solicitation announcement.

11

determined level are less of a concern to the managers of soliciting targets. In contrast,
given that the managers of unsolicited targets are unlikely to foresee the takeover bids,
their incentives for income smoothing should not be different from managers of non-
target ﬁrms. This discussion leads to my second hypothesis:
HYPOTHESIS 2. Ceteris paribus, soliciting targets become less prone to income
smoothing than unsolicited targets as approaching the solicitation announcement.
That is, I predict a signiﬁcantly negative association between discretionary
accruals and pre-managed earnings in unsolicited targets but a weaker negative
association in soliciting targets prior to takeover, because the managers of soliciting
targets do not expect to reap much beneﬁt from income smoothing.8 Instead, they could
take conservative measures, such as a balance sheet cleanup, to ease the takeover process.
One could argue that the managers of soliciting targets do not act in the best
interest of target ﬁrms’ shareholders (e.g., to encourage and facilitate takeover) if they
desire to retain control over their companies. However, “golden parachutes”9 reduce the
likelihood that these managers will take measures to fervently resist takeover offers. A
recent news article in the New York Times comments that top managers welcome mergers
because they beneﬁt from the large payouts at the completion of mergers (Morgenson,
2004). The chief executive of SouthTrust, for example, was expected to receive $59

million termination awards and an additional annual pension of about $3.8 million. Such

 

8 I do not entirely rule out the possibility of income smoothing in soliciting targets. Barnea et a1. (1975),
for example, suggest that smoothed earnings enhance investors’ ability to predict future cash ﬂows. If this
argument holds true, the managers of soliciting targets may have incentives to smooth earnings in order to
help the acquirer achieve a better estimate of their ﬁrm’s fair value. However, I believe such incentive only
has a second-order effect on managers’ income smoothing behavior in soliciting targets, because being
honest is more highly valued by the acquirer.

9 A clause in an executive's employment contract specifying that he/she will receive large beneﬁts in the
event that the company is acquired and the executive's employment is terminated. These beneﬁts can take
the form of severance pay, a bonus, stock options, or a combination thereof.

12

pecuniary beneﬁts may outweigh the beneﬁts received by managers of target ﬁrms by
continuing their careers, and thereby, motivate them to complete the takeover. For now, I
hold the assumption that, in general, the managers of soliciting targets do not oppose
takeover offers. However, entrenched managers may indulge their preference over non-
value-maximizing behavior (via their substantial equity ownership) and deliberately
inﬂate earnings to frustrate takeover attempts. I will discuss the impact of management

ownership on managers’ incentives to manage earnings in Chapter 6.

l3

CHAPTER 3: SAMPLE AND DESCRIPTIVE STATISTICS

3.1. Sample selection and description

In order to identify a sample of takeover targets who solicit takeover offers, I
searched the SDC database for public U.S. targets whose deal status is “Seeking Buyer”
or whose initial reception to a deal is “Solicited”.IO The sample selection procedure is
summarized in Table 1. The initial sample of soliciting targets is based on 853
announcements made between 1990 and 2003 by target ﬁrms seeking buyers. I eliminate
announcements made by ﬁrms without relevant Compustat ﬁnancial data (191), without
relevant CRSP stock return data (10), and those who are banks and ﬁnancial institutions
(137, SIC code 6000-6700). I delete banks and ﬁnancial institutions because existing
models of discretionary accruals are not applicable for these ﬁrms, and ﬁnancial
institutions’ ability to manage earnings might differ from that of industrial ﬁrms due to
greater regulation. To ensure that the soliciting targets were public ﬁrms when they
sought buyers, I eliminate the announcements made by ﬁrms prior to their listing in a
stock exchange (3) or aﬁer their delisting (188) from a stock exchange. For ﬁrms making
two or more solicitation announcements during 1990-2003, I focus on the initial
announcement and eliminate 16 subsequent announcements.” For the remaining 308
soliciting targets, I reviewed news articles on F activa News Service for their initial

announcement seeking buyers. A further 32 observations are deleted reﬂecting cases

 

'0 A SDC support person conﬁrmed that SDC codes all ﬁrms publicly announcing plans to seek out buyers
as “seeking buyer”. If the ﬁrm receives one or more bids following the announcement, SDC re-codes the
ﬁrm as “solicited” targets and updates its deal status as “Completed”, “Withdrawn” or “Pending”.

” 12 targets made two public announcements and 2 targets made three.

14

where the deal is a rumor (8), the deal involves only a partial sale of the company (10), or
the targets operate under the protection of Chapter 11 (14). '2

This study focuses on a sample of ﬁrms who put themselves on the block.
Therefore, for inclusion in the ﬁnal sample, the solicitation announcements must be made
by the board of directors or the ﬁrm management. However, the SDC also codes
negotiations initiated by other parties as “Seeking Buyer”. To remove this noise, I
searched F activa and eliminated cases where the sale negotiation was initiated by major
stakeholder(s) (25) or where the objective of the sale was to seek a “white knight” (1 1).”
'4 After removing these observations, the ﬁnal sample consists of 240 soliciting targets.

Table 2 lists the reasons disclosed in the news articles for the solicitations. It
indicates that the most frequently cited motivation to sell is to enhance ﬁrm value; a large
portion of the soliciting targets claimed that the purpose of the sale is to maximize
stockholders’ value (69 ﬁrms) or to cash out (8 ﬁrms), or believed that they were
undervalued by the market (15 ﬁrms). Persistent operating or ﬁnancing difﬁculties seem
to encourage ﬁrms to seek buyers. Twenty-two soliciting targets wanted to go out of
business because they were disappointed by their performance. In 24 cases ﬁrms were
selling the whole company because they were experiencing severe ﬁnancing problems,
and an additional 13 ﬁrms needed a buyer to provide additional capital input. An
additional reason for solicitation was to enhance the targets’ business opportunities.

Thirteen soliciting targets sought a merger partner in response to the trend of

 

'2 I suspect that ﬁrms operating under the Chapter 11 have different incentives (due to different monitoring
environment) when making accrual accounting choices. Therefore, including these targets would add noise
to my analyses.

'3 For example, Plains Petroleum Co. authorized its ﬁnancial adviser to study strategic alternatives,
including a sale of the company, to fend off an unwanted merger proposal initiated by Cross Timbers Oil
(“Plains Pete Authorizes Advisers To Seek Alternatives”, Dow Jones New Service, October 20, 1994).

'4 For 18 ﬁrms, 1 cannot ﬁnd news articles related to their solicitation announcements. However, deleting
these ﬁrms does not affect my conclusion.

15

consolidation in their industries. Apparently these ﬁrms believed that forming a larger
company through merger would improve their competitiveness. Moreover, 12 ﬁrms
looked for an acquirer to support strategic growth. Finally, for 29 ﬁrms no reason was
cited in their news releases, while a further 15 ﬁrms claimed that they have identiﬁed
potentially interested buyers.

Table 2 also reports the average cumulative abnormal returns (CAR) over a 3-day
window surrounding the initial solicitation announcement. I obtain stock return data
from CRSP. Daily returns for 210 days (covering period from day —255 through day —
46) are used to estimate market model parameters for each soliciting target’s CAR. I
require a minimum of 60 daily returns. On average, soliciting targets earn a signiﬁcant
8.02% abnormal return (t statistic = 5.68) for the announcement period [-1, +1]. A closer
inspection of their CAR by categories reveals that only ﬁnancially distressed ﬁrms suffer
from a signiﬁcantly negative return of 17.44%. Firms in all other categories earn positive
abnormal returns, suggesting that the solicitation announcement, like other takeover
announcements, conveys good news about targets.

To obtain a sample of unsolicited targets, I searched the SDC for takeover targets,
where the takeovers were coded as either “Hostile” or “Unsolicited”. The sample
selection procedure is similar to that of soliciting targets (see Table 1). I obtain 528
hostile or unsolicited takeover offers received by public U.S. targets between 1990 and
2003. To be included in the control sample, the surprised targets must have relevant

Compustat and CRSP data, must not be a ﬁnancial institution, and must be a public ﬁrm

l6

when receiving the bid. I further eliminate multiple bids, bankrupt ﬁrms and rumor
deals. This screening procedure yields 277 unsolicited targets.IS

I deﬁne the initial takeover announcement day (day 0) as the date when the ﬁrst
public announcement of the acquisition intent is made by either the target or the bidder
(for soliciting targets, it is the date when they publicly announce to seek buyers; for
unsolicited targets, it is the date when the bidders ﬁrst publicly announce their takeover
intent). This date is obtained from the SDC and veriﬁed by reviewing news articles from
Factiva. The distribution of initial announcements by soliciting and unsolicited targets
over calendar years is summarized in Table 3 panel A. I compare the calendar year
distributions of the initial announcement in both the initial sample (853 soliciting targets
and 528 unsolicited targets) and the ﬁnal sample (240 soliciting targets and 277
unsolicited targets) to examine if my screening rules result in any calendar time selection
bias. However, neither soliciting targets nor unsolicited targets show heavy clustering in
a speciﬁc year. In addition, a similar pattern emerges in the ﬁnal sample, indicating that
the sample truncation does not materially alter the calendar year distribution of the initial
takeover announcement.

I also examine the 2-digit SIC industry classiﬁcation of soliciting and unsolicited
targets and ﬁnd that both groups of target ﬁrms appear to cluster in high tech industries.
Table 3 Panel B shows that the most prevalent 2-digit SIC codes in my sample are
chemical and allied products (SIC code 28), computer equipment (SIC code 35),

electrical equipment (SIC code 36), measuring instruments, photography and watches

 

‘5 I do not apply other screening criteria to unsolicited targets given that the target ﬁrms are unaware of the
takeover offers until the bidders announce their merger and acquisition plans. It is unlikely that the offers
involve a liquidation, spin-off or MBO. Moreover, by deﬁnition unsolicited takeover offers are always
initiated by the bidder, not by the target ﬁrm or any third parties (e.g., shareholders).

17

(SIC code 38), and business services (SIC code 73). Moreover, I ﬁnd that soliciting
targets show signiﬁcantly more clustering than unsolicited targets in industries with SIC
codes 13, 27, 48, 57 and 78, whereas unsolicited targets are more clustered in industries
with SIC codes 20, 33, 37 and 73. To control for potential industry effects, I subtract the
2-digit SIC industry medians from each observation for each variable and focus on the

industry-adjusted variables in all my analyses.

3.2. Descriptive statistics

Table 4 presents descriptive statistics on soliciting and unsolicited targets in the
year prior to the initial takeover announcement. I compare the two sets of targets on the
following dimensions: ﬁrm performance, total accruals, size, ﬁnancial leverage, growth
opportunities, age, management ownership and institutional holdings. I focus on these
variables because they likely have an impact on a ﬁrm’s discretionary accrual choices
(Warﬁeld et al., 1995; Becker et al., 1998; Raj gopal et al., 1999; Beneish, 1999; Raj gopal
et al., 1999; McNichols, 2000). All accounting data are obtained from Compustat and
data on stock returns from CRSP. Except for management and institutional ownership,
all variables are 2-digit SIC industry adjusted.'6 The number of observations for each
variable varies with the data availability.

Table 4 panel A shows that unsolicited targets are substantially larger than
soliciting targets in terms of market value of equity (MVE). The mean market value of
equity for unsolicited targets is more than six times as large as that of soliciting targets.

This result is consistent with soliciting targets’ desire to increase their ﬁrm size through a

 

'6 I also carried out an analysis using unadjusted data. The inferences are virtually identical to those based
on the industry-adjusted data.

18

merger to improve their competitiveness. However, LEVERAGE, measured as long-
term debt over the sum of market value of equity and long-term debt, is not statistically
different across the two groups. This suggests that ﬁnancial leverage is not a major
determinant of target ﬁrms’ decision to solicit buyers.

