Cash is King –

Revaluation of Targets after Merger Bids I

Ulrike Malmendiera, Marcus M. Oppb, Farzad Saidic

aDepartment of Economics and Haas School of Business, University of California at Berkeley, 501Evans Hall, Berkeley, CA 94720. Phone: +1-510-642-5038. Fax: +1-510-642-6615.

bHaas School of Business, University of California at Berkeley, 545 Student Services Building #1900,CA 94720-1900. Phone: +1-510-643-0658. Fax: +1-510-643-1420.

cDepartment of Economics, New York University, 19 West 4th Street, New York, NY 10012. Phone:+1-212-992-9773. Fax: +1-212-995-4186.

Abstract

We provide evidence that a significant fraction of the returns to merger announcementsreflects target revaluation rather than the causal effect of the merger. In a sample ofunsuccessful merger bids from 1980 to 2008, we find that targets of cash offers are revaluedby 15% after deal failure, which is a sizeable portion of the average announcement effectof 25%. In contrast, stock targets revert to their pre-announcement levels. These resultshold for deals where failure is exogenous to the target’s stand-alone value. We showthat the differential revaluation of cash and stock targets is not explained by differencesin future takeover activity. We also find that the values of cash bidders revert to pre-announcement levels, while stock bidders fall below. Our revaluation estimates implyeconomically large mismeasurement when using naıve announcement-based estimates ofthe causal effect of mergers.

Keywords: mergers & acquisitions, synergies, revaluation, medium of exchange.JEL classifications: G14, G34, D03, D82.

IThis paper significantly benefited from very insightful discussions by Matthew Rhodes-Kropf andPavel Savor, and helpful comments by Viral Acharya, Nihat Aktas, Yakov Amihud, Malcolm Baker,Alberto Bisin, Lauren Cohen, Francesca Cornelli, Julian Franks, Xavier Gabaix, Gerard Hoberg, Vic-toria Ivashina, Boyan Jovanovic, Owen Lamont, Alessandro Lizzeri, Gustavo Manso, Atif Mian, AdairMorse, Stewart Myers, Terrance Odean, Alessandro Pavan, Antoinette Schoar, Andrei Shleifer, DavidSraer, Jeremy Stein, Daniel Wolfenzon, Jeffrey Wurgler, David Yermack, as well as seminar participantsat Berkeley, Princeton, LSE, NYU Stern, Ohio State, Harvard, MIT Sloan, ESMT Berlin, ColumbiaGSB, University of Houston, the 2012 Kellogg Junior Finance conference, the 2011 NBER Summer In-stitute, the 2012 EFA meeting, and the 2012 WFA Meeting. Research assistance by Javed Ahmed andcollaboration with Marlena Lee on an earlier draft of the paper were greatly appreciated.

Email addresses: ulrike@econ.berkeley.edu (Ulrike Malmendier), mopp@haas.berkeley.edu(Marcus M. Opp), saidi@nyu.edu (Farzad Saidi)

1. Introduction

Mergers are among the largest and most disruptive events in a corporation’s lifetime.The proper assessment of their value implications has been of foremost interest to policy-makers and academic researchers alike. Much of the research on mergers and acquisitionsaims to assess to what extent these transactions create, or destroy, shareholder value, in-cluding a recent debate about “massive wealth destruction” through mergers (Moelleret al. (2005)).

Empirically, the measurement of the causal effect of mergers is challenging. The stan-dard approach in the literature is to use stock-market reactions to merger announcements.Various studies document a small positive combined announcement return of targets andbidders, and interpret this finding as evidence in favor of value creation.1 This approachbuilds on a number of assumptions, including the assumptions that markets are efficient,that mergers are unanticipated and unlikely to fail, and that merger bids reveal littleabout the stand-alone values of the merging entities.

In this paper, we argue that a large portion of the announcement effect reflects tar-get revaluation rather than value created through mergers, and that this portion varieswith the type of payment: targets of cash offers are revalued by +15%, but there is norevaluation of stock targets. We also find significant negative revaluation effects for stockbidders, but no revaluation of cash bidders. Our results imply that the widespread useof announcement effects may distort the assessment of mergers.

Our empirical analysis is motivated by existing theories that predict that acquirers’private information is reflected in their bids and their choice of payment.2 While the neg-ative information effect of stock bids on acquirers ’ stand-alone values is well understoodat least since Myers and Majluf (1984), we focus on information revelation about targets.Our identification strategy exploits deal failures to measure information revelation: wecompare target values prior to the announcement and after failure of a bid. To see theintuition for our approach, consider the following stylized example. A target currentlytrades at $100 and receives a cash offer of $125, increasing its price to $125 per share.Shortly afterwards, the deal fails for exogenous reasons, and the target then trades at$115. After deal failure, the information that the bid revealed about the target’s stand-alone value is no longer confounded with the merger effect. Hence, we can attribute $15of the original $25 announcement effect to the revision of beliefs about the stand-alonevalue of the target.

We collect data on unsuccessful merger bids in the U.S. between 1980 and 2008 fromthe SDC database. We also collect detailed information about the failure reasons fromLexisNexis, which allows us to identify the subsample of bids where failure was exogenous

1 See Jensen and Ruback (1983); Jarrell et al. (1988); Andrade et al. (2001).2 Examples include Fishman (1989), where bidders use cash to preempt competing offers, as well as

Hansen (1987) and Eckbo et al. (1990), where the choice of payment reflects private information aboutthe bidder’s own value.

1

to the stand-alone values of targets and acquirers. In the overall sample, we document aninitial announcement effect of 15% for stock deals and 25% for cash deals (including the25-trading-days run-up). By the time of deal failure, however, the value of stock targetsfalls below the pre-announcement level while the value of cash targets remains at 15%cumulative abnormal returns, relative to the pre-announcement level. All results holdcontrolling for a host of deal- and firm-level characteristics, including the target size, rela-tive deal size, offer premium, and hostility of the bid. We also show that, consistent withprevious studies, stock acquirers trade at significantly lower prices post-failure (−17%)while cash acquirers return to their pre-announcement levels.3 Extrapolating from theseestimates to the sample of all mergers during our sample period, approximately $78mof the announcement effect (in 2010 dollars) should be attributed to target revaluationrather than the merger in the average cash deal.4

The most important caveat for this interpretation is sample selection: deal failure isgenerally not exogenous to target and bidder valuations. In most of our analysis, however,we estimate differential revaluation effects between cash and stock offers. Differentialestimates are unbiased as long as cash and stock offers are exposed to the same selectioninto the sample of unsuccessful bids. We show that, in our sample, the type of paymentdoes not predict failure. In other words, within the (self-selected) sample of failed bids,there is no additional sorting by means of payment.

To address the remaining concern of differential selection bias between cash and stockoffers, we identify the subsample of deals in which failure was exogenous to the stand-alone value of the target or acquirer, similar to the approach of Savor and Lu (2009). Forthe target analysis, we isolate negative shocks to the bidder, such as regulatory interven-tion, financing problems, or other negative news on the side of the bidder, that can bereasonably interpreted as being exogenous to the target’s stand-alone value. For the bid-der analysis, we isolate negative shocks to the target, such as the uncovering of negativeinformation in the due-diligence process, that are exogenous to the bidder’s value. Wethen replicate and confirm our results on the respective subsamples. Moreover, our re-sults hold directionally in every single failure category other than “market-wide shocks,”though the statistical significance varies given the small subsamples. This robustnessimplies that our results hold irrespective of the precise choice of classification criteria forexogenous deal failures, consistent with the lack of differential selection bias.

We also show that the revaluation effect is not explained by more frequent or morelucrative future takeover offers: failed cash deals are no more likely than failed stockdeals to be followed by another takeover attempt in subsequent years, nor are futureoffer premia higher. These results hold controlling for the amount of revaluation. Hence,the estimated revaluation differences between cash and stock targets appear to reflect

3 See Rhodes-Kropf et al. (2005) and Dong et al. (2006) for similar findings. Savor and Lu (2009)compare unsuccessful bids with completed deals, and find that unsuccessful bidders perform worse.

4 The average market capitalization of targets in all successful deals was $0.91bn, and in all successfulcash deals $0.52bn, both in 2010 dollars; see Panel A of Table 1 and Appendix-Table A.1, leftmostcolumns. We apply the 15% revaluation effect to the latter (smaller) number.

2

the revaluation of the stand-alone entities.

Why, then, do cash bids, but not stock bids, induce the market to positively revaluethe target? We present a simple theoretical framework to illustrate how the data andall of our empirical results can be understood through the lens of one simple learningchannel. We embed rational market updating, similar in spirit to the misvaluation modelof Rhodes-Kropf and Viswanathan (2004), into the framework of Shleifer and Vishny(2003). Bidding firms have private information about themselves and about their targets,which is not reflected in stock prices. Upon public announcement of a bid, the marketupdates its beliefs about the stand-alone values of acquirer and target. Market updatingis only partial, relative to the knowledge of an informed bidder, due to the presence ofother bidders, who bid for rationales other than mispricing. The presence of these otherbidders allows informed bidders to benefit from their information, even after taking intoaccount the rational response of the market to their offers.

In this framework, informed bidders make cash offers when the combined entity isundervalued in order to exclude target shareholders from the long-run price appreciation.Conversely, they make stock offers when the combined entity is overvalued so as to sharethe long-run price decline with target shareholders. As a result, rational market updatingimplies that firms involved in cash deals are revalued upwards relative to firms involvedin stock deals. On an absolute scale, a cash offer is always positive news about thestand-alone value of the target, and a stock offer is always negative news about thebidder.

Our empirical findings have significant implications for the welfare assessment ofmergers, i.e., for the question whether mergers create surplus, as in the Q-theory model ofJovanovic and Rousseau (2002), or whether they merely have distributional consequences,as in the misvaluation-driven model of Shleifer and Vishny (2003). Our findings implythat a large fraction of stock-price declines after stock bids is due to revaluation ratherthan the merger itself. Hence, announcement-effect-based estimates of the causal effectof stock-financed mergers are downward biased. In contrast, stock-price increases at theannouncement of cash deals reflect, to a large extent, target revaluation rather thansurplus creation, and are upward biased.

This paper relates to a large literature on the returns to mergers. Most studies usingannouncement returns find that tender offers generate small overall value with virtuallyall the gains accruing to the target; see, for example, the overview papers by Jensenand Ruback (1983), Andrade et al. (2001), and Betton et al. (2008). Using probabilityscaling methods, Bhagat et al. (2005) find larger estimates of combined value creation(7.3%) than the conventional CAR estimate (5.3%) reported by Bradley et al. (1988) foran earlier sample.5

Following Travlos (1987), studies of announcement returns distinguish between cashand stock transactions, and typically attribute the negative announcement returns for

5 In contrast to the focus of our paper, Bhagat et al. (2005) do not adjust for target revaluation.

3

stock bidders to the bid revealing negative information about them, consistent with thepecking order model of Myers and Majluf (1984).6 Huang and Walkling (1987) documenthigher announcement returns for cash targets than for stock targets, including the initialrun-up. Their point estimates, 14.4% for stock and 29.3% for cash, are remarkablyclose to the (initial) announcement effects in our sample, despite the different samplecomposition (all deals versus failed deals) and the different sample periods (six years,from 1977 to 1982, versus 29 years, from 1980 to 2008). The similar magnitudes aresuggestive of targets in failed and in successful deals not being too dissimilar.

Most closely related to our analysis are earlier papers by Dodd (1980), Bradley et al.(1983), Davidson et al. (1989), and Sullivan et al. (1994), who also examine failed acqui-sitions. For example, Davidson et al. (1989) find that targets of unsuccessful bids tradehigher than before the announcement because they are more likely to become future tar-gets, while targets that do not obtain a future offer revert to pre-announcement levels. Incontrast to our analysis, their study does not distinguish between cash and stock trans-actions. Conditioning on payment, we are able to separate targets for which (almost) noinformation is revealed (stock targets) from targets that are revalued significantly (cashtargets). Our theoretical framework and empirical findings are, instead, consistent withearlier evidence on the raw return differences in a small sample of 36 stock and cashtargets from the 1980s (Sullivan et al. (1994)).7

Our analysis is also closely related to Savor and Lu (2009), who analyze the perfor-mance of bidders by comparing successful and failed transactions, and to Malmendieret al. (2010), who compare the returns of competing bidders (winners and losers) incontested mergers, and show that winners underperform in the long run.

The remainder of the paper is organized as follows. Section 2 provides a simple theo-retical framework that motivates our analysis and spells out the identifying assumptions.In Section 3, we describe our data. Section 4 contains the main results of our empiricalanalysis and a broad range of robustness checks. Section 5 discusses various informationchannels for revaluation. We discuss the implications of our results for the quantitativeassessment of value creation in mergers in Section 6, and Section 7 concludes.

2. Theoretical Motivation

We motivate our empirical approach in a simple model, in which firms are fairlyvalued on average, but individual firms may be mispriced. Merger bids reflect private in-formation about misvaluation, and the market rationally updates upon observing mergerbids. To highlight the effectiveness of this learning channel in explaining our empirical

6 Alternatively, Jovanovic and Braguinsky (2004) generate bidder discounts in a Q-theory framework.7 The increase in sample size to 249 bids in our data (or, when including non-public bidders, from

66 to 662 pure cash/stock bids; see Table A.5) allows us to use a controlled regression framework ratherthan looking at raw differences in means and, most importantly, to address identification concerns byanalyzing subsamples with exogenous deal failures.

