SSRN Id1895473
SSRN Id1895473
SSRN Id1895473
Shuping Chen
University of Texas at Austin
[email protected]
Jake Thomas
Yale University
[email protected]
Frank Zhang
Yale University
[email protected]
We thank Shane Heitzman, Henock Louis, Phil Shane, Lakshmanan Shivakumar (editor), two
anonymous referees, and participants at workshops at London Business School, University of
Minnesota, Nanyang Technological University, University of Southern California, College of
William and Mary, and Yale University for helpful comments and suggestions. We are grateful for
financial support from the McCombs Research Excellence Fund of the University of Texas at Austin
and Yale School of Management. Brett Cantrell and Kristen Valentine provided excellent research
assistance.
Abstract
We find evidence that performance—reflected in earnings and cash flows—is transferred from
targets to acquirers around acquisitions. Using a sample of 2,128 completed deals from 1985-
2010, our results suggest that targets depress performance when investor attention declines once
the deal parameters are set, and much of that performance understatement is transferred to boost
post-acquisition acquirer performance. Evidence of variation across subsamples provides
additional confirmation: transfers are more visible for large deals (with transfers large enough to
be detected), and muted for pooling transactions (with lower incentives to transfer). We
contribute to the earnings management literature by showing that earnings and cash flows are
transferred not just within firms but also across firms, and to the mergers and acquisitions
literature by documenting that performance is managed not only before but also after deals are
announced.
Keywords: Earnings management, cash flow management, merger and acquisition, accruals
Mergers and acquisitions (M&A) are arguably the most significant investment decisions
that firms make. Research has shown that earnings management around acquisitions can have a
substantial economic impact because of the associated wealth transfers among stakeholders. This
literature focuses mainly on earnings management by acquirers and targets, before acquisitions
are announced. 1 We focus on the interim period, after acquisitions are announced but before
transactions are completed, to investigate whether targets understate earnings and cash flows,
feature of our study is that we investigate performance transfers across firms, from targets to
A Fortune article by Herb Greenberg, dated 4/1/2002 illustrates our motivation. The
article discusses various efforts by Tyco to depress target performance during the interim period,
with the intent of “spring-loading” Tyco’s post-acquisition performance. The quote below refers
to Tyco’s acquisition of Raychem, which was completed on Aug. 12, 1999, for $2.9 billion:
While the example cited above emphasizes improved post-acquisition cash flows, other
examples in the article refer to improving Tyco’s post-acquisition earnings by managing target
accruals. For example, the quote below refers to overstating Raychem’s reserves:
“One former Raychem division controller recalls how a Tyco official specifically
discussed inflating reserves. She would literally ask, ‘How high can we get these things.
1
Prior research analyzes the earnings of acquirers (e.g., Erickson and Wang 1999) and targets, both those taken over
by public acquirers (e.g., Raman et al. 2008; Anilowski et al. 2009; and Marquardt and Zur 2010) and via
management buyouts (DeAngelo 1986; Perry and Williams 1994).
1
How can we justify getting this higher?’ The employee … says he personally inflated
reserves for such items as workers' compensation, medical insurance, and pensions by at
least $10 million.”
interim-period performance understatement require that both acquirers and targets perceive the
associated benefits to exceed corresponding costs. 2 The main incentive for acquirers to engage in
performance transfers is that they get “free” performance boosts. Unlike the typical case of
another, here cumulative (over all years) acquirer performance is increased. 3 The incentives to
increase reported performance are accentuated for acquirers, as they are eager to link
The incentives for targets to understate performance are not as clear, however. While the
costs may be muted because understatement occurs during a “lame-duck” interim period after the
deal parameters are set, the benefits of cooperating with acquirers seem to also be low, especially
if there is uncertainty about target managers being retained. We note that most of the deals in our
sample are friendly, which suggests that target managers are likely to cooperate with acquirers as
their interests have been considered. If, for example, target managers forgo bonuses when
performance is understated during the interim period, it seems likely that acquirers would
provide alternative compensation in a friendly deal. On the other hand, as suggested by the
2
While purposeful understatement of performance is viewed negatively, not all earnings and cash flow management
is illegal. The SEC does not pursue cases of real earnings management—such as accelerating maintenance and
advertising. Even accruals management is often not contested, especially if it is disclosed and supported by a
reasonable basis. Some performance transfers from targets to acquirers, however, do attract the attention of the
SEC. For example, the SEC alleged that Medaphis directed a target in 1996 to create reserves during the interim
period with the intent of reversing them after the acquisition (http://www.sec.gov/litigation/admin/34-43570.htm).
3
Unconditionally, firms prefer more cumulative earnings to less for a variety of reasons. For example, most firms
reported higher cumulative earnings by electing not to expense stock option compensation when expensing was
voluntary. Reasons to boost reported performance typically rely on the assumption that the relevant stakeholders
are unlikely to adjust fully. For example, stock prices will be higher if investors do not adjust fully for overstated
cumulative performance (suggested by our evidence in Section 4.3).
2
Fortune article, target employees might resist efforts to engage in performance transfers because
This tension regarding the incentives for targets to cooperate is the main motivation to
Tyco’s behavior observed for other acquisitions? First, targets manage earnings and operating
cash flows downward during the interim period. Second, acquirer performance is boosted during
the quarters immediately following the acquisition. Third, target interim period performance
overstatement.
Figure 1 illustrates the three predictions. Assume that targets experience positive growth
for reported unmanaged performance measures before the interim quarter, indicated by the rising
profile from U-5 to U-1. Similarly, acquirers experience positive growth after deal completion,
corresponding to the profile depicted by U+1 to U+9. Per the first prediction, targets manage
performance down (from U0 to M0) during the interim quarter. Under the second prediction,
understatement of target performance is spread across the first few post-acquisition quarters
(e.g., M+1 is higher than U+1). The third prediction relates the extent of target performance
understatement (U0 – M0) during the interim quarter to the extent of acquirer overstatement
We document the following results, based on a sample of 2,128 M&A deals completed
between 1985 and 2010. Consistent with the first prediction, target earnings and operating cash
flows are substantially lower during the interim quarter. Whereas seasonally-differenced
quarterly earnings and abnormal cash flows are 0.52 percent and 2.20 percent of equity book
value in the quarter before the interim quarter, they drop to −0.06 percent and 1.27 percent
3
Our results are partially consistent with the second prediction. We find evidence
suggesting significant overstatement of earnings and analyst forecast errors, but not cash flows,
during the first post-acquisition year. This overstatement of earnings is observed for the second,
third, and fourth post-acquisition quarters. Perhaps overstatements that occur in the first post-
acquisition quarter, which includes the closing date, are offset by acquisition-related charges.
The third prediction, which links the first two predictions, is supported by the strong
positive cross-sectional relations we find between understated interim period target earnings
(cash flows) and overstated post-acquisition earnings (cash flows). These positive relations are
significant for the first and second post-acquisition quarters. In terms of economic significance,
our coefficient estimates suggest that 20.5 (23.7) cents of each dollar of earnings (cash flow)
understatement by targets during the interim quarter is used to overstate acquirer performance
during the first two post-acquisition quarters. Evidence consistent with this third prediction
provides the strongest support for performance transfers from targets to acquirers.
subsamples in the extent of transfers for two attributes. First, transfers should be muted for
transactions using the pooling method for business combinations. Because pooling combines pre-
acquisition target performance from the fiscal year of the acquisition with the acquirer’s
performance, depression of target performance before the acquisition will not boost acquirer
performance in that year. Second, our ability to detect performance transfers to acquirers should
increase with target size, relative to acquirer size. Our results support both predictions.
We recognize that there are alternative explanations for our results. For example, lower
target performance during the interim period could be due to (i) acquisition-related charges, such
targets before the interim period, intended to raise bids (e.g., Anilowski et al. 2009). And it is
4
possible that this understatement is related to higher post-acquisition performance for reasons
other than performance transfers. We find little support for these and other explanations. Overall,
we conclude that targets understate interim period performance by substantial amounts, and
acquirers benefit by reversing that understatement soon after completion of the acquisition.
Our study contributes to the earnings management and M&A literatures. First, we show
how performance management can occur across firms, not just within firms. Second, the
acquirers and targets also engage in substantial cash flow management. Finally, we highlight that
performance both before and after the deal is struck (e.g., estimates of synergy based on reported
We organize the rest of the paper as follows. Section 2 reviews relevant literature and
presents our empirical predictions. Sections 3 and 4 describe our sample and empirical results,
earnings management by acquirers and targets before the acquisition is announced. Researchers
find that acquirers manage earnings upward prior to stock-for-stock acquisitions, to increase their
stock price and reduce the number of shares issued in exchange (e.g., Erickson and Wang 1999).