I compare soliciting and unsolicited targets using three performance measures:
accounting earnings (EARNINGS), measured as net income before extraordinary items,
operating cash ﬂows (CASHFLOW), and market-adjusted buy-and-hold abnormal
returns (BHAR). Both earnings and operating cash ﬂows are scaled by total assets at the
beginning of the year. It appears that soliciting targets are less proﬁtable than unsolicited
targets. The mean (median) earnings are ~16.6% (-1.8%) of total assets for soliciting
targets compared to -2% (0.2%) for unsolicited targets. The t and Wilcoxon signed
ranked tests indicate that the differences are statistically signiﬁcant. The tests on mean
and median CASHF LOW indicate that cash ﬂow from operations tends to be lower for
soliciting targets than unsolicited targets. Moreover, although the mean buy-and-hold
abnormal return earned by soliciting targets during the year prior to the initial takeover
announcement is not distinguishable from that earned by unsolicited targets (p = 0.17),
the median value is signiﬁcantly lower for soliciting targets (p = 0.00). Overall, these
results suggest that soliciting targets underperform unsolicited targets preceding their
solicitation announcement. I 7

Prior literature predicts that managers of poorly-performing ﬁrms have incentives

to overstate earnings through their accrual accounting choices. Nevertheless, I ﬁnd that

 

‘7 I also compare the performance of soliciting targets to that of unsolicited targets over years —4 to —2. The
untabulated results indicate that in year —2 soliciting targets signiﬁcantly underperform unsolicited targets
in terms of earnings and market-adj usted abnormal returns. However, for years —4 and —3, none of the
performance measures is statistically different across the two groups.

19

soliciting targets have signiﬁcantly lower mean (median) total accruals than unsolicited
targets at the two-tailed level of 0.05. '8 There are two interpretations of this result. The
ﬁrst is that managers of soliciting targets make economic choices in response to
deteriorating performance; that is, the negative total accruals of soliciting targets are a
manifestation of poor performance. The second is that the managers of soliciting targets
deliberately make income-decreasing accounting choices to signal that their ﬁrms are
credible targets. I further explore the accrual choices of soliciting targets in Chapter 5.

Soliciting targets have signiﬁcantly lower book-to-market ratio (BTM) and larger
Tobin’s Q (TOBIN’S Q), a measure of investment opportunities that is widely accepted
in the ﬁnancial economics literature,19 indicating that soliciting targets have more growth
potential than unsolicited targets. Furthermore, I compare the age of soliciting and
unsolicited targets, where age is deﬁned to be the time lag (in months) from the ﬁrm’s
earliest listing date (BEGDAT in the CRSP Daily Header ﬁle) to its initial announcement
day. The industry-adjusted mean (median) age of soliciting targets is 1.2 (-23.5) months,
while it is 42.1 (-13.3) for unsolicited targets.20 The difference is signiﬁcant at the two-
tailed 0.05 level, indicating that soliciting targets are relatively younger than unsolicited
targets. Recall that the soliciting targets indicated that ﬁnancial distress, capital needs
and greater risk of failure were important reasons for seeking acquirers. The high

concentration of younger ﬁrms that solicit takeover offers is consistent with the argument

 

'8 Total accrual is deﬁned as the change in working capital plus the change in short-term debt minus sum of
the change in cash and depreciation.

'9 Following Song and Walkling’s (2000, p. 163) approach, Tobin’s Q is calculated as:

Tobin’s Q = (Market Value of Common Stock + Book Value of Preferred Stock + Book Value of Value of
Long-term Debt + Current Liabilities — Current Assets)/Book value of Total Assets.

20 The age of 2-digit industry ﬁrms equals the time lag (in months) from the earliest listing date to the initial
announcement day of corresponding target ﬁrms.

20

made in prior research that the ﬁnancial positions and capital needs of younger ﬁrms
create pressure on their managers to maintain earnings growth (Beneish, 1999).

I jointly test the differences across soliciting and unsolicited targets in a
multivariate setting. The probit regression results presented in Table 4 Panel B indicate
that only coefﬁcients on ﬁrm size and book-to-market ratio remain signiﬁcant and in the
same direction as the univariate test results. Contrary to the univariate test, when other
variables are controlled, I ﬁnd a positive coefﬁcient on EARNINGS but it is not
signiﬁcant at conventional levels (p = 0.16), implying that the accounting proﬁtability is
indistinguishable between soliciting and unsolicited targets. Overall, the evidence
suggests that soliciting targets tend to be smaller ﬁrms with higher growth potential
compared to unsolicited targets.

In Panel C of Table 4 I compare the ownership structure between soliciting and
unsolicited targets. Management ownership, MGTOWN is the percentage of shares held
by ofﬁcers and directors as a group.21 This measure is collected fromIO-K ﬁlings and
proxy statements. However, I am unable to identify management ownership information
from 10-K or proxy statements for 77 (32%) soliciting targets and 42 (15%) unsolicited
targets. I ﬁnd that the average (median) management of soliciting targets owns 28.1%
(21 .8%) of their ﬁrms’ stock, compared to 16.0% (10.4%) management holdings in
unsolicited targets. The difference is statistically signiﬁcant at the 0.01 level, suggesting
greater management control in soliciting targets. INSTOWN measures the percentage of
shares held by money managers. Institutional holdings data are collected from the

Spectrum database, which consists of all 13-f ﬁlings between 1990 and 2003 ﬁled by

 

2' Although this measure includes holdings by both ofﬁcers and directors, director holdings account for
only a small portion of MGTOWN. Therefore, I expect that this measure largely reﬂects management’s
incentives rather than directors’.

21

institutions with greater than $100 million in equity securities. I lose 25 observations
(10%) for soliciting targets and 23 (8%) for unsolicited targets due to missing data on
institutional ownership from Spectrum. Panel C of Table 4 shows that the difference in
institutional ownership across soliciting and unsolicited targets is not signiﬁcant at the

0.05 level.

22

CHAPTER 4: MEASURING DISCRETIONARY ACCRUALS

This dissertation considers two measures of discretionary accruals. The ﬁrst
measure is based on the modiﬁed version of the accruals expectation model proposed by
Jones (1991 ). For each year and each two-digit SIC industry, I estimate discretionary
accruals from the following model (the corresponding Compustat data items in

parentheses):22

ACCRUALS“, 1 PPE. AADJREV

: + + 7 ""L + '1' +8
ASSETW [30” ﬂ'” ASSETUH a, ASSETUH ’63-” ASSET. '1'

Ijt -I

 

 

where:

i = ﬁrm index.

j = two-digit SIC industry index.

I = year (and quarter) index.

ACCRUALS“, total accruals calculated as ACA (annual #4, quarterly #40) - ACL
= (annual #5, quarterly #49) - 11th (annual #1, quarterly #36) + ASTD
(annual #34, quarterly #45) — Depreciation (annual #14, quarterly #5).
ASSETijt-l = total assets (annual #6, quarterly #44).

PPEijt = gross property, plant and equipment (annual #8, quarterly #42).

AADJREVU: = change in cash-based revenues equals change in sales (annual #12,
quarterly #2) minus change in receivables (annual #2, quarterly #37)

8th = error term.

 

22 Consistent with Kothari et al. (2004), I include a constant [30,. in the estimation model. Dropping this
constant makes my discretionary accrual estimates even more negative.

23

Following Jones (1991), the discretionary accruals (DA) are deﬁned as the error term

from the above regression.

The second measure of discretionary accruals is the performance-adj usted
discretionary accruals proposed by Kothari et a1. (2004). Kothari et al. Show that the
existing accrual expectation models are biased toward detecting earnings management
when managerial incentives are associated with performance. This is particularly true for
takeover targets because targets generally perform poorly before takeovers (DeAngelo,
1988; Christie and Zimmerman, 1994). To adjust the discretionary accruals for ﬁrm
perfoMance, 1 construct ten portfolios for each industry and ﬁscal year (ﬁscal quarter) by
sorting data into deciles based on lagged return on assets, ROA, which is calculated as
earnings before extraordinary items over total assets at the beginning of the year
(quarter). The performance-adjusted discretionary accruals for a given ﬁrm are the
discretionary accruals from the modiﬁed Jones model minus the median discretionary

accruals of its corresponding performance—matched portfolio (excluding the sample ﬁrm).

24

CHAPTER 5: ANALYSIS RESULTS

5.1. Accrual analysis

I hypothesize that soliciting targets “clean up” their balance sheets before seeking
buyers in attempt to encourage and/or facilitate acquirers to bid for their companies. To
test this hypothesis, I examine the discretionary accruals of soliciting targets prior to the
initial solicitation announcement. An inherent limitation is that it is difﬁcult to precisely
pinpoint the time when such earnings management will begin to occur. Deﬁning year 0
as the year when either the soliciting targets make the initial announcement to seek
buyers or the bidders of unsolicited targets make the initial announcement of takeover
intent, I examine the time-series of discretionary accruals from year 0 back to year —4. I
believe this time-series is sufﬁcient to capture the effect of any takeover-motivated
attempts to clean up balance sheets.23

Table 5 panel A reports mean and median discretionary accruals for soliciting and
unsolicited targets along with the signiﬁcance levels for tests of differences across the
two groups.24 An analysis on the quarterly discretionary accruals over quarters [-4, 0]
relative to the initial announcement quarter (Qtr 0) is also provided. I ﬁnd that neither
soliciting targets nor unsolicited targets manage earnings upward via their accrual choices
in any year prior to and in the year of the initial takeover announcement. This ﬁnding is
inconsistent with the argument that target ﬁrms inﬂate reported earnings via accrual

choices to increase takeover price. For unsolicited targets, the insigniﬁcant result could

 

23 DeAngelo et al. (1994) point out that a shortcoming of this approach is that it may fail to detect earnings
management because different subsets of ﬁrms manage income in different years.

2“ Not all ﬁrms have data in each of the ﬁve years because of newly entering ﬁrms, exiting ﬁrms, and ﬁrms
with missing data from the Compustat. Limiting my sample to ﬁrms with complete reporting history of
ﬁve years would greatly reduce my sample size and result in a survivorship bias.

25

be due to the fact that these ﬁrms do not have sufﬁcient time to manage earnings upward
before the initial takeover announcement. For soliciting targets, the result suggests that it
is prohibitively costly for these ﬁrms to use positive discretionary accruals to
opportunistically manage earnings prior to solicitation. Apparently soliciting targets do
not want to risk losing takeover offers at a critical time.

Consistent with my ﬁrst hypothesis, the mean and median discretionary accruals
are signiﬁcantly negative for soliciting targets in year —1. The mean (median)
discretionary accruals account for —l .3 5% (—0.80%) of their total assets. A comparison
to unsolicited targets reveals that soliciting targets have signiﬁcantly more negative
discretionary accruals. Given that the discretionary accruals of soliciting targets are not
signiﬁcantly different from zero in year —4 though year —2, the negative discretionary
accruals in year —1 is likely to be conditional on their solicitation decision instead of a
manifestation of their poor performance.25 Moreover, the negative discretionary accruals
of soliciting targets do not reverse in the following year. Overall, these ﬁndings imply
that soliciting targets take discretionary accounting charges to credibly signal to potential
bidders their desire to consummate a sale. 26

The analysis of quarterly data shows that soliciting targets have negative
discretionary accruals in quarters —4, —2 and —1. Thus, soliciting targets proactively

clean up their balance Sheets (at least) four quarters prior to their initial announcement to

 

25 This pattern of discretionary accruals also suggests that soliciting targets time their solicitation
announcement.

26 I also examine speciﬁc accruals including change in sales revenue, account receivables, inventory,
payables, short-term debt, net property, plant and equipment, and special items in year —1. However, a
comparison between soliciting and unsolicited targets provides no evidence that soliciting targets have used
any of these speciﬁc accruals to manage earnings downward. It might be the case that different soliciting
targets make varying speciﬁc accrual choices to achieve balance sheet cleanup.

