Journal of Accounting and Public Policy 23 (2004) 321349

www.elsevier.com/locate/jaccpubpol

Using nancial accounting information in

the governance of takeovers: An analysis
by type of acquirer
Joseph Aharony

a,1

, Ran Barniv

b,*

Faculty of Management, Tel Aviv University, Israel

Department of Accounting, Graduate School of Management, Kent State
University, P.O. Box 5190, Kent, OH 44242-0001, United States

Abstract
The substantial changes in the corporate governance mechanism of acquired rms
that take place during the periods surrounding corporate acquisitions lead investors
and other corporate nanciers to an intensive search for nancial accounting inputs
for decision making. We examine whether nancial accounting information on takeover
targets provides useful input in the corporate governance mechanisms of US publicly
traded takeovers in these periods. Our analysis is by four dierent types of acquirers:
foreign rms, publicly traded US rms, private US acquirers, and leverage buyouts
(LBOs). We expect that certain rm-specic nancial accounting characteristics of takeover targets by type of potential acquirer aect valuation. To examine this expectation
we construct a probability summary-value measure, composed of eight nancial
accounting variables, based on the type of acquirer. We also expect the probability

Corresponding author. Tel.: +1 330 672 1112; fax: +1 330 672 2548.
E-mail addresses: Josepha@smu.edu.sg, yossia@tauex.tau.ac.il (J. Aharony), rbarniv@bsa3.
kent.edu (R. Barniv).
1
Visiting professor of Accounting, Singapore Management University.
0278-4254/$ - see front matter 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jaccpubpol.2004.07.002

322

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

summary-value measure to be useful for determining investment strategies in acquired

rms. The empirical results strongly support our expectations.
2004 Elsevier Inc. All rights reserved.
JEL codes: G34; M14; M41
Keywords: Corporate acquisitions; Corporate governance; Accounting information; Value relevance; Investment strategy

1. Introduction
During periods surrounding corporate acquisitions, the corporate governance mechanism of acquired rms changes substantially. In particular, new
potential nanciers such as the new investors and acquiring rms extensively
scrutinize publicly available information in search of inputs. A major research
question that has been partially ignored in prior literature is how relevant
accounting information is during these periods. In these settings, nancial
accounting information may play a more prominent role in corporate governance mechanisms and should therefore have greater value relevance to investors and other nanciers of the rm, increasing their demand for such
information when making decisions.
In this study, we examine the value relevance of information available from
the nancial statements of US publicly traded acquired rms prior to the
acquisition. Our focus is on the value relevance of the accounting information
to dierent types of acquirers and other investors in target rms. We identify
four types of acquirers: foreign rms, publicly traded US rms, private US
acquirers, and leverage buyouts (LBOs). 2 We expect the accounting information of the acquired rms to dier across types of acquirers and provide dierent inputs and signals to investors. In addition, we expect that the accounting
information is useful for investment strategy in choosing among acquired
rms.
Extant accounting studies use earnings and book value of equities to explain
prices and returns. Rather than focusing only on earnings and book values,
which may not provide sucient information at the end of the rms life cycle,
we construct a probability summary-value measure (Ou and Penman, 1989),
composed of other components of accounting data. This measure includes specic characteristics of accounting information used by each type of potential
major nancier in the corporate governance of takeovers, and diers, on average, across acquired rms classied by the four types of acquirers.

2
Bradley et al. (1988), Kim and Lyn (1991), Opler and Titman (1993), Barnes et al. (1996) and
Gonzalez et al. (1998).

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

323

We use three methodologies to test our research expectations. First, we employ logistic regressions to generate the probability measure. While constructing the probability measure, we nd that the logistic regression performs well
and certain rm-specic nancial accounting characteristics of the acquired
rms are useful in identifying the type of acquirer. 3 Second, we employ
OLS price and return regressions. We nd that the probability summary-value
measure is value relevant in price and return models. In particular, we show
that the probability summary-value measure is signicant while neither earnings nor changes in earnings are signicant in return models. Third, we use univariate statistics to examine whether the probability summary-value measure is
useful for investment strategy in choosing among acquired rms. We nd that
the probability measure is a very powerful investment instrument. In sum, the
empirical results strongly support our expectations.
The remainder of the paper is organized as follows. The next section provides further empirical background. The third section details the motivation
and incremental contribution of our study, and provides a literature review.
Section 4 deals with the research methods, and Section 5 presents the data.
Section 6 analyzes the results, and Section 7 provides a summary and
conclusions.

2. Further empirical background

We examine the impact of nancial accounting information on prices and
returns for investors in takeover rms, classied according to the four types
of acquirers we have dened, from 250 trading days prior to the acquisition
announcement date through the last day of trading. To assess the cross-sectional likelihood of acquisition by a certain type of acquirer, we identify other
rm-specic nancial accounting characteristics of takeover targets used in
prior studies (e.g., Palepu, 1986; Hasbrouck, 1999), rather than the earnings
and book value of equities used in valuation models. Specically, eight significant nancial accounting variables are used in a ve-group multi-logit regression to extract the probability summary-value measure. We use the four groups
of acquired rms by type of acquirer and a fth control group of manufacturing rms not being acquired during the entire sample period. These ve groups
are used to construct the probability summary-value measure. Then, prices and
returns are regressed on earnings, book value of equities, and the probability

3
Based on prior literature on the market for corporate control (e.g., DeAngelo et al., 1984;
Gompers et al., 2003), we expect the nancial accounting characteristics of acquired rms to dier
across the four types, providing dissimilar signals as inputs for eective operation of the governance
mechanisms of US takeover rms.

324

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

measure. Finally, we show that the probability measure is useful for investment
strategies in target rms.
We obtained an initial list of acquired rms from Hall (1990, 1993) and supplemented the information using Compustat, the Wall Street Journal Index,
and Mergers & Acquisitions (19782001). Financial accounting and market
data were obtained from the Compustat and CRSP, respectively. Our nal
sample consists of 1578 acquired US manufacturing rms that exited the stock
exchanges from 1978 to 2001. The control group consists of a nal sample of
932 manufacturing rms with complete data.
We nd that investors in rms acquired by either foreign rms or by publicly
traded US rms gain, on the average, about 66% returns, from 250 trading days
prior to the initial takeover announcement date through the last trading day,
while investors in rms taken over by private US acquirers gain, on the average,
46% returns, followed by even lower returns of 42% gained, on the average, by
investors in LBOs. 4
Our results provide strong evidence that earnings, book value of equities,
and the probability measure representing the other nancial accounting
information are signicantly value relevant in explaining prices. Furthermore,
the probability measure is a signicant factor in explaining returns for investors making choices among acquired rms, while, earnings and change in
earnings are insignicant. These specic ndings suggest that nancial information other than earnings and book value is increasingly more important to
investors in acquired rms. Finally, the results show that the probability
summary-value measure is very useful for investment strategies, providing
high security returns for investments in acquired rms. In sum, the results
indicate that nancial accounting information, in addition to book value of
equity and earnings, provides an important input in the governance mechanisms of US takeovers and has a signicant impact on valuation of acquired
rms.