Other studies have examined pre-acquisition earnings management by targets as well as the
impact of target earnings quality on different economic attributes of the acquisition. Our study
expands the focus in three ways: a) we investigate performance after, not before, acquisitions are
5
announced; b) we consider management of cash flows in addition to earnings; and c) we examine
performance transfers across firms, not just over time within the same firm.
Acquirers should in general benefit from performance transfers, especially given that
post-acquisition performance boosts are obtained from targets and do not entail subsequent
reversal. 4 Investor scrutiny of the validity of claimed synergies heightens further the incentives
for acquirers to boost post-acquisition performance. Prior research has shown that acquirers are
sensitive to this scrutiny and are willing to incur substantial costs to boost performance (Lys and
Vincent 1999), with the sensitivity increasing further if investors are skeptical about projected
synergies (Bens et al. 2012). 5 The desire to overstate performance is strongest in the period
immediately following the acquisition, because a) investor attention and links to synergies fade
over time and b) earnings are smoothed if boosted performance offsets the acquisition-related
charges recognized soon after acquisition (Graham et al 2006). Also, as a practical matter, we
expect target performance understatement to reverse relatively soon. For example, the benefits of
accelerating maintenance and advertising costs into the interim period should last for only a few
The costs to targets of understating performance during the interim period should be
relatively low, because the payments to be received by target shareholders are already set and
many contracts that rely on reported performance will either expire or be renegotiated. Also,
4
A large literature offers different motivations for firms to report higher performance (e.g., Healy and Wahlen
1999). We recognize there are instances where acquirers face incentives to suppress reported performance (e.g., to
avoid scrutiny from regulators), but those instances should be relatively infrequent.
5
Higher post-acquisition earnings appear to be an important objective when designing M&A transactions. For
example, acquirers go to great lengths, often to the extent of incurring substantial costs, to be able to account for
acquisitions under the pooling of interests method. When describing the billions in value destroyed during
AT&T’s acquisition of NCR, Lys and Vincent (1995) state: “We find that AT&T paid a documented $50 million
and possibly as much as $500 million to satisfy pooling accounting, thus boosting EPS by roughly 17% but
leaving cash flows unchanged.”
6
targets face substantially reduced scrutiny from market participants during this “lame duck”
which requires that the above-mentioned costs of understating target performance during the
interim period are lower than the benefits to target managers. One important factor that affects
these benefits is whether target managers are retained in the combined entity. Most relevant to
this question is our finding that more than 95 percent of deals in our sample are classified as
friendly takeovers. Presumably, target managers’ interests have been considered regardless of
whether or not they are retained. However, it is possible that managers and lower-level
employees that actually execute the understatement might be reluctant to participate if they feel
such performance understatement is unprofessional, and possibly even illegal. 7 This tension
motivates our empirical investigation of the three predictions, which we summarize below for
convenience:
P3 is the most important prediction, as it is easier to think of alternative explanations for P1 and
P2. Also, evidence consistent with P3 provides indirect support for P1 and P2. In addition, we
expect variation in these predictions across subsamples along the two dimensions mentioned
6
Public trading in target shares will cease soon, which reduces the incentives for market participants to monitor and
gather information on the target. The level of scrutiny by market participants and the associated costs of
understated performance are even lower during the stub portion of the interim period, because no financial
statements are filed. Reduced scrutiny by market participants and increased opportunities to manage earnings have
been shown by Du and Zhang (2013) to arise for stub periods that are created when firms change fiscal year-end..
7
Participants in experimental trust games are not only unwilling to participate if they feel “cheated”, they are
willing to pay money to punish transgressors (Pillutla and Murnighan 1996).
7
earlier. First, we expect weaker results for all three predictions for acquisitions using the pooling
method. Second, we expect stronger results for P2 and P3 for large deals, because the impact of a
small target on the performance of the combined entity will likely be too small for us to detect.
Our sample consists of 2,128 mergers and acquisitions completed between 1985 and
2010 from the Security Data Corporation (SDC) database. We require both the acquirer and
target to be U.S. corporations and the transaction value to be at least $100 million. We also
require that the acquirer obtain complete control of the target upon deal completion, to increase
the odds that acquirers are able to induce targets to cooperate and transfer performance. We
require that at least one fiscal quarter ends during the interim period (to obtain data from 10-Q
reports on target performance during the interim period) and that seasonally-differenced earnings
per share (EPS), before special items, are available for quarters in the interim period and the
quarter before. The sample drops to 1,404 mergers and acquisitions when we test P2 and P3
because we require non-missing acquirer data. We obtain financial data from Compustat, stock
Table 1 provides descriptive statistics for our sample (see Appendix for details of
variable definitions). The first row in Panel A shows that the mean (median) pre-acquisition
target market value, T_MV, in our sample is $1,738 million ($352 million), which is larger than
the corresponding mean (median) market value of $1,524 million ($88 million) for the
Compustat universe over the same 1985 to 2010 period, as reported in the second block in Panel
A. The mean (median) target book-to-market ratio, T_BM, is 0.60 (0.54), which is similar to the
corresponding values reported for the Compustat universe. The mean (median) transaction value
is $2,380 million ($488 million). Transaction values exceed pre-acquisition target market values
8
because of the premiums paid by acquirers. On average, cash (stock) represents 39 (51) percent
The mean (median) length of the interim period between announcement and completion,
Ndays, is 174 days (151 days); i.e., on average the acquisition is completed within about six
months after the acquisition is announced. Note that our requirement of a quarter-end during the
interim period eliminates transactions with relatively short gaps between announcement and
completion of the deal. Our interim period begins before the true interim period for two reasons:
a) the first quarter in the interim period includes days before the acquisition was announced, and
b) we use the date when the successful acquirer made its first bid rather than the date that the
deal parameters were settled, because the SDC database provides the former but not the latter
date.
The third and fourth blocks in Panel A provide representative levels of performance for
targets and acquirers, respectively. We describe the quarter before the interim period for targets
and the fifth quarter after the acquisition for acquirers, because target performance in the interim
period might be understated and acquirer performance in the first four post-acquisition quarters
performance management in later Tables. To maintain consistency across all our tests, we focus
on one measure of earnings—before special items—and deflate all variables by equity book
value.
Table 1, Panel B, describes the distribution of the number of quarters ending during the
interim period for our sample. For a majority of our sample (65 percent) the interim period
consists of only one quarter. For observations where the interim period spans multiple fiscal
quarters we report the average of each measure across those fiscal quarters.
9
Panel C of Table 1 describes the year-by-year distribution of acquisitions in our sample.
Consistent with prior research, the flow of transactions varies over time, increasing during the
merger booms of the mid 1980s, mid to late 1990s and mid-2000s. The intervening years,
especially the period of the recent financial crisis, are associated with fewer transactions. Given
that our minimum deal size of $100 million is not adjusted for inflation, earlier years in our
4 Empirical results
4.1 Tests of P1: target performance management during the interim period.
4.1.1 Abnormal earnings and cash flow for the interim quarter
To examine whether target firms report lower performance during the interim period, we
compare abnormal earnings and cash flows during the interim quarter with those in the quarter
before (quarter 0 versus quarter -1 in Figure 1). We believe that same-firm prior quarters
represent better controls than peer firms from the same quarter, as we may not be able to identify
peer firms that resemble targets in all relevant ways. Stated differently, the variables that explain
which firms become successful targets may be unobservable, and therefore not available to
match on.
we take two differences: we first compute seasonal differences for quarters 0 and -1, and then
investigate their difference. The seasonal difference for the quarter before (U-1 minus U-5 in
Figure 1) represents the expected growth for quarter 0, or the growth in quarter 0 if targets did
not suppress interim period performance (U0 minus U-4). If, however, targets suppress interim
period performance (say, from U0 to M0), actual growth (M0 minus U-4) observed during the
8
For P1, we also consider performance eight quarters ago rather than the four-quarter-ago benchmark implied by
seasonal differences. Similarly, for P2 and P3 we consider performance eight quarters later to calculate forward
seasonal differences. Our results are qualitatively similar to those reported in the paper.
10
interim quarter would be lower than that for the quarter before. We note that our use of earnings
before special items excludes any earnings understatement that is reflected in the one-time items
We use cross-sectional models to estimate abnormal or managed cash flows and accruals.
Specifically, we use the model in Barth et al. (2001) and Shivakumar (2006) for cash flows, and
the performance-matched Jones model (Kothari et al. 2005) for accruals. Abnormal cash flows
(T_abCFO) is given by the in-sample residuals (et) from equation (1) estimated each quarter.