26

publicly solicit takeover offers. To simplify my following analyses, I focus on annual
discretionary accruals in year —1.

Although I lose a few observations each year due to performance matching, the
analysis on performance-matched discretionary accruals reported in Table 5 panel B
presents a similar picture. This suggests that the results in Panel A are not due to any
performance differences between soliciting and unsolicited targets. Given that the
correlation coefﬁcient between the annual discretionary accruals and performance-
matched discretionary accruals in my sample is 0.95, I focus on the discretionary accruals
estimated from the modiﬁed Jones model in the following analyses. However, using
performance-matched discretionary accruals in these tests generate similar results.

AS a supplemental test, I investigate if soliciting targets are less likely to have
anti-takeover devices than unsolicited targets. Malatesta and Walkling (1988) suggest
that a poison pill is an important measure employed by managers to fend off takeover
attempts. Using data from the SDC, I compare three types of poison pill defense: back-
end plans, ﬂip-over plans and voting plans.27 I ﬁnd that none of the soliciting targets had
any poison pill defenses in place, while 35 (12.6%) unsolicited targets had at least one of
the three poison pill defenses. The difference in the proportions is statistically signiﬁcant
(t statistics = 6.32, p < 0.001), which lends support to my assumption that soliciting
targets do not actively oppose takeover offers. Instead, they take measures (such as

balance sheet cleanup) to invite interested buyers.

5.2. Income smoothing analysis

 

27 Malatesta and Walkling (1988) discuss four principle types of poison pill defenses. However, data on the
fourth type, preferred stock plans, are not available from the SDC. The deﬁnitions of the poison pill
defenses can be found in Malatesta and Walkling (1988, p. 351-355).

27

This chapter examines if the managers of soliciting targets have a lower
propensity to smooth earnings than the managers of unsolicited targets. To identify
managers’ income smoothing behavior, I follow the approach of Gaver et a1. (1995) and
DeFond and Park (1997), which deﬁnes current pre-managed earnings as current period
earnings minus discretionary accruals. Using a pooled cross-sectional analysis, both
studies ﬁnd that, when the current pre-managed performance is relatively poor (good),
the managers have incentive to increase (decrease) earnings through positive (negative)
discretionary accruals in order to maintain a bonus or to reduce the risk of job dismissal.
Therefore, a negative association between the pre-managed earnings and discretionary
accruals is consistent with managers’ income smoothing behavior.

To test if income smoothing is less prevalent in soliciting targets, I estimate the

following model for each year from year —4 to 0:

DA, = ,60 + ,BlPREMGT, + ﬂzSOLICIT, + ,63PREMGT, * SOLICIT, + ﬂﬂ-IIGHQ,
+ ,BSHIGHQ, * PREMGT, + ,BbSIZE, + [37 LE VERAGE, + 8,

where:

PREMGT = Current earnings before extraordinary items divided by lagged
total assets (subtracting 2-digit industry median) minus
discretionary accruals.

SOLICIT = 1 if soliciting targets and 0 if unsolicited targets.
HIGHQ = 1 if industry-adjusted Tobin’s Q is larger than 1, 0 otherwise.

SIZE = Industry-adjusted natural logarithm of total assets.

LEVERAGE = Long-term debt/(Market value of equity + Long-term debt),
subtracting 2-digit industry median.

28

The coefﬁcient on PREMGT is a measure of income smoothing by unsolicited targets.28
Consistent with prior literature, I predict this coefﬁcient to be negative because the
managers of unsolicited targets, by deﬁnition, should not behave differently from the
managers of non-target ﬁrms. The interaction PREMGT*SOLICIT measures the
differential income smoothing effects between soliciting and unsolicited targets. A
positive coefﬁcient on the interaction would be consistent with less income smoothing by
the managers of soliciting targets. Moreover, the coefﬁcient on the interaction, combined
with the coefﬁcient on PREMGT, captures the income smoothing behavior by soliciting
targets. According to my second hypothesis, I expect this combined coefﬁcient to
become less negative as soliciting targets approach their solicitation announcement.
Three control variables are included in the regression. F irm size (SIZE) and
ﬁnancial leverage (LEVERAGE) are included because they have been found to be
associated with ﬁrms’ discretionary accruals (Becker et al., 1997; DeFond and Park,
1997). McNichols (2000) ﬁnds that long-term earnings grth is important in explaining
earnings management behavior, so I add a variable HIGHQ to proxy for a ﬁrm’s long-
term earnings growth. Speciﬁcally, I deﬁne a ﬁrm to be HIGHQ ﬁrm if its (industry-
adjusted) Tobin’s Q ratio is more than 1. In addition, the managers’ incentive to smooth
earnings in the current period may be dampened in high growth ﬁrms. For example, if
the stock market imposes substantial costs on high growth ﬁrms when they decelerate,

these ﬁrms will adopt accounting techniques to maintain the high growth rate rather than

 

23 This research design implicitly assumes 2-digit industry median earnings to be the earnings target of
sample ﬁrms. I also tested if soliciting targets have weaker incentives to smooth earnings when pre-
managed earnings deviate from analysts’ predictions compared to unsolicited targets. Although I lose more
than 50% of the observations due to lack of analyst forecasts on l/B/E/S, the results are qualitatively similar
to those reported in Table 6.

29

smooth earnings.29 Therefore, to control for the possibility that high growth ﬁrms are
less likely to smooth earnings, 1 add an interaction PREMGT* HIGHQ in the regression.30

To analyze the differential trend in the income smoothing behavior of soliciting
and unsolicited targets over time, I estimate the regression model for each year from year
—4 to 0. Strictly speaking, the negative correlation between discretionary accruals and
pre-managed earnings in the annual cross-sectional regression does not necessarily
demonstrate a smoothed time-series trend in the reported earnings of sample ﬁrms.
Instead, it indicates that managers select income-increasing (-decreasing) discretionary
accruals when pre-managed earnings are exceptionally poor (good) in that year.
However, consistent negative correlations between discretionary accruals and pre-
managed earnings across years would indicate that managers make accrual choices to
achieve income-smoothing objectives. In a similar vein, a less negative correlation over
time would be supportive of managers’ declining incentives to smooth earnings.

The regression results are reported in Table 6 Panel A. The intercept, which
measures the discretionary accruals of unsolicited targets when PREMGT=0, is
insigniﬁcant across all years except in year 0. Recall that PREMGT equals targets ﬁrms’
current pre-managed earnings minus 2-digit SIC industry median. The insigniﬁcant
result suggests that when unsolicited targets have the same level of pre-managed earnings

as their industry peers, their discretionary accruals are not different from zero. However,

the signiﬁcantly negative intercept in year 0 suggests that the surviving unsolicited

 

29 Fridson (1991, p. 7-8) states: “almost invariably, companies try to dispel the impression that their growth
is decelerating, since that perception can be so costly to them.”

30 DeFond and Park (1997) predict and ﬁnd that managers consider both current performance and expected
future performance when making accrual accounting choices. In addition, the inverse relation between
discretionary accruals and current pre-managed earnings is conditional on the expected future performance.
Tobin’s Q measures a ﬁrm’s market power from existing assets and future investment opportunities. Due
to its forward-looking nature, Tobin’s Q could be considered a proxy for future ﬁrm performance.

30

targets may also pursue income-decreasing accounting practices to hasten the takeover
process.3 I The coefﬁcient on SOLICIT measures the difference in discretionary accruals
between soliciting and unsolicited targets when PREMGT=0. Consistent with the
univariate result in Table 5, the coefﬁcient on SOLICIT is signiﬁcantly negative in year —
1, indicating that the soliciting targets report more negative discretionary accruals than
the unsolicited targets. Nevertheless, this coefﬁcient is not different from 0 in all other
years. The sum of the coefﬁcient on SOLICIT and the intercept reﬂects the discretionary
accrual choices of soliciting targets when other variables are controlled for. It is negative
in all years but Signiﬁcant only in year —1 (-0.022, p value=0.002). Thus, the univariate
test results reported in Table 5 continue to hold in the multivariate setting.

The coefﬁcient on PREMGT is signiﬁcantly negative across all years, with its
magnitude ranging from —0.40 to —0.26. This result indicates that the managers of
unsolicited targets use discretionary accrual choices to smooth reported earnings.
However, the signiﬁcantly positive coefﬁcient on PREMGT*SOLICIT in year —1 and 0
reﬂects the weaker incentives of the managers of soliciting targets to smooth earnings
around their solicitation decision. To help explain the weakening smoothing behavior of
soliciting targets from year —4 to year 0, I add the coefﬁcient on PREMGT to the
coefﬁcient on PREMGT“ SOLICIT to measure the overall income smoothing effect for
soliciting targets. I report the sum separately in Panel A. It demonstrates that the
magnitude of smoothing by soliciting targets drops sharply from a statistically Signiﬁcant
0.477 in year —4 to an insigniﬁcant 0.030 in the year of the solicitation announcement,

compared to a slight drop from 0.395 to 0.263 for unsolicited targets. I further test the

 

3 ' The 174 surviving unsolicited targets have insigniﬁcant mean (median) discretionary accruals -0.06%
(0.05%) in year —1 with a two-tailed p value equaling 0.93 (0.83). Therefore, these ﬁrms do not proactively
clean up their balance sheets before the unsolicited takeover announcement.

31

change in the coefﬁcient on PREMGT + PREMGT*SOLICIT across years and present
the results in the last row of Panel A. Except in year —2, the increase in the income
smoothing effect of soliciting targets is statistically signiﬁcant in all years. This result
supports my second hypothesis that the managers of soliciting targets have lower
incentive to smooth earnings due to the desire to ﬁnd an acquirer.

The signiﬁcantly positive coefﬁcient on PREMGT*HIGHQ indicates that high
growth ﬁrms report more positive discretionary accruals to increase earnings given
income smoothing is not their priority. Consistent with my conjecture, ﬁrms with more
growth opportunities opt for higher earnings growth rate rather than a smoother income
series. However, during year 0, high growth ﬁrms do not behave differently from other
ﬁrms in terms of both discretionary accrual choices and earnings smoothing. This result
should be interpreted with caution because it is based on a group of surviving target
ﬁrms. SIZE and LEVERAGE do not have explanatory power in all years except year 0,
indicating that ﬁrm size and debt capacity are not major predictors of the levels of
discretionary accruals in my sample.

In this study, I calculate the pre-managed earnings by backing out the
discretionary accrual component from the reported earnings and interpret its negative
relation to discretionary aceruals as evidence of earnings management. Both Gaver et al.
(1995) and DeFond and Park (1997) caution that backing out discretionary accruals is
likely to create a spuriously negative correlation between pre-managed earnings and
discretionary accruals due to the measurement error in estimating discretionary accruals.
Elgers et al. (2003) and Lim and Lustgarten (2002) demonstrate that the estimation error

in discretionary accruals biases the correlation between pre-managed earnings and

32

discretionary accruals toward accepting the anticipatory income smoothing hypothesis,
even in the absence of any earnings management. Recall that I estimate annual
regression models from year —4 through 0. One possibility is that the bias in the
correlation between pre-managed earnings and discretionary accruals due to the
measurement error in discretionary accruals is of similar magnitude across time and
target type. If that is the case, then my ﬁnding that soliciting targets have Signiﬁcantly
less negative correlation between discretionary accruals and pre-managed earnings than
unsolicited targets in year —1 and 0 compared to other years is consistent with the
managers of soliciting targets becoming less inclined to smooth earnings. For the bias
caused by measurement error to explain my results, it must systematically increase from
year —4 to year 0 for soliciting targets, but remain at about the same level for the
unsolicited targets. However, I am not aware of any reason why the bias should display
such a pattern across time and target type.32 To further address the concern over the
measurement error in discretionary accruals, I use operating cash ﬂows as an alternative
measure of pre-managed earnings in the regression model. Speciﬁcally, to avoid any
mechanical correlation between total accrual measures obtained from the balance sheet
approach and operating cash ﬂows measured as earnings minus total accruals, I obtain

operating cash ﬂows directly from the statement of cash ﬂows (Compustat data item

 

32 Lim and Lustgarten (2002) analytically show that the magnitude of the bias increases with the variance
of income smoothing component (discretionary accruals in the current setting) and decreases with the
variance of smoothing error (the deviation of reported earnings from earnings target; assuming the median
industry RCA to be the smoothing target, smoothing error equals EARNINGS in the current setting).
However, I ﬁnd no trend in the variance of DA and EARNINGS of soliciting and unsolicited targets from
year —4 to 0, so the inter-temporal changes in these variances do not seem to spuriously contribute to the
less negative correlation between discretionary accruals and pre-managed earnings in my soliciting target
sample.