3. Motivation, literature review and incremental contribution

In excellent overviews of the explicit and implicit uses of accounting information in corporate governance mechanisms, Bushman and Smith (2001)
and Sloan (2001) claim, among other things, that nancial accounting systems
provide direct and indirect inputs for corporate control mechanisms and that

4
While demonstrating annual negative gains for investors in many acquiring rms one year
after the acquisition, Henry (2002) documents average excess returns during the two weeks
surround the announcement of 9% over the S&P 500 index for investors in the 15 largest acquired
companies from 1998 to 2000.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

325

the most important role of accounting information is its implicit use. For
example, Sloan highlights its role in facilitating corporate takeovers, and points
out Palepus (1986) contribution in showing that nancial accounting data are
useful in predicting takeovers.
Our study goes a step further in focusing specically on the issue of whether
nancial accounting information provides useful input for the corporate governance mechanisms of US publicly traded acquired rms facilitated by four
dierent types of large acquirers, and the extent to which this information is
instrumental in investment decisions by various nanciers.
Jensen (1984, 2000) argues that shareholders benet most from acquisitions
because of the increase in the value of the acquired shares (the synergy hypothesis). Takeovers protect shareholders from mismanagement of corporations
since they allow alternative managers to compete for the right to manage,
though they are also motivated by the self-interest of the members of the acquirers management, who can use the free cash ows to increase the size of
their rm. We further argue that the input provided by accounting information
turns out to be of the utmost importance under such circumstances, as managers, investors and other nanciers participate in the governance of target rms
and in the search for ways to improve the returns to various stakeholders of
both the acquired rm and the acquirer. 5 When large investors do not have
a clear voting majority on the board, they may have to take more drastic action
such as takeover (Sloan, 2001). 6
The corporate governance mechanism changes dramatically during the periods surrounding corporate acquisition announcements, with a concomitant
increasing demand for nancial statement information by the various corporate stakeholders who evaluate the acquired rms. In such an environment,
earnings and book value may be insucient for valuation. By using other components of nancial statements, our probability summary-value measure may
provide incremental value relevance for stakeholders competing in the market
for corporate control; it may also provide relatively costless inputs for eective
corporate governance in mergers and acquisitions.
Fig. 1, adapted from Sloan (2001), describes the role of both the accounting information and the participants in the corporate governance mechanisms

5
Agrawal and Jae (2003) nd little evidence of underperformance for acquired companies,
even for sub-samples of takeovers where managers are more likely to be disciplined. Bittlingmayer
(1998) shows that if managers of the acquiring rms act on behalf of their shareholders, investors
gain the highest value, but if managers serve their own interests rather than those of shareholders,
the market for corporate control determines the means by which managers may be disciplined or
replaced.
6
Holmstrom and Kaplan (2001) argue that internal corporate governance mechanisms of
incentive-based compensation and activist boards of directors and shareholders are playing an
increasing role in mergers and acquisitions.

326

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Corporate
Governance
Mechanism

Alternative New Financiers:

1. Foreign Acquirer,
2. US Publicly Traded
Acquirer,
3. US Private Acquirer,
4. LBO.

Managers
Financial
Accounting
Input

Other Financiers:
Investors:
Insiders Shareholders,
Outsiders Large Investors,
Small Investors
Creditors:
Current Creditors,
New Creditors providing
borrowed funds to the acquirer.

Adapted from Sloan (2001).

Fig. 1. Acquired rms corporate governance mechanism during the period of acquisition.

during the period of acquisition. The right-hand side of the gure lists the
four types of potential acquirers as new players among the nanciers, distinguishing them from other participants, including new investors seeking shortterm gains and new debt holders who may lend funds to the acquirer for
nancing the acquisition. For example, most capital provided for LBOs
usually comes from borrowed funds. Following the nal acquisition, the nanciers consist of a single type of acquirer, other remaining minority shareholders, if any, current creditors, and the new creditors. The management
of the acquired rm may be retained or a new team of executives may be
appointed.
Our study focuses on the role of nancial accounting information as an input for major nanciers during the acquisition period, suggesting that acquirers
use this information to target potential acquisitions. Thus, for example, research and development intensities reported by rms acquired by foreign rms
and by publicly traded US acquirers are about 10 times greater than those reported by rms acquired either by private US acquirers or through LBOs. We
also show that the probability measure, which summarizes the role of the
accounting information in distinguishing the type of acquirer, is useful for
investors in acquired rms.
In the process of corporate governance, managers of LBO rms may benet
at the expense of other shareholders by implementing various techniques, including low-ball bidding, altering rm economic decisions, earnings decrease manipulations, and compensation shifts to minimize prices (DeAngelo et al., 1984;

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

327

Perry and Williams, 1994; Wu, 1997; Jensen, 2000). 7 Further studies show that
returns to investors in LBOs are signicantly smaller than those for investors in
publicly traded US rms acquired by publicly traded US acquirers (Torabzadeh
and Bertin, 1992; Barnes et al., 1996). 8
While LBOs and acquisitions by private acquirers are considered alternative
methods of going-private transactions (Opler and Titman, 1993), there is little
empirical evidence on the returns to investors in publicly traded US rms acquired by private acquirers (Lehn and Poulsen, 1989). We conjecture that private bidders that acquire publicly traded US rms and transfer them to the
private domain may lack the nancial depth to pay high premiums. Hence,
lower returns are expected to investors in publicly traded US takeover targets
acquired by private US acquirers compared to those obtained in takeovers by
foreign acquirers or by publicly traded US rms. The perspective of the acquired rms management in this type of acquisition is less clear. On the one
hand, managers of the target rms may resist the takeover, in particular if they
have reason to believe that their compensation or wealth will be reduced and
that a new management team may replace them. On the other hand, they
may support the acquisition if they feel that their wealth and executive position
will remain secure.
Bris and Cabolis (2002) show that when acquiring foreign rms are from
countries with better corporate governance, investors wealth increases, suggesting that cross-border mergers provide an alternative mechanism for the
contractual transfer of corporate governance. 9 While searching for undervalued rms as targets for their acquisitions, managers of foreign acquirers
have stronger incentives to obtain nancial information, and higher wealth
premium for investors is expected (Gonzalez et al., 1998). Indeed, previous
7

Kaplan (1992) shows that, between 1979 and 1986, about 63% of all LBOs remained privately
owned and about one-third either subsisted as public rms or were acquired by other publicly
traded rms.
8
LBOs and other going-private modes are manifestations of the new emerging corporations
that may manage resources more eciently than the publicly traded rms (Jensen, 1997). Further
studies show that LBOs and other going private modes possess nancial and market characteristics
which are signicantly dierent from those of other publicly traded rms (Maupin et al., 1984;
Lehn and Poulsen, 1989; Kim and Lyn, 1991), and that managers of LBOs manipulate
discretionary accrual earnings in the year prior to the announcement (Perry and Williams, 1994).
9
Black (2000) argues that the 20th century wave of US takeovers may be considered the last,
and that the growing percentage of cross-border takeovers indicates a future wave of international
mergers. Shleifer and Vishny (2001) present a theoretical model of mergers and acquisitions based
on stock market misvaluation of the rms. Among other things, the model explains who acquires
whom and what are the valuation consequences of the mergers; the model is also consistent with
prior empirical evidence about the characteristics and returns of merging rms. One empirical
prediction is that managers and shareholders of acquired rms are likely to have shorter horizons
than the acquirers. We conjuncture that this prediction indicates that the role of acquirers in the
corporate governance of acquired rms is dierent than the role of other nanciers.

328

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

studies show that managers of foreign acquirers tend to pay higher premiums for publicly traded US takeover targets than do publicly traded
US acquirers (Michel and Shaked, 1986; Shaked et al., 1991; Harris and
Ravenscraft, 1991; Kang, 1993; Dewenter, 1995; Cheng and Chan, 1995;
Eun et al., 1996; Chen and Su, 1997). Thus, investors and managers of foreign acquirers of US rms not only have reason to search for potential
acquisitions with certain attributes reported by the accounting information;
they also have strong incentives to seek and obtain more nancial accounting information. The probability measure, which is obtainable from publicly
available nancial statements at a relatively small cost, may be a useful
instrument for these stakeholders. Since the expected gains from this type
of acquisition is higher than from other types of acquisitions, foreign acquirers may have incentives to incur the additional cost involved in obtaining
such information.