𝐶𝐶𝐶𝑡 =
𝛽6 𝑂𝑂𝑂𝑂𝑂𝑡−1 + 𝑒𝑡 (1)
where CFO is cash flow from operations, ΔAR is the change in accounts receivable, ΔINV is the
change in inventory, ΔAP is the change in accounts payable, DEP is the depreciation and
amortization expense, and OTHER is the aggregate of other accruals (OTHER = Earnings – CFO
– ΔAR – ΔINV + ΔAP + DEP). All variables in this model are scaled by lagged book value of
equity. 10 Equation (1) is estimated using all firm-quarters with available data on Compustat.
subtract the discretionary accruals of a matched firm from the same two-digit SIC code and
quarter with the closest return on assets. Our estimate of discretionary accruals is the in-sample
residual (et) from the following regression, estimated each quarter for all firm-quarters with
11
𝐴𝐴𝐴𝑡 = 𝛽0 + 𝛽1 1/𝐴𝐴𝐴𝐴𝐴𝐴𝑡−1 + 𝛽2 ∆𝑆𝑆𝑆𝑆𝑆𝑡−1 + 𝛽3 𝑃𝑃𝑃𝑡−1 + 𝑒𝑡 (2)
where ΔSALES is the change in sales, and PPE is net property, plant, and equipment, both scaled
The results of our comparisons of the interim quarter and the quarter before are provided
in Table 2, with mean and median values reported in Panels A and B, respectively. We scale
seasonal differences for per share target earnings before special items (T_E) and analyst forecast
error (T_FE) by the book value of equity per share from four quarters ago (TBV). All variables
are Winsorized at the 1st and 99th percentiles of the cross-sectional distributions to mitigate the
impact of outliers. The number of observations varies across performance metrics due to data
availability.
period. The mean (median) T_∆E reported in Row 1 of Panel A (Panel B) is −0.06 (0.08) percent
of lagged equity book value per share during the interim period versus 0.52 (0.28) percent in the
quarter before, resulting in a difference of −0.58 (−0.20) percent. Not only are these differences
statistically significant, they are economically significant relative to typical levels of quarterly
earnings, which are around 2 percent of equity book values (indicated by a mean value of 2.12
percent for T_E in the third block of Table 1, Panel A). Results based on analyst forecast errors
The difference between quarters 0 and −1 for abnormal cash flows (T_abCFO) suggests
that target cash flows are also substantially understated during the interim period. 11 The mean
(median) difference is −0.93 (−0.45) percent of lagged equity book value per share, both of
which are statistically significant. They are also economically significant, relative to the mean
11
The cash flow statement was not mandatory until 1988 under SFAS95, which raises the potential for bias due to
self-selection for acquisitions where targets voluntarily provided cash flow data during the 1985-1987 period. As a
robustness check, we delete those observations and find similar results.
12
level of quarterly cash flow (T_CFO) of 5.46 percent of book value of equity, reported in the
third block of Table 1, Panel A. In contrast to cash flows, neither the mean nor median difference
between quarters 0 and −1 for abnormal accruals (T_abACC) is significantly different from zero.
These results suggest that the lower earnings reported by target firms during the interim period
are due to lower cash flows, rather than more negative accruals. One interpretation of this finding
is that the requirement to mark assets/liabilities to fair values under the purchase method limits
the scope of accruals management, and targets emphasize real earnings management (e.g.,
acquirer performance. 12
Next, we use the subsample of observations under the pooling method as a falsification
test. Under the hypothesis that target performance understatement is designed to transfer
performance to acquirers, we expect weaker evidence when pooling is used. Our results in Panel
the performance measures for large deals, which are defined as cases where the target is more
than 50% of the acquirer’s market value of equity. Panel D shows similar results to those in
Panel A.
12
To the extent that assets and liabilities are marked to fair value under the purchase method, accruals that accelerate
target expenses or defer revenues would be adjusted at acquisition, eliminating the opportunity to transfer earnings
to acquirers. To be sure, there remains room for accrual management as acquirers retain discretion in determining
fair values of non-traded assets/liabilities. In this case, acquirers could justify leaving the asset/liability values
unchanged, as the estimates are recent and made by targets, an independent party.
13
One concern regarding the pooling sample is that its smaller sample size may drive differences between Panels A
and C. To address this concern, we construct a sample with 338 purchase transactions that matches each pooling
transaction by year and relative deal size (RELSIZE), where the purchase transaction’s RELSIZE is between 80%
and 120% of that for the pooling transaction. We continue to find strong evidence of earnings and cash flow
management for this smaller purchase sample. For example, the mean differences in T_∆E and T_abCFO between
the interim period and the quarter before are −2.88% (t=−5.19) and -5.14% (t=-3.03), respectively.
13
We view our results in Table 2 as providing evidence of understated target performance
during the interim period: targets report substantially lower earnings and cash flows during the
interim quarter, relative to the quarter before. We see little evidence of earnings understatement
using accruals.
An alternative explanation for our Table 2 results is that target managers overstate
earnings in the quarter immediately preceding the interim period. That is, the results reported in
Table 2 might reflect overstated performance for quarter −1, rather than understated performance
for quarter 0. To confirm that quarter −1 is indeed an appropriate benchmark, we examine each
of the eight quarters before the interim period. In Panel A of Table 3, we tabulate the time-series
of seasonally-differenced earnings (T_∆E) and abnormal cash flows (T_abCFO) for the interim
The results from both Panels of Table 3 suggest that the findings reported for quarter −1
and cash flows in Panel A are relatively stationary in all eight quarters leading up to the interim
period. 14 Regardless of which prior quarter is considered as a benchmark, both earnings and cash
4.1.3 Target earnings management during the stub portion of the interim period
Given that no financial reports are filed with the SEC for the stub portion of the interim
period, between the date of the last filed 10K/10Q and the completion date, managers of both
14
While the sample sizes for quarters t and t-1 are the same, the number of observations declines as we go back each
quarter from t-1 to t-8. For each lag we report statistics based on the subset of targets with available data.
15
To rule out the possibility that performance in quarters −2 to −8 are also overstated, similar to overstatement in
quarter −1, we confirm that similar results are observed for earnings and cash flow surprises for up to 20 quarters
before the interim period. For example, median earnings surprises are very stable, ranging between 0.33% and
0.44% from quarter t-20 to t-9.
14
targets and acquirers likely face lower costs and higher benefits from understating stub portion
performance, relative to the more visible interim quarter. As a result, we expect considerably
greater performance understatement to occur during the stub portion. Although targets provide
no stub period disclosures, some acquirers provide partial “as-if” statements that describe sales
and earnings that would have been observed if the target had been acquired at the beginning of
the fiscal year. We can infer stub portion earnings and sales for the target as the earnings and
sales provided in the as-if disclosures minus the sum of actual earnings and sales reported for the
acquirer and target over the same period. We note that acquisitions are often associated with
divestiture of one or more divisions of the target or acquirer. If the post-acquisition as-if
disclosure relates to a smaller entity than the combination of the pre-acquisition target and
acquirer, our estimate of stub portion sales and earnings will be biased downward.
Post-acquisition filings exhibit wide variation in the format and content of such
disclosures, and each transaction has to be analyzed manually on a case-by-case basis. Given the
cost of collecting and analyzing such data, we conduct an exploratory analysis on a subsample
(the first 300 observations from our full sample, sorted by acquirer CUSIP). We find that most
acquirers do not provide the necessary “as if” data. 16 We are able to calculate stub portion
earnings and sales for 106 and 93 observations, respectively. We then obtain quarter-equivalent
numbers by dividing stub portion earnings and sales by the number of calendar days in the stub
Table 4 compares seasonal differences for earnings and sales for the stub portion of the
interim period with surprises for the interim quarter and the quarter before. The main finding is
that there is considerable understatement of income during the stub portion, even more than that
16
Firms acquiring larger targets are more likely to report “as if” numbers. The median deal size is 43.8 percent of
acquirer’s market value (in the last quarter prior to deal announcement) for firms reporting “as if” numbers,
compared with 10.37 percent for firms not reporting such numbers.
15
during the interim quarter. Specifically, the mean (median) difference between seasonally-
differenced target earnings before extraordinary items (T_∆ΝΙ) during the stub portion and the
quarter before, reported in Row 5, represents an earnings understatement of 4.96 (3.59) percent
of equity book value. Not only is that understatement substantial relative to the normal level of
quarterly earnings (approximately 2 percent of equity book value), it is many times the mean
(median) understatement during the interim quarter of 0.32 (0.32) percent of equity book value,
reported in Row 4.