33

308).33 The regression results using operating cash ﬂows reported in Table 6 Panel B are
qualitatively similar to those reported in Panel A. Overall, I believe that my income
smoothing analysis results are not very susceptible to the limitations of the backing-out

technique (see Appendix for additional discussions on the backing out method).

 

33 I acknowledge that operating cash flows and total accruals are negatively related for various economic
reasons (Dechow et al. 1998). To the extent the discretionary accrual model is not well speciﬁed, any
negative correlation between operating cash ﬂows and discretionary accruals might be a reﬂection of the
economic forces faced by the ﬁrms and may not be a reﬂection of income smoothing. However, my focus
is not on the level of negative correlation between operating cash ﬂows and discretionary accruals, but how
this negative relation changes over time as well as how the negative relation is different in soliciting targets
versus unsolicited targets.

34

CHAPTER 6: OWNERSHIP STRUCTURE

Differential equity ownership structure can have an important inﬂuence on the
accounting accrual choices made by managers. In this chapter I examine the possible
effects of management ownership and institutional ownership on discretionary accruals in
soliciting and unsolicited targets.

The separation of ownership and control that characterizes the corporate form
creates potential conﬂicts between managers and investors (Jensen and Meckling, 1976;
Fama, 1980; Fama and Jensen, 1983). Increased management ownership aligns the
interests of shareholders and managers, thereby encouraging managers to adopt value-
maximizing strategies because they bear a higher fraction of the cost of poor decisions
(Jensen and Meckling, 1976). This favorable effect of managerial ownership is called
convergence-of-interest. On the other hand, increased management ownership insulates
managers from internal and external monitoring efforts through their voting power (Denis
et al., 1997). Such managers may pursue private beneﬁts (including shirking and
perquisite-taking) at the expense of outside shareholders. This unfavorable ownership
effect is called managerial entrenchment. As a result, Morck et a1. (1988) and McConnell
and Servaes (1990) argue that the relation between ﬁrm value and management
ownership depends on which of the two effects dominates.

To the extent that managers use their discretion in accrual choices to obtain
private beneﬁts, the two opposing forces of management ownership implies that the
relation between management ownership and managers’ accrual choices is nonlinear.

Prior research (e.g., Yeo e] at., 2002) ﬁnds that this relation varies over different ranges

35

of management ownership. So I assign each soliciting target to one of three portfolios.
Portfolio LOW comprises ﬁrms with management ownership lower than the 25th
percentile (7.0%), while portfolio HIGH comprises ﬁrms with management ownership
high than the 75th percentile (45.6%). The rest are assigned to the MIDDLE portfolio. I
repeat the same procedure for unsolicited targets.

Table 7 panel A reports the mean and median discretionary accruals in year —1 for
each of the three portfolios. For soliciting targets, the mean and median discretionary
accruals for the LOW portfolio are not different from zero, suggesting that the managers
of soliciting targets lack the incentive to expedite a sale given their immaterial equity
ownership. The mean (median) discretionary accruals for the MIDDLE portfolio is -
3.62% (-3.55%) relative to total assets at the beginning of the year and signiﬁcant at the
level of 0.01. Given that the sale of the company improves shareholder wealth, this result
suggests that managers in the MIDDLE portfolio have interests aligned with Shareholders
so that they use income-decreasing discretionary accruals to encourage and/or facilitate
the takeover. Managers in the HIGH portfolio inﬂate earnings by recording income-
increasing discretionary accruals. The signiﬁcantly positive 4.36% (1.87%) discretionary
accruals are higher than those of LOW or MIDDLE portfolios. This result is consistent
with managerial entrenchment; managers with substantial fraction of the ﬁrm’s equity
may want to extract private beneﬁts from continuing their control over the ﬁrm. As a

comparison, unsolicited targets have insigniﬁcant discretionary accruals in all three

36

portfolios, suggesting that the above results are contingent on the ﬁrms’ solicitation
decision.34

Institutional ownership is another key component of corporate governance.
Previous studies argue that institutional investors have incentives to actively monitor and
discipline management because they have greater ﬁnancial interest and their larger
stockholdings allow them to take corrective actions more easily (Shleifer and Vishny,
1986, 1997; Bushee, 1998; Ajinkya et al., 2003; Raj gopal et al., 1999). These studies
ﬁnd that high institutional ownership leads to efﬁcient monitoring of management and
less managerial opportunism.

I examine the association between institutional ownership and discretionary
accruals by assigning soliciting and unsolicited targets to LOW, MIDDLE, and HIGH
portfolios according to their 25th and 75‘h percentile of money manager holdings. Table 7
panel B shows that, for soliciting targets, discretionary accruals decrease monotonically
with institutional ownership: the mean (median) discretionary accruals decrease from an
insigniﬁcant 0.80% (-0.24%) in the LOW portfolio to a statistically signiﬁcant —4.06% (—
2.14%) in the HIGH portfolio. The difference between the LOW and HIGH portfolios is
statistically signiﬁcant. However, I ﬁnd a similar association between institutional
ownership and discretionary accruals for unsolicited targets, except that unsolicited

targets start with signiﬁcantly positive discretionary accruals in the LOW portfolio.35

Therefore, my results are consistent with the ﬁndings of Raj gopal et al. (1999) suggesting

 

3‘ 1 repeat this analysis by dividing each group of soliciting targets and unsolicited targets into three equal
portfolios based on management ownership. The results on the LOW and MIDDLE portfolios remain the
same, but the discretionary accrual in the HIGH portfolio becomes insigniﬁcant.

35 The correlation between institutional ownership and discretionary accruals is —0. 143 for soliciting targets
and —0. 141 for unsolicited targets. Both correlations are statistically signiﬁcant but not different from each
other.

37

that institutional investors are sophisticated investors who effectively reduce managers’
incentives to manage earnings aggressively. However, this result is unlikely to result
from soliciting targets’ decision to seek acquirers.36

I further test the impact of ownership structure on the level of discretionary
accruals in a multivariate setting. For management ownership, I create a variable denoted
LOWMGTOWN, which equals one if the target ﬁrm belongs to the LOW or MIDDLE
portfolio of management ownership and zero otherwise. Likewise, for institutional
ownership I create a variable, denoted HIGHINSTOWN, which identiﬁes the target ﬁrms
in the HIGH portfolio of institutional ownership. Table 8 presents the regression results.
Consistent with the univariate test results, managers with relatively low equity ownership
decrease discretionary accruals in an attempt to facilitate takeover (regressions 1 and 5),
and the signiﬁcantly negative coefﬁcient on SOLIC*LOWMGTOWN indicates that the
inverse relation only emerges in soliciting targets (regressions 2 and 6). Institutional
investors appear to compel managers to report more negative discretionary accruals
(regression 3). However, the differential effect of institutional ownership on
discretionary accruals between soliciting and unsolicited targets
(SOLIC*HIGHINSTOWN) is not signiﬁcantly different from zero (regressions 4 and 6).
The above results hold after control for pre-managed earnings (PREMGT), ﬁrm size
(SIZE), ﬁnancial leverage (LEVERAGE) and grth opportunities (HIGHQ).37‘ 38 In
addition, I substitute continuous variables (i.e., MGTOWN and INSTOWN respectively)

for LOWMGTOWN and HIGHINSTOWN and ﬁnd that the inferences are unchanged.

 

3° 1 repeat this analysis by dividing each group of soliciting targets and unsolicited targets into three equal
portfolios of institutional ownership. The inferences are unaffected.

7 Because neither ownership variable is adjusted for industry effects, to be consistent, I use industry-
unadjusted control variables in these regressions.
38 Adding interactions SOLIC*PREMGT and HIGHQ*PREMGT as in Table 6 does not alter the results.

38

Overall, my ﬁndings suggest that management ownership plays an important role in
inducing the managers of soliciting targets to clean up their balance sheets via
discretionary accrual choices. Although I do not ﬁnd similar results on institutional
ownership, my analysis is somewhat weak in that I treat all institutional investors as a

9
homogeneous group.3

 

39 Bushee (1998, 2001) classiﬁes institutional holders into three groups based on their expected time
horizon: institutions with high portfolio turnover and highly diversiﬁed portfolio holdings (“transient”
institutions), institutions with extremely low turnover and large average investments (“dedicated”
institutions) and institutions with low turnover but diversiﬁed holdings (“quasi-indexers”). Dedicated
institutions are the most likely to commit to ﬁrms’ long-term value and actively monitor managers’
behavior. Therefore, a more powerful test of the institutional investors’ impact on managers’ accrual
behavior would focus on dedicated institutions.

39

CHAPTER 7: CONCLUDING REMARKS

Previous studies argue that target ﬁrms manage earnings upward prior to takeover
in an attempt to increase the takeover price but fail to ﬁnd convincing results in support
of this prediction. In this dissertation I examine the accrual choices of a group of ﬁrms
that publicly solicit takeover offers (soliciting targets). This research context is unique in
that soliciting targets have sufficient time to make a decision on earnings management
before takeover and they are likely to face substantial costs if opportunistic earnings
management is detected. To control for earnings management incentives unrelated to
takeover, I include a group of ﬁrms that receive unsolicited takeover bids (unsolicited
targets).

My accrual analysis demonstrates that neither soliciting targets nor unsolicited
targets manage earnings upward before a takeover. However, the soliciting targets make
income-decreasing accrual choices in the year preceding their solicitation announcement.
In addition, I ﬁnd that soliciting targets are less likely to have poison pill defenses in
place to hinder takeover attempts than unsolicited targets. These results are consistent
with the argument that soliciting targets proactively clean up their balance sheets to
credibly signal that they are worthy targets. Further investigation reveals that the pre-
takeover downward earnings management is concentrated in the soliciting targets with
relatively lower management ownership, suggesting that those managers have interests
aligned with outside investors to enhance ﬁrm value. Thus, my ﬁndings partially explain
the mixed evidence documented in prior literature on accrual behavior of target ﬁrms:

while some target managers may be motivated to inﬂate earnings to inﬂuence the

40

takeover premium, target managers willing to sell their companies may make
conservative accrual choices to invite takeover offers.

This dissertation also contributes to existing income smoothing literature by
showing that, when approaching the solicitation decision, the managers of soliciting
targets are less inclined to smooth earnings than the managers of unsolicited targets. This
complements the ﬁndings of Gaver et al. (1995) and DeFond and Park (1997) that
managers smooth reported earnings to maximize their bonus or to reduce the threat of job
dismissal. As the long-term beneﬁts derived from future compensation or job security
become less important to the managers of soliciting targets, they do not smooth earnings
as much as other managers do. Instead they make income-decreasing accounting choices
to facilitate the takeover process.

