Do Firms Set Free Cash Free?
Misallocation vs. Opportunism in Acquisitions
Lee Pinkowitz, Jason Sturgess, and Rohan Williamson*
March 2010
Abstract
We examine the impact of excess cash on the decision and method of payment in acquisitions.
Consistent with prior literature, we find that firms that have excess cash do make more
acquisitions. However, those acquisitions frequently use stock as the method of payment. We
investigate this finding further and show that cash-rich firms use overvalued equity to make
acquisitions. Moreover, there is no evidence that firms waste excess cash on acquisitions; cashrich firms that pay with cash actually acquire undervalued firms. Our results suggest that with
respect to acquisitions, managers are opportunistic with the firm’s financial resources; both its
equity and its cash.
Keywords: Cash holdings, method of payment, acquisitions
______________________
*Respectively, Associate Professor and Beit Research Fellow, Georgetown University, Assistant
Professor, Georgetown University, and Associate Professor and Stallkamp Research Fellow, Georgetown
University. We thank seminar participants at George Mason University and Georgetown University. We
also thank the Capital Markets Research Center for Research Support.
Do Firms Set Free Cash Free?
Misallocation vs. Opportunism in Acquisitions
One of the primary questions addressed in the finance literature is whether or not managers act in
the best interest of their shareholders. In particular, mergers and acquisitions are a key avenue for
exploring Jensen’s (1986) free cash flow problem, which argues that firms with free cash flow will make
poor investments. Harford (1999) finds evidence that appears consistent with this argument by showing
that high-cash firms make more acquisitions than other firms, and that those acquisitions have lower
announcement returns. Perhaps naturally, it is assumed that acquisitions of the type discussed by Harford
are made using the firm’s cash. Yet we observe firms that have high levels of cash holdings, but use
stock as the method of payment. For instance, on August 8th, 2003, USA Interactive successfully
acquired Lending Tree Inc for nearly $730 million. Despite having almost $4 billion in cash and
marketable securities, USA Interactive paid for the deal solely with equity.
Using a sample of 58,117 firm years from 1984-2005, we find 5,968 firm years where at least one
bid occurred. We confirm the results in the extant literature by showing that cash-rich firms are more
likely to make acquisitions. However, we document that being cash-rich does not make a firm more
likely to use cash as the method of payment. In fact, we find higher levels of cash in the 1,354 bidders
who paid solely with equity, than the 3,517 bidders who paid solely with cash. This suggests that the
USA Interactive example is more the rule than the exception and raises the question: Why might highcash firms resort to using stock in the acquisition?
In this study we address this question and, in so
doing, revisit whether the acquisition decisions of high-cash firms destroy shareholder value.
The finding that high-cash firms use equity as the method of payment can be consistent with
either self-interested managers squandering shareholder value, or managers acting in the best interests of
existing shareholders. If one believes the former, as implied by Jensen (1986), firms with excess cash
flow will waste the cash on negative NPV investments; which naturally extends to investment in negative
NPV acquisitions. However, Jensen’s argument need not imply that cash-rich firms use cash to pay for
acquisitions. Instead, a self-interested manager may choose to use stock because she values continued
1
flexibility and freedom from capital market discipline (Easterbrook (1984) and Jensen (1986)). Thus, a
self-interested manager may choose to use stock to make acquisitions and reserve the cash for other value
destroying investments.
Alternatively, if managers act in the interest of shareholders, they should opportunistically use
equity to make acquisitions when the stock is overvalued.1 This line of argument borrows from a rich
literature examining the role of overvalued stock in acquisitions.2 Particularly relevant to this evaluation
are Rhodes-Kropf and Viswanathan (2004) and Shleifer and Vishny (2003) who argue that overvaluation
is one of the key drivers in merger waves.
The use of “cheap” stock when cash is available is consistent with the mechanisms described in
Stulz (1990) and Myers and Majluf (1984), where managers stockpile excess cash to prevent future
underinvestment due to costly external funds. Survey responses in Graham and Harvey (2001) support
this, while Mikkelson and Partch (2002) provide empirical evidence that managers hold excess cash while
acting in the best interest of shareholders. They find that the operating performance of high-cash firms is
comparable to or greater than the performance of firms matched by size, industry, or prior performance. If
such managers are better able to identify acquisition opportunities, we should observe the empirical
association between high cash and takeover activity.
The two explanations have distinct empirical predictions. The first – which we label the
misallocation hypothesis – predicts the likelihood of cash bidding increases with cash levels and
acquisitions will be negative-NPV investments. It also predicts that managers will acquire overvalued
targets and are just as likely to pay with undervalued equity as overvalued equity. As such, cash-rich
firms will have lower announcement returns, regardless of their method of payment.
The second hypothesis – the opportunism hypothesis – implies that managers act
opportunistically with both the firm’s cash and its equity. Under the opportunism hypothesis, even if the
likelihood of bidding increases with cash levels, acquisitions will be positive NPV investments.
1
This
While the existing discussion focuses on overvalued equity, there are other reasons using stock as the method of
payment can benefit existing shareholders. For instance, Hansen (1987) argues that equity is beneficial when
information asymmetry is high as it leads to risk sharing between the bidder and target.
2
See Rhodes-Kropf, Robinson, and Viswanathan (2005), Dong, Hirshleifer, Richardson, and Teoh (2006), Ang and
Cheng (2006), and Fu, Lin, and Officer (2008) for example.
2
hypothesis also predicts that the use of stock increases with the overvaluation of the acquirer. Lower
announcement returns will be limited to deals where the firm uses equity as the method of payment, thus
signaling overvaluation.
To determine which mechanism is at play, this paper uses the methodologies employed in
Harford (1999) and Opler et al (1999) to identify firms with excess cash, and combines these with the
methodology in Rhodes-Kropf, Robinson and Viswanathan (RKRV) (2005) to identify overvalued firms.
Specifically, we examine the valuations of bidders and targets, for both cash-rich and non-cash-rich firms.
With this approach, we are able to determine whether acquisitions by cash-rich firms are in the best
interests of shareholders, or if they decrease shareholder value.
Consistent with the prediction of managerial opportunism, firms with excess cash use stock to
make acquisitions when their equity is overvalued - rather than spending their excess cash. Though
announcement returns are negative for acquisitions by cash-rich firms, the negative returns are isolated to
the deals where equity is the method of payment. Further, we document that when cash-rich firms pay
with cash, they actually pursue undervalued targets, and do not overpay.
The findings in this study are consistent with managers being opportunistic with the firm’s
financial resources; both its equity and its cash.
Managers appear to act in the best interest of
shareholders by using the cheapest source of capital to fund acquisitions. Our results suggest that
acquisitions are not a mechanism that managers use to waste firm cash holdings.3
The rest of the paper is organized as follows. Section 1 discusses the firm’s excess cash holdings
and the role it plays in acquisitions along with a motivation of the hypotheses to be tested. Section 2
describes the data. Section 3 discusses the results on the impact of liquid assets on acquisitions. Section 4
discusses and tests possible explanations. Section 5 examines robustness, and section 6 concludes.
1. Cash Holdings and Acquisitions
The finance literature has generally approached the impact of cash holdings on acquisitions and the
method of payment from two perspectives. Jensen (1986) argues that firms with free cash flow will tend
3
However, we do not entirely rule out over-investment in the spirit of Jensen (1986). Though managers do not use
acquisitions as a channel for inefficient behavior, they could waste free cash flow elsewhere (i.e. instance capital
expenditures, research and development, perquisite consumption, or simply inefficient operation of the firm).
3
to waste it, and acquisitions are one mechanism for doing so. The assumption in Jensen (1986) is that
self-interested managers will potentially undertake acquisitions which are negative NPV investments. The
stylized assumption of this literature is that these acquisitions will be made with the excess or free cash
flow of the firm; hence, limiting cash available to managers protects shareholders. Supporting this idea,
Harford (1999) shows that firms with excess cash tend to make more acquisitions, and those acquisitions
have lower announcement returns. Additionally, Harford et al (2008) show that managers of firms with
weaker corporate governance use cash to overinvest.
An alternate stream of literature is ambivalent on whether managers act in the interest of
shareholders; rather it examines the role of method of payment in acquisitions. Specifically, the choice of
method of payment will be determined by transaction characteristics related to investment opportunities,
market valuations, and risk sharing.4 Myers’ (1984) pecking order theory argues that firms will finance
investments in a manner that is cheapest and which maintains the financial flexibility of the firm.
Focusing on acquisitions, Hansen (1987) shows that a bidder may prefer to pay with stock in place of
cash if the target has better information about its investment opportunities than does the bidder. In this
case, using stock would force the target shareholders to share the risk of a revaluation of its investment
opportunities after the acquisition has been made. Martin (1996) empirically examines the method of
payment in acquisitions and finds that higher acquirer growth opportunities, greater pre-acquisition
acquirer returns, and greater pre-acquisition market returns increase the likelihood of stock being used as
the method of payment. Martin also examines whether ownership structure affects method of payment.
More recently, studies have claimed that the value of both acquirer and target equity are important.
Both Rhodes-Kropf and Viswanathan (2004) and Shleifer and Vishny (2003) argue that the method of
payment in mergers is driven by the relative valuations of the stock of the acquirer and the target. RhodesKropf and Viswanathan (2004) appeal to incomplete information, while Shleifer and Vishny (2003) apply
a more behavioral approach; but each reaches the conclusion that acquirer overvaluation drives the
4
A small literature has also examined the influence of ownership. Stulz (1988) and Jung, Kim, and Stulz (1995)
argue that managers will be reluctant to use stock to finance acquisitions if doing so will dilute their control and lead
to outside intervention. Martin (1996) finds weak supportive evidence. He shows that acquirer managerial
ownership is (negatively) related to the probability of stock over a middle range of ownership only.
4
decision to bid as well as the method of payment. In both models, acquirers with overvalued equity will
more likely pay with stock and will likely make acquisitions of overvalued targets.
The main difference in the two streams of the literature is the perceived motivation of management.
One takes an agency cost approach where management is acting in its own best interest while the second
assumes that managers are acting in shareholders’ best interest. We take these two perspectives to
motivate the hypothesis to be investigated in this paper.
1.1. Method of Payment and Agency
The arguments in Jensen (1986) and Harford (1999) imply an agency problem where managers
engage in empire building (Jensen and Meckling (1976)) at the expense of investing in positive-NPV
projects. While this suggests that managers would use their free cash as the method of payment in an
acquisition, management may have other objectives which can play a role in their decision on how to pay.
Self-interested managers would prefer to maintain freedom from future capital market discipline. If
management’s objective were to maintain financial independence, in addition to empire building, then
this could be met through using stock for acquisitions and reserving cash to spend post acquisition or on
perquisite consumption, as in Harford et al (2008).
As such, high cash would be a predictor of
acquisitions even if it were not the method of payment. These arguments result in our first hypothesis.
Misallocation Hypothesis: Self-interested managers have an incentive to misallocate the financial
resources of the firm; both its cash and its equity. The likelihood of cash bidding increases with cash
levels and acquisitions will be negative-NPV investments. Managers will acquire overvalued targets
and are just as likely to pay with undervalued equity as overvalued equity. Cash-rich firms will have
lower announcement returns, regardless of their method of payment.
1.2. Method of Payment, Opportunistic Managers and Stock Acquisitions
If managers act in the best interest of shareholders they should choose the method of payment
opportunistically. Consistent with this, both Rhodes-Kropf and Viswanathan (2004) and Shleifer and
5
Vishny (2003) argue that managers acting in the best interest of shareholders should prefer to use
overvalued stock as the method of payment. Analogously, the arguments in Baker and Wurgler (2002) on
market timing and firm capital structure claim managers use the cheapest funds for financing investments.
Further, as predicted Myers and Majluf (1984) and by pecking order, financial flexibility allows
the firm to take advantage of merger synergies, and fund future positive NPV projects without accessing
the capital markets. Graham and Harvey (2001) report that managers prefer financial flexibility as well as
attempt to time the market with security issues. Hence, a firm where management acts in the interest of
shareholders by using overvalued stock to make acquisitions, while maintaining financial flexibility, will
exhibit both high levels of cash and stock acquisitions.