4

results, we shut down other channels of theoretical interest and use the simple frame-work of Shleifer and Vishny (2003). (As discussed further below, the results are alsoconsistent with a richer theoretical framework such as Rhodes-Kropf and Viswanathan(2004). The proofs of all results are in Appendix A.) In our model, the informed bidder’sequilibrium choice of the medium of exchange depends on the perceived misvaluations ofthe combined company, which generates the following testable predictions:

1. Target revaluation is positive in cash deals and indeterminate for stock deals.2. Bidder revaluation is positive in cash deals and negative in stock deals.3. The difference between cash and stock revaluations is positive, both for the target

and the bidder.

Somewhat surprisingly, a stock offer may reveal negative information about the target,since the market understands that the bidder would have chosen cash if his informationabout the target had been sufficiently positive.

Consider a bidder B who has private information about his own long-run value andthat of a potential target T . B aims to maximize his long-run value, while T is onlyconcerned about the current stock price, which reflects all publicly available information.8

In addition, there are other bidders who choose cash or stock for rationales other thanprivate information about valuation. For simplicity, we assume that these bidders choosethe medium of exchange randomly, and we denote them as noise bidders B. Since noisebidders affect informed bidders solely through their effect on market learning, we firstsolve for the optimal decisions of informed bidders given an exogenous market updatingfunction. In a second step, we solve for the equilibrium market updating function.

We express market valuations of B, T , and of the merged firm M as valuationmultiples qi per capital stock in place Ki, with Vi = qiKi for i ∈ {B, T,M}, andKM = KB + KT . We can decompose the market valuation of the merged firm M into the sum of bidder and target values plus synergies s per capital stock of the target:

VM = qMKM = qBKB + qTKT + sKT . (1)

For ease of exposition, we assume that the bidder has no private information about thesynergies s and that the capital stock is observable to the market. Hence, potential mis-valuations must be caused by misperceived valuation multiples. We denote the initialmarket valuations as qi(0) and the updated (intermediate) market valuations, after ob-serving the behavior of the bidder, as qi(a), where a = C in case of a cash bid, a = S incase of a stock bid, and a = N in case of no acquisition. Finally, we denote the long-runvaluations of bidder and target as qB and qT , respectively. The long-run values are drawnfrom two independent continuous distributions, with means equal to the initial marketvaluations qi(0). Thus, bidders and targets are, on average, fairly valued initially.

The timing is as follows:

8 Whether targets have access to the same private information as the bidder is irrelevant since theyonly care about the short run.

5

At t=0, market valuations are qi(0) and, conditional on becoming informed, bidderB learns the long-run multiples qi.

At t=1, the bidder chooses one of three possible actions a: a cash bid C, a stock bidS, or no acquisition N . After observing a, the market updates the estimated valuationmultiples to qi (a) = E (qi| a).

At t=2, the merger bid succeeds with exogenous probability p.9

At t=3, market prices adjust to their long-run values Vi = qiKi.

Since target management is only concerned about short-run valuations, it will acceptany offer price PKT as long as P ≥ qT (a). In stock bids, the resulting target share ofthe merged company, x, is:

x · qM (S) ·KM = PKT , (2)

where we assume identical premia under cash and stock bids.10

Proposition 1. Given a market updating function qM (S), an informed bidder submitsa merger bid if and only if the long-run value of the target plus the synergies are higherthan the long-run cost of acquiring the target:

qT + s ≥ P min

(qM

qM (S), 1

). (3)

Conditional on submitting a bid, the bidder chooses stock over cash if and only if themerged entity is overvalued after a stock bid:

qM ≤ qM (S) . (4)

Note that the effective long-run cost of acquiring the target depends on the medium ofexchange even though, at announcement, the market values cash and stock offers equally,namely, at the offer price P per unit of capital. However, since target shareholders instock transactions participate in the future prospects of the combined company, theeffective long-run cost of acquiring a target with stock amounts to P qM

qM (S). Hence, a

long-term oriented informed bidder cares about temporary misvaluations in stock deals,as determined by discrepancies between qM and qM (S). Temporary misvaluations incash deals, as determined by discrepancies between qM and qM (C) , are irrelevant since

9 This simplifying assumption is motivated by our empirical evidence: Our empirical results hold inthe sample of exogenously failed deals, and deal failure is not related to the medium of exchange.

10 The predictions about the use of cash versus stock are qualitatively unaffected if we allow Pto be chosen endogenously such that, for example, the target’s participation constraint is met withequality, P (C) = qT (C) and P (S) = qT (S). In this case, premia would be higher for cash deals sinceqT (C) > qT (S) (see Proposition 2). The simplifying assumption above follows Shleifer and Vishny(2003) and is consistent with our empirical evidence, as premia do not vary significantly between cashand stock bids in our data.

6

the long run cost of a cash offer is always P .

0 1 2 3 40

1

2

3

4

5

6

−

Stock OfferCash OfferNo Offer = ()

Fig. 1. Optimal Bidding Decision. The figure plots the optimal bidding decision of an informedbidder as a function of the pair (qB , qT ), given the updating function of the market, for the parametervalues qM (S) = 2.15, KB/KM = 0.6, P = 3, s = 0.25.

Figure 1 illustrates this intuition. The figure plots the optimal bidding behaviorof the informed bidder as a function of his information qB and qT , holding constantmarket updating at qM(S) = 2.15. Price and synergies are fixed at P = 3 and s = 0.25,respectively. The relative size of bidder and target is chosen so that it roughly correspondsto the respective empirical moment in our sample, with KB/KM = 0.6, so that biddersare on average 50% larger than targets.

A bid is made for all pairs (qB, qT ) above the (kinked) solid line, i.e., if the informationabout the target is sufficiently good. No offer is made for all pairs below the solid line,i.e., if the information about the target is sufficiently low.

The informed bidder chooses stock if the combined firm will be overvalued, relativeto its long-run value, after a stock bid, i.e., qM ≤ qM (S) (light gray region). Otherwisethe bidder chooses cash as to prevent target shareholders from participating in the longrun price appreciation (dark gray region). Note that some transactions that would beunattractive as a cash deal (P > s+ qT ) can still create value for the bidder if structuredas a stock deal, namely if P > s + qT ≥ P qM

qM (S). If the private information about

bidder’s value, qB, is sufficiently negative, such deals allow the bidder to exploit hiscurrent overvaluation.

7

In order to retrieve the rational updating function of the market and calculate theannouncement effect, we need to specify the propensity of uninformed bidders to engagein mergers. We consider the case that the probability of an informed bid, ρ, is identical forboth cash and stock bids, i.e., the relative frequency of cash and stock bids by uninformedbidders matches the choice of informed bidders. In that case, the updating function forstock bids becomes:

qi (S) = ρE (qi|S,B) + (1− ρ)E(qi|S, B

)(5)

= ρE (qi|S,B) + (1− ρ) qi (0) . (6)

That is, the conditional expectation of the valuation multiples after observing a stockoffer is equal to the weighted average of the conditional expectation, assuming the bidderis informed, and of the unconditional expectation. Since the choice of cash versus stockby an informed bidder depends itself on qT (S) and qB (S), summarized by qM (S), weare looking for a fixed point in which the market updating function is given by equation(6) and the decision of the bidder is optimal given the market updating function. Oncewe have determined the relevant valuation multiple for stock offers qM (S), the updatingfunction for cash offers is simply:

qi (C) = ρE (qi|C,B) + (1− ρ) qi (0) . (7)

In the example shown in Figure 1, the updating functions can be obtained by integratingover the respective regions using the postulated joint probability density functions. Whilewe cannot obtain closed-form expressions for most distributions, we can determine thedirection of updating:11

Proposition 2. Bidders of stock transactions are revalued downwards, and targets ofcash transactions are revalued upwards at announcement of the bid:

qB (S) < qB (0) (8)

qT (C) > qT (0) . (9)

Cash offers signal better information about all entities i ∈ {B, T,M} than stock offers:

qi (C) > qi (S) . (10)

Consider again the example plotted in Figure 1 and further assume that bidder val-uations are drawn from an exponential distribution with mean qB (0) = 2.5 and targetsfrom an independent exponential distribution with mean qT (0) = 2, both in line with

11 The market also updates in response to observing “no offer.” Since our empirical analysis conditionson the existence of a bid, the characterization of qi (N) is not of interest here.

8

our empirical moments. Then, the equilibrium valuation multiples are given by:

(qB (S) , qB (C)) = (2.09, 2.62) (11)

(qT (S) , qT (C)) = (1.98, 2.61) . (12)

In other words, stock bidders are revalued downwards by −16.4%, from 2.5 to 2.09,whereas cash bidders are revalued upwards by 4.8%, from 2.5 to 2.62. Targets of stockoffers suffer slight downwards revaluation (−1%), while targets of cash offers are revaluedupwards by 30.5%. The example reveals that, surprisingly, obtaining a stock offer mightconvey negative news about the target. While any offer conveys positive news about thetarget in that it provides a lower bound on its value (see solid line in the graph), a stockoffer also puts an upper bound on the valuation of the target (dashed line) since a bettertarget would have obtained a cash offer. Rational market updating takes both boundsinto account.

Empirically, we can measure revaluation as the total return from the announcement tothe (exogenous) failure of the deal.12 As a result, the returns at the initial announcementare the weighted average payoffs upon merger completion (with probability p) and dealfailure (with probability 1− p).

Proposition 3. The gross announcement returns RA,i for entity i ∈ {B, T}, are alwayshigher for cash than for stock bids:

RA,i (C) > RA,i (S) . (13)

Note that even if cash and stock targets are offered the same premium P (as is roughlythe case in our sample), cash targets have a higher announcement return. The reason isthat their fall-back price upon deal failure, the updated stand-alone value, is higher.

In the remainder of the paper, we take these predictions to the data.

3. Data

Our main source of data is the SDC Mergers and Acquisitions database, mergedwith stock market and accounting data from CRSP and Compustat. We focus on targetreturns in the sample of all failed merger bids and in the subsample of exogenously failedmerger bids. We also analyze bidder returns and the causes of deal failure in the largersample of all successful and unsuccessful merger bids. To research deal synopses andfailure reasons, we run news searches in LexisNexis.

Our initial sample contains all merger agreements and tender offers between 1980 and2008 that meet the following data requirements. First, we require a valid announcement

12 From a theoretical perspective, it would be interesting to allow for endogenous merger cancelations.This is, however, beyond the scope of this paper.

9

date and a valid completion or failure date within 5 to 250 trading days after the an-nouncement. Second, we exclude bids with competing offers, i.e., deals for which SDCreports a competing deal number, to avoid capturing returns to the competing bid.13

Third, we require a match in the merged CRSP/Compustat database. We use the six-digit CUSIP provided in the SDC database to merge the data. When matched with morethan one CRSP CUSIP, we choose the CUSIP with the lowest 7th digit (typically 1).The initial sample consists of 7, 078 successful and 3, 182 unsuccessful bids.

For the regression analysis, we refine the data as follows: First, we require data onthe medium of exchange (cash, stock, and other) and the deal premium. We extrapo-late missing deal premia by regressing the available SDC premia on transaction valuesdivided by the target’s market capitalization one month prior to the bid, and predictout-of-sample premia based on transaction value and market capitalization (where avail-able). This data requirement reduces the sample of failed bids to 1, 168 observations (993without the extrapolation). As suggested by Officer (2003), we truncate deal premia be-low 0 and above 200%. Second, we require target stock data for at least 25 days prior toannouncement and 25 days post-failure, which reduces the failed-bids sample to 1, 024observations. The use of 25 days for the run-up period is motivated by the finding ofSchwert (1996) that relevant run-ups do not occur until 21 days before announcement.Third, we require acquirer stock data for the same +/− 25-day window, which reducesthe sample to 345 observations. Our results continue to hold if we include non-publicacquirers without Compustat data (see Table A.5), but the restriction to public acquirersis economically meaningful since it allows to address plausible alternative explanations.In particular, we control for leveraged buyouts, hostile bids, market value of equity, trans-action value over market value of acquirer’s equity, as well as acquirer and target q ratios(market value of equity plus assets minus the book value of equity, all divided by assets).The additional controls reduce the sample to 249 unsuccessful merger bids (194 of whichare pure-cash or pure-stock deals). We will refer to this sample as the “regression sam-ple.” When we also include Kaplan and Zingales (1997) indices as a measure of financialconstraints, we yield 217 unsuccessful merger bids (171 pure-cash or pure-stock deals).We use the four-variable version of the KZ index given in Lamont et al. (2001) and Bakeret al. (2003):

KZit = −1.002CFit

Ai,t−1− 39.368

DIVitAi,t−1

− 1.315Cit

Ai,t−1+ 3.139LEVit, (14)

where CFit, DIVit, Cit, and LEVit denote cash flows, cash dividends, cash balances, andleverage, respectively, and Ai,t−1 is the firm’s lagged assets.

Summary Statistics. The summary statistics for the regression sample are in Panel Aof Table 1, separately for successful and unsuccessful bids. Successful and unsuccessfuldeals in our sample are remarkably similar along many dimensions, including the per-

13 We refine the competing-bid flag in SDC based on our news search in LexisNexis. In several cases,our searches reveal competing bids that were not flagged in the SDC data, or indicate that a deal wentthrough shortly after it was recorded as canceled in the SDC data.

10

centage of cash payment offered (with almost identical means), the offer premium, thetransaction value, and the relative deal size. Successful and unsuccessful bids differ inthe percentage of stock and other payment offered, with less stock and more other pay-ments in successful deals. (We will see later that this difference is explained by otherdeal characteristics, most notably relative deal size.) We also see that deals take longerto be completed than to be withdrawn or rejected. LBOs and hostile deals are also lesscommon among successful deals, while tender offers are more common. Not surprisingly,experienced and financially less constrained acquirers have a higher chance of being suc-cessful. Finally, there are marginally significant differences in the q ratio of the target andthe likelihood of target and acquirer being in the same industry; both are higher amongsuccessful bids. There are no significant differences, though, in the proportion of bids forwhich the acquirer’s q is greater than the target’s q. Moreover, the acquirer’s q ratio andthe target’s KZ index do not vary significantly between the samples of successful andunsuccessful bids.