A portion of this observed decline in stub portion earnings could be due to declines in
sales, rather than efforts to depress earnings, which in turn may be due to the above-mentioned
downward bias in our estimates of earnings and sales caused by divestitures. Our results in Row
5 of Table 4 suggest that sales declined slightly during the stub portion, indicated by a mean
(median) value of sales surprise difference, T_∆S, of 2.42 (2.45) percent of equity book value,
multiply the observed sales changes by a typical net income margin (ratio of net income to sales)
of 5 percent. As the resulting income impact is small, we conclude that much of the earnings
decline occurring in the stub portion is due to expense overstatement, rather than sales declines.
Observing stronger results in the stub portion than in interim quarters confirms our intuition that
the results in Table 2, which exclude the stub portion, underestimate actual performance
management.
understatement by targets during the interim period. Whereas the evidence in Table 4 suggests
that the extent of target performance understatement is even greater during the stub portion than
that reported for the interim quarter, there is an important caveat. As described earlier for interim
quarters, some of the performance reduction during the stub portion may represent acquisition-
16
related charges, not understatement. While we are able to control for such charges in interim
quarters by focusing on earnings before special items, lack of available data prevents us from
4.2 Tests of P2 and P3: Acquirer financial performance after completion of acquisition
following the closing date. Note that the acquirer’s post-acquisition performance refers to the
performance of the combined entity, as our sample is limited to transactions where the acquirer
hampered by a) the lack of a sharply defined period over which acquirer performance is likely to
be boosted; b) the absence of comparable pre-acquisition data for the combined entity; and c) the
presence of acquisition-related items, such as write-offs and amortization of goodwill, which can
challenges that arise when investigating post-acquisition performance of the combined entity.
First, performance is compared to levels from four quarters later to estimate overstated
more likely to be reflected in the first four post-acquisition quarters than in subsequent quarters,
given the heightened incentives to boost immediate performance to justify the acquisition. 18
17
The sum of the acquirer’s and target’s pre-acquisition performance is not an appropriate benchmark for post-
acquisition performance. Pre-acquisition performance is too low a benchmark because it excludes synergies
created by the acquisition. On the other hand, post-acquisition performance could be systematically lower than the
sum of the two stand-alone companies for acquisitions accounted for using the purchase method, because of
additional write-downs and amortization/depreciation associated with goodwill and assets that have been marked
up to fair values. Similarly, the divestiture of some target and acquirer assets will reduce the base of operating
assets available to generate performance after the acquisition. Finally, transactions between the acquirer and target
will be eliminated after the acquisition, further reducing post-acquisition performance.
18
We manually search 10-Qs for the quarters surrounding the acquisition for the first 100 acquirers (sorted by
CUSIP) in our sample. We find that the majority of the acquisition-related charges are taken in quarter +1, the first
quarter after the completion date. For example, the median value of acquisition-related charges is $23.2 million in
quarter +1, compared to $7.25 million in quarter +2 and $4.1 million in quarter +3.
17
These design choices have the following implications for our results. Because of normal
growth, subtracting future levels of earnings from current levels results in negative values, which
suggests performance declines to the casual reader. To avoid reporting negative values, we use
“inverted seasonal differences”, which equal performance in quarter q+4 minus that in quarter q.
indicate lower inverted seasonal differences (U+5 minus M+1) relative to the case of no
The second design choice—assuming that performance boosting is limited to the first
four post-acquisition quarters—allows us to use the next four quarters, from +5 through +8, as an
observable proxy for the unobservable, no overstatement case. That is, average inverted seasonal
differences observed for quarters +5 through +8 (U+9 minus U+5 through U+12 minus U+8)
proxy for the no overstatement case (U+5 minus U+1 through U+8 minus U+4) for quarters +1
through +4. Evidence of performance overstatement will be indicated by lower inverted seasonal
differences than those for the benchmark quarters. If performance overstatement continues past
quarter +4, which is quite likely, this second design choice biases against observing acquirer
and abnormal cash flows for acquirers over the first four quarters following the acquisition. The
average of the benchmark quarters (quarters +5 through +8) is provided in the bottom row. Panel
A presents results for all firms in our sample, whereas Panel B presents results for 518 large
19
For example, we should not find significant results for P2 and P3 if performance overstatement is evenly
distributed across quarters +1 to +8.
18
deals, where deal values exceed 50 percent of acquirer market value. Panel C presents results on
the subsample of large deals that used pooling to account for the acquisition.
We expect the results to be stronger in Panel B than Panel A because we are more likely
to detect the impact on acquirer post-acquisition performance for larger targets with larger
pooling limits the opportunity to avail of “free” earnings and cash flows from the target. Note we
equity per share (CBV) because pre-acquisition book values for the combined entity are not
available.
Results from Table 5 are partially consistent with P2. First, results in both columns (1)
and (2) suggest overstated earnings in the first post-acquisition year. Compared to the benchmark
quarters (+5 to +8), the combined entity’s seasonally differenced earnings are overstated in
quarters +2 and +4. Also, analyst forecast errors (C_FE) are less negative for quarters +2, +3,
and +4. Not only are earnings reported in the first year overstated relative to benchmarks derived
from earnings reported in the next year, they are overstated relative to benchmarks based on
analyst forecasts. Apparently, analysts did not fully anticipate these overstatements. Second,
results reported in column (3) for abnormal cash flows (C_abCFO) do not support P2. 20 None of
the C_abCFO’s in the first four quarters are significantly higher than the benchmark quarters.
We find this evidence puzzling given that targets understate both earnings and cash flows. The
20
As the Barth et al. (2001) model requires lagged cash flows to estimate abnormal cash flows, C_abCFO is
associated with additional measurement error in the first post-acquisition quarter. In untabulated analyses, we use
inverted seasonally differenced cash flows (C_∆CFO) to measure abnormal cash flows, using the same approach
followed for seasonally differenced earnings (C_∆E), and find similar results.
19
Third, as expected, overstatement is greater for large deals (Panel B) and the results in
Panel C for the pooling subsample are largely insignificant. These results suggest that evidence
transactions using the purchase method, and the subset using pooling plays a limited role. 21
Finally, we note that evidence consistent with P2 is not observed in quarter +1 in either Panels A
or B. One possibility is that the first post-acquisition quarter includes substantial acquisition-
related charges that are not separately identified by Compustat as special items (see footnote 17).
Our key prediction, P3, links cross-sectional variation in the first two predictions: greater
target understatement during the interim period surfaces as greater acquirer overstatement after
the acquisition. The results reported in Table 6 describe regressions of the three performance
measures for the combined entity analyzed in Table 5 on target performance understatement
during the interim quarter, in the presence of control variables. Panels A1 and A2 relate to the
first and second post-acquisition quarters for all acquisitions, and Panels B1 and B2 report results
on the subsample of large deals. 22 Target performance understatement (T_∆EU and T_abCFOU)
The scaling variable for target performance, end-of-quarter equity book value per share
for the combined entity (CBV), is the same as that for the dependent variables measuring acquirer
21
The extent of overstatement for each quarter is the excess of the benchmark in the bottom row over the respective
performance measures for that quarter. For example, the excess amount for quarter +4 in column 1 is 0.40 percent
of equity book value in Panel A (=0.54% − 0.14%), and 0.96 percent in Panel B (=0.99% − 0.03%). Summing up
the corresponding estimates across quarters +2 to +4 suggests that the total overstatement by acquirers is
economically significant, especially for the larger targets described in Panel B: the mean level of quarterly
earnings before special items reported in Table 1 is 2.32%. The results reported below in Section 4.2.2 offer an
alternative indication of economic significance by comparing magnitudes of acquirer performance overstatement
with levels of target performance understatement.
22
Results for the third and fourth post-acquisition quarters are typically in the same direction, but often not
statistically significant.
20
performance. This approach allows us to interpret the coefficients on target understatement as the
combined entity.
We include as control variables the fraction of deal value paid as stock (PctStock),
whether the acquisition was accounted for as a pooling of interests (POOLING), the logarithm of
deal value [Log(DEAL)], deal value relative to that of the acquirer (RELSIZE), the book-to-
market ratio of the acquirer (A_BM), and abnormal annual accruals made by the acquirer over the
should be positive in the C_∆E regression and negative in the C_FE regression, while the
coefficient estimates on cash flow understatement (T_abCFO) should be negative. More negative
values of target earnings and cash flows during the interim quarter should be associated with
higher post-acquisition acquirer earnings and cash flows, as captured by less positive inverted
differenced earnings and higher analyst forecast errors and abnormal cash flows. To assist
The results in Panel A1 and A2, relating to the first and second post-acquisition quarters
for all observations, are largely consistent with P3. The coefficients on targets’ earnings
understatements (T_∆EU) are significantly positive based on C_∆E in both panels. The results
based on analyst forecast errors for the combined entity (C_FE) yield similar inferences. Finally,
the combined entity’s cash flow performance is also significantly positively associated with
targets’ cash flow understatement during the interim period (T_abCFOU) in Panel A1, but not in
Panel A2.