Finally, as argued in Shleifer and Vishny (2003), target owners may prefer overvalued equity
since they may receive more for the firm than they might receive in a cash offer and thus gain from the
transaction. This implies that even where the target is overvalued, bidder shareholders can still benefit by
using relatively more overvalued stock for the acquisition. These arguments lead to our next hypothesis.
Opportunism Hypothesis: Managers acting in the best interests of shareholders are opportunistic
with both the firm’s cash and its equity. Managers make positive-NPV acquisitions and hold excess
cash to achieve financial flexibility for the firm. Managers prefer to use stock as the method of
payment only when the firm’s stock is overvalued, or relatively overvalued. Lower announcement
returns will be limited to deals where the firm uses equity as the method of payment.
2. Data
2.1. Financial Data
Financial data are drawn from the Compustat database. All publicly traded firms with sales, assets,
and market capitalization (in year 2000 dollars) greater than $5 million, stock price greater than $1, and
non-missing data in the period 1984 to 2005 are included in the sample.
As our focus is on the cash holdings of firms, we construct several measures of corporate liquidity.
First, we use raw cash holdings deflated by assets. Second, we construct three different measures of
6
abnormal cash holdings: residuals from regressions similar to those in Harford (1999), residuals from
regressions similar to those in Opler, et al (1999), and deviation from the industry average. The appendix
describes the construction of the excess cash variables.
For our empirical analyses, we segment the universe of firms based on whether firms have high
levels of (excess) cash. As such, we are more interested with relative liquidity among firms each year
rather than the absolute level of liquidity. We assess this relative liquidity level in two ways. First, we
construct dummy variables if a firm is in the highest quartile of (excess) cash in a given year. Second, we
standardize all four of our liquidity measures to have zero mean and unit standard deviation. The reason
for this transformation is that we attempt to remain agnostic as to which model of excess cash, if any, is
better. Standardizing all our measures allows us to more directly compare across the liquidity variables.
In addition to our cash variables, we include control variables to capture the impact of other firm
characteristics on the bidder’s decision to acquire.
These variables, suggested by the literature on
acquisitions (Ambrose and Megginson (1992), Comment and Schwert (1996), Harford (1999), among
others), include the abnormal return prior to the bid, sales growth, non-cash net working capital, priceearnings ratio, market to book ratio, leverage, and firm size. Following Harford (1999), all control
variables are measured as four year averages from year t-4 to year t-1. Variable definitions and Compustat
data item codes are reported in Table 1.
To mitigate the effect of outliers, all our variables are
winsorized, by year, at the 1% tails.
2.2. Acquisition Data
Our sample of acquisition attempts is drawn from Thompson’s Securities Data Corporation (SDC)
data set. We examine bids from 1984 to 2005 subject to the following criteria:
1.
2.
3.
4.
5.
6.
The bidder owns less than 50% of the target prior to the bid and seeks to own 100%.
The bidder is classified by SDC as a public corporation, but is not a financial company or utility.
Both the bidder and the target are US firms.
The deal type is classified as a disclosed value merger, an LBO, or a tender offer.
The form of the deal is listed as merger, acquisition of assets, or acquisition of minority interest.
All repurchases, equity carve outs, and limited partnerships are excluded.
Because SDC contains data on very small acquisitions, we impose constraints on deal size. We
identify bidding firms as any firm making one or more bids within a year, where the sum of all bids in a
7
year is at least: one percent of the bidder’s prior-year assets, one percent of the bidder’s prior-year market
capitalization, and ten million dollars in size (deflated to year 2000 dollars using the CPI). The resulting
data set includes 10,820 bidder years.
To determine the method of payment for the bid years we confirm the SDC classification and
combine this with hand-collected data. This is necessary because Faccio and Masulis (2005) report that
there are frequent discrepancies within the SDC data field CONSID_STRUCTURE, which simply
classifies deals as cash, stock or hybrid. To ensure we have classified the method of payment as
accurately as possible, we examine our data in two ways. First, we use SDC’s data fields describing the
value of consideration offered based on consideration type (i.e. cash, common stock, convertible debt,
etc.)5 Second, we rely on SDC’s text field describing the consideration offered (data item CONSIDOC).
We classify cash consideration as payment offered in the forms of cash, debt (including convertible debt),
assets, or capital infusions.6 We classify equity consideration as payment offered in the forms of common
or preferred stock (including convertible preferred), warrants, rights or options.7 We ignore consideration
offered in the form of earnouts or royalties as they are contingent future payments. We then classify bids
as cash, stock, or mixed based on both of the above methods and examine by hand those cases where
there is a discrepancy. When possible, we go to the financial statements and news stories to determine the
composition of consideration listed as “other” or “undisclosed” by SDC, as well as to hand-correct errors
in the method of payment data.8
When we are unable to specifically identify the composition of
consideration, we assume it is paid in an equal mixture of cash and stock.
Ultimately, we classify firm years as a Stock Bid (Cash Bid) if all of the consideration offered within
that year is comprised of equity (cash) components. Bids are classified as Mixed if both equity and cash
are offered in acquisitions bids.9 This categorization serves as the most conservative way to segment
method of payment to isolate cases where firms used equity exclusively.
5
The SDC data items are: VCASH, VCOM, VCDEBT, VCPFD, VPFD, VWAR, VOTH, VUNDIS
The following codes are classified as cash consideration: CASH, A, NOTE, BOND, DEB, DBTRED, and LOAN
7
The following codes are classified as stock consideration: COM, STK, PFD, WAR, OPT, UNIT, PREF, SEC,
ORD, TRACKING, TSH, ADS, and, APPRECRTS
8
A list of data corrections is available from the authors.
9
This includes firms who make a single bid in a year and offer some combination of cash and equity as well as firms
which make multiple bids in a year where some of the bids were entirely equity while others were entirely cash.
6
8
We merge our SDC acquisition data with Compustat data and make two additional changes based on
industry, defined using the Fama-French 30 industry definitions. First, the Tobacco Industry (Industry 3)
has 64 firm year observations, but zero bids so we eliminate those firm years because the industry effect
perfectly predicts no bids will occur.10 We also drop the coal industry with 78 firm years but only a single
bid. Our final sample includes 58,117 firm years comprised of 5,968 firm years where at least one bid
occurred and 52,149 firm years where no bid was recorded.
For the 5,968 firm years with at least one acquisition attempt, approximately one quarter are classified
as financed entirely with equity. Our categorization results in 1,354 stock bids, 3,547 cash bids, and
1,067 mixed bids. Table 2 shows the distribution of our sample across time and method of payment. We
have a large sample of bids in each year from 1984-2005, ranging from a low of 131 bids in 1990 to a
high of 478 bids in 1997. We also have a reasonable sample of bids based on method of payment. We
classify 22.7% of our bids as all stock, ranging from a low of 6.2% in 2004 to a high of 38.9% in 1999.
Our proportions of all equity bids are slightly lower than those reported in Betton, Eckbo, and Thorburn
(2008) owing to the conservative nature by which we classify stock bids.
3. Liquidity and Acquisition Activity
3.1. Summary Statistics
The summary statistics for our full sample are shown in Table 3. Of particular importance are the
measures of liquidity. On average, firms hold 14% of their assets in the form of cash, while the median
firm has 6.7%. The different excess cash measures show a similar right skew in the data with means
between 3 and 6 percent of assets. The median firm has negative excess cash according to the Harford
measure, positive excess cash according to the OPSW measure, and approximately zero excess cash when
measured relative to the industry.
However, because our purpose is not to address the magnitude of excess cash holdings across
firms, but rather to examine relative liquidity, we focus on two measures to describe high-cash firms.
First, we create dummy variables signifying a firm as having high (excess) cash if it is in the top quartile
10
Our results are nearly identical if we include these firm years.
9
of firm liquidity in a given year. As must be the case, one quarter of our firm years are defined as being
high cash across all the measures. Second, we standardize our variables such that each year they have
zero mean and unit variance. The benefit of these measures is that we can remain agnostic as to which of
the models employed, if any, accurately predicts the level of excess cash. We need only to rely on the
assumption that a model will provide a relatively accurate ranking of firms based on (excess) liquidity.
Correlation coefficients for our liquidity measures suggest that using our measures makes sense.
For instance, the Pearson correlation coefficients between the raw OPSW excess cash measure and the
raw Harford excess cash measure is actually negative at -0.10. The correlation between the OPSW
measure and the industry excess cash measure is only 0.09. Standardizing our variables increases those
correlations to 0.07 and 0.24. More importantly, the Spearman rank correlations are 0.59 and 0.73,
respectively.11 The correlations suggest that while using different models of “normal” cash holdings allow
us to control for a variety of factors which may impact cash holdings, using the top quartile of each
measure should provide us with a sample of firms which tend to have high levels of cash, regardless of
one’s priors about the levels of cash holdings predicted by the models.
Table 4 provides the first evidence on the relation between liquidity and acquisition activity.
Panel A segments our sample by whether or not the firm launched a bid in a given year, without regard to
method of payment. In the 5,968 firm years where at least one bid occurred, the level of cash holdings
are significantly higher at both the mean and median than for the 52,149 firm-years where no bid was
launched. The results are similar for the excess cash measures. While those results rely on the magnitude
of our liquidity measures, we also focus on firms in the highest quartile. Of the bidders, 27.3% were
ranked in the highest quartile of cash holdings, while only 24.7% of the non-bidders were in the top
quartile. The proportions are significantly different at the 1% level. We find similar results for our
rankings based on the excess cash measures.
There are also significant differences in our control variables, providing support for including
them in additional tests. Bidders are more likely to have greater abnormal stock returns and sales growth
in the years leading up to the takeover attempt. Bidders also have significantly higher Price to Earnings
11
Correlation tables are available upon request from the authors.
10
(PE) and market to book (MB) ratios, which can be indicative of equity overvaluation as an incentive to
undertake acquisitions. Lastly, bidding firms have significantly lower levels of non-cash net working
capital and leverage, while also being larger in size.
Overall, our summary statistics confirm Harford’s (1999) result that cash-rich firms are more
likely to undertake acquisition activity. However, when we separate our sample on the basis of method of
payment, the impact of cash holdings on acquisition activity becomes more interesting. Panel B of Table
4 examines only the 5,968 firm years in which at least one bid occurred and documents that high cash
firms appear more likely to undertake stock acquisitions rather than use their cash.
In the 1,354 cases where equity was the sole consideration offered, bidders had an average of
19.8% of their assets in cash. In contrast, for the 3,547 years where cash was the sole consideration
offered, bidders held an average of 13.0% of their assets in the form of cash. This difference is
statistically significant at the 1% level and indicates that while the presence of cash makes firms more
likely to bid, it may not mean that they are spending that cash. We find similar results using medians and
for our excess cash measures.12 When examining our ordinal rankings, we find that stock bidders are
nearly twice as likely to be in the top quartile of cash holdings as firms which pay with cash. Roughly
40% of the stock bidders were in the highest quartile of cash holdings, while only 21.5% of cash bidders
were. This difference between the proportions is significant at the 1% level and is similar using all of our
excess cash measures. While our main comparisons are between all equity and all cash bids, it is also
interesting to notice that liquidity levels are higher in mixed bids than in the all cash bids.
Interestingly, while cash-rich firms are more likely to be involved in acquisition activity, it does
not appear that they use cash to pay for the acquisitions. In fact, firms which bid completely with cash
have less liquidity than firms which use solely equity, or a mixture of equity and cash.
We also observe significant differences in control variables across sample splits by method of
payments. Cash bidders tend to be larger and have greater leverage than those which pay with equity.
12
With the OPSW measure, cash bids have more liquidity at the mean, but not the median suggesting there are some
observations with very high positive levels of excess cash. This provides further argument for using our ranking of
cash-rich firms rather than the raw level.
11
Conversely, firms that bid with stock have higher prior abnormal returns, sales growth, PE, and MB ratios
indicating that the value of bidder shares may be an important determinant of method of payment.