Panel B of Table 1 provides more details on pure-cash and pure-stock deals withinthe subsample of unsuccessful bids.14 These deals make up roughly four-fifths of thetotal regression sample. Given the conclusion of Boone et al. (2011) that mixed-paymentdeals and pure deals are potentially different, it is important to note that all of our resultsare robust (or even stronger) in the sample of pure deals. Within the pure-deals sample,there are only few significant differences between cash and stock transactions. Cash dealsare more likely to be hostile or to be tender offers, and both bidders and targets havelower q ratios. There are no significant differences in the acquirer-to-target q ratios.

Failure Reasons. We categorize the failure reasons of unsuccessful deals based onthe deal synopses in SDC and our detailed news search in LexisNexis. Table 2 showsthe main categories. The category “Alliance” denotes cases in which the bidder andthe target entered another type of cooperation, instead of the merger. The next threecategories—“Price too low,” “Target rejection,” and “Target news,”—summarize casesin which the deal failed due to a negative response of the target to the bid or becausethere was negative news about the target. Specifically, “Price too low” indicates thatthe parties could not agree on the transaction price. “Target rejection” indicates thattarget shareholders or the target board prevented the merger, for example by adoptingpoison pills, via a repurchase of shares by the target from the bidder (greenmail), orby deliberately breaching merger covenants. “Target news” usually refers to cases inwhich the bidder discovered negative information about the target in the course of thedue-diligence process. (One deal failed due to positive news about the target: in August1996, US Diagnostic Labs called off the acquisition of Alliance Imaging because of arun-up in Alliance’s stock price.)

The next category, “Market problems,” summarizes failures due to real shocks affect-ing both the acquirer’s and the target’s industry, or even market-wide downturns, mostly

14 For completeness, we also show the characteristics of successful pure-cash and pure-stock bids inAppendix-Table A.1.

11

the “October 1987” crash, “September 11,” and the subprime crisis.

The last five categories are all cases in which deal failure can be interpreted as exoge-nous to the target’s, but not necessarily to the bidder’s, stand-alone value.“Regulator”refers to lack of regulatory approval as revealed by our news searches, whereas “Regulator(SDC)” are deals for which the SDC data (but not the news searches) indicate regulatoryintervention. One example from our sample is General Electric’s proposed acquisition ofHoneywell in October 2000. The deal was blocked by the European Commission, in adecision that deviated from the U.S. Department of Justice’s view. “Management terms”refers to cases where target management and acquirer management cannot agree on or-ganizational issues, such as the nomination of a CEO of the future company. “Bidderproblems” summarizes failures due to financing problems of the bidder or other negativenews about its business. “Bidder acquired” are sudden cancelations because the bidderitself became the target of an acquisition.

Note that multiple categories could be assigned to a single deal. We were unable toretrieve any information on the failure reason for 37 of 249 deals, and had no informationbeyond which party canceled the merger for another 52 deals.15

We will employ the classification of failure reasons to address potential endogeneityconcerns affecting our identification and the interpretation of our results. To this end,we will re-estimate all regressions separately for each failure category as well as for allcategories jointly where the cancelation was most likely to be orthogonal to the stand-alone value of the entity in question. We will denote as sample C the subset of deals forwhich we have robust information about a failure reason that is exogenous to the target’svalue, i.e., regulatory intervention (“Regulator” and “Regulator (SDC)”), disagreementon management terms or positions (“Management terms”), and unexpected issues onthe side of the bidder (“Bidder problems” and “Bidder acquired”). Alternatively, we usea larger but less restricted sample, sample N , which only excludes bids whose failurewas clearly endogenous to the target’s value, i.e., cancelations due to “Target news” and“Market problems.” Since, as we will show, our coefficient estimates have the same signin all subsamples (other than “Market problems”), our decision to label certain categoriesas exogenous will turn out to be of little importance.

4. Empirical Analysis

4.1. Identification

The idea behind our empirical approach is as follows. The change in the combinedmarket values of bidder B and target T in response to a merger announcement can bedecomposed into expected synergies S plus the revaluation of the hypothetical respective

15 This explains why the number of deals in Table 2 does not add up to (at least) 249 as in ourregression sample.

12

stand-alone values SVB and SVT :

∆VB + ∆VT = S + ∆SVB + ∆SVT . (15)

If a bid is successful, we observe only ∆VB and ∆VT ; the decomposition is not identified.If a bid fails, match-specific synergies are no longer priced and we observe ∆SVB and∆SVT . Moreover, if deal failure is exogenous to the target ’s stand-alone value, ∆SVTreflects updating about the target’s value due to the bid. If deal failure is exogenous tothe bidder ’s stand-alone value, ∆SVB reflects the updating about the bidder’s value dueto the bid. Finally, if deal failure is not exogenous with respect to stand-alone values,but there is no differential selection into the sample of failed cash and failed stock deals(i.e., deal failure is exogenous with respect to ∆V Cash

T −∆V StockT ), we can still extract the

difference in revaluation due to the bid, ∆SV CashT − ∆SV Stock

T . We will entertain (andtest) this weaker assumption when we analyze the full set of failed bids rather than thehand-collected data of exogenously failed deals.

We measure valuation changes as cumulative abnormal returns (CAR), buy-and-holdreturns (BHR), and buy-and-hold abnormal returns (BHAR), defined as follows:

CARit =t∑

j=1

(rij − rmj) (16)

1 +BHRit =t∏

j=1

(1 + rij) (17)

BHARit =t∏

j=1

(1 + rij)−t∏

j=1

(1 + rmj) , (18)

where rij and rmj denote firm i’s equity return and the CRSP value-weighted marketreturn at time j, respectively. Our analysis focuses on the CAR, but our main findingis robust to using buy-and-hold returns, buy-and-hold abnormal returns, and to usingindustry rather than market returns for the specification of abnormal returns.16 Notethat cumulative abnormal returns can be compared across deals with different windowlengths as long as the underlying equilibrium asset pricing model is correctly specified.Moreover, due to the relatively short length of the event window, the misspecification ofthe asset pricing model to compute “normal” returns is a second-order concern.17

13

-20 -10 B 20% 40% 60% 80% F 10 20-25%

-15%

-5%

5%

15%

25%

35%

-25%

-15%

-5%

5%

15%

25%

35%

Trading Days

CAR

Cash TargetStock TargetCash BidderStock Bidder

Fig. 2. Announcement Effects at Bid and at Failure (+/− 25 days). Cumulative AbnormalReturns (CARs) from 25 trading days pre-announcement of the initial bid (B) to 25 trading days afterdeal failure (F). The sample consists of 86 pure-cash and 108 pure-stock deals.

4.2. Graphical Evidence

Figure 2 previews our key empirical result. It plots the evolution of cumulativeabnormal returns for pure-stock and pure-cash failed merger bids in the U.S. between1980 and 2008, separately for acquirers and targets. The graph plots the returns from25 days prior to the announcement to 25 days after failure. Deal failure is normalizedto occur 50 synthetic trading days after the announcement of the initial bid, whichcorresponds to the median window length in our sample.18 Relative to a conventionalplot of average CARs against actual trading days, our procedure ensures that we do notmix deals that have already failed with those that have not yet failed.

16 Ideally, one would want to measure the enterprise value of the firm, i.e., including the market valueof debt. Given the difficulty of obtaining daily market values of debt, we follow the literature in usingequity market values. Our approximation mistakes are likely of second order since debt is less sensitiveto information.

17 See, e.g., Barber and Lyon (1997), Fama (1998), and Brav (2000) on the statistical issues involvedwhen calculating long-run returns.

18 We use a linear interpolation to normalize deals with different window lengths to the same numberof synthetic trading days; see Appendix B.

14

The pre-announcement and post-failure differences are striking. For targets of bothstock and cash offers, we observe strong (cumulative) announcement effects, of 15% and25% respectively. Both effects are consistent with earlier evidence by Huang and Walkling(1987). At the time of deal failure, however, the value of stock targets falls below the pre-announcement level, to which it ultimately returns. The value of cash targets, instead,remains significantly higher than prior to the bid: cash targets earn 15% cumulativeabnormal returns relative to the pre-announcement level. In other words, despite a smallupward trend for both cash and stock targets after deal failure, stock targets remainmore than 15% below cash targets. The graph also reveals that stock acquirers trade atsignificantly lower prices post-failure (−17%), whereas cash acquirers return to their pre-announcement levels, consistent with findings in previous studies. Note that the gradualdecline of stock bidders reflects the averaging of “jumps” of individual bidders, and it isinteresting to see that the gradual decline of stock targets corresponds to the decliningdeal value caused by the gradual decline of stock bidders. Finally, we confirmed thatthe graph looks very similar when using different window lengths, e.g., 100 days pre-announcement and to 100 days post-failure.

In the next subsection, we test whether the graphical evidence holds up in a controlledregression environment, and whether it is robust to the subset of cases in which dealfailure was exogenous to the stand-alone value of the target firm.

4.3. Regression Analysis

In Table 3, we first regress the target CAR from 25 days before announcement ofthe bid until 25 days after failure, CAR (B − 25, F + 25), on the fraction of cash offeredwithout further controls, simply replicating the graphical evidence and providing (robust)standard errors. In the full sample (column 1), we estimate a cash coefficient of 20.5%,and in the pure-cash/pure-stock sample (column 4) a coefficient of 20.4%. The estimatesare highly significant in both cases.

We then add control variables for deal characteristics that are correlated with themedium of exchange and might reflect the target’s stand-alone value: offer premium(over the target’s share price one month prior to the bid), target q, indicators for lever-aged buyouts (which are naturally in cash and involve well-informed bidders) and forhostile bids (which are likely to be in cash [see Panel B of Table 1] and might have apositive impact on target revaluation)19 as well as industry and year fixed effects. Thecoefficient estimates remain similar, 19.3% in the full sample (column 2) and 22.8% inthe pure sample (column 5). Other than the medium of exchange, only the deal pre-mium consistently matters for target CARs. The economic and statistical significanceof the medium of exchange are also confirmed in the larger sample including non-publicacquirers in which we cannot control for acquirer characteristics (see Table A.5).

Next, we include the full range of control variables as well as interactions of cash withthe premium control and with the q ratio of the target. As shown in columns 3 and

19 Further correlates of cash deals are investigated in Appendix-Table A.2.

15

6, the cash coefficient more than doubles. The economic magnitude of the cash effectfor the mean deal remains the same given the (insignificantly and significantly) negativeinteraction effects. It is interesting to note that the hostility dummy loses its explanatorypower (insignificant in the subsample of pure cash and stock deals, marginally significantin the full sample). We also see that the cash effect is stronger for targets with low qratios, both in the full and the pure sample. This may reflect stronger revaluation ofpotentially undervalued targets when they receive a cash offer.

All results are virtually identical when we use alternative return specifications: industry-adjusted abnormal returns, buy-and-hold abnormal returns, and buy-and-hold returns.As shown in Table 4, the raw cash effects in the subsample of pure cash and stock deals(cf. columns 1, 3, and 5) match the magnitude encountered in Figure 2. And, simi-larly, the estimated cash coefficient in the full regression model (columns 2, 4, and 6) (ofcolumn 6 in Table 3) for all three return definitions.

Finally, we re-estimate the cash effect on the subsample of bids where deal failurewas plausibly exogenous to the stand-alone value of the target. As outlined above, ourestimates on the sample of all failed bids are unbiased even if there is selection into thefailure sample but no differential selection into the failure sample across cash and stockbids. To address the concern about differential sample selection, we use two samplerestrictions outlined in Section 3: The most restrictive sample, C, consists only of dealsthat failed due to regulatory intervention, news about the bidder and disagreement onmanagement terms. The less restrictive but larger sample N excludes deals that faileddue to news about the target and market problems, which (also) affected the target.The results are shown in Table 5. In the most restrictive sample C, we estimate againa very similar univariate cash coefficient of 24.0%, significant at p < 0.05 (column 1).The coefficient estimate remains marginally significant (p < 0.08) also after including thecontrols discussed above, as shown in columns 2 and 3, despite the small sample size.Also the economic significance remains large and very similar to our previous estimates.Including interaction terms naturally changes the coefficient (to 0.614), but the economicimplications are unchanged after accounting for the interaction effects. These results arealso confirmed for the larger sample N , in which we estimate the results with and withoutinteractions for all deals (columns 4 and 5) and pure deals only (columns 6 and 7). In allinstances, we estimate economically very similar and statistically significant coefficientsfor the medium of exchange.

As a further step in addressing concerns about the endogeneity of deal failure, were-estimate the cash coefficient for every single failure category identified in our newssearches. We show the resulting coefficient estimates separately for each failure categoryin the second column of Table 2. While the small sample size for each failure categoryrestricts us to univariate regressions, the results further support our findings: The cashcoefficient is always positive except for one category, “Market problems.” (The revalua-tion estimates for individual companies during market crashes like “September 11” areextremely volatile, despite the adjustment for the market return.) The results also holdin the total sample of 160 deals for which we have identified failure reasons, and the re-maining 89 deals for which we were unable to find information about the failure reason,

16

as shown in the last two rows.

The robustness of our results across exogenous and endogenous failure categories isan interesting finding in and of itself. One interpretation is that, even if deal failureis endogenous, there is no differential selection between stock and cash deals. In otherwords, endogeneity affects stock and cash deals equally and, thus, does not affect the(differential) cash effect. To shed more light on this interpretation, we perform an aux-iliary analysis of the determinants of failure. We estimate a linear probability model fordeal failure as a function of a continuous cash variable (the fraction of payment offeredin cash). The results are presented in Table 6. The impact of the medium of exchangeon deal failure is insignificant, regardless of the set of controls included, whereas therelative deal size is a major determinant of deal failure. Not surprisingly, hostile dealannouncements are also less likely to be successful. In the last two columns, we includethe target’s announcement return for the bid, CAR (B − 25, B + 1), in addition to ourusual set of controls. The additional controls should capture market expectations of dealfailure that are based on all other publicly available information at the time of the an-nouncement (but are unavailable to the econometrician).20 Again, the cash coefficientremains insignificant. Our results are also robust to reducing our sample to the subsetof pure cash and stock deals only (cf. Appendix-Table A.3).