4. Research methods
4.1. Univariate statistics and logistic regression
We identify several rm-specic nancial accounting attributes of acquired
rms (other than the earnings per share (EPS) and book value of equities that
are included in the price and return models), which dier across the four
types of acquirers. Several accounting and nancial variables suggested in
previous studies (e.g., Harris et al., 1982; Palepu, 1986; Opler and Titman,
1993; Hasbrouck, 1999) are selected for this purpose. These variables are presumed to be value relevant for the dierent types of acquirers and indicative
of an increase in value that might be captured in an acquisition. Consequently, we assert that these variables are useful for assessing the likelihood
of a takeover in general, and the likelihood of a takeover by one of the four
dierent types of acquirers, in particular. Specically, we assume that acquirers, investors in general, and other nanciers of acquired rms use these
nancial variables to assess the probability, PROBi,j, that rm j will not be
acquired (designated by i = 0), or that rm j will be acquired by acquirer type
i (where i = 1, . . ., 4). This probability is used as a summary-value measure of
nancial accounting information. If dierences in rm-specic attributes are
found to be signicant in a classication by type of acquiring rm, it may
illuminate the dierent motivations for acquisition of each type of potential
acquirer in the corporate governance of takeovers prior to the nal
consolidation.
Using both univariate analysis and a two-group logit regression (not tabulated), we identify a nal set of eight signicant nancial variables representing
nancial information other than EPS and book value of equities used in value-

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

329

relevance analyses. 10 Then, a ve-group logistic regression is performed using

the eight measures as explanatory variables, where the acquired rms are
grouped by the four types of acquirers: foreign, US publicly traded, US private,
and LBO, a fth group of non-acquired rms being used as a control sample.
The following ve-group logit model is used:
TDUMi;j a01 a02 a03 a04 a1  RDIj a2  RSj a3  ADIj
a4  LEVj a5  TQj a6  LIQj a7  LSIZEj
a8  GDUMj ej ;

where TDUMi, j is an empirical grouping of both non-acquired and acquired

rms that equals 0, 1, 2, 3, 4, respectively, for the control group and for the
four types of acquirers: foreign, US publicly traded, US private, and LBO;
and j = 1, . . ., Ni, where Ni is the number of rms in group i = 0, . . ., 4. 11
The independent variables are estimated using the last nancial statements
available prior to the initial acquisition announcement date. 12 The economic
hypotheses regarding the characteristics of a target rm that make it attractive
to a particular type of acquirer are articulated in the Appendix A, which distinguishes target rms in terms of their relative attractiveness to dierent types
of buyers. Based on the untabulated results reported in footnote 10, we assume that these characteristics are also useful to distinguish among the ve
groups. The denitions of the independent variables are as follows (the hypothesized expected signs of the independent variables are presented in parentheses,
while a detailed articulation of the reasons is provided in Appendix A):
RDI
RS

research and development intensity, measured as research and development expenditure/net sales (a negative sign is expected);
net income/net sales (a negative sign is expected);

10
Several two-group comparisons are performed. The major purpose of these analyses is to
identify relevant and signicant nancial variables used in prior studies. The entire research period
is examined to obtain variables that remain relatively stable from 1978 to 2001, and might be used
for the estimation and testing periods examined in the ve-group logit model. We use 22 variables
for our sample of 1578 acquired manufacturing rms and a control sample of 1777 manufacturing
rms, with at least three consecutive years of data on Compustat, that were not acquired or listed as
target rms during the research period. The control rms are randomly assigned to each sample
year without repetition, under the restriction that the same proportions of merged and control rms
from the entire samples are kept for each year, and each rm is presented in the analysis no more
than once. The nal eight variables used for the nal ve-group logit regression are those that are
signicant in the univariate comparisons between the acquired and control rms or in the twogroup logit regression or in both.
11
Similarly, TDUMi, j is a binary (0, 1) dependent variable for the respective pair of acquired
and control rms for the two-group logit regression described in footnote 10.
12
The averages of the three-year data prior to the acquisition announcement date are also used,
but not tabulated because of missing values for several rms and empirical results for the reduced
sample that resemble the ndings based on the last nancial statements.

330

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

ADI

advertising intensity, measured as advertising expenses/net sales (the

sign is not predictable);
LEV
nancial leverage, measured as total debt/total assets (a positive sign is
expected);
TQ
Tobins Q ratio, measured as (market value of equity + book value of
debt)/total assets 13 (a negative sign is expected);
LIQ
liquidity, measured as (cash + marketable securities)/total assets
(a positive sign is expected);
LSIZE rms size, measured as ln (sales) (a negative sign is expected);
GDUM Palepu, 1986) (a positive sign may be expected).

4.2. Value-relevance OLS regressions

We examine whether the probability summary-value measure of the type of
acquirer (PROBi,j) is incrementally value relevant for investors in acquired
rms, given that the impact of both earnings and book value of equities are already included in the model. We use the following price model (Collins et al.,
1997; Francis and Schipper, 1999):
P jt b0 b1 BVPSjt b2 EPSjt ejt:

2a

and then add the probability summary-value measure as follows:

P jt b0 b1 BVPSjt b2 EPSjt b3 PROBjt ejt ;

2b

where Pjt is the price per share for acquired rm j at specied dates before or
after the initial acquisition announcement date t; BVPSjt is the book value per
share for acquired rm j in the last annual nancial statements available prior
to the initial acquisition announcement date t; EPSjt is the earnings per share
for acquired rm j in the last annual nancial statements available prior to
the initial acquisition announcement date t; PROBi,j is the probability summary-value measure, generated by the logit regressions, for a non-target rm
j not being acquired (designated by i = 0), or for a takeover rm j being acquired by acquirer type i (where, i = 1, . . ., 4). PROB = [1/(1 + eax)], where a
is the vector of the estimated coecients and x is the vector of the eight independent variables described in Section 4.1.
Similarly, the return models are:
Rjt c0 c1 EPSjt c2 CEPSjt ejt ;

3a

and
13
The Q-theory of mergers suggests that the rms investment rate increases with Q and may
explain and motivate the acquisition (Jovanovic and Rousseau, 2002).

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

331

Rjt c0 c1 EPSjt c2 CEPSjt c3 PROBjt ejt ;

3b
where Rjt is the returns commencing one year prior to the initial announcement
of the jth acquisition through either one day following the acquisition
announcement date or through the last trading date prior to the acquisition;
CEPSjt is the change in annual earnings per share for acquired rm j in the last
annual nancial statements available prior to the initial acquisition announcement date t.
Other variables are as dened above, and EPS and CEPS in the return models are deated by market price per share at the beginning of the period (i.e.,
250 trading days prior to the initial announcement date).
4.3. Using the probability summary-value measure for a simple investment
strategy among acquired and non-acquired rms
PROBi,j, the probability summary-value measure of nancial accounting
information, dened in Section 4.1, may also be used to construct a trading
strategy. For this purpose we use a ve-step procedure. First, using the logistic
regression (Eq. (1)), we estimate PROBi,j for each control and for each acquired rm during the estimation period, 197890. Second, the PROBi,j values
are ranked and the outcome is divided into three probability regions: (1) low
deciles (30% of the rms, i.e., the lowest three deciles of PROB); (2) medium
(the next four deciles, from 30% to 70% of the rms); (3) high (the top three
deciles of PROB). Third, a PROBi,j value is calculated for each acquired rm
in the testing period, 19912001. Fourth, we construct three investment portfolios during the testing period. The rst portfolio consists of rms with PROB
values in the low region; the second portfolio consists of rms with PROB values in the medium region; and the third portfolio consists of rms with PROB
values in the high region. Finally, for each portfolio we calculate and report
mean (median) returns resulting from a buy-and-hold strategy for three intervals from 250 trading days prior to the initial acquisition announcements
through the nal trading date. For the testing period, we report results for
(1) all acquired rms, (2) only for rms correctly classied as acquired, and
(3) for a mixed portfolio of correctly classied acquired rms and non-target
rms misclassied as acquired rms.
5. Data
We obtained our initial list of acquired manufacturing rms from Halls
Manufacturing Sector Master File Research data (1990 and 1993), and extended it using information from the Wall Street Journal Index and Mergers
and Acquisitions (19782001). Financial accounting and market data were obtained from the Compustat and CRSP, respectively. The initial sample was