3.2. Cash Holdings and the Probability of Bidding
The prior evidence suggests that while higher liquidity is associated with frequency of acquisition
activity, it is driven by equity acquisitions. However, the prior evidence is univariate and it is important
to control for other firm attributes as well. Since bidders have several choices in paying for acquisitions,
we use multinomial logistic regressions to evaluate the choices.13 Table 5 documents the results from
multinomial logistic regressions to determine not just if a firm is likely to be a bidder, but how it is likely
to pay. Our dependent variable is set to zero if the firm does not undertake a bid in the following fiscal
year, set to 1 if they launch an all stock bid, set to 2 if they launch an all cash bid, and set to 3 if they
undertake a mixed acquisition attempt. The regressions are estimated controlling for industry and year
effects. All p-values are based on standard errors clustered at the firm level (see Petersen (2009)).
Regression (1) confirms that being in the top quartile of cash to assets makes a firm significantly
more likely to bid than not. The p-values in brackets are based on a test of the difference in the
coefficients against the outcome of not bidding. Not surprisingly, the coefficients for all three bidding
outcomes: stock, cash, or mixed are significantly positive. Interestingly, the coefficient of 0.096 on cash
bids shows that high cash firms are only marginally more likely to undertake a cash bid than to not bid at
all. Moreover, the coefficient is 0.345 for all-stock bids, which is significantly larger than the coefficient
on cash bids at better than the 1% level. The interpretation is that while being in the highest quartile of
cash holdings makes a firm more likely to bid with cash than not bid at all, they are actually more likely
to bid with stock than with cash. Marginal effects imply that the economic significance is large as well.14
The unconditional probability of a firm being a stock bidder is 139 basis points. Being in the top quartile
of cash holdings increases the unconditional probability of being a stock bidder by 50 basis points, an
increase of nearly 40%.
13
In unreported results, we also estimate simple logistic regressions. Consistent with prior literature, we find that
high (excess) cash holdings significantly increase the likelihood of being a bidder. Results are available upon
request.
14
Marginal effects are calculated in Stata using the prchange command. See Long and Freese (2006) for details.
12
We observe similar results in Regression (2) using our standardized cash variable to measure
liquidity. Using the continuous measure, not only does an increase in cash make a firm more likely to bid
with stock than bid with cash, but the coefficient on all cash bids is insignificant. This indicates that high
cash firms are actually no more likely to spend their cash on an acquisition than not bid at all. Increasing
cash holdings by one standard deviation around the mean increases the likelihood of undertaking a stock
bid by roughly 20% (an increase of 27 basis points from a base of 139 basis points). As in the summary
statistics, we see that in both regressions the likelihood of making a mixed bid is greater than that of
paying with cash as liquidity increases.
The coefficients on the control variables are similar across both specifications and mostly
consistent with the evidence in Harford (1999). Bidders tend to be larger, have higher abnormal returns,
and greater sales growth than non-bidders. The positive relation between working capital that Harford
(1999) demonstrates seems to be driven by the cash bids, while the impact of market to book is only clear
when segmented by method of payment. Harford found that MB was insignificant, but this is likely
driven by the fact that it is significantly positive for stock bids, negative for cash bids, and insignificant
for mixed bids. When the method of payment is not broken out, those effects work in opposite directions
resulting in an insignificant estimate.
From an overvaluation viewpoint, it is interesting to note that the coefficients on the average
abnormal returns, while always significantly positive, do not differ across all three methods of payment.
On the other hand, the coefficient on market to book are positive for stock bids and negative for cash bids,
with mixed bids falling in the middle. These results suggest that firm-specific overvaluation is more
likely to be a determinant of method of payment than an industry or sector run up over the past few years.
Section 4 examines this more closely.
While the first two specifications use raw liquidity, Panel B of Table 5 examines our excess
measures of liquidity. For brevity, we only report the coefficients on the liquidity measures and omit the
control variables. For both the Harford and industry adjusted measures, we find the same results as in
Panel A. High liquidity makes firms more likely to be bidders, but the results are driven by firms which
use equity as the method of payment. The OPSW measure does not provide the same conclusions. While
13
the stock coefficient is higher than that on cash for firms in the top quartile, the difference is not
significant. In addition, using the continuous measure, the coefficient on cash bids is marginally larger
than that for stock acquisitions.
Nonetheless, the results in Table 5 indicate that the increase in takeover probability for cash-rich
firms is not driven by firms using their cash. Rather, the results indicate that cash-rich firms are actually
more likely to use equity as a method of payment. While our results are completely consistent with those
in Harford (1999), segmenting by method of payment allows for a different interpretation of the empirical
evidence. Since cash-rich firms are not using cash as the method of payment, the evidence is more
supportive of the opportunism hypothesis rather than the misallocation hypothesis.
3.3. Cash Holdings and the Method of Payment
The results in Table 5 show that cash-rich firms are more likely to be bidders than other firms, but
they tend to use equity as the method of payment. However, it is possible that the results are affected by
the exclusion of deal specific characteristics which are likely correlated with the choice of method of
payment. For instance, Martin (1996) shows that higher investment opportunities of the bidder and target
lead to stock financing of acquisitions, while bidder size is unimportant.
In an effort to examine the robustness of our results, we estimate conditional multinomial logistic
regressions using only the 5,968 firm years where a bid occurred. By excluding firm years where no bid
was attempted, we can include not only bidder characteristics, but also deal specific characteristics in our
regressions. Because many bidders in our sample launched multiple bids in a single year, we focus on the
first bid in a year when examining deal specific characteristics. Table 6 shows our results.
In addition to the bidder characteristics we previously examined, we also include variables which
measure: the size of the deal, the degree to which it is met with resistance, and the identity of the target.
Specifically, we construct the following variables. Relative value is defined as (deal value / (bidder
market capitalization + deal value)), Deal Size is the natural logarithm of deal value deflated to year 2000
dollars using the CPI, Unsolicited is an indicator variable which is set to one if SDC describes the deal
attitude as starting off unsolicited, Auction is an indicator variable set to one if there is more than one
bidder in the control contest, Defense is set to one if the target uses some sort of takeover defense, Private
14
is set to one if SDC indicates that the target is private, Subsidiary is set to one if SDC indicates that the
target is a subsidiary, and Outside Industry is set to one if the bidder and target are from different
industries as defined by the Fama-French 30 classification.
Regression (1) shows our results when we measure liquidity using the top quartile of raw cash
holdings. The first column shows the comparison of stock bids versus cash bids, while for completeness,
we also report the comparison between mixed bids and cash bids in the second column. The coefficient
on High Cash is 0.258 with a p-value below the 1% level. In Table 6, the p-values in brackets are the
relevant measure since we are addressing the difference between stock and cash deals, and are not
concerned with a difference between stock and mixed deals. The coefficient on High Cash indicates that
even conditioning on a bid occurring, firms in the top quartile of cash holdings are more likely to pay with
equity rather than cash. Additionally, the marginal effect is large. Being in the top quartile of cash
increases the probability of paying with equity by 282 basis points from an initial probability of 15.40%
It is reassuring that the coefficients on the firm-level control variables are consistent with those
from the unconditional regressions in Table 5 even with the additional deal-specific control variables. For
instance, stock deals have higher sales growth and MB, while having lower NWC and leverage than cash
deals. Results in Table 6 show that the coefficients on growth and MB are both positively significant,
while that on leverage is significantly negative. While the coefficient on NWC is negative, it is not
significant when deal variables are included. We also notice that the coefficient on prior abnormal returns
is significantly positive, indicating that bidders who offer stock had a larger run-up in their stock prices in
the recent past.
Regarding the deal-specific variables, we find that firms tend to pay stock when the deal is larger
both in absolute and relative size.15 Bidders are more likely to offer cash than equity if the deal is
unsolicited, if there are multiple bidders for the target, or if the target undertakes defensive action. These
are all consistent with the idea that when there is competition or target shareholders do not want to sell,
they are more likely to be persuaded by offering cash rather than equity.
15
Bidder size is omitted from the Table 6 regression due to the inclusion of deal size and relative size.
15
One of the main reasons that a bidder might use equity is to avoid overpayment through risk
sharing as argued in Hansen (1987). This is more likely to be the case when information asymmetries are
relatively large. We empirically control for information asymmetries by looking at whether the target was
privately held and whether it is in a different industry than the bidder. We expect that if information
asymmetries are larger in private targets and those outside of the bidder’s industry, equity would be more
likely to be the method of payment. Surprisingly, we find that the coefficient on Private is negative and
significant, indicating that controlling for other factors firms are more likely to offer cash for private
targets. While we do find a positive coefficient on Outside Industry, it is statistically insignificant.
We find similar results when we use our standardized measure of cash holdings in regression (2).
Firms with higher cash are more likely to choose equity as the method of payment. Panel B shows the
results of regressions using our alternative measures of liquidity. For the Harford and Industry excess
cash measures, we find identical results. For the OPSW excess cash measure, we again see insignificant
results using the indicator variable, and significantly negative results with the continuous measure.
While the majority of the results are consistent with the high cash firms using stock for
acquisitions, the inconsistency of the OPSW measure requires further investigation. If the misallocation
hypothesis was dominant, we should see firms with positive excess cash holdings using that cash as the
method of payment. However, finding that a continuous measure of cash holdings is related to the
propensity to use cash doesn’t necessarily imply that. While the result could be driven by high cash firms
using their cash; alternatively, it could be evidence that low cash firms have to use equity. To distinguish
between these explanations, we segment our standardized (excess) cash measures by whether they are
positive or negative. We create two mutually exclusive variables: Standardized Positive cash equals the
level of standardized cash if greater than zero, and zero otherwise; Standardized Negative cash equals the
absolute value of standardized cash if less than zero, and zero otherwise.
If firms were using excess cash to finance acquisitions, we would expect to see negative and
significant coefficients on the Positive cash variables, with no prediction regarding how firms with little
cash would behave. We estimate our multinomial logistics using these breakpoints and find coefficients
greater than zero across all four measures of Positive Cash, indicating that when firms have excess cash
16
holdings, they are more likely to use equity, not their cash. This result is inconsistent with firms
misallocating cash due to agency problems.
Instead, we notice that the reason for the flip in sign with the OPSW continuous measure relates
to firms with negative excess cash. The significantly positive coefficient on Standardized Neg. OPSW
indicates that firms with little OPSW excess cash are more likely to use equity – possibly because they are
cash constrained. Conversely, we find significantly negative coefficients for our other measures of
Negative Cash, indicating that the firms that pay with cash actually hold little cash in the year before the
bid. This is consistent with the idea that many cash deals are actually financed by issuing debt or equity
(Schlingemann (2004)).
3.4. Cash Holdings and Announcement Returns
We have demonstrated that while cash-rich firms are more likely to be bidders, in fact, they are
significantly more likely to pay equity for the target.
While these results are consistent with the
opportunism hypothesis, they are not necessarily inconsistent with the misallocation hypothesis if selfinterested managers simply retain their cash to maintain freedom from future capital market discipline.
The cleanest way to distinguish between the two hypotheses is to examine the bidder
announcement returns. Harford documents that bidder announcement returns are lower for cash-rich
bidders which seems supportive of the misallocation hypothesis. Once again, segmenting by method of
payment allows for further insights.
Table 7 shows regressions using cumulative abnormal returns from day -1 to day 1 as the
dependent variable. Abnormal returns are calculated using the market model with parameters estimated
over the period -370 to -253 as in Harford (1999). Our control variables include both bidder-specific and
deal-specific characteristics. With our data requirements, we have 5,900 observations in the full sample.
The first column shows the results of estimating our regression on the full sample of bids, without
regard to method of payment. The coefficient on High Cash is negative and significant at better than the
one percent level confirming Harford’s results that the market reaction to a bid announcement is worse for
high cash bidders. Moreover, it is reassuring that the signs on our control variables are consistent with
17
those found in the prior literature. For instance, as in Moeller, Schlingemann and Stulz (2004), bidder
announcement returns are negatively related to the Q of the bidder and the size of the deal. Conversely,
as in (Chang (1998), Fuller, Netter and Stegemoller (2002), Faccio, McConnell and Stolin (2006)), bidder
returns are higher if the target is a private company or a subsidiary.