Another potential selection concern is that cash and stock deals fail differentially withrespect to post-announcement news. In particular, one might be worried that good newsabout the target might make cash deals more likely to fail because a financially con-strained bidder is unable to increase his bid in cash, leading to deal failure.21 We addressthis in the first regression specification of Table 7, which includes the acquirer’s KZ indexfor financial constraints compared to the third column in Table 3. The interaction of themedium of exchange with the acquirer’s KZ index is insignificant. We also control for thetarget’s KZ index, as its financial constraints might impact its attractiveness to potentialbuyers, but the respective coefficient is not significant either.

5. Revaluation Channels

5.1. Future Takeover Activity

Our results so far suggest that a cash bid reveals positive private information of thebidder about the target. To interpret our estimates, we next distinguish whether this

20 The announcement return should approximately be given by the return captured if the deal goesthrough, i.e., the premium, weighted by the probability of a successful takeover, p, plus the return thatresults from learning if the deal does not go through, i.e., CART (a) ≈ p·Premium+ (1− p) ·Learning(a)where a ∈ {cash, stock} denotes the medium of exchange. Since we control for the premium and learningencoded in the choice of the medium of exchange, variations in CAR should capture variations in thedeal probability.

21 Empirically, we could only identify one instance in which good news about the target led to dealfailure. The remaining target-news categories identified in Table 2 are negative.

17

information truly reflects the stand-alone value of the target rather than anticipated fu-ture takeover activity. Due to differential legal treatment of future offers under DelawareLaw (“Revlon standard”), a positive correlation between cash offers and future takeoveractivity is plausible.22 Moreover, earlier empirical evidence by Davidson et al. (1989)suggests that revaluation is related to expected future takeover activity, although theydo not condition on the medium of exchange.

To see how we can isolate the effect of future takeover activity, consider the followingsimple illustration. Assume that a cash and a stock target both initially trade at 100 anda first (unsuccessful) takeover attempt leads to a revaluation of VC for the cash targetand VS for the stock target. We can decompose the respective values into the weightedaverage of the stand-alone value SVi and the expected future takeover price TPi:

Vi = (1− τi)SVi + τiTPi, (19)

where τi denotes the probability of a future takeover offer and i ∈ {C, S}. This rep-resentation separates the “anticipation effect” of future takeovers (as in Edmans et al.(2012)) from revaluation of the stand-alone value. Note that we allow the stand-alonevalue, takeover probability, and takeover price to vary with the medium of exchange. Foridentification purposes, it is useful to write this as:

(1− τi)Vi = (1− τi)SVi + τi (TPi − Vi) . (20)

This decomposition suggests an easily interpretable sufficient condition under which dif-ferential revaluation between cash and stock deals is entirely driven by updating aboutthe stand-alone value. If the likelihood of future takeovers is identical for cash and stockbids – i.e., τC = τS – and so are expected future deal premia – i.e., τC (TPC − VC) =τS (TPS − VS) – then we have that VC − VS = SVC − SVS.

Table 8 shows the empirical test of both conditions. In the first three columns, we useas the dependent variable an indicator variable for future merger bids occurring within thesubsequent two years (and after at least half a year, to avoid capturing bidding wars).23 Inthe last three columns, we replace the dummy variable with the actual takeover premiumin future deals (and zero otherwise) while using the same set of control variables as in thefirst three regression specifications. We find no evidence that cash targets are any morelikely to receive future merger bids, nor that they receive higher future takeover premia.Note in particular that we include the Target CAR (B − 25, F + 25) as an additionalindependent variable in the second and third specifications of each set of regressions

22 A cash-merger bid is legally considered a change-of-control transaction, while an all-stock bid is not.In particular, if the target board has entered into an all-cash merger agreement, the board is deemed tohave decided to sell the firm, and must consider a higher future bid should one materialize after dealfailure. In an all-stock agreement, however, the board is not deemed to have decided to sell control, andcan ignore future offers.

23 Our results are robust to other specifications of the grace period (such as one month, six weeks, orthree months).

18

(columns 2, 3, 5, and 6) and, hence, test directly for a relation between the revaluationeffect and future merger activity, as in Davidson et al. (1989). The coefficient estimateis insignificant in all specifications. Interestingly, there is tentative evidence that targetswith low q ratios are more likely to receive future takeover offers. Most importantly, allspecifications are strongly consistent with the hypothesis that the medium of exchange isnot related to future takeover activity. This auxiliary finding supports the interpretationof our main result: the differential revaluation estimate between cash and stock dealsreflects information about the stand-alone value of the target.

5.2. Undermanagement

Another interpretation of our findings is that cash bids signal “undermanagement,”rather than undervaluation, and motivate target management to improve their company’sperformance. While the undervaluation interpretation suggests that a cash bid signalsbetter operating performance of the target, the undermanagement story implies that thecash bid induces better operating performance. In both instances a bidder is more likelyto choose cash to prevent the target shareholders from participating in the upside.

An actual improvement of operating performance of an undermanaged target may beachieved through either (i) superior management skills of the bidder or (ii) a reductionin target managerial slack present at the time of the bid. Since we focus on failed deals,superior management skills of the bidder are unlikely to be be the source of operationalimprovement. However, target management may improve operational performance due tothe disciplinary effect of a merger bid. It is also plausible that such disciplinary pressureis higher for cash bids than for stock bids as cash bids are more likely to be hostile (seeAppendix-Table A.2).24 However, we already showed that the effect of cash on post-failure target CARs remains significant after controlling for hostile and LBO bids (seeTables 3 and 7). Thus, unless one interprets the cash variable as a better proxy for thedisciplinary channel than “hostility,” this evidence points in favor of the undervaluationexplanation rather than any disciplinary effect of cash bids. Anecdotally, one of the mostprominent value investors in our sample with a series of well-performing cash acquisitions,Warren Buffett, famously tends to seek out “undervalued,” well-managed companiesrather than undermanaged ones.

Finally, we aim to test directly for (observable) operational changes of the target.One prominent restructuring activity that may be induced by a disciplining bid is recap-italization as measured, for example, by the increase in the target’s leverage ratio. Thismay be interpreted as a way to reduce managerial slack (Jensen and Meckling (1976))or as evidence that target managers lever up to deter (further) takeover attempts, asdescribed in Hirshleifer and Thakor (1992).25

24 Consistent with this line of reasoning, Mikkelson and Partch (1997) provide evidence of a positiverelationship between takeover activity and top-management turnover during the hostile-merger wavein the 1980s. In turn, Schwert (2000) does not find any distinguishable difference between hostile andfriendly deals in terms of stock-market and accounting data.

25 While the first explanation is most likely associated with a value increase, the second rationale is

19

We regress post-failure changes in the target’s debt on our cash variable and theusual host of control variables (see Appendix-Table A.6). The estimates reveal thathostile bids and LBOs are associated with greater changes in target leverage, though theeffects are not significant in all specifications. Most importantly, cash does not predicthigher leverage.

5.3. Information Sources

Can we provide direct evidence on the information channel? Proxies for privateinformation are naturally difficult to obtain. As illustrated in our model in Section 2, theideal information proxy should capture variation in the probability of being informed,ρ, but not variation in any other relevant variables. Empirically, we use (i) whether thebidder is in the same industry (according to the two-digit SIC code), (ii) whether he has atoehold in the target, (iii) whether he is an experienced acquirer, and (iv) whether he hasa large cash balance to spend. The first three proxies should be positively associated withthe probability of an informed bid, while “cash balance” should be negatively related toρ as acquirers with a large cash balance might be less selective in choosing targets in acash deal. For each proxy, we re-estimate our baseline regression (cf. third column inTable 3), adding the respective proxy variable as well as its interaction with cash offered.As shown in the second to fifth columns of Table 7, the evidence is mixed, with onlyone out of four level effects pointing in the expected direction and all interaction effectsbeing insignificant. The mixed nature of the evidence could reflect the imperfection ofthe proxies, which might not capture variation in ρ well or simultaneously affect othervariables relevant for revaluation.

The decision to use cash vs. stock might also reflect accounting interests (see Section3.2 in Betton et al. (2008)). Until 2001, an accounting motivation for stock deals usedto be the pooling-of-interests method in order to avoid the creation and subsequentamortization of goodwill (as required under the purchase method of accounting).26 This“accounting incentive” implies two opposing effects on target revaluation. On the onehand, it induces the bidder to make a stock bid for a target, for which it otherwise wouldnot have made an offer at all, i.e., targets with low qT . On the other hand, the accountingincentive also favors the use of stock, even for good targets, i.e., targets with high qT .

Empirically, we try to address this in the following way. As goodwill is increasingin the offer premium, we interact the interaction between the premium and our cashvariable with a dummy for the pre-2001 period (the 1980s and 1990s). Indeed, in the lastcolumn of Table 7, we find that the triple interaction is significantly negative, indicatingthat targets of stock offers involving high premia in the 1980s and 1990s were not asovervalued as the more recent ones in the 2000s.

presumably either value-neutral or destructive.26 See Lys and Vincent (1995) for an extreme case – AT&T’s acquisition of NCR – of the bidder’s

interest in having the acquisition qualify as a pooling of interests.

20

6. Implications for the Returns to Mergers

Our evidence of large revaluation effects implies that announcement-effect estimatesof the causal effect of mergers in successful deals might be significantly biased, and thatthe direction of the bias depends on the medium of exchange offered. We next provide aback-of-the-envelope calculation for the adjustment needed when calculating returns tomergers.

To this end, we first compute the average joint announcement return of the bidderand the target on the [−25, 1] window for the sample of successful deals separately for allcash and all stock deals. Secondly, we need to compute joint revaluation effects. Here, werestrict ourselves to our most conservative revaluation estimates, i.e., based on sampleC for targets (column 1, Table 5) and an analogous sample C for the bidder.27 Therevaluation estimates for the bidder are shown in Appendix-Table A.7. Column 4 showsthe relevant sample C coefficient of 15.6%, the difference between the revaluation estimatefor pure cash bidders of −1.1% and −16.7% for pure stock bidders. Using the averagerelative size of the bidder and the target in the sample of successful pure-cash and pure-stock deals, we can then compute the average estimated revaluation. The following tableprovides an overview of joint announcement returns and revaluation estimates (measuredas a fraction of the joint market capitalization). We denote the differences as the impliedsynergies. In addition to the full sample of successful deals, we consider two additionalsubsamples in which we require the target share in the combined company to be at least10% and 30%, respectively. This filter governs the importance of the target in the deal.

Cash deals Stock dealsMinimum target size 0% 10% 30% 0% 10% 30%Announcement return 4.9% 10.0% 15.3% 3.8% 5.0% 6.7%Revaluation estimate 0.2% 2.0% 4.1% -16.4% -16.1% -15.8%Synergies s 4.7% 8.0% 11.2% 20.2% 21.1% 22.5%N 1,019 370 135 1,232 677 252

Given that the revaluation in cash deals is driven primarily by targets, the portionof the announcement effect that reflects revaluation (rather than the causal effect of themerger) is increasing in the minimum relative target size. For example, for deals involvingtargets that make for at least 30% of the total market capitalization, the revaluation effect(4.1%) accounts for 27% of the announcement effect (15.3%). Even for targets as small as10% of the total market capitalization, the revaluation effect accounts for one-fifth of thejoint announcement return, which implies that synergy estimates based on announcementreturns of cash deals might be significantly upward biased. Conversely, for stock deals,we find that announcement returns lead to an underestimation of synergies by as muchas 16 percentage points.

27 Following the logic of the sample C construction for targets, we only include deals that are exogenousto the bidder’s stand-alone value, i.e., deal failures due to regulatory intervention, target news, anddisagreement on management terms/positions.

21

7. Conclusion

We have shown that the medium of exchange in merger bids reveals economicallyand statistically significant information about the target: targets of cash offers trade15% above pre-announcement levels, whereas targets of stock offers are not revaluedupwards. Prior literature has primarily focused on information revealed about the bidder.Our findings imply that when assessing the value created in merger transactions, it isimportant to account for the implied target revaluation. Building on our back-of-the-envelope synergy estimates for completed deals provided in Section 6, an important nextstep would be the refined quantification of these estimates using advanced statisticaltechniques, such as in Bhagat et al. (2005).

Appendix A. Proofs

Proof of Proposition 1. We start from the conditional choice of cash versus stock.Since the failure probabilities and payoffs are the same for cash and stock bids, thebidder only needs to compare the long-run market value of shares owned by the bidder’sshareholders upon a completed deal, i.e., qMKM − PKT if a = C and (1− x) qMKM ifa = S. Hence, the bidder chooses stock if and only if:

(1− PKT

qM (S)KM

)qMKM ≥ qMKM − PKT .

Simple algebra yields inequality (4). Comparing the long-run values upon completionand in case of no bid, qBKB, delivers the decision to make a bid.

Proof of Proposition 2. In order to determine the direction of the updating, we simplyhave to determine the updating process conditional on an informed offer, E (qi| a,B).If E (qi| a,B) > qi (0) then qi (a) = ρE (qi| a,B) + (1− ρ) qi (0) > qi (0) and, thus, ifE (qi|C,B) > E (qi|S,B) then qi (C) > qi (S).