332

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Table 1
Sample selection of acquired manufacturing rms by type of acquirer: 1978 through 2001
Type of acquiring rm

Panel A: 19782001
Acquired rms with data available
on Compustata
Less: missing rms or data on CRSP
Firms with data available on
Compustat and CRSPb
Panel B: 19781990
Acquired rms with data available
on Compustata
Less: missing rms or data on CRSP
Firms with data available on
Compustat and CRSPb
Panel C: 19912001
Acquired rms with data
available on Compustata
Less: missing rms or data
on CRSP
Firms with data available on
Compustat and CRSPb

Total
rms
acquired

Foreign

US publicly
traded

US private

LBO

317

1190

217

93

1817

32
285

137
1053

53
164

17
76

239
1578

158

543

183

84

968

11
147

67
476

50
133

12
72

140
828

159

647

34

849

21

70

99

138

577

31

750

Acquired manufacturing rms with data available on Compustat for the nancial statement
variables used in our study.
b
Acquired manufacturing rms with data available on Compustat and CRSP for all variables
used in our study.

limited to rms for which acquisitions were successfully completed, that is, the
rms were fully acquired and exited the stock market. Moreover, nancial data
were available on Compustat for at least three years prior to the acquisition
announcement dates, and subsequent to the acquisition announcements prices
were available till the last day of trading. The sample selection of acquired
manufacturing rms is presented in Table 1. After excluding rms with missing
data, our nal sample, presented in Panel A, consists of 1578 acquired US
manufacturers that exited the stock exchanges between 1978 and 2001. Of
these, 285 (18%) were acquired by foreign rms, 1053 (67%) by publicly traded
US rms, 164 (10%) by private US acquirers, 14 and 76 (5%) through LBOs.
The initial control group was drawn from the population of US publicly
traded manufacturing rms that were active during the 1980s and remained

14
Private US acquirers include private rms, private investor groups and private investors, but
exclude LBOs.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

333

active throughout the year 2002. It consists of all rms with some available
data on Compustat from 1980 to 2002. These requirements yielded a control
sample of 1015 manufacturing rms. After excluding rms with missing data,
our nal control sample consists of 932 manufacturing rms.
We split our sample into two sub-periods: (1) estimation period (19781990)
for estimating the coecients (Eq. (1)) used to derive the probability summaryvalue measure, and (2) test period (19912001) for computing the probability
summary-value measure for each rm. The partition of the 828 (750) rms acquired during the estimation (test) period, among the four types of acquirers, is
presented in Panel B (Panel C). During the test period, the number of acquisitions by US publicly traded rms was considerably larger than during the estimation period, while the opposite is true for going-private acquisitions, either
by private acquirers or through LBO. 15

6. Results
6.1. Using nancial accounting input to determine the likelihood of a takeover by
type of acquirer
6.1.1. Descriptive statistics of nancial accounting variables
Comparative univariate descriptive statistics of the eight nancial variables
used to estimate the probability (PROBi,j) of rm j being acquired by an acquirer of type i in the estimation period, 19781990, are presented in Table
2. For each variable, the sample mean and median are presented across takeover targets by each type of acquirer and separately for the control rms. To
test whether for each variable the four groups and separately the ve groups
(including the control group) have equal central distributions (mean and median), we conducted ANOVA and KruskalWallis tests. Table 2 shows that
RDI, LEV, TQ, LSIZE, and GDUM dier signicantly and RS and LIQ dier
marginally across the four groups. Our comparison among the ve groups
shows signicant dierences among all variables except for ADI. These results
suggest that the variables dier across the four groups of acquired rms, not
just when the control sample of non-acquired rms is included in a-ve group
comparison. We further identify the sources of the dierence using t-tests and
Wilcoxon rank-sum tests on pair-wise comparisons (not reported).
The results (including the untabulated pair-wise comparisons) indicate that
both foreign acquirers and publicly traded US acquirers target rms having, on
the average, statistically signicant larger research and development intensity
15
Holmstrom and Kaplan (2001) provide evidence that LBOs and other going-private takeovers
signicantly decreased during the 1990s. They argue that these modes of acquisitions were no
longer needed due to the changes in corporate governance in the 1990s.

334

Table 2
Summary statistics for the explanatory variables of Eq. (1) (estimation period: 19781990)

RDIb
RSb
ADI
LEV
TQ
LIQ
LSIZE
GDUM
a

Type of acquirer
Foreign

US publicly
traded

US private

LBOs

Four-group p values
(F-test) [Kruskal
Wallis v2]

0.1041c
(0.0137)d
0.0112
(0.0345)
0.0130
(0)
0.4445
(0.4224)
1.4461
(1.1581)
0.1268
(0.0538)
10.11533
(9.7729)
0.1987
(0)

0.1045
(0.0151)
0.0002
(0.0359)
0.0129
(0)
0.4205
(0.4143)
1.6712
(1.2021)
0.1417
(0.0639)
9.2198
(8.9776)
0.2481
(0)

0.0107
(0)
0.0161
(0.0279)
0.0100
(0)
0.4903
(0.4425)
0.9332
(0.9233)
0.1108
(0.0604)
8.4401
(8.3964)
0.2824
(0)

0.0122
(0.0003)
0.0262
(0.0360)
0.0147
(0)
0.5053
(0.4623)
0.8535
(0.4619)
0.1168
(0.0640)
10.4892
(10.5956)
0.2667
(0)

(0.0004)
[0.0001]
(0.0489)
[0.394]
(0.6685)
[0.3089]
(0.0007)
[0.0007]
(0.0001)
[0.0001]
(0.0582)
[0.6048]
(0.0001)
[0.0001]
(0.0177)
[0.1365]

Control group of
non-target rms

Five-group
p values (F-test)
[KruskalWallis v2]

0.0355
(0.0138)
0.0030
(0.0138)
0.0210
(0)
0.4916
(0.4659)
1.9262
(1.3439)
0.1514
(0.0837)
11.5097
(11.5113)
0.1901
(0)

(0.0001)
[0.0006]
(0.0019)
[0.0326]
(0.5551)
[0.0517]
(0.0428)
[0.0251]
(0.0001)
[0.0001]
(0.0001)
[0.0001]
(0.0001)
[0.0001]
(0.0144)
[0.0115]

Variables are estimated based on the last nancial statements prior to the initial acquisition announcement dates. The control group includes 932
active publicly-traded manufacturing rms, obtained from the 2002 Compustat les, that were traded in the early 1980s. RDI is research and
development intensity, measured by research and development expenditures/net sales; RS is net income/net sales; ADI is advertising intensity, measured
by advertising expenses/net sales; LEV is nancial leverage, measured by total debt/total assets; TQ is Tobins Q, measured by (market value of
equity + book value of debt)/total assets; LIQ is liquidity, measured by (cash + marketable securities)/total assets; LSIZE is rms size, measured by Ln
(sales); GDUM is a mismatch growth index, measured as a growth dummy variable (Palepu, 1986).
b
Because a handful of outliers have a major impact on the means of both RDI and RS, the results reported for each of these variables are
winsorized at 1% and 99% for all rm-year observations.
c
Mean.
d
Median.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Explanatory variablea