Although my income smoothing analysis results appear to be robust to alternative
measures of pre-managed earnings, 1 cannot entirely rule out the bias arising from the
backing-out method as an alternative explanation for my income smoothing ﬁndings.
The most convincing result I document is the differential inter-temporal smoothing
behavior between soliciting and unsolicited targets. To the extent that the backing-out
bias inﬂuences the measured smoothing effect similarly for soliciting and unsolicited
targets, it cannot explain the declining smoothing behavior of soliciting targets as they
approach their solicitation decision.

Another limitation of this study is that I am unable to identify the speciﬁc accruals
managed by soliciting targets prior to their solicitation announcement. Examining
speciﬁc accruals Should provide more insight into how the negative earnings

management is achieved. For example, Marquardt and Wiedman (2004) ﬁnd that

41

management buyout ﬁrms manage earnings downward by reporting signiﬁcantly lower
unexpected account receivables prior to a buyout. However, the preliminary results
suggest that there is no signiﬁcant difference between soliciting and unsolicited targets
along a dimension of speciﬁc accruals. It might be the case that soliciting targets in
various industries employ different income-decreasing accounting techniques to clean up
their balance sheets. For example, anecdotal evidence suggests that banks generally write
off bad loans (e. g., European American Bancorp), whereas insurance companies increase
reserves for potential claim losses (e.g., Beneﬁcial Corp). I leave this issue to future
research.

My results have implications for research on the cost-beneﬁt trade-off involved in
ﬁrms’ accrual decisions. Schipper (1989) identiﬁes two conditions giving rise to
earnings management: (1) there are two distinct groups of stakeholders, one group
beneﬁts from earnings management at the expense of the other; and (2) information
asymmetry exists so that the group which bears the cost cannot undo the effects of
earnings management. Existing literature, however, rarely considers this trade-off,
focusing on the beneﬁts as the main factor driving the ﬁrm’s accrual decision. This study
provides a setting in which soliciting targets face both the beneﬁts arising from inﬂated
earnings and the cost of losing its credibility as a target. My results suggest that the cost
plays an important role in the ﬁrms’ accrual decision and may outweigh the beneﬁts if
soliciting targets’ priority is to consummate the sale of their companies. Future studies
can explore this issue by investigating the accrual decision in settings with various levels

of costs and beneﬁts.

42

APPENDIX

INCOME SMOOTHING USING THE BACKING OUT APPROACH

43

This Appendix provides a further analysis of the backing out approach that
potentially creates a spurious correlation in testing the income smoothing hypothesis. I
ﬁrst analyze how the measurement error resulting from estimating discretionary accruals
imparts a negative bias in the smoothing coefﬁcient. Second, based on a randomization
approach, I generate an empirical distribution for the smoothing coefﬁcients under the
null hypothesis and ﬁnd that the negative bias from randomization does not alter the tenor
of the results reported in Table 6. Finally, I argue that substituting operating cash ﬂows
for pre-managed earnings mitigates the bias caused by the backing out approach.
Spurious correlation created by measurement error

I adopt the model in Elgers et al. (2003) and Lim and Lustgarten (2002) that
analyzes the spurious correlation between discretionary accruals and pre-managed
earnings created by backing out. Let X,- be the earnings of ﬁrm i, and DA i. be the

discretionary accruals added to the true pre-managed earnings PX; Then
PX: = X , — DA: . Let T be an earnings target, the test of income smoothing is based on
the following regression:

DA] = no + ,6, (PX: — T) + Control, + ,u, (A1)
where C ontrol, includes variables that might affect DAK. If PX; and DA ,. were both
directly observable, then equation (A1) would provide a valid test in which [31 < 0 implies

income smoothing. However, neither PX," nor DA ,' is directly observable, so a proxy for

DA: is estimated from the modiﬁed Jones model (call it DA,). Denoting the estimation

error from the discretionary accruals model as a, DA, = DA: + 8. Then the test of

income smoothing becomes

DA, = ,60 +ﬂ,(X, — DA, —T)+Control, +p. (A2)

44

Because X, — DA,- — T = X,- — DA; — c— T = PX: — T — 8, equation (A2) is equivalent to

DA: + 8 = ,60 + ,6, (PX: — T —.9) + Control, +,u. (A3)
If there is income smoothing, one could expect a negative correlation between DA ,l and
(PX; — 7). However, even without income smoothing, the estimate of [31 is likely to be
negative given that the measurement error a is on both sides of equation (A3) with
opposite signs. In other words, the backing out approach cannot distinguish between this
spurious correlation and any real relationship between discretionary accruals and the
deviation of pre-managed earnings from the target.
The randomization approach

To assess the effects of the negative bias arising from the backing out approach on
the results reported in Table 6 Panel A, I randomly select reported earnings of ﬁrm j
(scaled by total assets of ﬁrm j) from all Compustat ﬁrm/years and assign it to each of my
target ﬁrms (target ﬁrm i) between year —4 and year 0. This random assignment is done
to isolate the spurious correlation due to the backing out approach given that the
correlation between reported earnings of the randomly chosen ﬁrm j and the discretionary
accrual measure of ﬁrm i is expected to be zero. I calculate pre-managed earnings by
backing out discretionary accruals DA,- from the randomly assigned reported earnings X,-
and using this measure in the income smoothing test:40

DA, = [30 + ,8, (X, — T - DA,)+ Control, + ,u (A4)

The spurious negative correlation under the randomization approach is due to DA,-

appearing on both sides of equation (A4). I repeat the random assignment and estimate

 

‘0 An alternative approach would be to randomly assign the discretionary accruals of fu'm j to the sample
ﬁrm i. I choose to randomly assign the earnings instead of discretionary accruals in order to preserve the
association between DA ,- and the control variables in the multivariate model.

45

my regression model 5000 times. I use the resulting parameter estimates from the 5000
iterations to generate their empirical distributions under the null hypothesis of no income
smoothing for the backing out approach.

Based on the empirical distributions, I adjust the smoothing coefﬁcients for
unsolicited targets (PREMGT) and soliciting targets (PREMGT+PREMGT*SOLICT) by
subtracting the mean of the empirical distributions from the coefﬁcient estimates as
reported in Table 6 Panel A. The adjusted smoothing coefﬁcients are presented in Table
A1. The p values reported in Table A1 are obtained from the empirical distributions of
the estimated smoothing coefﬁcients for the soliciting and unsolicited targets. Table A1
indicates that the adjusted coefﬁcient on PREMGT remains negative across time, but its
magnitude is slightly smaller than that reported in Panel A of Table 6. This ﬁnding
suggests that although the backing-out approach induces a negative bias in the smoothing
coefﬁcient, this bias cannot fully explain the negative coefﬁcient on PREMGT for my
sample of unsolicited targets. In addition, the smoothing coefﬁcient for soliciting targets
(PREMGT+ PREMGT*SOLIC) increases sharply from a signiﬁcantly negative —0.424 in
year —4 to an insigniﬁcant 0.031 in year 0.41 Therefore, even with the adjustment for the
potential bias from randomization, the differential income smoothing trend between the
two groups of targets still holds.

Operating cash ﬂows substitute
Another approach I take is to use operating cash ﬂows obtained from the

statement of cash ﬂows (OCFi) as a substitute for pre-managed earnings. The rationale

 

4' Table 6 Panel A shows that the income smoothing effect (before adjusting for backing out bias) of
soliciting targets diminishes signiﬁcantly as approaching solicitation announcement. Given that the mean
bias is relatively constant across years (ranging from -0.061 to -0.050), a test of change in the adjusted
smoothing coefﬁcient for soliciting targets would not be necessary.

46

behind the substitution is that operating cash ﬂows are free of measurement errors
resulting from the estimation of discretionary accruals. Using operating cash ﬂows, the
regression model of testing income smoothing is

DA, = ,60 + ,6, (OCF, — T) + Control, + y . (A5)
Let NDA: be the true nondiscretionary accruals, then

X, = OCF, + NDA,‘ + DA,‘ = PX,‘ + DA,‘, (A6)
and equation (A5) is equivalent to

DA,‘ + s = ,6, + p, (PX: — T — NDAI) + Control, + ,u . (A7)
Whether the substitution of operating cash ﬂows leads to an improved test of income
smoothing depends on the correlation between NDA,‘ and DA)... The test would be
unbiased if the two are uncorrelated, but would be conservative if the two are negatively
correlated. Only when NDA,‘ and DA; are positively correlated the test is likely to over
reject the null hypothesis of no income smoothing. Moreover, this positive correlation
must decrease over time in my sample of soliciting targets to explain my ﬁndings. When
I use OCF, as a proxy for pre-managed earnings, the results reported in Table 6 Panel B
are qualitatively similar to those under the backing out approach. Overall, my analysis
suggests that the test results of income smoothing are not entirely caused by the spurious

correlation due to the backing out approach.

47

Table A1

Randomization-adjusted income smoothing coefﬁcients of soliciting and unsolicited targets
from year —4 to year 0 (one-tailed p values in parentheses)

 

 

 

Year —4 Year —3 Year —2 Year —1 Year 0
PREMGT -0.361 -0.285 -0.250 -0.292 -0.201
(<0.001***) (<0.001***) (<0.001***) (<0.001***) (0.027“)
PREMGT+ -0.424 -0.258 -0. 190 -0.096 0.031
PREMGT*SOLICIT (<0.001***) (0.015") (0.021 **) (0.034“) (0.514)

 

48

ﬁes—9:8 0.53 Sonia—con 05 coca maﬁa—Eco 05 :3 9 £33 unease—=8 E8358
05 momma m E 53222 .3882: .3 95cm a .8 Amroumo>£ .3 @3895 8.8 2 Ba Amy—0335 .38me .3 0.553% .«o 880 w mops—2: bemoan. 2F c
.535 05 3 Boom Eu 29:8 05 not? 5 coEmEcom _ 28 Mac 5% v £53333. 5:23 v .092 _ woos—2: bowoao £5 $.8qu @5628 con .

 

 

 

 

A _ m Rm SN 29:3 .95
2 s c a semi sea waves sea a3
GS 38 A. EoEowmcaEESB 2: :9: 55¢ momtmm .3 28 3 oh ”$3
8: a: _. 28a 350 “a3
2 s 5 5V ease ”a3
3 3 cuv $3 6583 283 :otooaoa 2 SEED .595 wcumcomo DEE ”mmoq
”match so 83:8 map: @3632 .3 wanEom $5.5m
as $3 6 3 3588585 29:32 ”so:
Comv 33 gm; veg—on wasp note 23 mo ESE wcmogoﬁa 95E Ammo)—
8: AD AG mesa: 8&3 28 .«o E85 memogocem SEE ”mmoq
Ammmv 83 Ann: $83355 3235.,— ”m8;
2.: as a: 9% ammo 83 28E #3
A33 as so: 9% 23838 83 WEE ”son
5m; wmm mm” 88.3% mats—u mEn
35:85 9:382 co moors xoom 8 wcmogoﬁa Etc 233 =<
ﬂowcﬁ muowg
Bee 8338: mascozom

 

MgN nun-a ¢QG~ HQOEQQ muuwo—QHI ﬁvumOm—amﬂb FhN nus-a mwawhah. wﬂmwmum—Om GVN MO ﬂOmuuﬁ—Om O—BEQW

n 033—.

49

.o amp .5 Sup E82 x83. wEmﬂE we: bowoﬁo £5 5 SE 0:0 _.