The lower market reaction to high cash firms could be explained by either the misallocation or
opportunism hypothesis. The decision to pay for an acquisition using equity signals to the market that the
bidder’s stock is overvalued (Travlos (1987) and Amihud, Lev, and Travlos (1990)). Because we have
shown that high cash firms tend to pay with equity, it is possible that the negative announcement effect
for high cash firms is really capturing the method of payment. To examine this possibility, the remaining
three columns show the results of estimating the regression separately based on the method of payment.
When we segment based on method of payment, the results clearly favor the opportunism
hypothesis. Those firms that pay with equity wholly drive the negative announcement effect for high
cash bidders. The coefficient is -0.021 for the stock bids while essentially zero for the cash bids. Not
only is the coefficient in the stock regression significantly negative, but it is significantly more negative
than the coefficient in the cash regression.
It is important to note that this finding is not simply capturing lower announcement returns which
are associated with equity offers, since that main effect will be embedded in the constants. That suggests
that the market does not unconditionally attribute equity overvaluation as the reason for an equity bid;
however, when a firm has high cash, but still chooses to pay equity, it serves as a strong signal that equity
is overvalued.
We also find evidence of the risk-sharing hypothesis in our abnormal return results.
The
announcement returns are higher for stock bids than cash bids if the target is outside the bidder’s industry
or either a private firm or subsidiary. Targets with those characteristics likely have higher information
asymmetries and thus using equity would be better for bidder shareholders (Martin (1996)).
Panel B provides evidence for our other liquidity measures and demonstrates the same pattern.
Again, for brevity, we omit the control variables, which, in all cases are similar to those reported in Panel
A. For each of our liquidity measures we find that announcement returns are significantly negative only
18
for the stock bids and the difference between stock and cash bids is always significant. When we use our
continuous measures, we find that for all but the OPSW measure, the negative announcement returns are
limited to the stock acquisitions. Overall, the results in Table 7 support the opportunism hypothesis rather
than the misallocation explanation. Essentially, it appears that if a firm offers to pay with equity by
choice, it is a far worse signal than paying equity because you have no other option.16
4. Cash Holdings and Equity Overvaluation
4.1 Do Cash-rich Bidders have Overvalued Equity?
The evidence so far demonstrates that while high cash firms are more likely to be bidders, they tend
to use equity as the method of payment. Additionally, the results suggest there may be a link between
high cash holdings and equity overvaluation. However the literature has shown that in many cases, the
market values cash holdings less than dollar for dollar (Faulkender and Wang (2006), Pinkowitz, Stulz
and Williamson (2006), Pinkowitz and Williamson (2007), Dittmar and Mahrt-Smith (2007), among
others). Thus it appears an inconsistency remains which encourages us to more directly examine the
relationship between cash holdings and overvalued equity.
While many papers attempt to measure equity overvaluation, we choose to use the empirical
methodology of Rhodes-Kropf, Robinson, and Viswanathan (2005), hereafter RKRV, that similarly
examines equity overvaluation in acquisitions.17 Their method decomposes the market to book ratio into
three components, a firm-specific error, a time-series sector error, and a long-run value to book ratio.
We follow RKRV’s methodology and employ the regression for firm market value:
mit = α 0 jt + α1 jt bit + α 2 jt ln( NI + )it + α 3 jt I ( < 0) ln( NI + )it + α 4 jt Levit + ε it
The variables m and b are the logarithms of market value and book value of equity, respectively. NI+
represents the absolute value of net income, I(<0) is an indicator variable set to one if the firm has a
negative net income in a given year, and Lev is the market leverage of the firm defined as {1 - (market
16
This is analogous to Cornett and Tehranaian (1994) who find that banks which raise equity voluntarily have lower
announcement returns than those which raise equity to maintain their capital requirement.
17
We are not the first to employ the RRKV methodology. For example, Fu, Lin and Officer (2008) use it to study
value creation in acquisitions.
19
value of equity/(market value of equity + book value of assets – deferred taxes – book value of equity))}.
The subscripts represent observations at the firm (i), industry (j), and year (t) level. We segment firms
into the Fama-French 12 industries defined as on Ken French’s web site. We estimate the above
regression cross-sectionally for each industry year. We then calculate the firm-specific error, time-series
sector error, and long-run value to book following RKRV (see Table 5 of their paper, page 579).
While our sample differs, both in size and time frame, from RKRV, we find qualitatively similar
parameter estimates as they report. We also find similar relationships as RKRV with respect to splits on
the basis of bidder versus non-bidder, and segmentation based on method of payment.18
In order to examine whether our results are simply driven by a consistent correlation of high cash
firms being overvalued, in Table 8, we report the MB decomposition based on our liquidity breakpoints.
For firms in the top quartile of cash holdings, average firm-specific error is -0.0692 which is significantly
negative. Conversely, firms in the other three quartiles of cash holdings have a significantly positive
firm-specific error of 0.0234. The nearly 900 basis point difference is highly significant as well. We find
similar results using our other measures of firm liquidity with the difference ranging from 900 basis
points to 1900 basis points using the OPSW categorization. Overall, it appears that cash-rich firms do not
have overvalued equity. Rather, the market seems to discount the equity of firms holding high amounts
of cash. This is consistent with existing evidence in Faulkender and Wang (2006), among others.
In contrast to the firm-specific errors, Table 8 indicates that firms are more likely to be in the top
quartile of cash holdings when their industry is overvalued relative to its long run fundamentals. It is
reasonable to believe that firm overvaluation should reflect not only the firm-specific error component,
but the impact of a hot industry. Hence, we define total error to be the sum of both the firm-specific error
and the time-series sector error. We find that even accounting for the overvalued sectors, cash-rich firms
are undervalued by the market relative to similar firms with lower levels of cash. This result holds across
all our liquidity measures with differences between 700 and 1900 basis points.
18
The notable exception is that we do not find evidence of their low buys high result. We find higher long-run value
to book for bidders than non-bidders. It is possible that the timing difference in our sample is a major factor in this
because our sample runs from 1984-2005, while RKRV use data from 1977-2000. While this suggests that more
research should be done on that phenomenon, our focus is on the firm-specific overvaluation so we leave that to
future research
20
Finally, we note that the long-run value-to-book estimates are considerably larger for high cash firms.
This is reasonable given the characteristics of firms which are known to hold high levels of (excess) cash;
namely smaller firms with greater growth opportunities (see Kim, Mauer and Sherman (1998), OPSW
(1999), and others). The consistency with prior literature is reassuring and suggests that using the RKRV
decomposition to examine liquidity is valid.
If high cash firms are not systematically overvalued, then what explains the link between high cash
firms and equity bidding? It is possible that while high cash firms are valued at a discount, on average,
we might still see acquisition activity from those which happen to have overvalued equity.
Table 9 addresses this issue by segmenting our sample not only by level of liquidity, but also by
bidding activity. Panel A shows this two-way split. The first row provides the number of firm-year
observations in each cell. For instance, of the 5,867 bid years, 4,254 are from bidders classified as not
being in the top quartile of cash holdings, while 1,613 are in the top quartile. Segmenting by liquidity, we
notice that non-bidders have similar firm-specific errors as the whole sample. Non-bidders in the top
quartile of cash holdings are significantly undervalued with a mean of -0.0852, while those not in the top
quartile have a significantly positive firm-specific error of 0.0155. The difference between the two is
roughly 1,000 basis points which is highly significant; as shown by the p-value in the last column, which
allows for comparisons across levels of liquidity holding bidding activity constant. Conversely, the pvalues beneath the rows allow for comparison across bidding activity, holding liquidity constant.
In contrast to the full sample and the non-bidders, we find that the difference in firm-specific errors
for bidders is less than 400 basis points. Not only is the difference in magnitudes considerably smaller,
but the average firm-specific error of the high-cash bidders is actually positive at 0.0567. It appears that
cash-rich firms which undertake bids are in fact overvalued rather than being discounted by the market.
This difference becomes more apparent when we examine the total errors segmented by bidding activity.
Accounting for both the firm and sector errors, we find that the large discount placed on cash-rich firms
completely disappears for our bidders. The cash-rich bidders are as overvalued as bidders not in the top
21
quartile of cash holdings. While not reported, we find similar results when segmenting firms using our
measures of excess cash.19
Panel B of Table 9 repeats the analysis using just bidders, segmented by method of payment. We
focus only on firms which offered stock or cash exclusively and eliminate the mixed deals. We find
substantial overvaluation for stock bidders, regardless of whether they are in the top quartile of cash
holdings. In fact, it appears that high-cash stock bidders are actually more overvalued than stock bidders
with less cash.
However, the difference for both the firm-specific and total error components is
insignificant. By contrast, for cash bidders, we again see a roughly 900 basis point discount for cash-rich
firms, which is highly significant. The results once again support the opportunism hypothesis rather than
the misallocation hypothesis.
4.2 Do Cash-rich Bidders Overpay for Targets?
The combined results in Tables 7 through 9 strongly support the argument that cash-rich firms
undertake acquisitions when their equity is overvalued. There appears to be little support for the idea that
agency costs of free cash flow are a motivating factor in acquisitions. However, it may be the case that
while overvaluation explains the reason for stock bids, high-cash firms may still be wasting their cash
when they use it as the method of payment.
While there was no evidence of negative announcement returns for cash-rich firms undertaking cash
bids, we attempt to examine whether those firms overpay for their targets. It is difficult to determine
whether the offer price for a target is “too much” because the true value of the target is unobservable.
Nonetheless, we attempt to address this issue in two separate ways.
First, we use the MB decomposition to determine whether the targets are overvalued.
The
methodology is exactly the same as in the prior section. Unfortunately, we can only decompose MB for
publicly traded targets as there is no available market price for private firms. However, as shown in
Chang (1998), announcement abnormal returns are positive when the target is private; therefore it seems
19
With the OPSW measure, we still find that cash-rich bidders are overvalued, but they are statistically less
overvalued than bidders not in the top quartile.
22
reasonable to assume that using only public targets should, if anything, bias us towards finding support
for the misallocation hypothesis.
Panel A of Table 10 shows the results of the MB decomposition for 672 publicly traded targets, split
almost evenly across stock and cash bids. The columns indicate a split based on the liquidity of the
bidder in the takeover contest. We find that for stock bidders which are not in the top quartile of cash
holdings, the average firm-specific error of the target is 0.0347. For stock bidders which are cash-rich,
the average firm-specific error of the target is 0.1526. This implies that stock bidders are pursuing targets
which are somewhat overvalued. However, Panel B of Table 9 shows that bidders are even more
overvalued than their targets. Combined, these results are consistent with the opportunism hypothesis and
the results in RKRV.
For the misallocation hypothesis to be supported, cash-rich bidders who use cash should target
overvalued firms. However, the average firm-specific error for the targets of those firms is actually
negative and marginally significant at -0.1281. This indicates that the cash-rich bidders appear to be
spending their cash on undervalued targets. Additionally, the estimate is significantly lower than the
valuation of the targets by the bidders who are not in the top quartile of cash holdings. The results are
even more robust when combining firm-specific and sector errors.
We also use bid premiums to examine whether cash-rich bidders are overpaying for their targets. We
define bid premium as (bid offer price / target stock price at time t), where we measure the premium
relative to the target price one day, one week, and four weeks prior to the initial announcement. Panel B
of Table 10 shows the bid premiums offered segmented by bidder liquidity and method of payment.
While it is always the case that bid premiums are higher for cash bids, the important comparison is
across the level of liquidity. We see that for cash bids, there is no difference in bid premium based on the
cash-richness of the bidder. Cash-rich firms do not overpay for targets, controlling for method of
payment. There is weak evidence that cash-rich firms offer higher bid premiums when they pay with
equity, but since they are not spending their cash, it becomes difficult to argue that is supportive of the
misallocation hypothesis.
23
For robustness, we repeat the analyses in Table 10 with all measures of excess cash and find similar
results. Instead of wasting cash on overvalued targets, cash-rich bidders appear to act opportunistically in
an effort to acquire undervalued assets. Further, this opportunism holds for cash-rich bidders regardless
of whether they pay stock or cash. Cash-rich bidders with overvalued equity choose to use it as cheap
currency in acquisitions. Cash-rich bidders with fairly valued equity use their liquidity to target relatively
undervalued firms.