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

335

(RDI) and Tobins Q (TQ) compared with rms taken over by private US
acquirers or through LBO. Foreign acquirers and publicly traded US acquirers
target rms having, on the average, statistically signicant larger research and
development intensity (RDI) than rms in the control group. In contrast, rms
taken over by private US acquirers or through LBO have, on the average, statistically signicant smaller RDI compared with rms in the control group.
Also, the control rms have statistically signicant larger TQ compared with
each of the four acquired groups, where rms acquired by private acquirers
or through LBO have the smallest TQ. The results also suggest that rms acquired through LBO are, on the average, larger and have higher nancial leverage than those taken over by the other types of acquirers. LBOs are, on the
average, signicantly smaller than the control rms, but have larger nancial
leverage. Private US acquirers also tend to acquire highly leveraged but smaller
rms relative to those targeted by either foreign acquirers, publicly traded US
acquirers, or by the non-target control rms. Foreign acquirers generally tend
to target larger rms than do publicly traded US acquirers and private US
acquirers. Finally, the mismatch growth index (GDUM) is signicantly larger
for rms acquired either by private US acquirers or through LBOs than for
rms acquired by either publicly traded US rms or foreign acquirers. GDUM
is signicantly smaller for the control group compared with any of the four
groups of acquired rms.
6.1.2. Five-group logit regressions
Table 3 shows the estimated coecients, the chi-square for log-likelihood,
and the Somers D for the estimation period 19781990. Six estimated coecients (RDI, LEV, TQ, LIQ, LSIZE, and GDUM) are statistically signicant,
and their signs are in the predicted directions. The overall model is highly signicant, and has a relatively high explanatory power. Overall, using cuto
points that minimize the number of misclassications, the model correctly classies 59% of the rms in the ve groups signicantly more than a random
chance of correctly classifying the outcomes among the ve groups. 16 In addition, the model correctly classies 70% of the rms as acquired or non-target
in the estimation period (88% of the acquired rms, but only 51% of the
non-target control rms). 17

16

The proportionate random ve-group classication criterion is 36.8% (9322 + 1472 + 4762 +
133 + 722)/1,7602 for the 197890 estimation period.
17
A classication of an acquired rm is considered to be correct when it is classied as an
acquisition by any of the four types of acquirers. For example, an actual acquisition by a foreign
acquirer is considered a correct classication if the model classies it as an acquisition by foreign
acquirer, by public US rm, by private US rm or through LBO; it is considered a misclassication
when classied as an observation in the non-target control group. This approach is used in the
analyses reported in Section 6.3 and in Panel C of Table 6.
2

336

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Table 3
Five-group logit regression results of Eq. (1) for non-acquired control rms and for acquired rms
classied by the four types of acquirers: 19781990a
Independent variableb
Intercept 1
Intercept 2
Intercept 3
Intercept 4
RDI
RS
ADI
LEV
TQ
LIQ
LSIZE
GDUM
Chi-square for log likelihood
Somers D

Expected signsc

()
()
(?)
(+)
()
(+)
()
(+)

Estimated coecients (19781990)

4.635**
4.174*
2.264*
0.886**
2.988*
0.995*
2.317
2.558*
1.211*
1.410*
0.284*
0.339*
667.9*
0.55

Signicant at p < 0.01.

Signicant at p < 0.05.
a
The non-acquired control rms and the acquired rms grouped by the four types of acquirersforeign, US publicly traded, US private and LBO, take the values of 0, 1, 2, 3, 4, respectively.
b
The variables are dened in Table 2.
c
Expected signs of coecients.

**

As expected, the estimated coecients for RDI, RS, TQ, and LSIZE are statistically signicant and negative, and those for LEV, LIQ, and GDUM are signicant and positive. 18 These results suggest that as research and development
intensity, Tobins Q, and rm size increase, the probability of being acquired by
foreign rms or by publicly traded US rms increases, while the probability of
being acquired by private US acquirers or through LBO decreases. Also, as the
nancial leverage, liquidity and the mismatch growth index increase, the probability of being acquired by private US acquirers or through LBO increases.
6.2. The value-relevance of the probability summary-value measure
Table 4 presents pooled cross-sectional regression results of the price model
(Eqs. (2a) and (2b)) using the sample for the test period 19912001. The regression results are reported for three alternative measures of the dependent variable (Pjt): (1) the price on the 60th trading day prior to the initial takeover
announcement date (Day-60); (2) the price at the initial takeover announce18
Signicant positive estimated coecients indicate that the probability of acquisition by
private US acquirers (group 3) and through LBO (group 4) increases. Signicant negative estimated
coecients indicate that the probability of acquisition by the control rms (group 0), foreign
acquirers (group 1) and by publicly traded US acquirers (group 2) increases.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

337

Table 4
The value relevance of accounting information: Pooled cross-sectional regressions of price on book
value per share (BVPS), earnings per share (EPS) and the probability of acquisition (PROB): 1991
2001
Trading days relative to
initial takeover announcement
date (day zero)

Estimated coecientsa
BVPS

EPS

Adjusted R2
b

PROB

Panel A: (Eq. (2a)) Model without probability measure

Day 60
0.892
3.222
(10.5)*
(8.32)*
Day 0
1.193
3.539
(10.6)*
(5.38)*
Last datec
1.307
4.286
(6.62)*
(8.75)*
Panel B: (Eq. (2b)) Model with probability measure
Day 60
0.954
(11.2)*
Day 0
1.295
(11.4)*
Last datec
1.426
(8.79)*

3.822
(8.77)*
4.463
(5.57)*
5.338
(6.32)*

0.317
0.197
0.203

20.102
(4.27)*
33.222
(3.02)*
28.687
(2.84)*

0.385
0.269
0.282

White adjusted t-statistics are in parentheses.

*
Signicant at p < 0.01.
a
Intercepts tend to be insignicant and are not reported.
b
The probability of acquisition, PROB, by type of acquiring rm is estimated based on the
logistic regression reported in Table 3. The lowest (highest) probability suggests an acquisition by
foreign rms (LBOs).
c
Last date is the last trading date.

ment date (Day 0), and (3) the price on the last trading date (Last date). Our
focus is on testing the incremental explanatory power of the additional nancial accounting information represented by PROB, the probability summaryvalue measure.
The results in Panel A indicate that for regression model (2a), the adjusted
R2s vary from 0.317, 60 days prior to the initial takeover announcement dates
to 0.197 on day 0. In comparison, the results in Panel B for the three-variable
model (Eq. (2b)) show greater explanatory power: the adjusted R2s vary from
0.385, 60 days prior to the initial announcement date, to 0.269 on day 0. The
estimated coecients for BVPS and EPS are positive (as expected) and highly
statistically signicant, and the estimated coecients for PROB are highly
signicant and negative (as expected). 19 These results suggest that PROB
19
We use a chi-square test to examine the incremental information of PROB when BVPS and
EPS variables are already included in the model. The signicance level is equal to that of a simple
t-test (the square root of the chi-square test with one degree of freedom) for testing the hypothesis
that the estimated coecient for each variable is equal to zero in the full model.

338

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Table 5
The value relevance of accounting information: Pooled cross-sectional regressions of returns on
earnings per share (EPS), changes of earnings per share (CEPS)a and the probability of acquisition
(PROB): 19912001
Trading days relative to initial
takeover announcement
date (day zero)

Estimated coecientsb
EPS

CEPS

Adjusted R2
c

PROB

Panel A: (Eq. (3a)) Model without probability measure

250 to +1
0.118
0.558
(0.33)
(1.12)
250 to Last dated
0.070
0.623
(0.12)
(1.30)
Panel B: (Eq. (3b)) Model with probability measure
250 to +1
0.615
0.255
(1.88)***
(0.53)
250 to Last dated
0.218
0.433
(0.34)
(0.93)

0.001
0.001

1.390
(3.71)*
1.021
(2.19)**

0.059
0.021

White adjusted t-statistics are in parentheses.