23
The results are robust to calculating abnormal accruals over one year or two years before the acquisition. In each
case, the coefficients on abnormal accruals are statistically insignificant.
21
The sum of the coefficients on interim quarter target understatement across Panels A1
and A2 reflect the fraction of that understatement that appears as acquirer performance
overstatement in the first and second post-acquisition quarters. That sum is 20.5 percent and 23.7
percent for earnings and abnormal cash flows, respectively. These computations are approximate
as we exclude target understatement from the stub portion as well as acquirer overstatement after
quarter +2. Moreover, as mentioned earlier, these estimates are biased down for a variety of
reasons (e.g., our measures of target performance understatement are noisy, measures of acquirer
performance overstatement are biased down, and we do not know when target understatement
will reverse and affect combined firm performance). While these constraints limit our ability to
provide a complete documentation of performance transfers, our results still indicate that a
acquirers.
Panels B1 and B2 tabulate the regression results for the subsample of 518 large deals. For
parsimony we only present the results for the key variables of interest, T_∆EU and T_ abCFOU.
These two panels yield even stronger results than those in Panels A1 and A2: the coefficients on
these variables suggest that 36.2 and 30.5 cents for each dollar of the target’s earnings and cash
flow understatement are reflected in the first two quarters in the post-acquisition period.
Finally, we turn to the coefficients on the interaction term between POOLING (set to 1
for pooling and 0 for purchase) and target underperformance measures (T_∆EU and T_
abCFOU) in all four subpanels of Table 6. If performance transfers are limited for transactions
using the pooling method we expect no relation between target performance understatement and
acquirer performance overstatement; i.e., we expect the sum of the coefficients on T_∆EU and
22
Consistent with that expectation, we find the interaction term coefficients generally have the
opposite signs to those on the corresponding target underperformance measures. F-tests confirm
that the coefficients measuring performance transfers from targets to acquirers are generally
The results in Table 6 not only support P3, they also provide support for P1 and P2. This
additional support is more relevant for the two areas of performance overstatement not supported
by the results in Table 5: earnings performance for the first post-acquisition quarter and cash
flow performance for all four post-acquisition quarters. Whereas acquirers’ average levels of
earnings surprises in quarter +1 and cash flow surprises in quarters +1 through +4 are not higher
earnings and cash flows are significantly associated with target interim-period performance
understatement. One explanation is that earnings in quarter +1 and cash flows in all four quarters
are lower after the acquisition (due to severance payments and other acquisition-related
activities). In Table 6, this lower level of earnings and cash flows is absorbed by the intercept,
Overall, the evidence regarding P1, P2, and P3 is consistent with the thesis posited by the
understatement from targets during the interim period. The one notable exception is our evidence
regarding cash flows, which does not support P2. Additional confirmation of this thesis is
provided by our finding that such performance transfers are more evident in larger deals but less
evident in acquisitions using the pooling method (when incentives to transfer are lower).
One motivation proposed for performance transfers is that investors pay little attention to
target performance during the interim period, and do not adjust fully to the portion of reported
23
post-acquisition earnings news that is overstatement based on performance transfers from targets.
That is, investor responses are only weakly related to forecast errors during the interim period,
following the acquisition, when they should ignore that portion of forecast errors.
Table 7 contains estimates obtained from regressing three-day market adjusted returns
(ARET) centered on relevant quarterly earnings announcement dates on earnings surprises (T_FE
and C_FE) of the respective quarters for targets in Panel A and acquirers in Panel B. Consistent
with investors ignoring forecast errors during the interim period, the coefficient on T_FE, which
is significant in the quarter before the interim quarter (left half of Panel A), is insignificant
In Panel B of Table 7, we regress ARET on forecast error for acquirers (C_FE) and our
acquirer performance overstatement in the first and second post-acquisition quarters. If investors
do not adjust for performance transfers (proxied by T_∆EU), they will respond fully to the
reported forecast error, C_FE, which includes both the true earnings surprise and the transferred
portion, and the coefficient on T_∆EU should be zero. However, if investors adjust for earnings
transfers and respond only to the true earnings surprise, the coefficient on T_ΔEU should be of a
similar magnitude to the coefficient on C_FE, but of opposite sign. Our results, which indicate
insignificant coefficients on T_∆EU for both the first and second post-acquisition quarters, imply
that the market does not adjust fully for earnings transfers from the target to the acquirer. These
results suggest a powerful incentive for acquirers to transfer performance from targets: they are
24
5. Alternative explanations and robustness checks
understatement represents a reversal of overstatement that occurs before the announcement date.
Our evidence is inconsistent with this explanation. First, if interim period target performance is
simply the reversal of prior overstatement, we should not observe post-acquisition performance
overstatement (P2), nor should we find that such overstatement is explained by interim period
Second, we investigate if our P1 results are robust to explicit controls for any reversal of
where ΔE is seasonally differenced target earnings, and D is an indicator variable taking the
Equation (3) is an alternative approach to the analysis underlying the results in Table 3
(left panel). The coefficient β1 in equation (3) describes the difference in ΔE between the interim
quarter and the quarters before, after controlling for autocorrelation in seasonally differenced
earnings. A significantly negative estimate for β1 in equation (3) suggests that our results
regarding P1 are robust to including this control for autocorrelation. A negative coefficient β3 in
equation (4) indicates incremental reversals of earnings overstatement that occurs in the quarter
before the interim quarter. As an additional robustness check, we expand equations (3) and (4) to
25
include three more lagged values of ΔE (ΔEt-2, ΔEt-3, and ΔEt-4) to test reversals from the prior
four quarters.
Table 8 reports the regression results. In all four models the coefficient β1 is significantly
negative, indicating target earnings understatement ranging from 0.75 to 0.85 percent of equity
book value. In addition, any reversal of potential prior overstatement accounts for about 1%
Models (3) and (4), we expand the regressions by adding three more ΔE lags. The estimate for β1
drops from −0.749 to −0.769 from Model (3) to Model (4). The results in Table 8 suggest that
the observed decline in target performance during the interim period is not due to reversal of
levels decline, due to factors such as reduced target manager effort, divestiture of target
divisions, and customers deferring purchases until after the acquisition is completed. If so, we
sales. To determine the portion of the earnings reduction that is explained by sales declines, we
multiply sales declines by the net income margin, or ratio of net income to sales, which is about
explanation. The mean difference between quarters 0 and −1 for T_∆S in Panel A is −0.10
percent of equity book value, which is statistically insignificant. The earnings impact of that
with the observed earnings understatement of about 0.58 percent of equity book value. These
26
results suggest that the substantially lower interim period target performance we document is
unlikely to be due to a corresponding decline in activity levels, which effectively rules out
alternative explanations based on factors such as decreased target management effort, divestiture
Another alternative explanation for declining target performance is that sales levels do
not decline but costs increase during the interim period. We anticipate a substantial increase in
intermediaries. It is also possible that recurring costs might rise during the interim period. For
example, suppliers might raise prices on goods and services consumed by the target because
suppliers with bargaining power, with limited incentives to maintain long-term relations, engage
Our performance measures, such as earnings before special items and analyst forecast
errors, exclude such non-recurring costs. More importantly, this alternative explanation makes no
predictions for post-acquisition performance (P2 and P3), which is a key focus of this paper.
Acquisitions are often associated with restructuring activities. For cases where target
restructuring is known when acquisitions are announced, targets may have created an allowance
for this restructuring in the pre-acquisition period. On the other hand, in cases where
restructuring arrangements are not agreed upon early on, restructuring charges could be observed
after the merger, causing a negative cross-sectional relationship between pre-acquisition target’s
earnings and post-acquisition earnings. This explanation should not affect our results, as they are
based on earnings before special items and analyst forecast errors, which typically exclude
27
restructuring charges. It is possible, however, that these earnings measures include some
restructuring charges.
It is possible that there is cross-sectional variation in the extent to which acquirers seek to
transfer performance from targets, and acquirers that do (do not) transfer performance do not
(do) provide supplemental disclosures that allow investors to gain a better understanding of the
underlying situation. If so, acquirers that provide pro forma disclosures in their post-acquisition
10-K footnotes should exhibit less earnings management. Consistent with this prediction, we find
that the 106 targets from our hand-collected sample of 300 firms for which acquirers provided
as-if disclosures report interim quarter earnings that exhibit only moderate declines relative to
the quarter before (mean of 0.32 percent of equity book value, as reported in row 4 of Table 4),
relative to the much higher declines reported for our overall sample (mean of 0.58% in row 1 of
Table 2).