To determine E (qi| a,B) we first characterize the relevant boundaries for the optimumaction a (see Figure 1). Define kB = KB

KM, kT = KT

KMand qwa (S) = kBqB (S) + kT qT (S).

Proposition 1 implies that an informed bidder makes a stock bid if and only if:

qBkB + qTkT ≤ qwa (S) (A.1)

xqBkB − (1− x) qTkT ≤ (1− x) skT . (A.2)

This can be written as simple linear constraints on qB and qT :

qT ≤ c− aqB (A.3)

qT ≥ −s+ eqB, (A.4)

where a = kBkT> 0, c = qwa(S)

kT> 0, and e = PkB

qwa(S)+skT−PkT> 0.

22

Similarly, a bidder makes a cash bid iff:

qT > c− aqB (A.5)

qT ≥ d, (A.6)

where d = P − s > 0.28

Since qT ≥ 0, we can solve for the cutoff value in qB for which −s+ eqB ≤ 0, i.e.,

q∗B =s

e. (A.7)

Moreover, we can solve for the cut-off value of qB for which stock-offers are alwaysdominated by cash offer, i.e., c− aqB = −s+ eqB or:

q∗∗B =c+ s

a+ e=c− da

. (A.8)

Since d > 0, we immediately have q∗∗B > q∗B. Thus, an informed bidder chooses actiona ∈ {C, S,N} with the following probabilities (determined by regions in Figure 1):

Pr (C|B) =

∫ q∗∗B

0

∫ ∞c−aqB

f (qB, qT ) dqBdqT +

∫ ∞q∗∗B

∫ ∞d

f (qB, qT ) dqBdqT (A.9)

Pr (S|B) =

∫ q∗B

0

∫ c−aqB

0

f (qB, qT ) dqBdqT +

∫ q∗∗B

q∗B

∫ c−aqB

eqB−sf (qB, qT ) dqBdqT (A.10)

Pr (N |B) = 1− Pr (C|B)− Pr (S|B) . (A.11)

Then:

E (qi|C,B) =

∫ q∗∗B0

∫∞c−aqB

qif(qB ,qT )dqBdqT+∫∞q∗∗B

∫∞d qif(qB ,qT )dqBdqT

Pr(C|B)(A.12)

E (qi|S,B) =

∫ q∗B0

∫ c−aqB0 qif(qB ,qT )dqBdqT+

∫ q∗∗Bq∗B

∫ c−aqBeqB−s qif(qB ,qT )dqBdqT

Pr(S|B). (A.13)

Thus, since target cash offers are made if qT > c− aqB and qT ≥ d, we obtain:

E (qT |C,B) > E (qT | qT > d) > qT (0) . (A.14)

Moreover, for stock bidders:

E (qB|S,B) < E (qB| qB < q∗∗B ) < qB (0) . (A.15)

For the following statements, we use the assumption of independent distributions. Forcash bidders, we have that E (qB|C,B) > qB (0) because conditioning on a cash offer,

28 We implicitly assume that the price paid for the target is greater than just the synergies, but smallerthan the value of the combined entity, i.e., qM (S)KM > PKT > sKT .

23

the region qB < q∗∗B is undersampled because of the tighter restriction on qT , i.e., qT >c− aqB ≥ d vs. qT ≥ d (for qB > q∗∗B ). Lastly, it follows that:

E (qT |S,B) < E (qT | qT > −s+ emin (qB, q∗∗B )) < E (qT | qT > d) < E (qT |C,B) .

(A.16)

Proof of Proposition 3. The announcement return for the target is given by theweighted average of the payoffs under completion and the fall-back value:

RA,T (a) =pP + (1− p) qT (a)

qT (0), (A.17)

so that

RA,T (C)−RA,T (S) = (1− p) qT (C)− qT (S)

qT (0)> 0. (A.18)

Since qT (C)− qT (S) > 0 by Proposition 2, the result immediately follows.

Similarly, the announcement return for the bidder is given by the weighted averageof the payoffs under completion and the fall-back value:

RA,B (a) =p (KM qM (a)− PKT ) + (1− p)KB qB (a)

KB qB (0), (A.19)

so that

RA,B (C)−RA,B (S) = pKM

KB

qwa (C)− qwa (S)

qB (0)+ (1− p) qB (C)− qB (S)

qB (0). (A.20)

Since qwa (C)− qwa (S) > 0 and qB (C)− qB (S) > 0 by Proposition 2, the result imme-diately follows.

Appendix B. Linear Approximation

The time interval between announcement of the initial bid and failure of the dealvaries across the sample. For the graphical illustration, we normalize this window to 50synthetic trading days. Each synthetic trading day corresponds to 1

50= 2% of the actual

time elapsed between initial bid and failure. The cumulative abnormal returns of deal iwith (actual) window length Ti after n synthetic trading days are thus equal to the CARafter n actual trading days, where n = nTi

50:

CARi (n) = CARi (n) . (B.1)

For example, the CAR after 10 synthetic trading days, CARi (10), for a deal with Ti =100, are equal to the CAR after 20 actual trading days. If n is not an integer number,

24

we use a linear approximation between the relevant integer numbers, i.e.,

CARi (n) =(1− w(i,n)

)CARi (bnc) + w(i,n)CARi (bnc+ 1) , (B.2)

where w(i,n) = n−bnc and bxc refers to the floor function. Hence, for a deal with Tj = 10,

the CAR after 8 synthetic trading days is given by: CARj (8) = 25CARj (1)+ 3

5CARj (2)

as n = nTj

50= 8× 10

50and w(j,n) = 8

5− 1.

25

References

Andrade, G., Mitchell, M., Stafford, E., 2001. New evidence and perspectives on mergers.Journal of Economic Perspectives 15, 103–120.

Baker, M., Stein, J.C., Wurgler, J., 2003. When does the market matter? stock pricesand the investment of equity-dependent firms. The Quarterly Journal of Economics118, 969–1005.

Barber, B.M., Lyon, J.D., 1997. Detecting long-run abnormal stock returns: The em-pirical power and specification of test statistics. Journal of Financial Economics 43,341–372.

Betton, S., Eckbo, B.E., Thorburn, K.S., 2008. Corporate takeovers, in: Eckbo, B.E.(Ed.), Handbook of Corporate Finance: Empirical Corporate Finance. Elsevier/North-Holland. volume 2. chapter 15, pp. 291–430.

Bhagat, S., Dong, M., Hirshleifer, D., Noah, R., 2005. Do tender offers create value?new methods and evidence. Journal of Financial Economics 76, 3–60.

Boone, A., Lie, E., Liu, Y., 2011. Is cash king? An analysis of mixed payments incorporate takeovers and the preferences of target shareholders. Mimeo. Texas A&MUniversity.

Bradley, M., Desai, A., Kim, E.H., 1983. The rationale behind interfirm tender offers:Information or synergy? Journal of Financial Economics 11, 183–206.

Bradley, M., Desai, A., Kim, E.H., 1988. Synergistic gains from corporate acquisitionsand their division between the stockholders of target and acquiring firms. Journal ofFinancial Economics 21, 3–40.

Brav, A., 2000. Inference in long-horizon event studies: A bayesian approach with appli-cation to initial public offerings. Journal of Finance 55, 1979–2016.

Davidson, Wallace N, I., Dutia, D., Cheng, L., 1989. A re-examination of the marketreaction to failed mergers. Journal of Finance 44, 1077–1083.

Dodd, P., 1980. Merger proposals, management discretion, and stockholder wealth. Jour-nal of Financial Economics 8, 105–137.

Dong, M., Hirshleifer, D., Richardson, S., Teoh, S.H., 2006. Does investor misvaluationdrive the takeover market? Journal of Finance 61, pp. 725–762.

Eckbo, B.E., Giammarino, R.M., Heinkel, R.L., 1990. Asymmetric information and themedium of exchange in takeovers: Theory and tests. Review of Financial Studies 3,651–675.

Edmans, A., Goldstein, I., Jiang, W., 2012. The real effects of financial markets: Theimpact of prices on takeovers. The Journal of Finance 67, 933–971.

26

Fama, E.F., 1998. Market efficiency, long-term returns, and behavioral finance. Journalof Financial Economics 49, 283–306.

Fishman, M.J., 1989. Preemptive bidding and the role of the medium of exchange inacquisitions. Journal of Finance 44, 41–57.

Hansen, R.G., 1987. A theory for the choice of exchange medium in mergers and acqui-sitions. Journal of Business 60, 75–95.

Hirshleifer, D., Thakor, A.V., 1992. Managerial conservatism, project choice, and debt.Review of Financial Studies 5, 437–470.

Huang, Y.S., Walkling, R.A., 1987. Target abnormal returns associated with acquisitionannouncements: Payment, acquisition form, and managerial resistance. Journal ofFinancial Economics 19, 329–349.

Jarrell, G.A., Brickley, J.A., Netter, J.M., 1988. The market for corporate control: Theempirical evidence since 1980. Journal of Economic Perspectives 2, 49–68.

Jensen, M.C., Meckling, W.H., 1976. Theory of the firm: Managerial behavior, agencycosts, and ownership structure. Journal of Financial Economics 3, 305–360.

Jensen, M.C., Ruback, R.S., 1983. The market for corporate control : The scientificevidence. Journal of Financial Economics 11, 5–50.

Jovanovic, B., Braguinsky, S., 2004. Bidder discounts and target premia in takeovers.American Economic Review 94, 46–56.

Jovanovic, B., Rousseau, P.L., 2002. The q-theory of mergers. American EconomicReview 92, 198–204.

Kaplan, S.N., Zingales, L., 1997. Do investment-cash flow sensitivities provide usefulmeasures of financing constraints? The Quarterly Journal of Economics 112, 169–215.

Lamont, O., Polk, C., Saa-Requejo, J., 2001. Financial constraints and stock returns.Review of Financial Studies 14, 529–554.

Lys, T., Vincent, L., 1995. An analysis of value destruction in at&t’s acquisition of ncr.Journal of Financial Economics 39, 353–378.

Malmendier, U., Moretti, E., Peters, F., 2010. Winning by Losing: Evidence on Over-bidding in Mergers. Mimeo. University of California, Berkeley.

Mikkelson, W.H., Partch, M.M., 1997. The decline of takeovers and disciplinary man-agerial turnover. Journal of Financial Economics 44, 205–228.

Moeller, S.B., Schlingemann, F.P., Stulz, R.M., 2005. Wealth destruction on a massivescale? a study of acquiring-firm returns in the recent merger wave. Journal of Finance60, 757–782.

27

Myers, S.C., Majluf, N.S., 1984. Corporate financing and investment decisions whenfirms have information that investors do not have. Journal of Financial Economics 13,187–221.

Officer, M.S., 2003. Termination fees in mergers and acquisitions. Journal of FinancialEconomics 69, 431–467.

Rhodes-Kropf, M., Robinson, D.T., Viswanathan, S., 2005. Valuation waves and mergeractivity: The empirical evidence. Journal of Financial Economics 77, 561–603.

Rhodes-Kropf, M., Viswanathan, S., 2004. Market valuation and merger waves. Journalof Finance 59, 2685–2718.

Savor, P.G., Lu, Q., 2009. Do stock mergers create value for acquirers? Journal ofFinance 64, 1061–1097.

Schwert, G.W., 1996. Markup pricing in mergers and acquisitions. Journal of FinancialEconomics 41, 153–192.

Schwert, G.W., 2000. Hostility in takeovers: In the eyes of the beholder? Journal ofFinance 55, 2599–2640.

Shleifer, A., Vishny, R.W., 2003. Stock market driven acquisitions. Journal of FinancialEconomics 70, 295–311.

Sullivan, M.J., Jensen, M.R.H., Hudson, C.D., 1994. The role of medium of exchange inmerger offers: Examination of terminated merger proposals. Financial Management23, pp. 51–62.

Travlos, N.G., 1987. Corporate takeover bids, methods of payment, and bidding firms’stock returns. Journal of Finance 42, 943–963.

28

Tab

le1.

Su

mm

ary

Sta

tist

ics

Tim

eto

com

ple

tion

/fail

ure

isin

trad

ing

day

s.T

ran

sact

ion

valu

esin

2010

$bn

are

bas

edon

his

tori

cal

tran

sact

ion

valu

es,

conve

rted

usi

ng

Con

sum

erP

rice

Ind

ex(C

PI)

Con

ver

sion

Fac

tors

.T

arge

tsi

zeis

the

targ

et’s

mar

ket

valu

eof

equ

ity

in20

10$b

n.

Rel

ativ

ed

eal

size

isth

etr

ansa

ctio

nva

lue

over

the

acq

uir

er’s

mark

etva

lue

ofeq

uit

y.O

ffer

pre

miu

mis

nor

mal

ized

by

the

targ

et’s

mar

ket

cap

ital

izat

ion

aton

em

onth

pri

orto

the

ann

oun

cem

ent

ofth

eb

id,

and

tru

nca

ted

bet

wee

n0

and

2.H

osti

led

eal,

LB

O,

and

Ten

der

offer

are

du

mm

yva

riab

les

ind

icat

ing

host

ile

bid

s,le

vera

ged

bu

you

ts,

an

dte

nd

eroff

ers,

resp

ecti

vely

.q

rati

ois

the

mar

ket

valu

eof

equ

ity

plu

sas

sets

min

us

the

book

valu

eof

equ

ity

all

over

ass

ets.

KZ

ind

exis

the

fou

r-va

riab

leve

rsio

nas

,e.

g.,

inL

amon

tet

al.

(200

1).

New

dea

lan

nou

nce

dw

ith

in2

year

s(d

um

my

vari

ab

le)

isco

ndit

ion

alon

the

dea

lb

ein

gan

nou

nce

dat

leas

th

alf

aye

araf

ter

the

pre

vio

us

one.