In both cases, the actions of management are in the best interests of existing
shareholders, and corroborate the opportunism hypothesis.
5. Robustness
5.1 Pooling of Interests Accounting
Prior to June 30, 2001, bidding firms could use pooling of interests (hereafter pooling) to account
for stock acquisitions.20 If bidders with high cash systematically prefer to use pooling and pooling also is
characterized by lower acquirer announcement returns, then our results may be explained by the option to
use pooling. While there is some evidence that pooling is valuable (see Robinson and Shane (1990) and
Ayers, Lefanowicz and Robinson (2002)), there is no direct evidence that high cash firms systematically
choose to use pooling. We nonetheless examine whether our results are robust to controlling for pooling.
We present three pieces of evidence that our findings our robust to pooling. First, our multinomial
logistic regressions are already robust since they account for the availability of pooling via the inclusion
of yearly dummy variables. Second, we estimate the abnormal return regressions from Table 7 including a
dummy variable if the year was prior to 2001, and hence pooling was available. Interestingly, we find
that the coefficient on the pooling available dummy is positive for stock bids. This indicates that acquirer
announcement returns were higher for stock bidders when pooling was available, without regard to
whether they actually chose to use pooling. This seems to support the evidence in prior literature that the
option to use pooling accounting could be valuable. More importantly though, our results regarding the
20
Bidders could account for acquisitions using either pooling of interests or the purchase method. While there were
a dozen conditions that needed to be met to qualify for pooling of interests, a key one is that the method of payment
had to be equity.
24
negative bidder announcement returns are unchanged; the lower returns for high cash firms are limited to
those transactions where stock was the method of payment.
Third, if pooling provided an incentive for firms to use equity as the method of payment, and
managers act opportunistically, then to use stock post pooling cash-rich bidders would have to be even
more overvalued. We examine our RKRV decomposition from Table 9, both pre- and post-2001 and find
this to be the case. When pooling was available, there is no significant difference between the
overvaluation of stock bidders based on liquidity (firm-specific error of 0.21 for cash-rich versus 0.21 for
not cash-rich). However, after pooling was disallowed, in order for a cash-rich bidder to use stock, they
needed to be far more overvalued (firm-specific error of 0.32 versus 0.12).
We further find that our
results from Table 10 showing that cash-rich firms which pay with cash do not seem to overpay for
targets holds across both sub-periods.
Overall, while the distribution of bids in Table 2 seems to suggest that the availability of pooling
of interests accounting had an impact on method of payment; our results are robust to controlling for it.
5.2 Do High-Cash Firms have Enough Cash to Finance Acquisitions?
Our primary analyses have examined the relative level of liquidity across firms so as to be
comparable with prior literature. However, two issues potentially arise using that characterization. First,
if in anticipation of undertaking an acquisition, a firm raised the cash through a debt offering, then
monitoring through the debt channel would arise (Schlingemann (2004)), which would bias against
finding support for the misallocation hypothesis.21 Second, even though a firm might be cash-rich, if the
target was large enough, the bidder might still be unable to finance it completely through internal funds
and thus paying through equity might not represent a choice.
To account for both issues, we examine the subset of acquisitions where the bidder had, in the
prior year, at least three times the amount of cash as the value of the acquisition. For simplicity, we refer
to these as the 3X bidders. Focusing on the 3X bidders relieves concerns that debt was issued to fund the
21
It seems unlikely that this would be the impetus for our results though as they are driven by firms which pay with
equity, not with cash. It is hard to believe that firms would raise cash through a debt offering and then pay for the
transaction solely with equity.
25
transaction, and that the firm’s high cash levels are not high enough to finance the transaction. There are
655 bid years that qualify as 3X, segmented into 130 stock bids and 525 cash bids. While we require the
firms to have a minimum of 3 times the cash to bid value, the 3X firms have a mean of almost 7 times and
a median of 5 times. The mere fact that 20% of firms with roughly 5 times the amount of cash needed to
pay, choose to solely use equity seems counter to the misallocation hypothesis.
Nonetheless, we repeat the tests in Table 6 and once again find that these firms are no more likely
to use cash as the method of payment.22 Next, we find that 3X firms which choose to pay with cash do
not have negative announcement returns. Finally, we examine the RKRV overvaluation measures using
the 3X sample. Our results are also similar: cash-rich firms which choose to pay with stock are
significantly overvalued, and more so than the firms which pay with cash. Overall, the 3X subsample
results tend to strongly support our main findings and suggest that the opportunism hypothesis fits the
data considerably better than the misallocation hypothesis.
6. Conclusions
In this paper we show that excess cash impacts the acquisition decision, but not necessarily the
method of payment in the acquisition. The results have important implications for the impact of cash
holdings on investment decisions. That is, though cash holdings do influence the acquisition decision,
managers do act in the best interest of shareholders when deciding on the method of payment.
We start by building on Jensen (1986) and Harford (1999). Jensen (1986) argues that firms should
disgorge free cash flow since managers will waste excess cash on negative NPV investments. Indeed,
Harford (1999) shows that firms with excess cash tend to make more acquisitions than firms without
excess cash; moreover, acquisitions by cash-rich firms have lower bidder returns. The prior interpretation
of this result was that it reinforces the agency arguments made by Jensen. This study first confirms
Harford’s findings, and then attempts to understand the inconsistency between this interpretation and the
observation that firms with excess cash frequently pay with stock.
22
Interestingly, with the 3X subsample, the coefficient on outside industry, which was insignificant in Table 6,
becomes significantly positive. This suggests that for firms which are relatively unconstrained in their choice of
using cash, equity is more likely when information asymmetries are larger, which supports Hansen (1987).
26
The paper examines two competing arguments to explain this inconsistency. The first is the
misallocation hypothesis, in which managers act in their own self-interest and misallocate capital to
negative NPV investments. The competing hypothesis is the opportunism hypothesis, where managers
act in the best interest of shareholders by opportunistically using the firm’s cash and equity.
To answer which is at work, we present three main findings. First, we show that cash-rich firms
are more likely to be bidders, but that the method of payment is more likely to be stock. Second,
controlling for acquirer-target characteristics we find that bidder returns are negative for high-cash firms
only when they pay with stock. These results suggest that firms use stock only when it is overvalued –
signaling this to the market in doing so – and use cash otherwise. Third, using the methodology in RKRV
(2005) we identify over- and undervaluation in both the acquirer and target for deals involving high-cash
bidders. Firms with excess cash use stock as the method of payment rather than using their excess cash
when their own stock is overvalued. Additionally, overvalued stock is used to acquire less-overvalued
targets. Finally, cash acquisitions by cash-rich firms tend to be for undervalued targets.
The results strongly support the opportunism hypothesis. The managers of cash-rich firms act in
the interest of their shareholders when making acquisitions. These cash-rich firms are more likely to make
acquisitions, but they do not appear to be negative-NPV investments. Consistent with managers using the
cheapest source of capital, stock is used only when it is overvalued or relatively overvalued, and cash is
used to target undervalued targets.
In providing evidence for the opportunism hypothesis, this paper focuses strictly on acquisitions
as the sole mechanism for firm investment of excess cash. Though it is possible that managers of highcash firms are wasting excess cash in other ways, they do not appear to do so through acquisitions.
The results in this paper add to the literature on cash holdings and the behavior of managers of
firms with excess cash. Future studies can examine the role of governance and other firm specific
characteristics and its impact on managers’ use of excess cash. Additionally, further examination of the
other channels in which managers may use excess cash would add to our understanding of managerial
behavior and firm cash holdings.
27
References
Ambrose, Brent and William Megginson, 1992, The role of asset structure, ownership structure, and
takeover defenses in determining acquisition likelihood, Journal of Financial and Quantitative
Analysis 27, 575-589.
Amihud, Yakov, Baruch Lev, and Nickolaos Travlos, 1990, Corporate control and the choice of
investment financing: The case of corporate acquisitions, Journal of Finance 45, 603-616.
Ang, James S. and Yingmei Chen, 2006, Direct evidence on the market-driven acquisition theory, Journal
of Financial Research 29, 199-216.
Ayers, Benjamin, Craig Lefanowicz, and John Robinson, 2002, Do firms purchase the pooling method?
Review of Accounting Studies 7, 5-32.
Baker Malcolm, and Jeffrey Wurgler, 2002, Market timing and capital structure, Journal of Finance 57,
1-32.
Betton, S., B. E. Eckbo, and K. Thorburn, 2008, Corporate Takeovers, working paper, Dartmouth
University.
Chang, S., 1998, Takeovers of privately held targets, methods of payment, and bidder returns,
Journal of Finance 53, 773-784.
Comment, Robert and G. William Schwert, 1996, Poison or Placebo? Evidence on the deterrence and
wealth effects of modern antitakeover measures, Journal of Financial Economics 39, 3-43.
Cornett, Marcia Millon and Hassan Tehranian, 1994, An examination of voluntary versus involuntary
security issuances by commercial banks: The impact of capital regulations on common stock returns,
Journal of Financial Economics 35. 99-122.
Dittmar, Amy and Jan Mahrt-Smith, 2007, Corporate governance and the value of cash holdings, Journal
of Financial Economics 83, 599-634.
Dong, Ming, David Hirshleifer, Scott Richardson, and Siew Hong Teoh, 2006, Does investor
misevaluation drive the takeover market? Journal of Finance 61, 725-762.
Easterbrook, Frank, 1984, Two agency-cost explanations of dividends, The American Economic Review
74, 650-659.
Faccio, Mara, and Ronald V. Masulis, 2005, The choice of payment method in European mergers and
acquisitions, Journal of Finance 60, 1345-1388.
Faccio, Mara, John J. McConnell, and David Stolin, 2006, Returns to acquirers of listed and unlisted
targets, Journal of Financial and Quantitative Analysis 41, 197-220.
Faulkender, Michael and Rong Wang, 2006, Corporate financial policy and the value of cash, Journal of
Finance 51, 1957-1990.
Fu, Fanjian, Leming Lin and Micah Officer, 2008, Acquisitions driven by stock overvaluation, working
paper, Loyola Marymount University.
Fuller, Kathleen, Jeffrey Netter, and Mike Stegemoller, 2002, What do returns to acquiring firms tell us?
Evidence from firms that make many acquisitions, Journal of Finance 57, 1763-1793.
28
Graham, John R., and Campbell R. Harvey, 2001, The theory and practice of corporate finance: Evidence
from the field, Journal of Financial Economics 60, 187-243.
Hansen, Robert G., 1987, A theory for the choice of exchange medium in the market for corporate
control, Journal of Business 60, 75-95.
Harford, Jarrad, 1999. Corporate cash reserves and acquisitions. Journal of Finance 54, 1969-1997.
Harford, Jarrad, Sattar A. Mansi, and William F. Maxwell, 2008, Corporate governance and firm cash
holdings in the US, Journal of Financial Economics 87, 535-555.
Jensen, M. C., 1986, Agency costs of free cash flow, corporate finance and takeovers, American
Economic Review, 323-329
Jensen, M. C. and W. H. Meckling, 1976, Theory of the firm: Managerial behavior, agency costs, and
ownership structure, Journal of Financial Economics 3, 305-360.
Jung Kooyul, Yong-Cheol Kim, and Rene Stulz, 1996, Timing, investment opportunities, managerial
discretion, and the security issue decision, Journal of Financial Economics 42, 159-185
Kim, Chang-Soo, David C. Mauer and Ann E. Sherman, 1998, The determinants of corporate liquidity,
Journal of Financial and Quantitative Analysis 33, 335-359.
Long, J. Scott and Jeremy Freese, 2006, Regression Models for Categorical Dependent Variables Using
Stata, 2nd edition, Stata Press, College Station, Texas.
Martin, K, 1996, The method of payment in corporate acquisitions, investment opportunities, and
management ownership, Journal of Finance 51, 1227-1246.
Mikkelson, W. H. and M. Partch, 2002, Do persistent large cash reserves hinder performance, Journal of
Financial and Quantitative Analysis, forthcoming.