Similarly, less disclosure can enable more performance transfers. If acquirers that
complete acquisitions earlier during the fiscal year provide less information about targets’ pre-
acquisition performance, they may have an opportunity to engage in more target performance
understatement. We partition our sample into three terciles based on the gap between the closing
date and the fiscal year-end. Consistent with our expectation, we find that the magnitude of
earnings understatement is substantially larger for the early tercile and then declines to a lower
level for the other two terciles. Specifically, the average declines in T_∆E equal 0.82 percent of
equity book value for the early tercile, and 0.43 and 0.50 percent for the middle and late terciles.
Finally, scrutiny of the acquirer’s financial statements might be higher if the target and
the acquirer are audited by the same firm (even if different offices are involved), which should
28
pairs that have the same auditor with pairs using different auditors. We find more performance
management for acquirer/target pairs with different auditors. For example, the mean difference in
T_∆E between the interim quarter and the quarter before is −0.65 percent of equity book value
for pairs with different auditors, which is statistically significant at the 5 percent level. In
contrast, the mean different in T_∆E is −0.48 percent and statistically insignificant for pairs
Taken together, the results of these additional analyses of variation across different
subsamples in target performance during the interim period support the view that in many cases
We use earnings before special items in our main analysis. As a robustness check, we
consider earnings before extraordinary items and earnings from I/B/E/S as alternative earnings
measures. The results are largely similar with one exception: Earnings before extraordinary items
tends to be much lower in the first post-acquisition quarter relative to four quarters later, as
acquisition-related charges are included as special items and reflected in this earnings measure.
We also observe similar results for alternative, simpler measures of unexpected accruals and cash
flows.
6 Conclusions
In this paper we investigate whether targets understate earnings and cash flows during the
interim period with the intent to transfer that understatement to boost post-acquisition
performance of the combined entity. The interim period extends from announcement of the
acquisition, when the deal terms are set, to formal completion of the deal. This type of
performance management is unique in the sense that the acquirer has opportunities to avail itself
29
of “free” earnings and cash flows transferred from another firm to boost its performance. While
there is anecdotal evidence suggesting that targets’ interim period income and cash flow are
understated and then transferred to acquirers, no empirical study has systematically examined
Our results, based on a sample of 2,128 acquisitions completed over a 26-year period
from 1985-2010, suggest that targets understate their interim period performance, and the
significant. Our overall evidence is inconsistent with alternative explanations for target
performance understatement, such as the reversal of earnings overstatement that targets might
engage in before the acquisition is announced, target managers’ decreased effort after merger
More important, our results suggest that target performance understatement assists
acquirers in boosting their performance after the acquisition. While empirical challenges limit
Additional support for performance transfers from targets to acquirers is provided by our finding
that the results are stronger for larger deals and for acquisitions accounted for using the purchase
method. Overall, the magnitudes of performance transfers are economically significant, relative
to both the magnitude of target understatement as well as levels of quarterly earnings and cash
flows.
Our results extend prior research on performance management around mergers and
acquisitions in two ways. While that research has focused on transfers of earnings before the
30
announced, and consider both cash flows and earnings. A novel contribution of our study is that
we document performance transfers across firms, whereas prior studies typically investigate
transfers of earnings across years within the same firm. Future research can shed further light on
cross-firm earnings management by investigating managerial incentives and the costs and
31
Appendix: Variable definitions
(Quarterly COMPUSTAT variable names are provided in parentheses under Description)
Variables Description
Acquirer’s book-to-market ratio at the end of the last fiscal quarter ending before the
A_BM
completion date.
Three day earnings announcement return centered on earnings announcement dates,
ARET
measured as raw returns minus value-weighted market returns.
Combined entity’s book value of equity per share (CEQQ/CSHOQ*AJEXQ), where
CBV*
combined entity represents the acquirer after the deal is completed.
CFO Cash flow from operations per share (OANCFY/CSHOQ*AJEXQ)
The combined entity’s abnormal cash flow, based on the residual from the Barth et al.
(2001) model, adjusted to deflate all variables by lagged equity book value. These
C_abCFO
residuals are multiplied by lagged equity book value and then deflated by outstanding
shares * CBV.
The combined entity’s inverted seasonal difference for earnings before special items,
C_∆E*
measured as (Eq+4 – Eq)/CBVq.
Combined entity’s EPS analyst forecast error (I/B/E/S actual minus median consensus
C_FE
forecast as of the last month of the fiscal quarter) scaled by CBV.
DEAL Deal size, value paid for target (in millions of dollars)
Earnings before tax-adjusted special items per share
E
[IBQ-SPIQ*(1-35%))/CSHOQ*AJEXQ]
number of days in the interim period, between the acquisition announcement date and
Ndays
the completion date
PctCash The percentage of cash to total consideration paid for target.
PctStock The percentage of stock to total consideration paid for target.
POOLING Indicator variable=1 if acquisition accounted for as pooling, and 0 otherwise
Abnormal accruals made by the acquirer during the 3 years prior to the acquisition,
computed as average abnormal annual accruals, where abnormal accruals is the
PRE_ACC
modified Jones model estimated for each 2-digit SIC industry. The deflator is the
lagged total assets for all variables in the modified Jones model.
Size of deal relative to acquirer market value at the end of the last fiscal quarter before
RELSIZE
completion of the acquisition.
S Sales per share (SALEQ/CSHOQ*AJEXQ)
TBV Target’s book value of equity per share (CEQQ/CSHOQ*AJEXQ)
Target’s abnormal accruals, measured as industry-adjusted and performance-matched
T_abACC abnormal accruals based on the Jones model (Kothari et al. 2005). The deflator is the
lagged book value of equity for all variables in the model.
Target’s abnormal cash flow, which is the residual of the model in Barth et al. (2001).
T_abCFO
The deflator is the lagged book value of equity for all variables in the model.
Measure of target cash flow understatement in interim period, defined as the average
target’s abnormal cash flow surprise in dollar amount (T_abCFO) for quarters during
T_abCFOU the interim period minus that for the quarter before the acquisition announcement date
multiplied by the number of quarters in the interim period, scaled by outstanding shares
* CBVq.
T_BM Target’s book-to-market ratio at the end of the quarter before the acquisition
32
announcement date.
T_∆E Target’s seasonally differenced earnings, measured as (Eq – Eq-4)/TBVq-4
Measure of target’s income understatement in interim period, defined as the average
target’s seasonally-differenced earnings before special items (Eq – Eq-4) for the interim
T_∆EU
period minus that for the quarter before the acquisition announcement date multiplied
by the number of quarters in the interim period, scaled by outstanding shares * CBVq.
Target’s analyst forecast error (I/B/E/S actual minus forecast) scaled by book value of
equity per share (TBV) four quarters ago, where the forecast is the average of
T_FE
individual forecasts made in the last month of the fiscal quarter, obtained from I/B/E/S
detail files.
Target’s market value of equity (in millions of dollars) at the end of the quarter before
T_MV
the acquisition announcement date. (CSHOQ*PRCCQ)
T_∆NI Target’s seasonally-differenced net income, measured as (NIq – NIq-4)/TBVq-4
T_∆S Target’s seasonally-differenced sales, measured as (Sq – Sq-4)/TBVq−4
* The combined entity (C_) refers to the acquirer after the deal is completed. For the combined entity, the
performance measures for cash flows (CFO) and different earnings proxies are inverted because we cannot use
performance from four quarters ago as a benchmark for the first four post-acquisition quarters (as the acquirer is not
comparable before and after the acquisition). In response, we compute seasonal difference (∆) for quarter q by
comparing quarter q with quarter q+4, rather than quarter q-4.
33
References
Anilowski, C.L., A. J. Macias, and J.M. Sanchez, 2009. Target firm earnings management and
the method of sale: evidence from auctions and negotiations, Purdue University and
University of Arkansas, working paper.
Barth, M., D. Cram, and K. Nelson. 2001. Accruals and the prediction of future cash flow. The
Accounting Review 76, 27-58.
Bens, D. A., T. H. Goodman, and M. Neamtiu. 2012. Does investment-related pressure lead to
misreporting? An analysis of reporting following M&A transactions. The Accounting
Review. 87 (3) 839-865).
DeAngelo, L., 1986. Accounting numbers as market valuation substitutes: A study of
management buyouts of public stockholders. The Accounting Review 67: 77-96.
Du, K. and F. Zhang. 2013. Orphans deserve attention: Financial reporting in the missing months
when corporations change fiscal year. The Accounting Review (forthcoming).
Erickson, M., and S. Wang. 1999. Earnings management by acquiring firms in stock for stock
mergers. Journal of Accounting & Economics 27: 149-167.