 

 

50

 

*3.“me Nod cam .80...
1mm .m as M: 958 He: #:0885552,
12.36 2.». _ 2 £925 35285 .QEEBOQ BEES:
***mm.m 3.9 on “Vote acme“: oZ
med 8.3 m 850
*5 _ .N Kg: 3 “cacao—pic $0525 memocﬁcm
iammﬁ :4: 2 8:805 3:309:00
Ivhm- 3.2- vm 3325 3365.35
med onto. m.— EoEQoEEw noscccoo c8 commaﬁ 3E8 warms—com
mwd 2% mm 00588.23 $353935
*zooé vim ~ 2 xooam nos—afcovcb
mo; _m.m_ m be: Em: 05 3 So :30
22.8.0 QM: co m 83> 522335 wEﬁmemE
ea 2+ .1 wee

352% a want .5>o ~20 CO 59:82 £832 3me

 

muowuuh «£3315 .8.— ﬂoobm Eon—853:2 .33:— 2: can mouse—om 9392 5 notazéom me 33.3% .830
N ozah

 

 

 

 

 

 

 

2: E 2: ca 2: mm 2: v3 Eﬁ
03 E m? m 8.0 mm 3a a 88
E .0 2 m; a new om S.” ﬁ 88
of. 2 SN 5 ”3 a a: S 58
8.x E M? 3 as R N3. a. 88
2 3 m? a NE: mm 3.». cm 82
as 8 N3 2 2 mm was 8 M32
3 mm m; a ”2 8 a; Q 82
NS ON 2.: mm as S S E 82
$2 8 9.2 mm on : 8 Rd 8 82
3 M: 2.: 3 m: m $6 3 $2
86 I ”3 m as 3 Km 2 82
ta 0 SN 5 mm mm 86 a. $2
8m : h _ .o mm 2 Ma 9; 3 52
£6 2 32 2 N3 2 RE No. 82

mEhM mo mgm mo maﬁa mo mic—m mo

95 “BE—AZ. AX“ 59:52 .XV aver—DZ o\o 59532

2.933 hmnuzeﬂs .08wa MEEQQM Sump: hmzumemrb £9wa MENBQQQ
038mm Em 038mm Ea:

 

3095. Eggs—5 :N “Ea wing—cw gm .8 Eon—3253.3 3.58.3. 3::— uo 23> guano—«U 3. .955

m «Bah.

51

..0>0. $0. .0 0:005:30. m. 8098 02.2.85 0:0 08.0..8 503000 5000:0050 590:. .0.0:0._0.m.0 0.0 0. 0000:... 0.8 m..00 0:... .

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0m... 0 0:57.303”. 0305.30.05... 00 N50 N06 0.83000: .002 M05 mm. mn.~ 3.. 00 0.2050 $0.0 $0.0 05% mm
00>

.01 .25.). .53. .03. 5:.” mm on... ma... 0.000.003-0050 030.5052 3. 000:... 000.00.“. 00.3. 0.0 .0003. S.

000.2% .0.00m mm owd mod 0.000.055-0050 038:0 3. 00:. 00.0.00. 0% m:.:_..0y. 0.0.. 0m

Au.00.>.0m 35:005.". N0 00.>._0m .0855 .000 0.0005 0.. 30:00:. 00...< 00 202E020 mm

m00.>:0m .200: 0% m:0.§0.:::=:00 m» 00...< ”a. 5.00.33. 0:05;. mm

50000000. .m::0E0m=E< 0m m00.>:0m 5.00.0558... «V 80:00:. 00...< 0:0 :05; 0m

8:30... 5002 «K no. 93 :.< mm. 5.08.5032... m.» 0050:“. 0:0 0.2.50... mm

0%:
.m00.>..0m £2.00. 05¢. E 0nd 0 5085028... .0003 3. 0m... mvd Es... xm. .00: 0003 0:0 .0051. VN
00:05:03

m00.>:0m mm0:.m=m K on... N00 .05.; EBB“. 55.). N». 0nd mm. 00.. 005.5... .050 0% .0593 mm

m000... w:.m00.. .050 0.0.0.. 3 who 9»... 5.05.0005... 05....5. 0» .5. mm. 80:00.5 ...S. 0:08... mm

:83. $005.38.: 3. .. _ .N mm. 00.55:. 000.352 00.2 0m «9....” mac 300005 00:53. 0:0 000... gm

8:803
« . .N wad 800.: 5.0.55 0:0 ”.50.“. .3. . .0. 3.0 H00 0.2... 55. 000$. 0M Que o 000.... 3.003-500.5200 u.
0:
HQN. 0.5m. d. 0.35:“. 0E”... A... .0...” a 50.5.0.5 5085005.; mm o 9.... 500.5050 .900: 3
005
3.0 NV... bomm000< 0:0 .059.“ 0% cm0 wo.« .05950 x0 0.50 30.0005 3. 0nd Nvd 5:005:00 m0.m. w.
0 N00 05.005 30 50.000 90§ ww mm... ”m... 300.: .0:. 0.50 .050500 a. 3....“ «0m 5008.5. 30 0:0 :O 2
N50 mm. 850m 000... E. «..N Nod b05009: x0 .908 0.50... Vm N50 0 w:_:.E .803. 0.
we. «0.. 0055 3.052.203. .8050 mm 0.0mm a, 00.55:. .802 ENEF. MM N50 N00 80.5 500200... 0.3.:0_:w< .0
. 0:52 . . 0502 . . 0:52

.05: .0m U.m .02: .0w U.m .023 .0m 2m

 

 

 

 

 

 

 

 

 

 

.. 8035—. 08.2.05: ««N 0:0 08.0.15 ova 5.— :003555 «50:0:— U—m «BR—.QSH. "m .05?—

m 03:“.

 

 

52

000.0 0... .0 m:.::.w0: 0... .0 0.0000 .05. 0:. .5 00:... 00. :0... .090. 000. 0505.0 5.3 50020000 :0 0050.0 . 0.0.0.050 0.... :. .

 

 

 

 

 

.000. .000. 000.0 000.0- 000.0 000.0- .000 .00 0.00.0. 02.2.80:
00.0- 0.0- 000.0 000.0- 0...0 0.0.0- 000.0- 000 0.00.... 00.22.00
0250000.
.000. .20. 00.0 000.0- 000.0 00.0- 0000- 000 0.00.... 02.2.0000
00.0- 00..- 000.0 000.0- 0.0.0 000.0- .0..0- 000 0.00.... 00.22.00
0<:m
.000. .000. 000.0 000.0- 00.0 ..0.0 000.0 000 0.00.0. 02.2.80:
0.0- 00.0- 000.0 000.0- 000.0 000.0- 00.0- 000 0500.00.22.00
30.00.05
.000. .000. 000.0 000.0- 00.0 000.0 000.0- 000 0.00.0. 02.2.80:
00.0- 00.0- 000.0 00.0- 0000 0.0.0- 00.0- 000 0.003.000.0000
002.7020.
.000. .000. 000.0 0.0.0- 000.0 000.0 00.0 000 0.00.... 0022.80:
00..- 00.0- 3.0 000.0- 0.0.0 000.0 000.0 000 0.00.... 00.22.00
0.9.00.3...
80.0. 20.0. 000.000 $0.00- 0000000 0.0.00 000.000 000 0.00.0. 0022.000:
00.0- .00- .000: .0000- 000.000 000.1 000.00. 000 0.00.... 000.260
0.2).
x050... R x050; 0w
000.0-$0. 00.20.02. 00 .0 0.0 22.0000 5.00.. 2
ﬁsh-905.5 0.0.2 0
a «Eon—09:52:00.

5500.0... .00.... 0... 0. 5.... 50> 0:0 0.0»..08 03.3—00.5 05. 95.3.5 .0 A0000=_.0<
.0550... 05 0.93-N. 00.00200000000 3.0000... 00.00.0m =0 00.00.:3m 0.5....002— .< .000.—

v 0.0.0...

53

 

 

A38 83: 832 83$- 8&5: :82- ”8.? OR 3&5 833335 M
m; 33- 80? 25.3- 83: 808- 32 SN 3&5 3528
mg
83: 25.8 89° Bod- £3 :3 33 SN 3&5 333335
3.? mg. 8.3 83- Rod 83. god ”mm 3&5 353:5
22
898 82: Rod 23- :5 $2- 08.? o2 3&5 528::
m3” 3m was 83¢. 93% 82- ES Rm 3&5 3.20:8
0 3258
339; E 3593 Q
5533» 3332: «a S Em 53: :8: 2

QQRQDNTR

3.2 ~

 

 

8.283 < .25 e 2...;

 

 

Rad Sod- nmmd Sod- m0<
_oc.ov :md- Sodv owmd- 2km
med mmod v36 god 0 mLémOH
ommd 3 _.c- ommd 3 ﬁo- Maxim
mm_ .o 3: d 304m2m<0
cc _ .o 5N _ .o mOEZm<m
Q: .o :35 av _ .o weed m0§m>mq
Sodv mm _ .o- Sodv mm _ .o- mN_m
0:6 owed Sumo mmod {EmomeZ—
33>-N 255i
hmEméih Emmumﬁmcb Vmeéik Emmugmcb ﬂoats;

 

 

 

$09.8 Exam—Oman t: ﬂ u
muowbﬁ wanza h _ H
033:; Botswana

 

Eon—3:353. hogs—ah 3::— 95 S 3th 23> 2.0 Accum=_1<-fum=u=_
U—m aux—-3 3:3.88230 .anaﬁh Bane—om .8 $933. coax—3:: 33.3 $5515 .3 macmmmouwom .39..— ”m .23.—
v 933,—.

55

 

 

83v 399 2.: N?“ on: 2.2 3.2 § mamas 8:268:

3. _- No. 7 8.3 ”an 32 on: #9 m a 3&5 wszéom
23057:

83: see 8.: o? 8.8 mi: 3.2 m8 mamas 8:282:

a} mom 8.? Nos 33 MK. a 3% MB $095 wcumozom
23052

353 Q «was»
33:62: a BEéE mm S Ba ES: :3: 2

2383:

3.3 ~

 

 

«583.5555. 333—3. .35: 2: 3 .Stm 28> 2.0 $832.22
-Eumacad Esp—2.30 3.83.58:— ena 9322.30 E25352 a: :omtaaﬁcu ”U .05?—

v 02.;

56

.Aﬂowmcﬁ: 3:08 sod E833 25:5ng 6353?: was $3388 “5::ng =w 3 Eu: 88% mo owﬂcoeoa M 230.55
.9.on a ma ﬂowoemu 98 Eco—.23 an Bo: 83% mo 093:8qu N 230902

.90me :38 we EﬁtawB .8:ch 333.89%“635 H mNE

Saw Eoﬁoogoccm 3:5 05 9 28 mag: 60:30 05 Bob was: >555 :w%-m wcmwcommotoo .8 :23st 562:

on“ mctombnsm 5% “5885558 REE 05 8 05 893: ammo 05 E 23 mag: 802.80 05 88,2 $558 :3 5:85 H mO<
.533 mo 02? gov—SE $>o Son 3%:anan £33 5888 we 2:5, xoom u 32b

.ﬁomm’x 30H .3 03m> xoomkmwommxx 32.50 I 332534 62.50

+ Boa govwcoq .«o 02m> xoom + 8% “83388 “—0on veranda mo 33> moom + €on coEEoU («o 38> “8.83:" 02502.
.80» 05 mo wcchmwon 05 3 33mm :38 as $638 52% 652980an coummoeaon I 953, “89988 mama—BEA

Eobsu E cows—05 EoQ< + 2% 883.2808 :3“? I Ann aﬁmzanUv 82:23»— Eot=U< I can 35:anan m3mm< EBBUd n
seaﬂooaoa I £§doo< E850 H mA<DMUU<

.59: .819: woﬁwmoabig 2o $838 38383 25:: £858 xoowm SE town—38:0 u M<Im

.05an oomv 2mEoo<

€2.50 3ng 2: gm 332588 mcosﬁoao 89a £53 233 x .52 Bowen ”$2 Awﬁwﬂoﬁv Btu Awomu “839:0an mcocﬁomO
8on 32m ammo ﬁasco 25E :30 wcuﬁomo use» 2: mo wﬁaﬁwon 05 8 33mm 139.55 255 :80 @58qu n BOIEEm/‘O
58% 2: mo wcmcﬁwon 05 3 3% 833808 333 3885 Am 3 “Ema—9:08 was: buEEombxm 883 088:— " m02~2m<m
.Cnov chaﬁwzoq + 333 mo 33> Bwaion uﬁmzmﬁovv Bow SunﬁwEJ H m0§m>mq

Agra “Swan—o0 x mg 98.013608 52:3 20 2:? 6x32 u m>2

"285.53. 033...;

5:53 a 2...;

57

 

 

 

 

 

 

Rd 1:. 3d wvd de wwd 5:855
Nwd u. o _ d mm d odd Ed “m2 9 Emma: 3:232: .m> 320:8
E uh: do 829 a Btu-03H

de- ONd mNd- mod Nod 5:52

ovd- de Ed 3d omd :82
v: om N m m N um N _NN Z 3&5 3:2323

5d *Iwm._- Rd NE- de 5:52

Rd- * VN._- mod 0 _ d dd. _ :32
of o 2 N SN mm: we: 2 Bums: wﬁzozom
o 50> T 30> N- 30> m- 30> v- Bu> 3:88:
b23886 :oaowwE-oogﬁcotom ”m ESE
mad 1 mod ovd BNd Nod :88259 $098
and ,.. mod dod cwd Nod “mo: H 322—03: .m> $5628
:_ E6 do 829» a coca-03,2.