Exp

erie

nce

dac

qu

irer

s(d

um

my

vari

ab

le)

app

ear

atle

ast

five

tim

esin

the

dat

ase

t.S

ame

ind

ust

ryis

anin

dic

ator

vari

able

for

wh

eth

erta

rget

and

bid

der

oper

ate

inth

esa

me

ind

ust

ry,

asd

efin

edby

the

two-d

igit

SIC

cod

es.

All

non

-dea

l-re

late

dva

riab

les

(e.g

.,q

rati

os)

are

mea

sure

dat

the

end

ofth

eye

arp

rior

toth

ed

eal’

sann

oun

cem

ent,

and

allq

and

KZ

vari

able

sar

ew

inso

rize

dat

the

1st

and

99th

per

centi

les.

Th

ep

-val

ues

inth

ela

stco

lum

nare

for

atw

o-s

ided

diff

eren

ce-i

n-m

ean

ste

st.

Bot

hp

anel

ssh

owth

ere

gres

sion

sam

ple

,i.

e.,

bid

sfo

rw

hic

hal

lco

ntr

olva

riab

les

are

avail

able

.In

Pan

elB

,th

esa

mp

leis

furt

her

rest

rict

edto

un

succ

essf

ul

pu

re-c

ash

and

pu

re-s

tock

bid

s.E

xce

pti

ons:

For

New

dea

lan

nou

nce

dw

ith

in2

yea

rs,

the

sam

ple

conta

ins

220

un

succ

essf

ul

dea

ls(i

nP

anel

A),

69ca

shan

d10

2st

ock

dea

ls(i

nP

anel

B).

For

KZ

ofac

qu

irer

,w

eh

ave

1,8

19

succ

essf

ul

an

d22

7u

nsu

cces

sfu

ld

eals

(Pan

elA

)an

d78

cash

and

102

stock

dea

ls(P

anel

B),

and

for

KZ

ofta

rget

1,82

9su

cces

sfu

lan

d23

3u

nsu

cces

sfu

ld

eals

(Pan

elA

)an

d78

cash

and

104

stock

dea

ls(P

anel

B).

Pan

el

A:

Su

cces

sfu

lan

dU

nsu

cces

sfu

lB

ids

Su

cces

sfu

lB

ids

Un

succ

essf

ul

Bid

sV

ari

able

Mea

nS

td.

dev

.M

inM

axM

ean

Std

.d

ev.

Min

Max

p-v

alu

e

Cash

in%

46.9

045

.77

010

044

.31

45.7

90

100

0.40

1S

tock

in%

45.7

545

.56

010

055

.23

45.8

50

100

0.00

2O

ther

pay

men

tin

%7.

3518

.15

010

00.

455.

780

100

0.00

0T

ime

toco

mp

leti

on/fa

ilu

re76

.90

43.8

05

245

63.6

848

.26

523

80.

000

Tra

nsa

ctio

nva

lue

in20

10

$bn

1.41

4.24

0.00

70.5

11.

596.

860.

0177

.04

0.56

7T

arge

tsi

zein

2010

$bn

0.91

2.93

0.00

65.1

61.

255.

460.

0056

.04

0.12

6R

elat

ive

dea

lsi

ze1.

4824

.68

0.00

830.

221.

112.

190.

0017

.57

0.81

3O

ffer

pre

miu

min

%45

.62

38.4

70

200

45.7

842

.51

020

00.

950

Hos

tile

dea

lin

%1.

6412

.72

010

013

.65

34.4

10

100

0.00

0L

BO

in%

0.10

3.20

010

00.

808.

940

100

0.01

5T

end

eroff

erin

%24

.60

43.0

80

100

10.8

431

.16

010

00.

000

qof

acq

uir

er2.

442.

370.

5115

.20

2.32

2.52

0.51

15.2

00.

432

qof

targ

et2.

031.

710.

509.

911.

841.

580.

509.

910.

094

qof

acq

uir

er>

qof

targ

etin

%62

.15

48.5

10

100

59.8

449

.12

010

00.

480

KZ

of

acq

uir

er-0

.09

1.47

-10.

463.

730.

121.

46-5

.26

3.73

0.04

4K

Zof

targ

et0.

121.

68-1

0.05

5.22

0.23

1.51

-6.4

44.

280.

347

Exp

erie

nce

dacq

uir

erin

%81

.61

38.7

50

100

69.8

845

.97

010

00.

000

%of

targ

etso

ught

95.0

717

.25

1.80

100

93.8

916

.55

15.4

010

00.

307

New

dea

lan

nou

nce

dw

ith

in2

yea

rsin

%n

/an

/an

/an

/a23

.18

42.3

00

100

n/a

Sam

ein

du

stry

(2d

igit

sS

IC)

in%

61.4

348

.69

010

055

.82

49.7

60

100

0.08

8

N1,

947

249

Pan

el

B:

Un

succ

essf

ul

Pu

reC

ash

an

dS

tock

Bid

s

Cas

hB

ids

Sto

ckB

ids

Vari

able

Mea

nS

td.

dev

.M

inM

axM

ean

Std

.d

ev.

Min

Max

p-v

alu

e

Tim

eto

fail

ure

61.6

545

.42

518

858

.83

46.7

25

232

0.67

3T

ran

sact

ion

valu

ein

2010

$bn

0.67

1.45

0.01

8.88

1.32

5.63

0.01

55.6

40.

296

Tar

get

size

in20

10

$bn

0.45

0.79

0.00

4.09

1.36

5.61

0.00

56.0

40.

140

Rel

ativ

ed

eal

size

1.17

2.26

0.00

11.4

20.

801.

150.

018.

490.

150

Off

erp

rem

ium

in%

51.6

638

.43

020

043

.94

43.1

80

200

0.19

6H

osti

led

eal

in%

26.7

444

.52

010

03.

7018

.97

010

00.

000

LB

Oin

%2.

3315

.16

010

00

00

00.

112

Ten

der

offer

in%

24.4

243

.21

010

01.

8513

.54

010

00.

000

qof

acq

uir

er1.

721.

570.

5113

.72

2.88

3.09

0.58

15.2

00.

002

qof

targ

et1.

471.

230.

509.

912.

241.

800.

559.

910.

001

qof

acq

uir

er>

qof

targ

etin

%61

.63

48.9

10

100

57.4

149

.68

010

00.

555

KZ

of

acq

uir

er0.

161.

38-3

.51

3.41

0.03

1.53

-4.5

53.

730.

567

KZ

ind

exta

rget

0.30

1.12

-2.2

92.

900.

001.

74-6

.44

4.28

0.18

3E

xp

erie

nce

dacq

uir

erin

%65

.12

47.9

40

100

70.3

745

.88

010

00.

438

%of

targ

etso

ught

92.3

617

.40

15.4

010

093

.97

17.1

519

.10

100

0.51

9N

ewd

eal

ann

oun

ced

wit

hin

2yea

rsin

%26

.09

44.2

30

100

19.6

139

.90

010

00.

320

Sam

ein

du

stry

(2d

igit

sS

IC)

in%

56.9

849

.80

010

057

.41

49.6

80

100

0.95

2

N86

108

Table 2. Failure Categories

The category Alliance denotes failed deals where bidder and target entered other cooperations.

Price too low indicates that the parties could not agree on the transaction price. Target rejection

refers to deals that failed because the shareholders and/or management or board refused the

bid. Target news to failed deals where the acquirer discovered (good or bad) information

in the due-diligence process. Market problems denotes deal failure due to shifting market

conditions (typically stock market plunges). Regulator refers to deal failure where the news

search revealed lack of regulatory approval, whereas Regulator (SDC) are deals for which SDC

(but no other retrievable source of news) indicates regulatory intervention. Management terms

describes all failed deals where acquirer and target were unable to agree on terms other than

the price (e.g., the nomination of a CEO of the future company). Bidder problems summarizes

deal cancelations due to financing problems or other bad news on the part of the bidder.

Bidder acquired are sudden cancelations triggered by the acquisition of the bidder. The column

entitled Average average percentage of the transaction value offered in cash. The columns Cash

coefficient target and Cash coefficient acquirer show the coefficient estimates from regressing,

respectively, the target’s or the acquirer’s CAR from 25 days before announcement to 25 days

after deal failure on the fraction offered in cash. */**/*** denote significance at the 10%/5%/1%

level, respectively.

Failure reason Avg. % cash Cash coeff. target Cash coeff. acquirer N

Alliance 35.9% 0.146 0.247 11

Price too low 54.4% 0.209* 0.049 28

Target rejection 59.0% 0.235 0.087 28

Target news 37.8% 0.367 0.100 29

Market problems 39.4% -0.387 0.526** 21

Regulator 50.6% 0.368* 0.162 21

Regulator (SDC) 46.2% 0.279* 0.310** 28

Management terms 35.4% 0.105 0.204 14

Bidder problems 20.5% 0.090 0.787* 22

Bidder acquired 33.3% 0.789 0.422 3

All bids with identified 43.2% 0.155* 0.244*** 160failure reasonAll bids (regression 44.3% 0.205*** 0.132** 249sample)

Table 3. Target Returns

OLS regressions with target CAR from 25 days before announcement to 25 days after deal failure

as the dependent variable. The sample All consists of all unsuccessful bids, the sample Pure of

all unsuccessful pure cash and stock bids. Cash is expressed as a fraction of the total payment

(and hence equal to a dummy for cash in the sample of pure cash and stock deals (Pure) in the

last three columns). Target size is the target’s market value of equity in 2010 $bn. Relative

deal size is the transaction value over the acquirer’s market value of equity. Offer premium is

normalized by the target’s market capitalization at one month prior to the announcement of

the bid, and truncated between 0 and 2. We include indicator variables for whether the bid

was hostile or a leveraged buyout. All non-deal-related variables (e.g., q ratios) are measured

at the end of the year prior to the unsuccessful deal’s announcement, and all q variables are

winsorized at the 1st and 99th percentiles. Industry fixed effects are based on one-digit SIC

codes. A constant term is always included in the absence of fixed effects. Robust standard

errors are in parentheses. */**/*** denote significance at the 10%/5%/1% level, respectively.

Target CAR (B-25, F+25)

Cash 0.205*** 0.193** 0.464*** 0.204*** 0.228** 0.552***(0.07) (0.08) (0.17) (0.07) (0.10) (0.20)

Log(target size) -0.031 -0.038*(0.02) (0.02)

Log(relative deal size) 0.011 0.007(0.02) (0.02)

Offer premium 0.347*** 0.398*** 0.337** 0.372***(0.11) (0.10) (0.14) (0.11)

Hostile 0.119* 0.133* 0.001 0.028(0.07) (0.07) (0.10) (0.10)

LBO 0.236 0.273 0.305 0.353(0.19) (0.21) (0.29) (0.31)

q of acquirer 0.022 0.031(0.02) (0.02)

q of target -0.010 0.009 -0.004 0.017(0.02) (0.03) (0.03) (0.03)

Cash × Premium -0.223 -0.228(0.24) (0.25)

Cash × q of target -0.090** -0.118**(0.04) (0.05)

Industry & year FE N Y Y N Y Y

Sample All All All Pure Pure PureN 249 249 249 194 194 194

Table 4. Target Returns – Alternative Return Specifications

OLS regressions with industry-adjusted target CAR (first two columns), buy-and-hold abnormal

return (third and fourth columns), and buy-and-hold return (last two columns) from 25 days

before announcement to 25 days after deal failure as the dependent variable. The sample

consists of all unsuccessful pure cash and stock bids. Target size is the target’s market value of

equity in 2010 $bn. Relative deal size is the transaction value over the acquirer’s market value

of equity. Offer premium is normalized by the target’s market capitalization at one month prior

to the announcement of the bid, and truncated between 0 and 2. We include indicator variables

for whether the bid was hostile or a leveraged buyout. All non-deal-related variables (e.g., q

ratios) are measured at the end of the year prior to the unsuccessful deal’s announcement, and

all q variables are winsorized at the 1st and 99th percentiles. Industry fixed effects are based

on one-digit SIC codes. A constant term is always included in the absence of fixed effects.

Robust standard errors are in parentheses. */**/*** denote significance at the 10%/5%/1%

level, respectively.

ICAR (B-25, F+25) BHAR (B-25, F+25) BHR (B-25, F+25)

Cash ∈ {0, 1} 0.197*** 0.548*** 0.201*** 0.296* 0.209*** 0.351**(0.07) (0.19) (0.07) (0.15) (0.07) (0.15)

Log(target size) -0.039* -0.008 -0.002(0.02) (0.02) (0.02)

Log(relative deal size) 0.007 -0.006 -0.011(0.02) (0.02) (0.02)

Offer premium 0.390*** 0.016 0.069(0.11) (0.16) (0.15)

Hostile 0.016 0.203** 0.205**(0.11) (0.09) (0.09)

LBO 0.204 0.513 0.580(0.29) (0.34) (0.35)

q of acquirer 0.032 -0.015 -0.016(0.02) (0.02) (0.02)

q of target 0.018 0.064** 0.061*(0.03) (0.03) (0.03)

Cash × Premium -0.284 0.164 0.100(0.24) (0.22) (0.22)

Cash × q of target -0.107** -0.146*** -0.152***(0.04) (0.05) (0.05)

Industry & year FE N Y N Y N Y

N 194 194 194 194 194 194

Table 5. Target Returns – Exogenous Failure Reasons

OLS regressions with target CAR from 25 days before announcement to 25 days after deal failure

as the dependent variable. Sample C consists of all deals that were canceled due to (potential)

regulatory intervention (i.e., even if only indicated in the SDC deal synopsis), bidder news, or

disagreement on management terms. Sample N is limited to all bids that were not withdrawn

due to market problems or news regarding the target. Cash is expressed as a fraction of the

total payment (and hence equal to a dummy for cash in the sample of pure cash and stock

deals (Pure) in the last two columns). Target size is the target’s market value of equity in

2010 $bn. Relative deal size is the transaction value over the acquirer’s market value of equity.