*
Signicant at p < 0.01.
**
Signicant at p < 0.05.
***
Signicant at p < 0.10.
a
The variables EPS and CEPS are deated by the market price per share at the beginning of the
returns interval (day 250).
b
Intercepts tend to be insignicant and are not reported.
c
The probability of acquisition, PROB, by type of acquiring rm, is estimated based on the
logistic regression reported in Table 3. The lowest (highest) probability suggests an acquisition by
foreign rms (LBOs).
d
Last date is the last trading date.

provides signicant incremental value-relevant information for investors in

general, and for each type of potential major nancier in the corporate governance of takeover targets.
Table 5 presents pooled cross-sectional regression results for the return
models (Eqs. (3a) and (3b)) using the testing period, 19912001. The regressions results are reported for two alternative return interval measures of the
dependent variable (Rjt): (1) from trading day 250 to day +1 relative to the
initial takeover announcement date, and (2) from trading day 250 to the last
trading day. Again, our focus is on testing the incremental explanatory power
of the additional nancial accounting information represented by PROB, the
probability summary-value measure, when the variables EPS and CEPS are
already included in the model.
The results in Panel A indicate that for the variables EPS and CEPS in
regression model (3a), the estimated coecients are not statistically signicant
and the adjusted R2s are only 0.001. In contrast, the regressions results in Panel
B for the three-variable model show adjusted R2s of 0.059 and 0.021 for the
two return intervals, respectively. Furthermore, the estimated coecients

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

339

for PROB are negative (as expected) and highly signicant for both return
intervals. These results suggest that PROB not only provides signicant incremental value-relevant information for each type of potential acquirer, but it is
also essentially the main signicant explanatory variable for the returns models. In summary, the results indicate that while earnings and book value of
equities alone may not provide sucient information content during the period
surrounding the acquisition, the additional nancial accounting input is value
relevant for investors in general and for each type of potential major acquirer.
Thus, the additional nancial information provides important input to nanciers participating in the corporate governance of acquired rms and to the
valuation of acquired rms.
6.3. Using the probability summary-value measure for investment strategy
Table 6 presents the mean (median) buy-and-hold returns for each of the
three portfolios resulting from implementing a buy-and-hold investment strategy in the second sample period, 19912001, as described in Section 4.3. The
results are shown for three return intervals: (1) from trading day 250 to
1, relative to the initial takeover announcement date; (2) from trading day
250 to +1; and (3) from trading day 250 to the last trading day. 20 In Panel
A we show that all acquired rms classied in the low PROB portfolio (the
three portfolios are dened in Section 4.3) gained, on the average, the highest
buy-and-hold returns for each of the three return intervals; rms classied in
the medium PROB portfolio gained, on the average, lower returns, and rms
in the high PROB portfolio gained, on the average, the lowest returns. Note
that the returns presented in Panel A also include those of acquired rms that
are misclassied as non-target rms.
Panel B shows similar results for the 580 acquired rms correctly classied
in the testing period (19912001) using a cuto point that minimizes the vegroup logistic misclassication rate in the estimation period. For example, for
the return interval 250 to Last date, the mean (median) returns are 70.3%
(48.1%), for the low PROB portfolio, 60.8% (43.2%) for the medium PROB
portfolio, and only 34.1% (38.1 %) for the high PROB portfolio. Panel C
shows similar results for a mixed portfolio of the correctly classied 580 acquired rms and a random 313 non-target control rms. The composition of
rms in this portfolio is based on applying the percentage of both correctly
classied acquired rms and misclassied non-target rms in the estimation

20
We also examine but do not report the mean (median) returns for shorter intervals
commencing on day 60. The results across the three portfolios resemble those reported in Table 6,
except that the returns are smaller.

340

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Table 6
Mean (median) buy-and-hold returns (%) by PROBa distribution: 19912001
Trading days relative
to initial takeover
announcement date
(day zero)

PROB distribution
Low (the bottom
three deciles of
the rms)

Medium (between the

fourth and seventh
deciles of the rms)

High (the top

three deciles of
the rms)

41.6*
(27.3)*
50.5*
(37.1)*
62.0*
(46.9)*

30.1*
(22.6)*
36.9*
(32.1)*
36.7*
(35.0)*

Panel B: Correctly classied acquired rms (n = 580)b

250 to 1
62.1*
45.9*
(33.6)*
(29.3)*
250 to +1
66.7*
50.5*
(44.7)*
(39.5)*
250 to Last date
70.3*
60.8*
(48.1)*
(43.2)*

31.9*
(25.6)*
34.7*
(29.3)*
34.1*
(38.1)*

Panel A: All acquired rms (n = 750)

250 to 1
59.6*
(30.2)*
250 to +1
63.6*
(41.2)*
250 to Last date
74.5*
(50.9)*

Panel C: Portfolio of correctly classied acquired rms and random non-target control rms
(n = 893)c
250 to 1
45.8*
35.3*
26.2*
(27.3)*
(24.5)*
(22.0)*
250 to +1
48.8*
38.3*
22.5*
(34.5)*
(31.1)*
(24.5)*
250 to Last date
51.1*
44.9*
27.6*
(48.1)*
(36.7)*
(30.2)*
*

Signicant at p < 0.01.

a
The probability of acquisition, PROB, by type of acquiring rm, is estimated based on the
logistic regression reported in Table 3. The lowest (highest) probability suggests an acquisition by
foreign rms (LBOs).
b
Firms correctly classied as acquired in the test period 19902001. We use a cuto point of
PROB from the estimation period based on the ve-group logistic regression and the method
explained in footnote 17.
c
Portfolio consisting of the 580 correctly classied acquired rms and a random sample of 313
non-target rms. The composition of rms in the portfolio is based on applying the percentage of
both correctly classied acquired rms and misclassied non-target rms for the estimation period,
19781990, to the testing sample in 19912001.

period, 19781990, to the testing sample in 19912001. The returns are similar but smaller than those reported in panels A and B because non-target
rms gain, on average, only 15.6 percent annual return during the 1991
2001 period.
Recall that a relatively low PROB suggests a higher likelihood that a foreign
acquirer will acquire a US publicly traded rm, whereas a high PROB implies a

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

341

Table 7
Mean buy-and-hold returns (%) for acquired rms classied by type of acquirer
Trading days relative
to initial takeover
announcement date
(day zero)

Type of acquiring rms

Foreign

US Publicly traded

US Private

LBO

Panel A: 19782001
250 to +1
250 to Last date

58.62*
66.30*

56.78*
66.11*

38.87*
46.22*

31.69*
41.80*

Panel B: 19781990
250 to +1
250 to Last date

59.79*
66.88*

55.24*
65.65*

38.26*
46.29*

32.23*
41.59*

Panel C: 19912001
250 to +1
250 to Last date

56.23*
65.72*

58.16*
66.49*

41.09**
45.94*

18.94**
45.79**

Signicant at p < 0.01.

Signicant at p < 0.05.

**

higher likelihood of a takeover through LBO. We conclude that the probability

summary-value measure, PROB, composed of eight components of nancial
data, is benecial for investment strategies in takeovers.
To check the validity of these conclusions for investment strategies, we also
calculate the ex-post mean buy-and-hold returns for four portfolios of takeover
targets classied by the type of actual acquiring rms: foreign, US publicly
traded, US private acquirers and LBOs. Table 7 shows the results for two return intervals: (1) from day 250 to day +1 relative to the announcement date;
and (2) from day 250 to the last trading day. The results are presented separately for the entire sample period 19782001 and for the two sub-periods
19781990 and 19912001. The results show signicantly higher buy-and-hold
returns for the portfolio of takeover targets acquired by either foreign rms or
publicly traded US rms compared with those acquired either by private US
acquirers or through LBO. 21 For example, the results in Panel A for the entire
sample period, 19782001, show that investors in rms acquired by either foreign rms or publicly traded US rms gained, on the average, 66% returns,
while investors in rms acquired by private US acquirers or through LBO
gained, on the average, returns of only 46.2% and 41.8%, respectively, in an
21
We also examine shorter intervals of both cumulative abnormal returns (CARs) and buy-andhold returns and nd comparable results. For example, for the 19782001 period, the 11-day CARs
surrounding the announcement are 21.9%, 21.7%, 15.4%, and 8.1% for investments in rms taken
over by foreign acquirers, US publicly traded acquirers, private US acquirers, and LBOs,
respectively. Buy-and-hold short-interval investments provide comparable but lower returns to
those returns that start at day 250 across the three portfolios.

342

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

interval from 250 trading days prior to the initial takeover announcement date
through the last day of trading. These results are consistent with those obtained
using the probability summary-value measure, PROB, for investment strategy
among takeover targets.