Graham, J., C. Harvey, and S. Rajgopal, 2006. Value destruction and financial reporting
decisions. Financial Analysts Journal: 62 (2): 27-39.
Healy P.M. and J.M. Wahlen. 1999. A review of the earnings management literature and its
implications for standard setting. Accounting Horizons, 365-383.
Kothari, S.P., A. J. Leone, and C. E. Wasley. 2005. Performance matched discretionary accrual
measures. Journal of Accounting & Economics 39 (1): 163-197.
Lys, T., L. Vincent. 1995. An analysis of value destruction in AT&T’s acquisition of NCR.
Journal of Financial Economics: 39:353-378.
Marquardt, C., and E. Zur. 2010. The role of accounting quality in the M&A market. Baruch
College, CUNY, working paper.
Perry, S., and T.H.Williams. 1994. Earnings management preceding management buyout offers.
Journal of Accounting & Economics 18: 157-179.
Pillutla, M. M., and J, K. Murnighan. 1996. Unfairness, anger, and spite: Emotional rejection of
ultimatum offers. Organizational Behavior and Human Decision Processes 68: 208-224.
Pourciau, S. 1991. Earnings management and nonroutine executive changes. Journal of
Accounting & Economics 16: 317-336.
Raman, K., L. Shivakumar, and A. Tamayo, 2008. Targets’ earnings quality and bidders’
takeover decisions, Bentley College and London Business School, working paper.
Shivakumar, L. 2006. Accruals, cash flows and the post-earnings-announcement drift. Journal of
Business, Finance, and Accounting 33, 1-25.
34
Table 1 Description of acquisition transactions
Panel A: Descriptive statistics
Mean Std. dev. Min. Q1 Median Q3 Max.
General data relating to the acquisition
Performance of the combined entity in the fifth quarter after the acquisition
Panel B: Distribution of the number of fiscal quarters ending during the interim period
The number of quarters Frequency Percent
1 1,375 64.61%
2 515 24.20%
3 143 6.72%
4 53 2.49%
35
Table 1 (continued)
Panel C: Distribution of the number of acquisitions by year
Year Frequency Percent Year Frequency Percent
The sample consists of 2,128 mergers and acquisitions completed between 1985 and 2010 covered by Security Data
Corporation, that satisfy the following conditions: both the acquirer and target are U.S firms, transaction values
exceed $100 million, the acquirer achieves complete control of the target, a fiscal quarter for the target ends during
the interim period (between announcement and completion of the deal) and seasonally-differenced target earnings
are non-missing for both the interim quarter and the quarter before. The data reported in Panel A for the Compustat
population are based on the same 1985 to 2010 period. The prefixes T_ and C_ in the variable names refer to the
target before the acquisition and the combined entity (acquirer) after the acquisition. MV and BM refer to market
value of equity and book-to-market ratio of equity. PctCash and PctStock refer to the percent of total consideration
paid as cash and stock, and Ndays refers to the number of days between the acquisition announcement date and
completion date. E and CFO refer to per share earnings before special items and cash flows from operations,
respectively. Additional details of variables are provided in the Appendix. All variables except for MV, PctCash,
PctStock, and Ndays are Winsorized at the 1st and 99th percentiles of their respective cross-sectional distributions.
36
Table 2 Target performance during the interim quarter versus the quarter before
Panel A: Mean performance
Predicted
Variable N Interim Quarter Quarter Before Difference (t-statistic)
Difference
T_∆E − 2,128 -0.06% 0.52% -0.58%*** (-2.70)
T_FE − 1,119 -0.39% -0.17% -0.22%** (-2.19)
T_abCFO ? 1,663 1.27% 2.20% -0.93%* (-1.72)
T_abACC ? 1,350 -0.60% -0.70% 0.10% (0.09)
Panel D: Mean performance of observations for large deals (>50% of acquirer market value)
Predicted
Variable N Interim Quarter Quarter Before Difference (t-statistic)
Difference
T_∆E − 539 0.16% 0.69% -0.53%** (-2.25)
T_FE − 438 -0.33% -0.02% -0.31%* (-1.80)
T_abCFO ? 365 2.55% 3.40% -0.85%* (-1.59)
T_abACC ? 299 -0.37% -1.82% 1.45% (0.77)
*, ** and *** indicate significant difference at the 10%, 5%, and 1% level, respectively, based on one-tailed tests.
This table compares performance measures for the interim quarter with those for the quarter before, where the
interim quarter includes all quarters ending between the announcement and completion of the acquisition. For
observations with more than one interim quarter, we report averages across those quarters. “Difference” equals the
interim quarter value of the measure minus that for the quarter before, with t-statistics and z-statistics reported in
parentheses for means and medians, respectively. The sample consists of 2,128 mergers and acquisitions between
1985 and 2010. The prefix T_ in the variable names refers to the target. The variable T_∆E refers to seasonal
differences in (per share) earnings before special items. T_FE is the forecast error relative to the consensus EPS
forecast as of the last month of the fiscal quarter. T_abCFO and T_abACC represent targets’ abnormal cash flows
and abnormal accruals, which are residuals generated from the cross-sectional Barth et al. (2001) model and the
performance-matched Jones model of Kothari et al. (2005), respectively. See Appendix for details of variables. All
variables are Winsorized at the 1st and 99th percentiles of their respective cross-sectional distributions.
37
Table 3 Target performance during the interim quarter and eight quarters before
*, ** and *** indicate a significant difference relative to the interim quarter performance reported in the bottom row
at the 10%, 5%, and 1% level, respectively, based on one-tailed tests.
This table compares two measures of target performance—seasonally differenced per share earnings before special
items (T_∆E) and abnormal operating cash flows (T_abCFO)—for the interim quarter with performance for the
prior eight quarters. The interim quarter includes all quarters ending between the announcement and completion of
the acquisition. For observations with more than one interim quarter, we report averages across those quarters.
T_abCFO is the residual from the cross-sectional Barth et al. (2001) model. See Appendix for details of variables.
The sample consists of 2,128 mergers and acquisitions between 1985 and 2010. All variables are Winsorized at the
1st and 99th percentiles of the cross-sectional distribution each period.
38
Table 4 Earnings and sales performance during the stub portion of the interim period:
a subsample analysis
T_∆NI T_∆S
Row
N Mean Median N Mean Median
The t-statistics associated with differences reported in the bottom three rows are shown below in parentheses, and *,
** and *** indicate significant differences at the 10%, 5%, and 1% level, respectively, based on one-tailed tests.
The top three rows in this table report the mean and median values of seasonally-differenced net income (T_∆NI/BV)
and sales (T_∆S/BV) for the quarter before, the interim quarter, and the stub portion of the interim period for a
subsample of deals. The bottom three rows report means and medians and associated t-statistics and z-statistics for
differences between pairs of the distributions in the top three rows. The interim period, which extends from the
beginning of the fiscal quarter that includes the announcement date to the completion date of the acquisition, can be
divided into the interim quarter and the stub portion (See Figure 1). The quarter before is the last fiscal quarter prior
to the acquisition announcement date. The interim quarter includes all quarters ending between the announcement
and completion of the acquisition. For observations with more than one interim quarter, we report averages across
those quarters. The stub portion is from the end of the last interim quarter to the acquisition completion date.
Because data for the stub portion is hand-collected and requires case-by-case analysis, we limit the subsample to the
first 300 observations in our full sample (ranked by acquirer CUSIP). We are able to calculate stub portion earnings
and sales for 106 and 93 observations, respectively. As no reports are filed by the target during the stub portion, we
infer stub portion sales as acquirer_as_if_sales minus (acquirer_actual_sales + target_actual_sales), where
acquirer_as_if_sales and acquirer_actual_sales are from the footnotes of acquirer’s 10-Q after the acquisition
completion date, and target_actual_sales are from target’s 10-Q prior to the acquisition completion date. A similar
approach is applied to obtain earnings during the stub portion. We get quarter-equivalent numbers by dividing stub
portion earnings and sales by the number of calendar days in the stub portion and multiplying by 91. All variables
are Winsorized at the 1st and 99th percentiles of the cross-sectional distribution each period. See Appendix for details
of variables.