Moeller, Sara B., Frederik P. Schlingemann, and Rene M. Stulz, 2004, Firm size and the gains from
acquisitions, Journal of Financial Economics 73, 201-228.
Myers, Stewart C., 1984, The capital structure puzzle, Journal of Finance 39, 575-592.
Myers, S. and N. Majluf, 1984, Corporate financing and investment decisions when firms have
information that investors do not have, Journal of Financial Economics 13, 187-221.
Office, Micah S., Annette B. Poulsen, and Mike Stegemoller, 2007, Information asymmetry and acquirer
returns, Review of Financial Economy forthcoming.
Opler, T., L. Pinkowitz, R. M. Stulz and R. Williamson, 1999, The determinants and implications of
corporate cash holdings, Journal of Financial Economics 52, 3-46.
Petersen, Mitchell A., 2009, Estimating standard errors in finance panel data sets: Comparing approaches,
Review of Financial Studies 22, 435-480.
Pinkowitz, Lee, René Stulz, and Rohan Williamson, 2006, Does the contribution of corporate cash
holdings and dividends to firm value depend on governance? A cross-country analysis, Journal of
Finance 51, 2725-2751.
Pinkowitz, Lee and Rohan Williamson, 2007, What is the market value of a dollar of corporate cash?
Journal of Applied Corporate Finance 19, 74-81.
29
Rhodes-Kropf, Matthew, and David T. Robinson, and S. Viswanathan, 2005, Valuation waves and
merger activity: The empirical evidence, Journal of Financial Economics 77, 561-603.
Rhodes-Kropf, Matthew, and S. Viswanathan, 2004, Market valuation and merger waves, Journal of
Finance 59, 2685-2718.
Robinson, John R. and Philip B. Shane, 1990, Acquisition accounting method and bid premia for target
firms, The Accounting Review 65, 25-48.
Schlingemann, Frederik P., 2004, Financing decisions and bidder gains, Journal of Corporate Finance
10, 683-701.
Shleifer, Andrei, and Robert Vishny, 2003, Stock market driven acquisitions, Journal of Financial
Economics 70, 295-311.
Stulz, Rene, 1988, Managerial control of voting rights: Financing policies and the market for corporate
control, Journal of Financial Economics 20, 25-54.
Stulz, Rene, 1990, Managerial discretion and optimal financing policies, Journal of Financial Economics
26, 3-27.
Travlos, Nickolaos G., 1987, Corporate takeover bids, method of payment, and bidding firms' stock
returns, Journal of Finance 42, 943-963.
30
Table 1
Variable Definitions Used in the Analyses
All variables are measured at time t unless otherwise subscripted.
Variable
Description
Calculation and Compustat Data Codes
Cash
Cash and Marketable securities to assets
che / at
Excess OPSW
OPSW excess liquid assets of the acquiring
firm deflated by its asset base
OPSW regression residual
(see appendix)
Excess Harford
Harford excess liquid assets of the acquiring
firm deflated by its sales
Harford regressions residuals
(see appendix)
Excess industry
Excess liquid assets based on median of
industry for the year
Cash/assets – Median Industry Cash/Assets
(see appendix)
High (Excess)
Cash
Top Quartile by year of (Excess) Cash
Measure
Standardized
(Excess) Cash
Liquidity measure standardized yearly to
have zero mean and unit variance
(Cash Measureit - μt) / σt
Abret
4 year average of 12 month buy and hold
return of the bidder above the 12 month buy
and hold return of the appropriate 5x5 FamaFrench size and MB portfolio
12 month buy and hold return – 12 month
buy and hold return of Fama-French 25
portfolio
Growth
4 year average of Growth rate in sales of the
acquiring firm
(Sale – Salet-1) / Salet-1
NWC
4 year average of Non-cash net working
capital deflated by assets
(act – lct – che) / at
PE
4 year average of Price Earnings Ratio
prcc_f / epspi
MB
4 year average of Market to book ratio of the
firm’s equity
(csho * prcc_f) / (at – lt – mib)
Leverage
4 year average of Book value of debt
divided by market value of equity
(dltt + dlc) / (csho * prcc_f)
Size
Size of the acquiring firm’s real asset base
(year 2000 dollars)
ln((at/CPI)*100)
31
Table 2
Distribution of 58,117 firms years through time
Firm years are classified as no bid if they do not undertake any acquisition in that year. We classify firm
years as a Stock Bid (Cash Bid) if all of the consideration offered for acquisition bids within that year is
comprised of equity (cash) components. Bids are classified as Mixed if both equity and cash are offered
in acquisitions bids
Year
No bid
All Stock Bids
All Cash Bids
Mixed Bids
Total Bids
% Stock
1984
2,124
26
93
19
138
18.84%
1985
2,195
25
120
20
165
15.15%
1986
2,225
24
96
27
147
16.33%
1987
2,127
25
131
27
183
13.66%
1988
2,194
27
120
21
168
16.07%
1989
2,240
25
97
21
143
17.48%
1990
2,127
29
78
24
131
22.14%
1991
2,155
41
78
24
143
28.67%
1992
2,209
50
112
41
203
24.63%
1993
2,352
79
129
53
261
30.27%
1994
2,444
107
164
54
325
32.92%
1995
2,603
146
181
67
394
37.06%
1996
2,749
122
214
86
422
28.91%
1997
2,762
139
253
86
478
29.08%
1998
2,816
136
220
73
429
31.70%
1999
2,820
133
154
55
342
38.89%
2000
2,520
64
149
57
270
23.70%
2001
2,392
37
171
58
266
13.91%
2002
2,299
37
198
53
288
12.85%
2003
2,376
37
260
68
365
10.14%
2004
2,307
22
259
72
353
6.23%
2005
2,113
23
270
61
354
6.50%
Total
52,149
1,354
3,547
1,067
5,968
22.69%
32
Table 3
Summary Statistics
Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression
predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the
industry median for that year, where industry is defined using the Fama-French 30 industries.
Standardized (excess) cash measures are transformed to have zero mean and unit variance each year.
High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a given year.
Abret is the 4 year average excess buy and hold return benchmarked against the appropriate 5x5 FamaFrench size and MB portfolio. Growth is the four year average of sales growth. NWC is the four year
average of non-cash net working capital deflated by assets. PE is the four year average price earnings
ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is the four
year average book value of debt deflated by market equity. Size is the natural logarithm of the firm’s real
assets deflated to year 2000 dollars using the CPI.
Panel A:
N
Mean
25th
Percentile
Median
75th
Percentile
Standard
Deviation
Cash
58,117
0.140
0.020
0.067
0.195
0.173
Excess Harford
58,117
0.033
-0.121
-0.047
0.025
0.582
Excess OPSW
58,117
0.056
-0.024
0.010
0.125
0.731
Excess Industry
58,117
0.043
-0.036
-0.001
0.095
0.155
Standardized Cash
58,117
0.000
-0.693
-0.422
0.338
1.000
Standardized Harford
58,117
0.000
-0.366
-0.166
-0.007
1.000
Standardized OPSW
58,117
0.000
-0.190
-0.055
0.114
1.000
Standardized Industry
58,117
0.000
-0.550
-0.282
0.342
1.000
High Cash
58,117
0.250
High Harford
58,117
0.250
High OPSW
58,117
0.250
High Industry
58,117
0.250
Abret
58,117
0.056
-0.169
-0.007
0.191
0.416
Growth
58,117
0.249
0.037
0.119
0.267
0.524
NWC
58,117
0.138
0.008
0.130
0.262
0.176
PE
58,117
15.304
3.501
13.412
22.642
28.500
MB
58,117
2.679
1.255
1.963
3.202
2.602
Leverage
58,117
0.532
0.063
0.241
0.621
0.857
Size
58,117
5.373
3.995
5.211
6.576
1.837
33
Table 4
Summary Statistics Split by Bidder Activity and Method of Payment
Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass regression
predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the
industry median for that year, where industry is defined using the Fama-French 30 industries.
Standardized (excess) cash measures are transformed to have zero mean and unit variance each year.
High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a given year.
Abret is the 4 year average excess buy and hold return benchmarked against the appropriate 5x5 FamaFrench size and MB portfolio. Growth is the four year average of sales growth. NWC is the four year
average of non-cash net working capital deflated by assets. PE is the four year average price earnings
ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is the four
year average book value of debt deflated by market equity. Size is the natural logarithm of the firm’s real
assets deflated to year 2000 dollars using the CPI.
Bidders (n=5,968)
Non – Bidders (n=52,149)
Mean
Median
Mean
Median
***
***
Cash
0.154
0.073
0.138
0.066
***
*
Standardized Cash
0.056
-0.401
-0.006
-0.425
Standardized Harford
0.024**
-0.153***
-0.003
-0.168
***
***
Standardized OPSW
0.046
-0.040
-0.005
-0.057
***
***
Standardized Industry
0.035
-0.254
-0.004
-0.287
***
High Cash
0.273
0.247
0.000
***
High Harford
0.283
0.246
0.000
***
High OPSW
0.272
0.247
0.000
**
High Industry
0.263
0.248
0.000
***
***
Abret
0.159
0.067
0.044
-0.016
Growth
0.309***
0.161***
0.242
0.116
***
NWC
0.108
0.097
0.141
0.135***
PE
18.01***
15.64***
14.99
13.14
***
***
MB
3.10
2.38
2.63
1.92
***
0.244***
Leverage
0.425
0.211
0.544
Size
6.23***
6.14***
5.28
5.09
***,**, and * indicate significant differences between bidders and non-bidders at the mean or median at
the 1, 5, and 10% levels respectively.
Panel A: Split by Bidder
Activity
34
Table 4 (Continued)
Summary Statistics Split by Bidder Activity and Method of Payment
Panel B: Just Bidders
(n=5,968)
Cash
Standardized Cash
Standardized Harford
Standardized OPSW
Standardized Industry
High Cash
High Harford
High OPSW
High Industry
Abret
Growth
NWC
PE
MB
Leverage
Size
All Stock Bids (n=1,354)
Mean
Median
***
0.198
0.121***
0.365***
-0.071***
0.181***
-0.130***
-0.024
-0.031***
0.231***
-0.141***
0.391***
0.360***
0.324***
0.350***
0.243***
0.125***
0.430***
0.241***
0.098
0.092
***
20.49
17.86***
4.05***
3.05***
0.277
0.120
6.02
5.88
All Cash Bids (n=3,547)
Mean
Median
0.130
0.058
-0.097
-0.499
-0.062
-0.164
*
0.052
-0.047
-0.063
-0.284
0.215
0.239
0.239
0.221
0.119
0.050
0.240
0.131
***
0.113
0.100
17.24
15.28
2.72
2.18
***++
0.479
0.259***+++
***+++
6.46
6.38***+++
Mixed Bids (n=1,067)
Mean
Median
+++
0.179
0.093+++
0.171+++
-0.296+++
0.109+++
-0.138+++
0.118++
-0.032+++
0.114+++
-0.232+++
0.318+++
0.329+++
0.315+++
0.295+++
0.184+++
0.074++
0.384+++
0.210+++
0.104
0.089
17.42
15.10
+++
3.14
2.36+++
0.432
0.207
5.72
5.60
***,**, and * indicate significant differences between all stock and all cash bids at the mean or median at the 1, 5,
and 10% levels respectively. +++,++, and + indicate significant differences between Mixed and All Cash at the 1, 5,
and 10% levels respectively.
35
Table 5
Multinomial Logistic analysis of determinants of bidding
Dependent variable set to 0 if there is no bid in the subsequent fiscal year; set to 1 if there is at least 1 bid
in the subsequent fiscal year and all are completely financed with stock; set to 2 if there is at least 1 bid in
the subsequent fiscal year and all are financed completely with cash; and set to 3 if there is at least one
bid in the subsequent fiscal year and the financing contains both cash and equity components. The
sample is 58,117 firm-years comprised of 1,354 stock bids, 3,547 cash bids, 1,067 mixed bids, and 52,149
non-bids. Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass
regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is
cash/assets less the industry median for that year, where industry is defined using the Fama-French 30
industries. Standardized (excess) cash measures are transformed to have zero mean and unit variance
each year. High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a
given year. Abret is the 4 year average excess buy and hold return benchmarked against the appropriate
5x5 Fama-French size and MB portfolio. Growth is the four year average of sales growth. NWC is the
four year average of non-cash net working capital deflated by assets. PE is the four year average price
earnings ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is
the four year average book value of debt deflated by market equity. Size is the natural logarithm of the
firm’s real assets deflated to year 2000 dollars using the CPI. All regressions are run including industry
and year dummy variables. Industries are defined as the 30 Fama-French industries. Standard errors are
clustered by firm. P-values based on the clustered standard errors are reported in brackets. ***,**, and *
indicate significant differences between All Stock and All Cash at the 1, 5, and 10% levels respectively.