7. Summary and conclusions

This study examines whether nancial accounting information on US publicly traded target rms provides useful input for the corporate governance
mechanisms of these rms, facilitated by four dierent types of acquirers, during the period surrounding the acquisition. For this purpose, we examine the
value relevance of information available from the nancial statements of US
publicly traded acquired rms prior to the acquisition. Our focus is on value
relevance of the accounting information to dierent types of acquirers and
other investors in acquired rms. The acquired rms are grouped into four
types of acquirers: foreign rms, publicly traded US rms, private US acquirers, and leverage buyouts (LBOs). We expect that accounting information of
the acquired rms dier across types of acquirers and provides dierent inputs
and signals to investors. In addition, we expect that the accounting information
is useful for investment strategy in choosing among acquired rms.
Rather than focusing only on earnings and book values, which may not provide sucient information at the end of the rms life cycle, we construct a
probability summary-value measure, composed of other components of
accounting data. This measure consists of specic characteristics of accounting
information used by each type of potential major nancier in the corporate
governance of takeovers. This probability measure diers across acquired rms
classied by the four types of acquiring rms.
We use logistic regression to construct the probability summary-value measure and nd that certain rm-specic nancial accounting characteristics of the
acquired rms are useful in identifying the type of acquirer. Then, we employ
OLS price and return regressions and provide strong evidence that the probability summary-value measure is value relevant in price and return valuation
models. Finally, we use univariate statistics to examine whether the probability
summary-value measure is useful to potential investors for investment strategy
in choosing among acquired rms. We nd that the probability measure is a
very powerful investment instrument. In sum, the empirical results strongly
support our expectations.
Our test sample consists of most of the population of US publicly traded
manufacturing rms acquired in the period 1978 through 2001. In addition,

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

343

we use a control sample drawn from the population of US publicly traded

manufacturing rms that were active during the 1980s and remained active
throughout the year 2002.
In sum, our results strongly indicate that the probability summary-value
measure provides investors with incremental value-relevant information beyond that provided by earnings and book values of equities, which alone
do not provide sucient information to explain returns during the period
surrounding the acquisition announcements. Overall, the results imply that
the nancial accounting characteristics of acquired rms dier across the
four types of acquirers, providing dissimilar signals as input for eective operation of the corporate governance mechanisms of US takeovers.
Thus, our study suggests the increasing importance of nancial accounting information during takeovers. Finally the probability summary-value
measure is shown to be benecial to the investment strategies in
takeovers.

Acknowledgments
We appreciate the research assistance of Yan Bao and Dave Cannon. This
research was partially funded by the Henry Crown Institute of Business
Research in Israel at Tel Aviv University. Data were obtained from sources
identied in the paper. All potential remaining errors are of course ours.

Appendix A. Reasons for expected signs of estimated coecients of Eq. (A.1)

TDUMi;j a01 a02 a03 a1  RDIj a2  RSj a3  ADIj
a4  LEVj a5  TQj a6  LIQj a7  LSIZEj
a8  GDUMj ej :

A:1

In the multi-group logit regression, TDUM is a grouping of the acquired

rms that equals 0, 1, 2, 3 for the four types of acquirers: foreign, US publicly traded, US private, and LBO, respectively. Accordingly, signicant positive estimated coecients indicate that the probability of acquisition by
private US acquirers (group 2) and through LBO (group 3) increases. Significant negative estimated coecients indicate that the probability of acquisition by foreign rms (group 0) and by publicly traded US rms (group 1)
increases.

344

Expected
sign

Reasons for expected signs of estimated coecients of Eq. (A.1)

RDI

Negative

RS

Negative

Research and development intensities are expected to be larger in rms acquired by foreign
rms, in support of the technology transfer hypothesis (Chen and Su, 1997), and smaller for
rms targeted by private acquirers, in support of the nancial distress theory (Opler and
Titman, 1993). Eun et al. (1996) show that shareholders gains from acquisitions by foreign
acquirers are signicantly greater than for acquisitions by US publicly traded rms, and that
foreign acquirers benet from the targets R & D capabilities. Harris and Ravenscraft (1991)
show that in R&D-intensive industries takeovers by foreign rms are more frequent than
takeovers by US acquirers. Conclusion: a negative sign is expected for the estimated
coecient of RDI.
Other things being equal, rms with lower prot margins have less ecient management,
lacking the ability to control costs properly. Such rms are more likely to become acquisition
targets and have their inecient managers replaced. The inecient management hypothesis
suggests that rms with lower protability will be less likely to be acquired by foreign
acquirers than by US publicly traded acquirers (Chen and Su, 1997; Gonzalez et al., 1998)
as the former are less able to analyze and identify the true operating situation of the rm.
Incumbent management may be aware of the situation and try to avoid hostile takeover by
purchasing their rms through LBO, in support of the incentive realignment and nancial
distress theories (Opler and Titman, 1993). Further, they may benet at the expense of
other shareholders by implementing various techniques, including low-ball bidding,
altering rm economic decisions, earnings decrease manipulations, and compensation shifts
to minimize prices. Conclusion: a negative sign is expected for the estimated coecient of RS.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Variable

LEV

Positive

TQ

Negative

Prior studies do not provide a theory or empirical evidence regarding dierences in

advertising intensity across types of acquirers or its impact on the probability of acquisition
by a certain type of acquirer. Conclusion: the sign of the estimated coecient of ADI is
unpredictable.
Firms acquired through LBO are typically highly leveraged both before and during the
acquisition. Since debt is the primary source of nancing LBOs, we expect them to have a
higher nancial leverage. Based on the debt-management and the co-insurance eect
hypotheses, Chen and Su (1997) suggest that highly leveraged rms are less likely to be
acquired by foreign and US public acquirers. Foreign acquirers are typically reluctant to
acquire highly leveraged rms. Conclusion: a positive sign is expected for the estimated
coecient of LEV.
The undervaluation-target hypothesis suggests that the probability of acquiring a US publicly
traded rm by a foreign acquirer increases with increasing Tobins Q (Chen and Su, 1997;
Gonzalez et al., 1998). The incentive realignment theory, the nancial distress theory, and the
unfavorable investment opportunities hypothesis suggest that the likelihood of an LBO
decreases with increasing Tobins Q (Opler and Titman, 1993). The ineective governance of
some corporations and the nancial incentive hypotheses predict a negative sign for the
Tobins Q (Weir and Laing, 2002). Kim and Lyn (1991) nd that going-private rms have
signicantly smaller market to book ratio . Conclusion: a negative sign is expected for the
estimated coecient of TQ.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

ADI

345

346

Variable

Expected
sign

Reasons for expected signs of estimated coecients of Eq. (A.1)

LIQ

Positive

LSIZE

Negative

The enhanced liquidity hypothesis suggests that, other things being equal, rms with better
liquidity may become more attractive takeover targets. Incumbent managers, who typically
have better inside information regarding current and prospective liquidity, may be induced,
other things being equal, to acquire their rm through LBO for liquidity reasons more than
other types of acquirers. Opler and Titman (1993) nd that LBOs are motivated by free
cash ow (the incentive realignment theory). Lehn and Poulsen (1989) show that the
likelihood of going private tends to increase with greater cash ows, in support of the free
cash ow theory associated with the agency problem. Chen and Su (1997) nd that the US
rms acquired by foreign acquirers have signicantly lower liquidity than those acquired by
other publicly traded US rms. Conclusion: a positive sign is expected for the estimated
coecient of LIQ.
Chen and Su (1997) provide evidence in support of the size hypothesis that the probability
of acquisition by a foreign acquirer increases with size. Other things equal, a larger rm is
more likely to be a foreign takeover target because of the higher cost of searching for a
desirable small-rm target. Kim and Lyn (1991) nd that rms going private are signicantly
smaller than rms not going private. Conclusion: a negative sign is expected for the estimated
coecient of LSIZE.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

Appendix A (continued)