NNd- mod- de mod Ed 5:62

mo. 2 - oNd _Nd Nod owd :32
EL EN om N OVN VNN Z Samba Exam—8:3

3d- : dwd- 3d 3.0- add 5:82

$2- .1wa Ed mNd No._ :82
mm. dNN de R; we: 2 womb: mﬁzozom
d Eo> T Eo> N- 30> m- 30> v- 30> 6:08 3:3 85on 59:

368:8 28:8: Egozﬂoma n< 6:8

 

ﬂow-3h Exam—can: 6:: Mugoaom u: aaauo< banoﬂohuan buoy—«=0 6:: 15::

m 033—.

58

 

 

 

 

 

wmd d2 d ... Sd NNd c 2 d 2.880223 30mg 232202285
cmd .1. mod _.. nod oNd 3., Nod 38 .2- .m> 9.222328 :2 “2.2%
do mo=23> 2 8293.952-
wdd- de- dod- d 2 d- m 2 d 32252
Nod N2 d- mod cvd- mNd 302
22222 dNN 2MN mMN NmN 72 m2ow32 2522022852
de- I Nvd- 2. Ned- mod ... mdd- 32232
206. :i :2- **mw.o- N25 ELVNA- =30:
32 002 32 2&2 m2: 2 $0332 w:222o22om
d :0 2- :0 N- .50 m- to v- :20 23:33 3:02220m22u
. donoHQE-ooqmgotom ”m 2232
mvd .w 0 — .o 3.. #66 BN6 .1. mod douse—$5 gums“ $362022:
MNd .2... 5d .1 mod mvd 2.... 5d 282 H .m> $222228 :2 H22:
.20 mug—23> n2 coca-03,2-
mod v 2 d- m Nd NNd- m 2 d ~32252
n 2 d mod- N 2 d VNd- wvd :82
22222 dNN 2mN mMN NMN Z 3&3: 2322022852
BN6- .1. Sad. I. ovd- BN6 .2" N. 2 .o- 28:52
odd- _.:.:.. do. 2- t. und- wdd .1. ”N. 2- :82
£2 32 32 2&2 32 Z m2ow32 w:222o22om
d :0 2- :20 N- to m- :5 v- :20 20288 3:02. 25$on 88.2 33:58

 

£3883 3:026:85 ”< 2232

 

maowhah 182922352 1:3 9.239222% .2: £an9< aha—.3298»:— buotaso 2:3 23:55.

€3.33 m as;

59

8.52 :83 n :55

8233085 3:28 3284 23:25 3:55: woman-:33 u .2_>m2m2d<<

2:06:23: :3 2:32: 5:228: mmb u Emma

$83 2328 u _-.2_<

x0232 £833: 238 3:» H2

33:: 2m 3:29-23 .- 2

3522:80 :0 023233 m2 33: 32038.2 8923 25.2 8.2 5232 :52 n 2
8223
I... 2-.....

2-5 22..

2. < ~. < ~ < 2 <
NVQ-T S 5%.? S :Q-T CG“ 2:
AMNSQE Mai 2 Q3230va

 

 

”20:08 w:2>>o22o.2 2: 88.2 338228 033 23353 3:028:82: n <9

:32 :1 Ba .03 £2: 35-23 2: a 8:853: 28%: *1 Ba f. n..-

60

Table 6
Panel A: Annual OLS Regressions Of Discretionary Accruals on Pre-Managed
Earnings And Control Variables From Year -4 To Year 0, Where Year 0 Is The Year
Of Initial Announcement (Two-Tailed P Values In Parentheses)

 

 

 

 

Year —4 Year —3 Year —2 a Year —1 Year 0
INTERC EPT -0.001 -0.002 0.004 -0.006 -0.020
(0.875) (0.803) (0.604) (0.362) (0055*)
PREMGT -0.395 -0.294 -0.270 -0.312 -0.263
(<0.001***) (<0.001***) (<0.001***) (<0.001***) (<0.001***)
SOLICIT 0.001 -0.003 -0.01 1 -0.016 0.007
(0.955) (0.780) (0.280) (0069*) (0.633)
PREMGT*SOLIC1T -0.082 —0.0 1 4 0.030 0.187 0.233
(0.031") (0.410) (0.464) (<0.001***) (<0.001***)
HIGHQ 0.035 0.045 0.006 0.014 0.026
(0.010***) (0.005***) (0.682) (0.300) (0.280)
PREMGT*HIGHQ 0.440 0.301 0.099 0.1 16 -0.06l
(<0.001***) (<0.001***) (0.004***) (<0.001***) (0.285)
SIZE 0.004 0.002 0.003 0.002 0.013
(0.225) (0.543) (0.273) (0.398) (0.003***)
LEVERAGE -0.038 -0.029 -0.025 0.010 -0.096
(0.164) (0.359) (0.326) (0.643) (0.005***)
OBS 371 420 450 479 325
Adj. 1?.2 0.29 0.09 0.22 0.15 0.11
PREMGT+ -O.477 -0.308 -0.240 -0. 125 -0.030
PREMGT*SOL1C1T (<0.001***) (<0.001***) (<0.001***) (<0.001***) (0.500)
Change in
PREMGT+ 0.169 0.068 0.1 15 0.095
PREMGT*SOL1CIT 0.012“ 0.235 0005*" 0059*

 

 

' 1 drop 4 extreme observations in this regression. Including these observations weakens the coefﬁcient on
PREMGT but does not change the overall inferences.

61

Table 6
Panel B: Annual OLS Regressions Of Discretionary Accruals On Operating Cash Flows
And Control Variables From Year —4 To Year 0, Where Year 0 Is The Year Of Initial
Announcement (T wo-Tailed P Values In Parentheses)

 

 

 

 

Year —4 Year —3 Year —2 Year -1 Year 0
INTERCEPT 0.007 -0.000 0.005 0.001 -0.017
(0.378) (1.000) (0.498) (0.892) (0.103)
CASHFLOW -0.431 -0. 193 -0.198 -0.289 -0.368
(<0.001***) (<0.001***) (<0.001***) (<0.001***) (<0.001***)
SOLICIT -0.003 -0.004 -0.007 -0.020 0.006
(0.803) (0.730) (0.489) (0.024“) (0.673)
CASHFLOW -0.068 -0.060 -0.048 0.151 0.340
*SOLICIT (0.101) (0.005***) (0.346) (<0.001***) (<0.001***)
0.030 0.036 0.008 0.010 0.033
HIGHQ (0.029") (0.022“) (0.586) (0.489) (0.152)
0.506 0.197 0.108 0.124 -0.060
PREMGT*HIGHQ (<0.001***) (<0.001***) (0.010***) (<0.001***) (0.387)
0.002 0.001 0.002 0.002 0.014
SIZE (0.655) (0.714) (0.506) (0.479) (0.002***)
-0.021 -0.012 -0.019 0.013 -0.104
LEVERAGE (0.458) (0.71 1) (0.498) (0.586) (0.002***)
371 420 450 479 325
OBS 0.24 0.07 0.14 0.08 0.13
Adj. R2
PREMGT+ -0.499 -0.253 -0.246 -0. 138 -0.028
PREMGT*SOLIC1T (<0.001***) (<0.001***) (<0.001***) (<0.001***) (0.606)
Change in
PREMGT+ 0.246 0.007 0.108 0.1 10
PREMGT*SOLICIT 0.001*** 0.917 0.031" 0069*

 

*, **, and *** indicate signiﬁcance at the two-tailed 10%, 5%, and 1% level.

62

Table 6 (cont’d)
Variable deﬁnitions:

PREMGT = Current earnings before extraordinary items over lagged total assets
(subtracting two-digit industry median) minus discretionary accruals.

SOLICIT = 1 if soliciting targets and 0 if unsolicited targets.

HIGHQ = 1 if industry-adjusted Tobin’s Q is larger than 1, 0 otherwise.

LEVERAGE = Long-term debt/(Market value of equity + Long-term debt), subtracting
two-digit industry median.

63

Table 7
Panel A: The Association between Management Ownership and Discretionary
Accruals in Year -1

 

 

 

 

 

 

 

 

 

 

Soliciting targets
N Mean (%) Median (%) Wilcon. Rank Z
(two-sided p (two-sided p t test for diﬁ? in test for diff in
value) value) mean median
LOW 37 -1.378 -0.819 Low to Median Low to Median
(0.319) (0.176) -1.15 -1.54
MIDDLE 79 -3.616 -3.545 Med to High Med to High
(0.003***) (0.000***) 3.78*** 407*“
HIGH 37 4.358 1.870 Low to High Low to High
(0.022") (0.010***) 2.52M 282*"
Unsolicited targets
N Mean (%) Median (%) W iIcon. Rank Z
(two-sided p (two-sided p t test for diﬂ in test for diﬁf in
value) value) mean median
LOW 50 -0.191 -0.458 Low to Median Low to Median
(0.849) (0.426) 0.93 1.1 l
MIDDLE 120 0.938 0.625 Med to High Med to High
(0.184) (0.231) -1.12 -1.53
HIGH 54 -0.011 -0.154 Low to High Low to High
(0.511) (0.355) 0.48 0.46

 

64

Table 7

Panel B: The Association between Institutional Ownership and Discretionary

Accruals in Year —1

 

 

 

 

 

 

 

 

 

 

Soliciting targets
N Mean (%) Median (%) W iIcon. Rank Z
(two-sided p (two-sided p t test for diﬂ in test for diff in
value) value) mean median
LOW 50 0.802 -O.236 Low to Med Low to Med
(0.663) (0.750) -1.06 -1.15
MIDDLE 100 -l .360 -0.875 Med to High Med to High
(0.138) (0088*) -1.69* -1.31
HIGH 50 -4.059 -2. 140 Low to High Low to High
(0.004***) (0.005***) -2.14** -1.98**
Unsolicited targets
N Mean (%) Median (%) Wilcon. Rank Z
(two-sided p (two-sided p t test for diﬂf in test for dijf in
value) value) mean median
LOW 58 2.738 1.603 Low to Med Low to Med
(0.027") (0.040") -1.75* -1.75"‘
MIDDLE 125 0.360 -0.359 Med to High Med to High
(0.627) (0.993) -1.16 -1.01
HIGH 62 -l .085 -1.263 Low to High Low to High
(0.265) (0.189) -2.49** -2.37**

 

*, **, and *** indicate signiﬁcance at the two-tailed 10%, 5%, and 1% level.