+++ ++
, , and + indicate significant differences between Mixed and All Cash at the 1, 5, and 10% levels
respectively.
(1)
Panel A:
High Cash
All Stock
0.345***
[0.000]
All Cash
0.096
[0.080]
(2)
Mixed
0.276+
[0.001]
Standardized
Cash
All Stock
All Cash
Mixed
0.156***
[0.000]
0.036
[0.177]
0.124++
[0.001]
Abret
0.578
[0.000]
0.524
[0.000]
0.521
[0.000]
0.571
[0.000]
0.524
[0.000]
0.518
[0.000]
Growth
0.200**
[0.000]
0.101
[0.012]
0.241+++
[0.000]
0.193*
[0.000]
0.100
[0.013]
0.235+++
[0.000]
NWC
-0.288**
[0.245]
0.369
[0.021]
-0.044
[0.857]
-0.171*
[0.510]
0.375
[0.022]
0.023
[0.926]
PE
0.001
[0.181]
0.001
[0.212]
0.001
[0.178]
0.001
[0.149]
0.001
[0.205]
0.001
[0.154]
MB
0.040***
[0.000]
-0.037
[0.000]
-0.010++
[0.388]
0.039***
[0.000]
-0.037
[0.000]
-0.010++
[0.391]
Leverage
-0.474***
[0.000]
-0.126
[0.000]
-0.093
[0.039]
-0.461***
[0.000]
-0.127
[0.000]
-0.088
[0.048]
Size
0.322
[0.000]
0.346
[0.000]
0.176+++
[0.000]
0.326
[0.000]
0.346
[0.000]
0.178+++
[0.000]
Pseudo R2
0.0783
0.0784
36
Table 5 (Continued)
Multinomial Logistic analysis of determinants of bidding
Panel B: Alternative Liquidity Measures
All Stock
All Cash
Top Quartile by Year
0.119
0.363***
High Harford
[0.000]
[0.017]
0.224
0.194
High OPSW
[0.001]
[0.000]
0.311***
0.089
High Industry
[0.000]
[0.079]
Continuous Measures, Standardized
0.048***
-0.029
Standardized Harford
[0.045]
[0.211]
0.054
0.122*
Standardized OPSW
[0.031]
[0.000]
***
0.144
0.013
Standardized Industry
[0.000]
[0.597]
37
Mixed
0.343+++
[0.000]
0.330+
[0.000]
0.196
[0.013]
0.052++
[0.063]
0.180
[0.000]
0.100++
[0.003]
Table 6
Multinomial Logistic analysis of determinants of Method of Payment for First Bids
Dependent variable set to 1 if the bid is completely financed with stock; set to 2 if financed completely
with cash; and set to 3 if the financing contains both cash and equity components. The sample is 5,968
bids comprised of 1,354 stock bids, 3,547 cash bids, and 1,067 mixed bids. The cash alternative is the
baseline. Cash is cash/assets. Excess Harford [OPSW] is defined as the residual from a first pass
regression predicting cash holdings based on Harford (1999) [OPSW (1999)]. Excess Industry is
cash/assets less the industry median for that year, where industry is defined using the Fama-French 30
industries. Standardized (excess) cash measures are transformed to have zero mean and unit variance
each year. High (excess) cash measures indicate the firm is in the top quartile of cash holdings within a
given year. Abret is the 4 year average excess buy and hold return benchmarked against the appropriate
5x5 Fama-French size and MB portfolio. Growth is the four year average of sales growth. NWC is the
four year average of non-cash net working capital deflated by assets. PE is the four year average price
earnings ratio. MB is the four year average of the market to book ratio of the firm’s equity. Leverage is
the four year average book value of debt deflated by market equity. Size is the natural logarithm of the
firm’s real assets deflated to year 2000 dollars using the CPI. Relative value is defined as (deal value /
(bidder market capitalization + deal value)), Deal Size is the natural logarithm of deal value deflated to
year 2000 dollars using the CPI, Unsolicited is an indicator variable which is set to one if SDC describes
the deal attitude as starting off unsolicited, Auction is an indicator variable set to one if there is more than
one bidder in the control contest, Defense is set to one if the target uses some sort of takeover defense,
Private is set to one if SDC indicates that the target is private, Subsidiary is set to one if SDC indicates
that the target is a subsidiary, and Outside Industry is set to one if the bidder and target are from different
industries as defined by the Fama-French 30 classification. All regressions are run including industry and
year dummy variables. Industries are defined as the 30 Fama-French industries. Standard errors are
clustered by firm. P-values based on the clustered standard errors are reported in brackets.
38
(1)
Panel A:
High Cash
Stock vs. Cash
0.258
[0.006]
(2)
Mixed vs. Cash
0.214
[0.026]
Standardized Cash
Abret
Growth
NWC
PE
MB
Leverage
Relative Value
Deal Size
Unsolicited
Auction
Defense
Private
Subsidiary
Outside Industry
Pseudo R2
0.349
[0.000]
0.350
[0.000]
-0.248
[0.447]
-0.001
[0.639]
0.096
[0.000]
-0.451
[0.000]
2.650
[0.000]
0.170
[0.000]
-1.341
[0.000]
-1.010
[0.000]
-1.811
[0.003]
-0.711
[0.000]
-2.509
[0.000]
0.051
[0.541]
0.229
[0.014]
0.372
[0.000]
0.095
[0.757]
0.000
[0.877]
0.041
[0.018]
-0.149
[0.023]
3.102
[0.000]
0.035
[0.264]
-0.272
[0.257]
0.079
[0.688]
-0.869
[0.039]
0.150
[0.164]
-1.064
[0.000]
-0.104
[0.204]
0.1935
Stock vs. Cash
Mixed vs. Cash
0.142
[0.001]
0.340
[0.000]
0.349
[0.000]
-0.147
[0.657]
-0.001
[0.682]
0.094
[0.000]
-0.434
[0.000]
2.622
[0.000]
0.174
[0.000]
-1.333
[0.000]
-1.019
[0.000]
-1.849
[0.003]
-0.713
[0.000]
-2.509
[0.000]
0.052
[0.532]
0.112
[0.010]
0.224
[0.017]
0.371
[0.000]
0.160
[0.608]
0.000
[0.845]
0.039
[0.024]
-0.141
[0.031]
3.079
[0.000]
0.038
[0.230]
-0.268
[0.266]
0.074
[0.707]
-0.900
[0.034]
0.149
[0.167]
-1.063
[0.000]
-0.104
[0.204]
0.1939
39
Panel B: Alternative Liquidity Measures
Stock vs. Cash
Mixed vs. Cash
Top Quartile By Year
High Harford
0.276
[0.002]
0.275
[0.001]
High OPSW
0.026
[0.773]
0.131
[0.138]
High Industry
0.229
[0.011]
0.164
[0.070]
Continuous Measures, Standardized
Standardized Harford
0.142
[0.000]
0.133
[0.001]
Standardized OPSW
-0.073
[0.056]
0.022
[0.601]
Standardized Industry
0.137
[0.000]
0.101
[0.010]
Continuous Measures, Split by Positive versus Negative
Standardized Pos. Cash
0.103
[0.062]
0.088
[0.136]
Standardized Neg. Cash
-0.299
[0.058]
-0.190
[0.198]
Standardized Pos. Harford
0.092
[0.053]
0.089
[0.069]
Standardized Neg. Harford
-0.453
[0.017]
-0.402
[0.029]
Standardized Pos. OPSW
0.009
[0.903]
0.029
[0.675]
Standardized Neg. OPSW
0.115
[0.035]
-0.018
[0.799]
Standardized Pos. Industry
0.086
[0.141]
0.046
[0.454]
Standardized Neg. Industry
-0.256
[0.027]
-0.226
[0.037]
40
Table 7
Regression Analysis of Abnormal Returns
Dependent variable is bidder cumulative abnormal returns over the -1 to 1 period using the market model
with parameters estimated from days -370 to -253. Cash is cash/assets. Excess Harford [OPSW] is
defined as the residual from a first pass regression predicting cash holdings based on Harford (1999)
[OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year, where industry is
defined using the Fama-French 30 industries. Standardized (excess) cash measures are transformed to
have zero mean and unit variance each year. High (Excess) Cash equals 1 if the firm is in the top quartile
of (excess) cash holdings in a given year and zero otherwise. Run up is the bidder cumulative abnormal
returns over the period -252 to -20, where day zero is the announcement of the bid. Leverage is the
bidder’s leverage. Low Q is an indicator variable set to one if the bidder’s market to book value of assets
is less than 1. Outside Industry is an indicator variable equaling 1 if the bidder and the target are in
different industries as measured by the 30 Fama-French industries. Deal Size is the logarithm of real
value of the deal, deflated by the CPI. Relative value is the value of the transaction deflated by the market
capitalization of the bidder. Unsolicited is an indicator variable set to one if the bid is indicated as hostile
or unsolicited by SDC. Tender Offer is equals 1 if the bid is through a tender offer and zero otherwise.
Auction is an indicator variable set to one if there is more than one bidder in the control contest. Defense
is set to one if the target uses some sort of takeover defense. Private target equals 1 if the target is listed
as private by SDC, and zero otherwise. Subsidiary is set to one if SDC indicates that the target is a
subsidiary. Unexpected Bidder is set to one if the acquisition is unexpected based on a first pass logistic
regression predicting bidders. Numbers in brackets are p-values based on robust standard errors clustered
at the firm level. ***,**, and * indicate significant differences between all stock and all cash bids at the
mean or median at the 1, 5, and 10% levels respectively. +++,++, and + indicate significant differences
between Mixed and All Cash at the 1, 5, and 10% levels respectively.
41
Panel A:
Full Sample
All Stock
All Cash
Mixed
High Cash
-0.008
[0.006]
-0.021
[0.001]
-0.001
[0.674]
-0.011
[0.143]
Run up
0.006
[0.001]
0.008
[0.019]
0.004
[0.127]
0.005
[0.224]
Leverage
0.001
[0.852]
-0.005
[0.774]
-0.002
[0.801]
0.018
[0.329]
Low Q
-0.006
[0.016]
0.002
[0.825]
-0.007
[0.008]
-0.009
[0.211]
Outside Industry
0.001
[0.635]
0.009*
[0.135]
-0.003
[0.160]
0.001
[0.846]
Deal Size
-0.005
[0.000]
-0.005
[0.012]
-0.003
[0.000]
-0.007
[0.006]
Relative Value
0.083
[0.000]
0.058
[0.010]
0.085
[0.000]
0.105
[0.000]
Unsolicited
-0.001
[0.830]
0.000
[0.996]
-0.012
[0.095]
0.001
[0.935]
Tender Offer
0.019
[0.000]
0.007
[0.657]
0.002
[0.668]
-0.013
[0.209]
Auction
-0.002
[0.679]
-0.012
[0.461]
-0.005
[0.400]
0.010
[0.316]
Defense
-0.017
[0.086]
-0.047
[0.305]
-0.016
[0.169]
-0.010
[0.591]
Private Target
0.032
[0.000]
0.056***
[0.000]
0.006
[0.145]
0.028++
[0.004]
Subsidiary
0.033
[0.000]
0.051***
[0.000]
0.008
[0.061]
0.049+++
[0.000]
Unexpected Bidder
0.004
[0.188]
0.004
[0.604]
0.005
[0.083]
-0.002
[0.808]
Constant
-0.005
[0.399]
-0.010
[0.487]
0.010
[0.118]
0.003
[0.863]
N
5,900
1,334
3,509
1,057
Adjusted R2
0.0611
0.0932
0.0475
0.0667
***
42
Panel B
Full Sample
All Stock
All Cash
Mixed
Top Quartile By Year
High Harford
-0.006
[0.026]
-0.020***
[0.001]
-0.000
[0.893]
-0.005
[0.487]
High OPSW
-0.004
[0.098]
-0.015**
[0.026]
0.001
[0.858]
-0.009
[0.242]
High Industry
-0.002
[0.467]
-0.013**
[0.048]
0.002
[0.470]
-0.002
[0.832]
Continuous Measures, Standardized
Standardized Cash
-0.003
[0.021]
-0.009**
[0.004]
-0.001
[0.413]
-0.002
[0.513]
Standardized Harford
-0.003
[0.011]
-0.006*
[0.001]
-0.001
[0.467]
-0.002
[0.602]
Standardized OPSW
0.001
[0.395]
0.002
[0.398]
-0.002
[0.370]
0.004
[0.241]
Standardized Industry
-0.003
[0.065]
-0.007
[0.029]
-0.001
[0.362]
-0.001
[0.781]
43
Table 8
Overvaluation Decomposition Segmented By Liquidity
Overvaluation is calculated using the RKRV (2005) methodology. Firm-specific error, Time-Series
Sector Error and Long-Run Value to Book are defined as in RKRV (see Table 5 of their paper, page 579).