Positive

This variable represents mismatch between growth and nancial resources. A rm with
higher growth rate of sales or total assets has more protable investment opportunities and
positive earnings prospects than does a rm with lower growth rate. Chen and Su (1997)
argue that the growth potential or redeployment of corporate capital hypothesis suggests
that rms with higher growth rate tend to be more attractive takeover targets, especially
for mature acquirers. They nd insignicant empirical dierences in growth between US
publicly traded rms acquired by foreign rms and by US publicly traded rms, and provide
no support for this hypothesis in comparing the two modes of acquisitions. Lehn and
Poulsen (1989) show that the likelihood of going private tends to increase with greater
sales growth, in support of the free cash ow theory associated with the agency problem.
Kim and Lyn (1991), however, nd that rms going private have signicantly lower growth
than rms not going private, but the growth coecient is insignicant in their logistic model.
Conclusion: overall, a positive sign may be expected for the estimated coecient of GDUM.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

GDUM

347

348

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

References
Agrawal, A., Jae, J.F., 2003. Do takeover targets under-perform? Evidence from operating and
stock returns. Journal of Financial and Quantitative Analysis 38 (December), 721746.
Barnes, P., Davidson, I., Wright, M., 1996. The changing nature of corporate control and
ownership structure. Journal of Business Finance & Accounting 23 (July), 651671.
Bittlingmayer, G., 1998. The market for corporate control, including takeovers. In: Bouckaret, B.,
De Geest, G. (Eds.), Printed in the Encyclopedia of Law and Economics. Edward Elgar and the
University of Ghent.
Black, B., 2000. The rst international mergers wave and the fth (and last U.S. wave). University
of Miami Law Review 54, 799818.
Bradley, M., Desai, A., Han Kim, E., 1988. Synergistic gains from corporate acquisitions and their
division between the stockholders of target and acquiring rms. Journal of Financial
Economics, 340.
Bris, A., Cabolis, C., 2002, Corporate governance convergence by contracts: Evidence from crossborder mergers. Working paper 02-32, Yale School of Management.
Bushman, R., Smith, A., 2001. Financial accounting information and corporate governance.
Journal of Accounting and Economics 32, 237333.
Chen, C., Su, R., 1997. Do cross-border acquisitions of U.S. targets dier from U.S. domestic
takeover targets? Global Finance Journal 8 (Spring/Summer), 7182.
Cheng, L.T.W., Chan, K.C., 1995. A comparative analysis of the characteristics of international
takeovers. Journal of Business Finance & Accounting 22 (July), 637657.
Collins, D.W., Maydew, E.L., Weiss, I.S., 1997. Changes in the value-relevance of accounting
earnings and book values over the past forty years. Journal of Accounting and Economics 24,
2967.
DeAngelo, H., DeAngelo, L., Rice, E.M., 1984. Going private: Minority freeze outs and
stockholder wealth. Journal of Law and Economics, 367401.
Dewenter, K.L., 1995. Does the market react dierently to domestic and foreign takeover
announcements. Evidence from the U.S. chemical and retail industries. Journal of Financial
Economics, 421441.
Eun, C.S., Kolodny, R., Scheraga, C., 1996. Cross-border acquisitions and shareholder wealth:
Tests of the synergy and internalization hypotheses. Journal of Banking & Finance 20
(November), 15591582.
Francis, J., Schipper, K., 1999. Have nancial statements lost their relevance? Journal of
Accounting Research 37, 353386.
Gompers, P.A., Ishii, J.L., Metrick, A., 2003. Corporate governance and equity prices. The
Quarterly Journal of Economics 118 (1), 107148.
Gonzalez, P., Vasconcellos, G.M., Kish, R.J., 1998. Cross-border mergers and acquisitions: The
undervaluation hypothesis. Quarterly Review of Economics and Finance 38 (Spring), 2545.
Hall, B.H., 1990 and 1993. The manufacturing sector master le research. National Bureau of
Economic Research, Washington DC.
Harris, R.S., Ravenscraft, D., 1991. The role of acquisition in foreign direct investment: Evidence
from the US Stock Market. Journal of Finance, 825844.
Harris, R.S., Stewart, J.F., Guilkey, D.K., Carleton, W.T., 1982. Characteristics of acquired rms:
Fixed and random coecients prot analyses. Southern Economic Journal, 164184.
Hasbrouck, J., 1999. The characteristics of takeover targets: q and other measures. Journal of
Banking and Finance, 351362.
Henry, D., 2002. MERGERS why most big deals dont pay o. Business Week (October 14), 6070.
Holmstrom, B., Kaplan, S., 2001. Corporate governance and mergers activity in the US: Making
sense in the 1980s and the 1990s. Working paper, Massachusetts Institute of Technology.

J. Aharony, R. Barniv / Journal of Accounting and Public Policy 23 (2004) 321349

349

Jensen, M.J., 1984. Takeovers: Folklore and science. Harvard Business Review (NovemberDecember). Available from <http://papers.ssrn.com/ABSTRACT=350425>.
Jensen, M.J., 1997. Eclipse of the public corporation. Harvard Business Review (SeptemberOctober 1989), revised 1997. Available from <http://papers.ssrn.com/abstract=146149>.
Jensen, M.J., 2000. Active investors, LBOs and the privatization of bankruptcy. In: Jensen, M.C.
(Ed.), Theory of the rm: Governance, residual claims and organizational forms. Harvard
University Press, and Journal of Applied Corporate Finance 2 (1 Spring) 1989. Available from
<http://ssrn.com/ABSTRACT=244152>.
Jovanovic, B., Rousseau, P.L., 2002. The q-theory of mergers. Working paper 8740. The National
Bureau of Economic Research.
Kang, J., 1993. The international market for corporate control. Journal of Financial Economics,
345371.
Kaplan, S., 1992. The staying power of leveraged buyouts. Working paper 3653, The National
Bureau of Economic Research.
Kim, W.S., Lyn, E.O., 1991. Going private: Corporate restructuring under information asymmetry
and agency problems. Journal of Business Finance and Accounting, 637648.
Lehn, K., Poulsen, A., 1989. Free cash ow and stockholder gains in going private transactions.
Journal of Finance, 771787.
Maupin, R.J., Bidwell, C.M., Ortegren, A.K., 1984. An empirical investigation of the characteristics of publicly-quoted corporations which change to closely-held ownership through
management buyouts. Journal of Business Finance & Accounting 11 (Winter), 435450.
Michel, A., Shaked, I., 1986. Multinational corporations versus domestic corporations: Financial
performance and characteristics. Journal of International Business Studies, 89100.
Opler, T., Titman, S., 1993. The determinants of leveraged buyout activity: Free cash ow vs.
nancial distress costs. Journal of Finance, 19851999.
Ou, J.A., Penman, S.H., 1989. Financial statement analysis and the prediction of stock returns.
Journal of Accounting and Economics 11 (November), 95329.
Palepu, K.G., 1986. Predicting takeover targets: A methodological and empirical analysis. Journal
of Accounting and Economics 8, 335.
Perry, S., Williams, T.H., 1994. Earnings management preceding management buyout oers.
Journal of Accounting and Economics 18 (September), 157179.
Shaked, I., Michel, A., McClain, D., 1991. The foreign acquirer bonanza: Myth or reality? Journal
of Business Finance and Accounting, 431447.
Shleifer, A., Vishny, R.W., 2001. Stock market driven acquisitions. Working paper 8439. The
National Bureau of Economic Research.
Sloan, R.G., 2001. Financial accounting information and corporate governance: A discussion.
Journal of Accounting and Economics 32, 335347.
Torabzadeh, K.M., Bertin, W., 1992. Abnormal returns to stockholders of rms acquired in
business combinations and leveraged buyouts. Journal of Accounting, Auditing and Finance 7
(2), 231239.
Weir, C., Laing, D., 2002. Going private transaction and corporate governance in the U.K.
Working paper, R. Gordon University.
Wu, Y.W., 1997. Management buyout and earnings management. Journal of Accounting, Auditing
and Finance 12 (4), 373389.