65

Table 8
OLS regressions of discretionary accruals on industry-unadjusted ownership
variables and control variables in year —1 (two-tailed p values in parentheses)

 

Regression Regression Regression Regression Regression Regression

 

1 2 3 4 5 6

INTERCEPT -0.007 -0012 -0.008 0.009 -0001 0.004
(0.698) (0.514) (0.547) (0.545) (0.955) (0.834)

souc 0.049 -0022 0.025
(0.011") (0.034") (0.192)

LOWMGTOWN -0.018 0.024 -0017 0.016
(0.1 17) (0.101) (0.124) (0.263)

SOLIC“ -0094 -0079
LOWMGTOWN (<0.001***) (<0.001***)
HIGHINSTOWN -0.024 -0.016 -0.016 -0.021
(0.020") (0.234) (0.130) (0.127)

501.10 -0.013 0.021
HIGHINSTOWN (0.507) (0.313)
PREMGT -0.082 -0.084 -0.031 -0.033 -0.064 -0.068
(<0.001***) (<0.001***) (0.004***) (0.002***) (<0.001***) (<0.001*")

5le 0.002 -0.003 ' 0.001 -0.001 0.002 -0.002
(0.630) (0.429) (0.649) (0.758) (0.429) (0.578)

LEVERAGE 0.002 0.018 0.007 0.013 -0.005 0.010
(0.944) (0.420) (0.758) (0.546) (0.826) (0.635)

HIGHQ 0.010 0.012 0.004 0.008 0.007 0.010
(0.296) (0.232) (0.668) (0.391) (0.452) (0.292)

OBS 376 376 443 443 359 359

Adj. R2 0.12 0.16 0.02 0.04 0.09 0.13

 

*, **, and *** indicate signiﬁcance at the two-tailed 10%, 5%, and 1% level.

Variable definitions:

LOWMGTOWN = 1 if the ﬁrm’s management ownership is below 75th percentile, and 0

otherwise.
HIGHMONEYOWN = 1 if the ﬁrm’s institutional ownership is above 75th percentile,
and 0 otherwise.

66

BIBLIOGRAPHY

Alexander, G., 2001. What is your company worth? Black Enterprise 31 (12): 103-110.

Ajinkya, B., S. Bhojraj and P. Sengupta, 2003. The governance effect of institutional
investors and outside directors n the properties of management earnings forecasts.
Working paper.

Barnea, A., J. Ronen, and S. Sadan, 1975. The Implementation of Accounting Objectives
- An Application to Extraordinary Items. The Accounting Review 50(1): 58-68.

Bartov, E., 1993. Timing of Asset Sales and Earnings Manipulation. The Accounting
Review 68(4): 840-855.

Beatty, A., S. Chamberlain, and J. Magliolo, 1995. Managing ﬁnancial reports of
commercial banks: The inﬂuence of taxes, regulatory capital and earnings.
Journal of A ccounting Research 33 (2): 231-261.

Becker, C., M. DeFond, J. Jiambalvo, K. R. Subramanyam, 1998. The effect of audit
quality on earnings management. Contemporary Accounting Research 15(1): 1-
24.

Beneish, M. D., 1999. Incentives and Penalties Related to Earnings Overstatement that
Violate GAAP. The Accounting Review 74(4), 425-457.

Bettinghaus, B., 2001. Earnings management by merger targets: Discretion over the loan
loss provision in commercial banks. Working paper (University of Missouri —
Columbia).

Bushee,]998. The inﬂuence of institutional investors on myopic R&D investment
behavior. The Accounting Review 73(3): 305-333.

, 2001. Do institutional investors prefer near-term earnings over long-run value?
Contemporary Accounting Research 18(2): 207-246.

Chamberlain, S., 1991. The effect of mergers on the operating performance and
discretionary earnings components of acquired banks. PhD Thesis, University of
Chicago.

Christie, A. A. and J .L.Zimmerman, 1994. Efﬁcient and Opportunistic Choices of

Accounting Procedures: Corporate Control Contests. The Accounting Review
69(4), 539-66.

Collins, J ., D. Shackelford, and J. Wahlen. 1995. Bank differences in the coordination of

regulatory capital, earnings and taxes. Journal of Accounting Research 33 (2):
263-291.

67

DeAngelo, LE, 1986. Accounting Numbers as Market Valuation Substitutes: A Study

of Management Buyouts of Public Stockholders. The Accounting Review 61(3):
400-420.

, 1998. Managerial Competition, Information Costs, and Corporate
Governance. Journal of A ccounting and Economics 10, 3-36.

DeAngelo, H., L.E. DeAngelo, and D. J. Skinner, 1994. Accounting choice in troubled
companies. Journal of Accounting and Economics 17: 113-143.

Dechow, P., Kothari, S.P., Watts, R., 1998. The relation between earnings and cash
ﬂows. Journal of Accounting and Economics 25(2): 133-168.DeFond, M. L., C.
W. Park, 1997. Smoothing income in anticipation of future earnings. Journal of
Accounting and Economics 23. 115-139.

Denis, D. J ., D. K. Denis and A. Sarin, 1997. Ownership structure and top executive
turnover. Journal of Financial Economics 45, 193-221.

Dhaliwal, D., M. Frankel and R. Trezevant, 1994. The taxable and book income
motivations for LIFO layer liquidations. Journal of Accounting Research
(Autumn): 278-289.

Easterwook C.M., 1997/1998. Takeovers and Incentives for Earnings Management: An
Empirical Analysis. Journal of Applied Business Research 14(1): 29-47.

Eddey, P. H. and S. L. Taylor, 1999. Directors’ recommendations on takeover bids and
the management of earnings: evidence from Australian takeovers. Abacus 7(1):
29-45.

Elgers, P.T., R.J. Pfeiffer Jr, and S.L. Porter, 2003. Anticipatory income smoothing: A
re-examination. Journal of Accounting and Economics 35: 405-422.

Erickson, M. and S. Wang, 1999. Earnings management by acquiring ﬁrms in stock for
stock mergers. Journal of Accounting and Economics 27, 149-176.

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

and MC. Jensen, 1983. Separation of ownership and control. Journal of Law
and Economics 26, 301-326.

Francis, J ., D. Hanna and L. Vincent, 1996. Causes and effects of discretionary asset
write-offs. Journal of Accounting Research (Supplement): 117-134.

. Fridson, M. S., 1991. Financial Statement Analysis: A practitioner ’s Guide. New York,
NY: John Wiley.

68

Gaver, J. J ., K. M. Gaver, J. R. Austin, 1995. Additional evidence on bonus plans and
income management. Journal of Accounting and Economics 19, 3-28.

Hand, J ., 1989. Did ﬁrms undertake debt-equity swaps for an accounting paper proﬁt or
true ﬁnancial gain? The Accounting Review 64: 587-623.

Healy, P. M., J. M. Wahlen, 1999. A review of the earnings management literature and
its implications for standard setting. Accounting Horizons 13(4), 365-383.

Hunton, J ., R. Libby, and C. Mazza. 2004. Financial reporting transparency and earnings
management. Working paper.

Jensen, M. and W. Meckling, 1976. Theory of the ﬁrm: Managerial behavior, agency
costs and capital structure. Journal of Financial Economics 3: 305-360.

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

Kothari, S. P., A. J. Leone and C. E. Wasley, 2005. Performance matched discretionary
accrual measures. Journal of A ccounting and Economics, forthcoming.

Kinney, M. and R. Trezevant, 1997. The use of special items to manage earnings and
perceptions. The Journal of Financial Statements Analysis (Fall): 45-53.

and W. Wempe, 2004. JIT Adoption: The Effects of LIFO Reserves and
Financial Reporting and Tax Incentives. Contemporary Accounting Research,
forthcoming.

Lim, SC. and Lustgarten S., 2002. Testing for income smoothing using the backing out
method: A review of speciﬁcation issues. Review of Quantitative Finance and
Accounting 19: 273-290.

Louis, H., 2004. Earnings Management and the Market Performance of Acquiring Firms.
Journal of Financial Economics 74(1): 121-148.

Malatesta, RH. and RA. Walkling, 1988. Poison pill securities: Stockholder wealth,
proﬁtability, and ownership structure. Journal of Financial Economics 20: 347-
376.

Marquardt, C.A., C.I. Wiedman, 2004. How Are Earnings Managed? An Examination of
Speciﬁc Accruals. Contemporary Accounting Research 21(2): 461-491.

Matsunaga, S., 1995. The effects of ﬁnancial reporting costs on the use of employee
stock options. The Accounting Review 70(1): 1-26.

McConnell, J .J . and Servaes, H., 1990. Additional evidence on equity ownership and
corporate value. Journal of Financial Economics 26, 671-688.

69

McNichols, M. and G. P. Wilson, 1988. “Evidence of earnings management from the
provision for bad debts.” Journal of Accounting Research 26: p.1-31.

, 2000. Research design issues in earnings management studies. Journal of
Accounting and Public Policy 19 (4, 5): 313.

 

Moehrle, S., 2002. Do Firms Use Restructuring Charge Reversals to Meet Earnings
Targets? The Accounting Review 77: 397-413.

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

Morgenson, G., No Wonder C.E.O.’s love Those Mergers. New York Times, July 18,
2004.

Moyer, S. 1990. Capital adequacy ratio regulations and accounting choices in
commercial banks. Journal of Accounting & Economics 13 (2): 123-154.

Perry, S. and R. Grinaker, 1994. Earnings expectations and discretionary research and
development spending. Accounting Horizons (December): 43-51.

, and T. Williams, 1994. Earnings management preceding management buyout
offers. Journal of Accounting and Economics 18: 157-179.

Raj gopal, S. and M. Venkatachalam, and J. Jiambalvo, 1999. Is institutional ownership
associated with earnings management and the extent to which stock prices reﬂect
future earnings? Working paper.

Ronen, J ., S. Sadan, 1981. Smoothing income numbers: Objectives, means and
implications. Adison-Wesley, Reading MA.

Schipper, K., 1989. Commentary: Earnings management. Accounting Horizons
(December), 91-102.

Scholes, M., G. Wilson, and M. Wolfson, 1990. Tax planning, regulatory capital
planning, and ﬁnancial reporting strategy for commercial banks. Review of
Financial Studies: 625-650.

Schrand, C. and B. Walther, 2000. Strategic benchmarks in earnings announcements:

The selective disclosure of prior-period earnings components. The Accounting
Review (April): 151-177.

Shivakumar, L., 2000. Do ﬁrms mislead investors by overstating earnings before
seasoned equity offerings? Journal of Accounting and Economics 29: 339-371.

Shleifer, A. and R.W. Vishny, 1986. Large shareholders and corporate control. Journal
of Political Economy 94(3): 461-488.

70

and , 1997. A survey of corporate governance. Journal of
Finance 52: 737-783.

 

 

Song, M. H., Walkling R. A., 2000. “Abnormal returns to rivals of acquisition targets: A
test of the ‘acquisition probability hypothesis’.” Journal of Financial Economics
22: 143-171.

Teoh, S. H., I. Welch, and T. J. Wong, 1998a. Earnings management and the post-issue
performance of seasoned equity offerings. Journal of Financial Economics
50(October): 63-99.

, , , 1998b. Earnings management and the long-term market performance
of initial public offerings. Journal of Finance 53 (December): 1935-1974.

 

T. J. Wong, and G. Rao, 1998. Are accruals during initial public offerings
opportunistic? Review of Accounting Studies 3: 175-208.

 

3

Warﬁeld, T.D., J .J . Wild and KL. Wild, 1995. Managerial ownership, accounting
choices, and informativeness of earnings. Journal of Accounting and Economics
20, 61-91.

Yeo, Gillian H.H., M.S. Tan Patricia, K.W. Ho and S. Chen. Corporate ownership
structure and the informativeness of earnings. Journal of Business Finance & accounting
29(7&8), 1023-1045.

71

   

1jljjjj‘jjjjjjjju