39
Table 5 Evidence of acquirer performance overstatement in the post-acquisition period
Panel B Large acquisitions, Deal value exceeds 50% of Acquirer market value (518 observations)
Post- acquisition C_∆E C_FE C_abCFO
Quarter 1 2 3
+1 1.41% -0.61% 1.01%
+2 0.04%*** -0.28%*** 1.82%
+3 0.37%*** -0.29%** 2.10%
+4 0.03%*** -0.27%** 2.83%
Prediction relative to
Lower Higher Higher
average of (+5, +8)
Average +5 to +8 0.99% -0.49% 2.35%
Panel C Large acquisitions (Deal value exceeds 50% of Acquirer market value) using the pooling
method (124 observations)
Post- acquisition C_∆E C_FE C_abCFO
Quarter 1 2 3
+1 2.84%** -0.51% 0.72%
+2 0.08% -0.14% 3.68%
+3 -0.23% -0.22% 2.08%
+4 0.10% -0.22% 3.21%
Prediction relative to
Similar Similar Similar
average of (+5, +8)
Average +5 to +8 0.25% -0.33% 2.37%
*, ** and *** indicate significant differences at the 10%, 5%, and 1% levels, respectively, based on one-tailed tests,
for comparisons of performance measures in each of the first four quarters relative to the average of the
corresponding measures for the next four quarters (5 to 8).
40
This table reports time-series patterns for the combined entity’s performance in the post-acquisition period. The
prefix C_ in the variable names refers to the combined entity (acquirer) after the acquisition. C_∆E refer to inverted
seasonal difference in per share earnings before special items, equal to earnings per share in q+4 minus earnings per
share in q scaled by the acquirer’s equity book value per share as of the end of quarter q. We invert seasonal
differences for earnings and compare quarter q with q+4 rather than q-4, because we cannot use performance from
four quarters ago as a benchmark for the first four post-acquisition quarters (as the acquirer plus target from before
the acquisition is not comparable to the post-acquisition acquirer). The deflator CBV refers to the acquirer’s equity
book value per share as of the end of quarter q. C_FE is the forecast error relative to the consensus forecast as of the
last month of the fiscal quarter. C_abCFO represents the acquirer’s abnormal cash flows, which are residuals
generated from the cross-sectional Barth et al. (2001) model. Assuming that post-acquisition overstatement is
limited to the first four quarters, we expect the inverted differences in the first four quarters to be smaller than the
average over quarters 5 to 8 (the benchmark quarters), reported in the bottom rows of each panel. The sample
consists of 1,404 mergers and acquisitions with non-missing C_∆E (for quarter +1) and seasonally-differenced
target earnings before special items, T_∆E (for quarters 0 and -1) between 1985 and 2010. All variables are
Winsorized at the 1st and 99th percentiles of their respective cross-sectional distributions. See Appendix for details of
variables.
41
Table 6 Transfer of target performance understatement to post-acquisition performance
42
Table 6 (continued)
Panel B Large acquisitions, Deal value exceeds 50% of Acquirer market value (518 observations)
Dep. Var. C_∆E C_FE Dep. Var. C_abCFO
*, ** and *** indicate a significant difference from zero at the 10%, 5%, and 1% level, respectively, based on one-
tailed tests.
This table reports results from regressing acquirers’ post-acquisition performance in quarters +1 and +2 on
corresponding performance measure understatement by targets during the interim quarter, and various control
variables. The interim quarter refers to all quarters ending between announcement and completion of the acquisition,
and values reported for the interim quarter are means for all quarters included in the interim period. The prefixes C_
(T_ ) in the variable names refer to the combined entity after the acquisition (target before the acquisition). T_∆EU
and T_abCFOU measure the target’s performance understatement during the interim quarter and refer to target’s
seasonally-differenced earnings before special items and abnormal cash flows during the interim quarter minus those
in the quarter before. For acquirers, C_∆E refers to inverted seasonal difference in earnings before special items,
measured as earnings in q+4 minus earnings in q. We invert seasonal differences for earnings for acquirers and
compare quarter q with q+4 rather than q-4 because we cannot use performance from four quarters ago as a
benchmark for the first four post-acquisition quarters (as the acquirer plus target from before the acquisition is not
comparable to the post-acquisition acquirer). C_FE is the forecast error relative to the consensus forecast as of the
last month of the fiscal quarter. C_abCFO represents the acquirer’s abnormal cash flows, which are residuals
generated from the cross-sectional Barth et al. (2001) model. T_∆EU, T_abCFOU, and all dependent variables are
deflated by CBV, which is the acquirer’s equity book value per share as of the end of quarter q. The control variables
we consider are as follows. POOLING is an indicator variable set to 1 when the acquisition is accounted for as a
pooling of interests, PctStock is the fraction of the deal value paid as stock, Log(DEAL) is the logarithm of deal
value, RELSIZE is deal value relative to that of the acquirer, A_BM is the book-to-market ratio of the acquirer, and
PRE_ACC is abnormal annual accruals made by the acquirer over the three years before the acquisition.
43
Higher target performance understatement results in more negative values of the target variables, and higher
acquirer performance overstatement results in smaller values for earnings and higher values for forecast errors and
abnormal cash flows. According to prediction P3, if target performance understatement explains acquirer
performance overstatement, the coefficients on target performance understatement should be positive for the
earnings measures and be negative for forecast errors and abnormal cash flows.
The sample consists of 1,404 mergers and acquisitions with non-missing A_∆E and T_∆E between 1985 and 2010.
In Panel A (Panel B), there are 1404, 1321, and 958 (455, 406, and 354) valid observations for C_∆E, C_FE, and
C_abCFO models, respectively. All variables are Winsorized at the 1st and 99th percentiles of the cross-sectional
distribution. See Appendix for details of variables.
44
Table 7. Market reaction to earnings announcements: observed vs. predicted if investors
anticipate and adjust efficiently for earnings transfers from targets to acquirers
The dependent variable is returns around earnings announcement (ARET), measured as raw returns minus value-
weighted market returns over the three-day [-1, 1] period, where day 0 is the earnings announcement date. T_FE
(C_FE ) is analyst forecast error for the target (the combined entity), measured as actual EPS minus analysts’
consensus forecast made in the last month of the fiscal quarter scaled by book value of equity per share four quarters
ago. MV is the market value of equity at quarter end. BM is the book-to-market ratio at the quarter end. The interim
quarter includes all quarters ending between announcement and completion of the acquisition. For observations with
more than one interim quarter, we report averages across those quarters. In Panel B, T_∆EU measures target’s
earnings understatement during the interim quarter and refers to seasonal differences during the interim quarter
minus seasonal differences in the quarter before. RELSIZE is deal value relative the acquirer’s market value of
equity. The sample consists of 2,128 mergers and acquisitions between 1985 and 2010 with non-missing seasonally-
differenced earnings data. All variables except ARET, MV, and RELSIZE are Winsorized at the 1st and 99th
percentiles of the cross-sectional distribution each period. Market reactions are in general expected to be related to
analyst forecast errors (_FE), indicated by positive signs on T_FE and C_FE in Panels A and B. That relation is
expected to be weakened for targets during the interim quarter in Panel A if investors believe that reported earnings
have been understated substantially. Similarly, if investors believe that target understatement is transferred to boost
reported post-acquisition earnings, they would discount the portion of the forecast error (C_FE) that is due to
understatement of target earnings (T_EU), suggesting a negative coefficient on T_EU. See Appendix for details of
variables.
45
Table 8. Is target interim-quarter earnings understatement explained by reversal of pre-
acquisition earnings overstatement?
The dependent variable is seasonally differenced target earnings scaled by equity book value from four quarters ago,
expressed as a percentage. D is an indicator variable taking the value of 1 for the interim quarter and 0 otherwise.
The results in Tables 2 and 3 suggest that target earnings are understated during interim quarters, relative to the
quarters before. To estimate the extent to which this understatement is simply a reversal of earnings overstatement
that occurred in prior quarters, we include seasonally-differenced earnings from lagged quarters and allow for
autocorrelations to vary across interim and prior quarters. The interim quarter includes all quarters ending between
announcement and completion of the acquisition. For observations with more than one interim quarter, we report
averages across those quarters. The sample consists of 2,128 mergers and acquisitions between 1985 and 2010 with
non-missing seasonally-differenced earnings data. For each acquisition, we include the interim quarter and the prior
eight quarters. There are 17,802 firm-quarter observations for Models (1) and (2) and 16,598 firm-quarter
observations for Models (3) and (4). All variables are Winsorized at the 1st and 99th percentiles of the cross-sectional
distribution each period. See Appendix for details of variables.
46
Figure 1
Illustration of Target and Acquirer performance (earnings or cash flows), before (U) and after (M) performance management.
(Quarter 0 indicates interim quarters between the deal announcement date and the deal completion date)
announcement date
acquirer performance (post-acquisition)
completion date
EARNINGS & cASH FLOWS of TARGET/ACQUIRER
U+9
U+8
U+7
U+6
M+4 U+5
M+3
M+2 U+4
M+1 U+3
U+2
U+1
U0
U-1
U-2
U-3
U-4
U-5 M0
-5
I I -3
I -2
I -1
I 0
I +1
I I +3
I +4
I +5
I +6
I +7
I +8
I I
-4 +2 +9
47