Offer premium is normalized by the target’s market capitalization at one month prior to the

announcement of the bid, and truncated between 0 and 2. We include indicator variables for

whether the bid was hostile or a leveraged buyout (there are no LBOs in Sample C, however).

All non-deal-related variables (e.g., q ratios) are measured at the end of the year prior to

the unsuccessful deal’s announcement, and all q variables are winsorized at the 1st and 99th

percentiles. Industry fixed effects are based on one-digit SIC codes. A constant term is always

included in the absence of fixed effects. Robust standard errors are in parentheses. */**/***

denote significance at the 10%/5%/1% level, respectively.

Target CAR (B-25, F+25)

Cash 0.240** 0.194* 0.614* 0.153** 0.311** 0.172** 0.320**(0.10) (0.11) (0.34) (0.06) (0.13) (0.07) (0.15)

Log(target size) 0.036 0.028 -0.024 -0.031 -0.030* -0.038(0.02) (0.04) (0.01) (0.02) (0.02) (0.02)

Log(rel. deal size) -0.019 0.042 -0.006 0.017 -0.001 0.015(0.03) (0.05) (0.02) (0.02) (0.02) (0.02)

Offer premium 0.445*** 0.352* 0.425*** 0.339*** 0.436*** 0.281***(0.13) (0.19) (0.08) (0.10) (0.08) (0.10)

Hostile -0.008 0.037 0.098* 0.059 0.046 -0.037(0.11) (0.16) (0.06) (0.07) (0.06) (0.10)

LBO N/A N/A 0.074* 0.286 0.084 0.297(0.04) (0.19) (0.05) (0.29)

q of acquirer 0.048* 0.004 0.010(0.02) (0.03) (0.03)

q of target -0.110** 0.018 0.018(0.04) (0.04) (0.04)

Cash × Premium -0.387 0.044 0.104(0.29) (0.16) (0.20)

Cash × q of tgt. -0.179 -0.082* -0.084*(0.13) (0.04) (0.04)

Ind. & year FE N N Y N Y N Y

Sample C C C N N N , Pure N , PureN 76 76 76 201 201 159 159

Table 6. Determinants of Deal Failure

OLS regressions with a dummy variable for deal failure as the dependent variable. Target size

is the target’s market value of equity in 2010 $bn. Relative deal size is the transaction value

over the acquirer’s market value of equity. Offer premium is normalized by the target’s market

capitalization at one month prior to the announcement of the bid, and truncated between 0

and 2. We include indicator variables for whether the bid was hostile or a leveraged buyout.

KZ index is the four-variable version in Lamont et al. (2001). Experienced acquirers (dummy

variable) appear at least five times in the data set. Target CAR is measured on the [-25, 1]

window around deal announcement. All non-deal-related variables (e.g., q ratios) are measured

at the end of the year prior to the unsuccessful deal’s announcement, and all q and KZ variables

are winsorized at the 1st and 99th percentiles. Full set of interactions with Cash indicates that

interactions of all independent are included. (All coefficients are insignificant and not shown.)

Industry fixed effects are based on one-digit SIC codes. A constant term is always included

in the absence of fixed effects. Robust standard errors are in parentheses. */**/*** denote

significance at the 10%/5%/1% level, respectively.

Deal failure

Cash ∈ [0, 1] -0.015 -0.014 -0.033 -0.138(0.01) (0.03) (0.03) (0.12)

Log(target size) -0.015*** -0.016*** -0.013*(0.00) (0.01) (0.01)

Log(relative deal size) 0.027*** 0.024*** 0.030***(0.00) (0.00) (0.01)

Offer premium -0.041* 0.020 0.021(0.02) (0.03) (0.04)

Hostile 0.341*** 0.286*** 0.309***(0.06) (0.06) (0.11)

LBO 0.232 0.063 0.061(0.22) (0.23) (0.24)

q of acquirer 0.007* 0.007* 0.003(0.00) (0.00) (0.00)

q of target 0.001 0.003 0.005(0.01) (0.01) (0.01)

KZ of acquirer 0.003 0.006(0.01) (0.01)

KZ of target -0.000 -0.005(0.00) (0.01)

Experienced acquirer -0.023 -0.022(0.02) (0.03)

Target CAR (B-25, B+1) -0.141*** -0.125**(0.04) (0.05)

% of target sought -0.001 -0.001(0.00) (0.00)

Cash × Premium 0.056 0.101** 0.100(0.04) (0.04) (0.07)

Cash × q of target -0.001 -0.003 -0.008(0.01) (0.01) (0.01)

Full set of interactions with Cash N N N YIndustry & year FE N Y Y Y

N 2,206 2,200 1,967 1,967

Table 7. Robustness Checks

OLS regressions with target CAR from 25 days before announcement to 25 days after deal failure as the

dependent variable. The sample is restricted to unsuccessful bids. KZ index is the four-variable version

in Lamont et al. (2001). Same industry is an indicator for whether target and bidder have matching two-

digit SIC codes. Toehold is equal to one if the acquirer owned a share of the target before announcement.

Experienced acquirers (dummy variable) appear at least five times in the data set. Cash of acquirer

is the acquirer’s cash position in 2010 $m on the balance sheet (as used in the construction of the KZ

index). All other variables are as in our baseline specification in Table 3. All non-deal-related variables

(e.g., q ratios) are measured at the end of the year prior to the unsuccessful deal’s announcement, and the

acquirer’s cash balance as well as all q and KZ variables are winsorized at the 1st and 99th percentiles.

Coefficient estimates for Offer premium, Hostile, Tender offer (dummy variable), LBO, and q of acquirer

and target are not shown; as in Table 3, only Offer premium (and Hostile) is significant. Industry fixed

effects are based on one-digit SIC codes. Robust standard errors are in parentheses. */**/*** denote

significance at the 10%/5%/1% level, respectively.

Target CAR (B-25, F+25)Cash ∈ [0, 1] 0.323** 0.402** 0.493*** 0.359** 0.467*** 0.458***

(0.16) (0.16) (0.17) (0.18) (0.17) (0.16)Log(target size) -0.030 -0.030 -0.037* -0.015 -0.041** -0.018

(0.02) (0.02) (0.02) (0.02) (0.02) (0.02)Log(relative deal size) 0.021 0.012 0.015 -0.000 0.020 0.003

(0.03) (0.02) (0.02) (0.02) (0.03) (0.02)KZ of acquirer -0.018

(0.04)KZ of target 0.027

(0.03)Same industry (2 digits SIC) -0.049

(0.08)Toehold 0.227**

(0.09)Experienced acquirer -0.238**

(0.10)Cash of acquirer 0.036**

(0.02)Cash × Premium -0.105 -0.215 -0.223 -0.252 -0.229 0.145

(0.23) (0.24) (0.24) (0.25) (0.24) (0.25)Cash × q of target -0.058 -0.091** -0.100** -0.104** -0.087** -0.093**

(0.04) (0.04) (0.04) (0.05) (0.04) (0.04)Cash × KZ of acquirer 0.007

(0.06)Cash × Same industry 0.095

(0.12)Cash × Toehold -0.102

(0.13)Cash × Experienced acquirer 0.197

(0.15)Cash × Cash of acquirer -0.030

(0.02)Cash × 1980s/1990s × Premium -0.515**

(0.25)Controls for Premium, Hostile, Y Y Y Y Y YTender offer, LBO, q of acquirer,and q of targetIndustry & year FE Y Y Y Y Y YN 217 249 249 249 249 249

Table 8. Future Takeover Activity

OLS regressions (i) with a dummy variable indicating another merger bid within the next two

years as the dependent variable in the first three columns, and (ii) with future offer premia in

case of another merger bid within the next two years (and zero otherwise) as the dependent

variable in the last three columns. We exclude observations with merger bids within half a

year after failure since their classification as competing bid (in the previous takeover attempt)

versus new bid is ambiguous. The sample All consists of all remaining unsuccessful bids, the

sample Pure of all remaining unsuccessful pure cash and stock bids. Cash is expressed as a

fraction of the total payment (and hence equal to a dummy for cash in the sample of pure cash

and stock deals (Pure) in the third and sixth columns). Target size is the target’s market value

of equity in 2010 $bn. Offer premium is normalized by the target’s market capitalization at

one month prior to the announcement of the bid, and truncated between 0 and 2. We include

an indicator variable for whether the bid was hostile. KZ index is the four-variable version in

Lamont et al. (2001). Target CAR (B-25, F+25) is the cumulative abnormal return from 25

days before announcement until 25 days after deal failure. All non-deal-related variables (e.g., q

ratios) are measured at the end of the year prior to the unsuccessful deal’s announcement, and

all q and KZ variables are winsorized at the 1st and 99th percentiles. Industry fixed effects are

based on one-digit SIC codes. A constant term is always included in the absence of fixed effects.

Robust standard errors are in parentheses. */**/*** denote significance at the 10%/5%/1%

level, respectively.

Dependent variable Bid Bid Bid Premium Premium Premium

Cash 0.068 0.061 0.033 0.026 0.006 0.009(0.06) (0.09) (0.09) (0.03) (0.06) (0.07)

Log(target size) 0.022 0.010 0.001 0.001(0.02) (0.02) (0.01) (0.01)

Offer premium -0.024 0.003 0.052 0.084(0.08) (0.10) (0.07) (0.09)

Hostile -0.084 0.019 -0.023 0.002(0.14) (0.17) (0.05) (0.07)

q of target -0.034** -0.027 -0.015** -0.018(0.02) (0.02) (0.01) (0.01)

KZ of target 0.010 -0.003 -0.008 -0.019(0.02) (0.03) (0.01) (0.01)

Target CAR (B-25, F+25) -0.056 -0.067 -0.003 0.017(0.06) (0.07) (0.03) (0.04)

Industry & year FE N Y Y N Y Y

Sample All All Pure All All PureN 220 206 161 213 199 156

Supplementary Appendix

to

Cash is King –

Revaluation of Targets after Merger Bids

by Ulrike Malmendier, Marcus M. Opp, and Farzad Saidi

Tab

leA

.1.

Su

mm

ary

Sta

tist

ics

–S

uccess

ful

Bid

s

Tim

eto

com

ple

tion

isin

trad

ing

day

s.T

ran

sact

ion

valu

esin

2010

$bn

are

bas

edon

his

tori

cal

tran

sact

ion

valu

es,

conve

rted

usi

ng

Con

sum

erP

rice

Ind

ex(C

PI)

Con

ver

sion

Fac

tors

.T

arge

tsi

zeis

the

targ

et’s

mar

ket

valu

eof

equ

ity

in20

10$b

n.

Rel

ativ

ed

eal

size

isth

etr

ansa

ctio

nva

lue

over

the

acq

uir

er’s

mark

etva

lue

ofeq

uit

y.O

ffer

pre

miu

mis

nor

mal

ized

by

the

targ

et’s

mar

ket

cap

ital

izat

ion

aton

em

onth

pri

orto

the

ann

oun

cem

ent

ofth

eb

id,

and

tru

nca

ted

bet

wee

n0

and

2.H

osti

led

eal,

LB

O,

and

Ten

der

offer

are

du

mm

yva

riab

les

ind

icat

ing

host

ile

bid

s,le

vera

ged

bu

you

ts,

and

tend

eroff

ers,

resp

ecti

vely

.q

rati

ois

the

mar

ket

valu

eof

equ

ity

plu

sas

sets

min

us

the

book

valu

eof

equ

ity

all

over

asse

ts.

KZ

ind

exis

the

fou

r-va

riab

leve

rsio

nas

,e.

g.,

inL

amon

tet

al.

(200

1).

New

dea

lan

nou

nce

dw

ith

in2

year

s(d

um

my

vari

ab

le)

isco

nd

itio

nal

onth

ed

eal

bei

ng

ann

oun

ced

atle

ast

hal

fa

year

afte

rth

ep

revio

us

one.

Exp

erie

nce

dac

qu

irer

s(d

um

my

vari

able

)ap

pea

rat

least

five

tim

esin

the

dat

ase

t.S

ame

ind

ust

ryis

anin

dic

ator

vari

able

for

wh

eth

erta

rget

and

bid

der

oper

ate

inth

esa

me

ind

ust

ry,

as

defi

ned

by

the

two-d

igit

SIC

cod

es.

All

non

-dea

l-re

late

dva

riab

les

(e.g

.,q

rati

os)

are

mea

sure

dat

the

end

ofth

eye

arp

rior

toth

ed

eal’

san

nou

nce

men

t,an

dallq

and

KZ

vari

able

sar

ew

inso

rize

dat

the

1st

and

99th

per

centi

les.

Th

ep

-val

ues

inth

ela

stco

lum

nar

efo

ra

two-s

ided

diff

eren

ce-i

n-m

ean

ste

st.

Th

e(r

egre

ssio

n)

sam

ple

isre

stri

cted

tosu

cces

sfu

lb

ids

for

wh

ich

all

contr

olva

riab

les

are

avai

lab

le,

exce

pt

for

KZ

ofacq

uir

er(6

78ca

shan

d61

8st

ock

dea

ls)

and

KZ

ofta

rget

(666

cash

and

625

stock

dea

ls).

Cas

hB

ids

Sto

ckB

ids

Vari

able

Mea

nS

td.

dev

.M

inM

axM

ean

Std

.d

ev.

Min

Max

p-v

alu

e

Tim

eto

com

ple

tion

58.9

037

.84

523

084

.80

39.3

516

239

0.00

0T

ran

sact

ion

valu

ein

2010

$bn

0.68

1.46

0.00

17.4

01.

595.

670.

0070

.51

0.00

0T

arge

tsi

zein

2010

$bn

0.52

1.09

0.00

11.4

01.