Total Error is the sum of Firm-Specific Error and Time-Series Sector Error. Cash is cash/assets. Excess
Harford [OPSW] is defined as the residual from a first pass regression predicting cash holdings based on
Harford (1999) [OPSW (1999)]. Excess Industry is cash/assets less the industry median for that year,
where industry is defined using the Fama-French 30 industries. High (Excess) Cash signifies the firm is
in the top quartile of (excess) cash holdings in a given year. Numbers in parentheses represent the
number of observations for each classification. ***,**, and * indicate the mean is significantly different
from 0 at the 1, 5, and 10% levels respectively, using a t-test. P-values within columns are calculated
using a t-test correcting for unequal variances.
Firm-Specific
Error
Time-Series
Sector Error
Total Error
Long-Run
Value to Book
High Cash (14,269)
Not High Cash (42,270)
-0.0692***
0.0234***
0.0280***
0.0080***
-0.0412***
0.0313***
0.9547***
0.5503***
P-value High vs. Not
[0.0000]
[0.0000]
[0.0000]
[0.0000]
High Harford (14,221)
Not High Harford (42,318)
-0.0732***
0.0246***
0.0198***
0.0107***
-0.0535***
0.0354***
0.8512***
0.5856***
P-value High vs. Not
[0.0000]
[0.0000]
[0.0000]
[0.0000]
High OPSW (14,211)
Not High OPSW (42,328)
-0.1454***
0.0488***
0.0217***
0.0101***
-0.1237***
0.0589***
0.8299***
0.5928***
P-value High vs. Not
[0.0000]
[0.0000]
[0.0000]
[0.0000]
High Industry (14,237)
Not High Industry (42,302)
-0.0760***
0.0256***
0.0195***
0.0109***
-0.0565***
0.0364***
0.9199***
0.5623***
P-value High vs. Not
[0.0000]
[0.0000]
[0.0000]
[0.0000]
44
Table 9
Overvaluation Decomposition Segmented By Liquidity and Acquisition Activity
Overvaluation is calculated using the RKRV (2005) methodology. Firm-specific error, Time-Series
Sector Error and Long-Run Value to Book are defined as in RKRV (see Table 5 of their paper, page 579).
Total Error is the sum of Firm-Specific Error and Time-Series Sector Error. Cash is cash/assets. High
Cash signifies the firm is in the top quartile of cash holdings in a given year. Firm years are classified as
non-bidders if they undertake no acquisition in that year. We classify firm years as a Stock Bid (Cash
Bid) if all of the consideration offered for acquisition bids within that year is comprised of equity (cash)
components. Numbers in parentheses represent the number of observations for each classification.
***,**, and * indicate the mean is significantly different from 0 at the 1, 5, and 10% levels respectively,
using a t-test. P-values within columns and across rows are calculated using a t-test correcting for
unequal variances.
Not High Cash
(N = 42,270)
High Cash
(N = 14,269)
Non – Bidders (50,672)
Bidders (5,867)
38,016
4,254
12,656
1,613
Non – Bidders
Bidders
0.0155***
0.0932***
-0.0852***
0.0567***
P-Value Bid vs. Non
[0.0000]
[0.0000]
Non – Bidders
Bidders
***
0.0041
0.0423***
0.0225***
0.0718***
P-Value Bid vs. Non
[0.0000]
[0.0000]
Panel A
# of Observations
Firm-Specific Error
Time- Series Sector
Error
Non – Bidders
Bidders
0.0196
0.1355***
-0.0628***
0.1284***
P-Value Bid vs. Non
[0.0000]
[0.0000]
Non – Bidders
Bidders
***
0.5357
0.6811***
0.9514***
0.9806***
P-Value Bid vs. Non
[0.0000]
[0.0145]
Not High Cash
(N = 3,542)
High Cash
(N = 1,277)
Stock Bid (1,332)
Cash Bid (3,487)
809
2,733
523
754
Stock Bid
Cash Bid
0.2155***
0.0601***
0.2199***
-0.0334*
P-Value Stock vs. Cash
[0.0000]
[0.0000]
Stock Bid
Cash Bid
***
0.0661
0.0340***
0.1100***
0.0410***
P-Value Stock vs. Cash
[0.0000]
[0.0000]
Stock Bid
Cash Bid
***
0.2816
0.0941***
0.3299***
0.0075
P-Value Bid vs. Non
[0.0000]
[0.0000]
Stock Bid
Cash Bid
***
0.7593
0.6654***
1.1043***
0.9028***
P-Value Stock vs. Cash
[0.0000]
[0.0000]
Total Error
Long-Run Value to
Book
***
Panel B
# of Observations
Firm-Specific Error
Time- Series Sector
Error
Total Error
Long-Run Value to
Book
45
P-value High
Cash vs. Not
[0.0000]
[0.0271]
[0.0000]
[0.0000]
[0.0000]
[0.6964]
[0.0000]
[0.0000]
P-value High
Cash vs. Not
[0.8932]
[0.0000]
[0.0004]
[0.3949]
[0.1833]
[0.0002]
[0.0000]
[0.0000]
Table 10
Examination of Valuation of Public Targets
Overvaluation is measured for the target firms and is calculated using the RKRV (2005) methodology.
Firm-specific error, Time-Series Sector Error and Long-Run Value to Book are defined as in RKRV (see
Table 5 of their paper, page 579). Total Error is the sum of Firm-Specific Error and Time-Series Sector
Error. High Cash signifies the bidding firm is in the top quartile of cash/assets in a given year. We
classify firm years as a Stock Bid (Cash Bid) if all of the consideration offered for acquisition bids within
that year is comprised of equity (cash) components. Bid premiums are measured using the offer price
relative to the target stock price at period t before the bid. Numbers in parentheses represent the number
of observations for each classification. ***,**, and * indicate significant differences from 0 at the 1, 5,
and 10% levels respectively, using a t-test. P-values across rows and columns are calculated using a t-test
correcting for unequal variances.
RKV Decomposition for Target
High Cash
(N = 161)
89
72
0.1526**
-0.1281*
[0.0073]
0.0671**
-0.0333
[0.0056]
0.2197***
-0.1614**
[0.0005]
0.8855***
0.7080***
[0.0346]
P-value high
cash vs. not
Stock Bid (331)
Cash Bid (341)
Stock Bid
Cash Bid
P-Value Stock vs. Cash
Stock Bid
Cash Bid
P-Value Stock vs. Cash
Stock Bid
Cash Bid
P-Value Stock vs. Cash
Stock Bid
Cash Bid
P-Value Stock vs. Cash
Not High Cash
(N = 511)
242
269
0.0347
0.0037
[0.5415]
0.0472***
0.0233**
[0.1287]
0.0820**
0.0270
[0.3031]
0.6866***
0.5329***
[0.0009]
Bid Premium of offer price to stock price at
period t before bid
Stock Bid (504)
# of Observations
Cash Bid (469)
Stock Bid
Cash Bid
1 Day
P-Value Stock vs. Cash
Stock Bid
Cash Bid
1 Week
P-Value Stock vs. Cash
Stock Bid
Cash Bid
4 Weeks
P-Value Stock vs. Cash
Not High Cash
(N = 704)
343
361
30.26
47.52
[0.0001]
33.65
52.46
[0.0005]
40.05
59.82
[0.0004]
High Cash
(N = 269)
161
108
35.61
44.99
[0.0678]
41.44
52.88
[0.0260]
47.11
67.32
[0.0853]
P-value high
cash vs. not
# of Observations
Firm-Specific
Error
Time- Series
Sector Error
Total Error
Long-Run Value
to Book
46
[0.1093]
[0.0951]
[0.4178]
[0.0201]
[0.1148]
[0.0240]
[0.0020]
[0.0133]
[0.1519]
[0.6883]
[0.0443]
[0.9479]
[0.1117]
[0.5373]
Appendix – Excess Cash Measures
We construct three measures of abnormal cash holdings which we refer to as Excess Cash: Harford,
OPSW, or Industry.
A.
Excess Cash - Harford
The Harford measure follows equation (1) of Harford (1999) where we run the following regressions:
Cash
Salesi,t
= a i + b1 NetCFO
+ b5 NetCFO
Salesi,t
Salesi,t+2
+ b 2 ΔRisk Premium t+1 + b3 Recession t + b 4 NetCFO
+ b6 M
Bi,t-1
Salesi,t+1
+ b 7 CFOVari + b8Sizei,t-1 + ε i,t
where: NetCFO is operating cash flow net of investments23, Risk premium is the difference between rates
of AAA and BAA bonds, Recession is a dummy variable set to one for years within recessions as defined
by the National Bureau of Economic Research, M/B is the firm’s market to book of assets ratio24,
CFOVar is the coefficient of variation using the standard deviation and mean of the firm’s cash flow for
up to the prior 10 years, and Size is the natural log of market value of the equity deflated to 2000 dollars
using the CPI. As in Harford (1999), the model is separately for each of the Fama-French industries and
is estimated with firm-specific fixed effects. The Excess Cash – Harford measure is constructed as the
deviation from the industry average prediction (see Harford (1999) page 1977).
B.
Excess Cash – OPSW
We also construct a measure of excess cash based on the model in Opler, et al (1999), where we run the
following regression:
ln ( Cash ) = α + β1MB + β 2Size + β3Cash Flow + β 4 NWC + β5 RD
+ β6 Indsigma + β7 Leverage + β8Capex + β9 Dividends + ε
where cash is defined as (cash and marketable securities / net assets). Net assets are defined as book
value of assets less cash. MB is (market value of equity – book value of equity + book value of assets) /
(net assets), Size is the log of assets deflated to 2000 dollars, Cash Flow is (operating income before
depreciation – interest – taxes – common dividends) / net assets, NWC is non-cash net working capital
deflated by net assets, RD is research and development expenses deflated by sales, Indsigma is the
median of the industry’s cash flow volatility for up to the prior 10 years, Leverage is defined as (short
term debt + long term debt) / net assets, Capex is capital expenditures deflated by net assets, and
Dividends is a dummy variable which equals one if the firm paid common dividends within the year.
Following Opler, et al (1999), when R&D is missing, it is set to zero. The model is estimated each year
to allow the coefficients to vary across time. The Excess Cash – OPSW measure is the deviation from the
exponential of the prediction of the model.
C.
Excess Cash – Industry
Finally, we construct a simple measure of excess cash holdings using the deviation from the annual
industry median. Industries are defined using the Fama-French 30 industry classifications.
23
Operating cash flow is operating income before depreciation – interest – taxes – change in noncash working
capital (ebitda – xint – txt – Δ(act – lct – che)). Investments are defined as capital expenditures (capx)
24
Constructed as (market value of equity + short term debt + long term debt)/ book value of assets.
47