104.

140.

0065

.16

0.00

0R

elat

ive

dea

lsi

ze0.

695.

830.

0013

7.03

0.58

3.17

0.00

79.3

80.

662

Off

erp

rem

ium

in%

46.3

935

.11

020

047

.49

43.0

20

200

0.60

4H

osti

led

eal

in%

2.56

15.8

00

100

0.75

8.67

010

00.

010

LB

Oin

%0.

285.

320

100

00

00

0.17

1T

end

eroff

erin

%49

.29

50.0

30

100

4.08

19.8

10

100

0.00

0q

ofacq

uir

er2.

081.

710.

5115

.20

3.28

3.22

0.51

15.2

00.

000

qof

targ

et1.

841.

390.

509.

912.

582.

270.

509.

910.

000

qof

acq

uir

er>

qof

targ

etin

%59

.80

40.0

60

100

63.6

948

.13

010

00.

140

KZ

of

acq

uir

er-0

.15

1.30

-6.0

33.

73-0

.36

1.66

-10.

463.

010.

011

KZ

ind

exta

rget

-0.0

41.

58-1

0.05

5.22

-0.2

41.

86-9

.57

4.70

0.03

6E

xp

erie

nce

dacq

uir

erin

%83

.24

37.3

80

100

78.6

741

.00

010

00.

031

%of

targ

etso

ught

91.7

822

.36

1.80

100

95.7

016

.04

7.70

100

0.00

0N

ewd

eal

ann

oun

ced

wit

hin

2yea

rsin

%n

/an

/an

/an

/an/a

n/a

n/a

n/a

n/a

Sam

ein

du

stry

(2d

igit

sS

IC)

in%

53.8

449

.89

010

064

.75

47.8

10

100

0.00

0

N70

466

1

Table A.2. Determinants of Cash Offers

OLS regressions include acquirer and target industry controls. Cash is expressed as a fraction

of the total payment (and hence equal to a dummy for cash in the sample of pure cash and

stock deals (Pure) in the last three columns). Target size is the target’s market value of equity

in 2010 $bn. Relative deal size is the transaction value over the acquirer’s market value of

equity. Offer premium is normalized by the target’s market capitalization at one month prior

to the announcement of the bid, and truncated between 0 and 2. We include indicator variables

for whether the bid was hostile or a leveraged buyout. KZ index is the four-variable version in

Lamont et al. (2001). Experienced acquirers (dummy variable) appear at least five times in the

data set. All non-deal-related variables (e.g., q ratios) are measured at the end of the year prior

to the unsuccessful deal’s announcement, and all q and KZ variables are winsorized at the 1st

and 99th percentiles. Industry fixed effects are based on one-digit SIC codes. Robust standard

errors are in parentheses. */**/*** denote significance at the 10%/5%/1% level, respectively.

Cash ∈ [0, 1] Cash ∈ {0, 1}Log(target size) 0.003 0.002 0.004 0.001

(0.01) (0.01) (0.01) (0.01)Log(relative deal size) -0.053*** -0.061*** -0.063*** -0.066*** -0.071*** -0.074***

(0.01) (0.01) (0.01) (0.01) (0.01) (0.01)Offer premium 0.060** 0.041

(0.03) (0.03)Hostile 0.232*** 0.288***

(0.05) (0.06)LBO 0.410*** 0.376***

(0.09) (0.11)q of acquirer -0.028*** -0.030*** -0.029*** -0.028***

(0.00) (0.00) (0.00) (0.01)q of target -0.032*** -0.034*** -0.037*** -0.038***

(0.01) (0.01) (0.01) (0.01)KZ of acquirer -0.001 0.002

(0.01) (0.01)KZ of target -0.014** -0.011

(0.01) (0.01)Exp. acquirer -0.006 -0.011 -0.017 -0.010 -0.008 -0.021

(0.02) (0.02) (0.03) (0.03) (0.03) (0.03)% of target sought -0.002*** -0.001** -0.002** -0.002*** -0.001* -0.001**

(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)Industry & year FE Y Y Y Y Y Y

Sample All All All Pure Pure PureN 2,200 2,200 1,967 1,563 1,563 1,417

Table A.3. Determinants of Deal Failure: Pure Cash and Pure Stock Bids

OLS regressions with a dummy variable for deal failure as the dependent variable. The sample

is restricted to unsuccessful pure cash and stock bids. Target size is the target’s market value of

equity in 2010 $bn. Relative deal size is the transaction value over the acquirer’s market value

of equity. Offer premium is normalized by the target’s market capitalization at one month prior

to the announcement of the bid, and truncated between 0 and 2. We include indicator variables

for whether the bid was hostile or a leveraged buyout. KZ index is the four-variable version

in Lamont et al. (2001). Experienced acquirers (dummy variable) appear at least five times

in the data set. Target CAR is measured on the [-25, 1] window around deal announcement.

All non-deal-related variables (e.g., q ratios) are measured at the end of the year prior to the

unsuccessful deal’s announcement, and all q and KZ variables are winsorized at the 1st and 99th

percentiles. Full set of interactions with Cash indicates that interactions of all independent are

included. (All coefficients are insignificant and not shown.) Industry fixed effects are based

on one-digit SIC codes. A constant term is always included in the absence of fixed effects.

Robust standard errors are in parentheses. */**/*** denote significance at the 10%/5%/1%

level, respectively.

Deal failure

Cash ∈ {0, 1} -0.033* -0.049 -0.058* -0.211(0.02) (0.03) (0.03) (0.14)

Log(target size) -0.015** -0.017** -0.014(0.01) (0.01) (0.01)

Log(relative deal size) 0.033*** 0.032*** 0.043***(0.00) (0.01) (0.01)

Offer premium -0.037 0.010 0.015(0.03) (0.04) (0.05)

Hostile 0.358*** 0.291*** 0.312**(0.07) (0.07) (0.14)

LBO 0.236 0.072 0.072(0.22) (0.22) (0.23)

q of acquirer 0.005 0.005 0.003(0.00) (0.00) (0.01)

q of target -0.003 0.002 0.005(0.01) (0.01) (0.01)

KZ of acquirer 0.009 0.011(0.01) (0.01)

KZ of target 0.003 0.002(0.00) (0.01)

Experienced acquirer -0.013 -0.011(0.03) (0.04)

Target CAR (B-25, B+1) -0.105** -0.091(0.05) (0.06)

% of target sought -0.000 -0.002(0.00) (0.00)

Cash × Premium 0.073 0.109** 0.100(0.04) (0.05) (0.07)

Cash × q of target 0.004 0.001 -0.006(0.01) (0.01) (0.01)

Full set of interactions with Cash N N N YIndustry & year FE N Y Y Y

N 1,566 1,563 1,417 1,417

Table A.4. Pure Cash and Pure Stock Deals, Robustness Checks

OLS regressions with target CAR from 25 days before announcement to 25 days after deal failure as

the dependent variable. The sample is restricted to unsuccessful pure cash and stock bids. KZ index

is the four-variable version in Lamont et al. (2001). Same industry is an indicator for whether target

and bidder have matching two-digit SIC codes. Toehold is equal to one if the acquirer owned a share

of the target before announcement. Experienced acquirers (dummy variable) appear at least five times

in the data set. Cash of acquirer is the acquirer’s cash position in 2010 $m on the balance sheet (as

used in the construction of the KZ index). All other variables are as in our baseline specification in

Table 3. All non-deal-related variables (e.g., q ratios) are measured at the end of the year prior to the

unsuccessful deal’s announcement, and the acquirer’s cash balance as well as all q and KZ variables are

winsorized at the 1st and 99th percentiles. Coefficient estimates for Offer premium, Hostile, Tender offer

(dummy variable), LBO, and q of acquirer and target are not shown; as in Table 3, only Offer premium

is significant. Industry fixed effects are based on one-digit SIC codes. Robust standard errors are in

parentheses. */**/*** denote significance at the 10%/5%/1% level, respectively.

Target CAR (B-25, F+25)Cash ∈ {0, 1} 0.345* 0.431** 0.554*** 0.424** 0.548*** 0.553***

(0.20) (0.20) (0.21) (0.19) (0.20) (0.19)Log(target size) -0.033 -0.037* -0.046* -0.016 -0.042* -0.018

(0.03) (0.02) (0.02) (0.02) (0.02) (0.02)Log(relative deal size) 0.020 0.008 0.007 -0.013 0.015 -0.007

(0.03) (0.02) (0.02) (0.03) (0.03) (0.02)KZ of acquirer -0.034

(0.04)KZ of target 0.044

(0.04)Same industry (2 digits SIC) -0.113

(0.10)Toehold 0.276***

(0.10)Experienced acquirer -0.309**

(0.12)Cash of acquirer 0.024

(0.03)Cash × Premium -0.096 -0.218 -0.194 -0.258 -0.240 0.177

(0.26) (0.26) (0.27) (0.26) (0.26) (0.27)Cash × q of target -0.057 -0.113** -0.125*** -0.140*** -0.113** -0.124***

(0.04) (0.05) (0.05) (0.05) (0.05) (0.05)Cash × KZ of acquirer 0.057

(0.06)Cash × Same industry 0.160

(0.14)Cash × Toehold -0.134

(0.14)Cash × Experienced acquirer 0.240

(0.16)Cash × Cash of acquirer -0.020

(0.03)Cash × 1980s/1990s × Premium -0.651**

(0.27)Controls for Premium, Hostile, Y Y Y Y Y YTender offer, LBO, q of acquirer,and q of targetIndustry & year FE Y Y Y Y Y YN 171 194 194 194 194 194

Table A.5. Target Returns: All Acquirers

OLS regressions with target CAR from 25 days before announcement to 25 days after deal

failure as the dependent variable. The sample All consists of all unsuccessful bids, including

those by non-public acquirers. The sample Pure consists of all unsuccessful pure cash and

stock bids, including those by non-public acquirers. Cash is expressed as a fraction of the total

payment (and hence equal to a dummy for cash in the sample of pure cash and stock deals

(Pure) in the last three columns). Target size is the target’s market value of equity in 2010

$bn. Offer premium is normalized by the target’s market capitalization at one month prior to

the announcement of the bid, and truncated between 0 and 2. We include indicator variables

for whether the bid was hostile or a leveraged buyout. All non-deal-related variables (e.g., q

ratios) are measured at the end of the year prior to the unsuccessful deal’s announcement, and

all q variables are winsorized at the 1st and 99th percentiles. Industry fixed effects are based

on one-digit SIC codes. A constant term is always included in the absence of fixed effects.

Robust standard errors are in parentheses. */**/*** denote significance at the 10%/5%/1%

level, respectively.

Target CAR (B-25, F+25)

Cash 0.100** 0.103** 0.227** 0.128*** 0.133** 0.346***(0.04) (0.04) (0.09) (0.05) (0.06) (0.11)

Log(target size) -0.027*** -0.033***(0.01) (0.01)

Offer premium 0.363*** 0.406*** 0.355*** 0.446***(0.06) (0.08) (0.06) (0.09)

Hostile 0.095*** 0.131*** 0.099*** 0.143***(0.03) (0.03) (0.03) (0.04)

LBO 0.017 0.023 0.017 0.017(0.04) (0.04) (0.05) (0.05)

q of target -0.020 0.007 -0.007 0.037*(0.02) (0.02) (0.02) (0.02)

Cash × Premium -0.118 -0.187*(0.11) (0.11)

Cash × q of target -0.053* -0.085***(0.03) (0.03)

Industry & year FE N Y Y N Y Y

Sample All All All Pure Pure PureN 872 872 872 662 662 662

Table A.6. Determinants of Target Debt Changes

OLS regressions with the one-year change (from the end of the year before deal announcement

to the end of the year of deal failure) in the log of the sum of the target’s long-term and short-

term debt as the dependent variable, which is winsorized at the 1st and 99th percentiles. All

samples are restricted to unsuccessful bids. The samples are as follows: (I) regression sample;

(II) no restrictions on sample; (III) regression sample, pure cash and stock deals; (IV) no

restrictions on sample other than pure cash and stock deals, i.e., including private acquirers.

Cash is expressed as a fraction of the total payment (and hence equal to a dummy for cash in

the sample of pure cash and stock deals in the last two columns). Offer premium is normalized

by the target’s market capitalization at one month prior to the announcement of the bid, and

truncated between 0 and 2. We include indicator variables for whether the bid was hostile

or a leveraged buyout. KZ index is the four-variable version in Lamont et al. (2001). All

non-deal-related variables (e.g., q ratios) are measured at the end of the year prior to the

unsuccessful deal’s announcement, and all q and KZ variables are winsorized at the 1st and

99th percentiles. Industry fixed effects are based on one-digit SIC codes. Robust standard

errors are in parentheses. */**/*** denote significance at the 10%/5%/1% level, respectively.

∆ lnDebt

Cash -0.002 -0.102 -0.107 -0.080(0.16) (0.09) (0.16) (0.11)

Offer premium 0.034 0.101 0.161 0.140(0.13) (0.07) (0.15) (0.09)

Hostile 0.199 0.185* 0.126 0.066(0.18) (0.10) (0.21) (0.10)

LBO 3.899*** 0.107 3.908*** 0.141(0.49) (0.09) (0.52) (0.11)

q of target 0.017 0.034 -0.090 0.006(0.05) (0.03) (0.06) (0.03)

KZ of target -0.013 -0.046** -0.079 -0.057**(0.05) (0.02) (0.05) (0.02)

Industry & year FE Y Y Y Y

Sample (I) (II) (III) (IV)N 203 687 160 525

Table A.7. Acquirer Returns

OLS regressions with acquirer CAR from 25 days before announcement to 25 days after deal

failure as the dependent variable. The sample All consists of all unsuccessful bids, and the

sample C consists of all deals that were canceled due to (potential) regulatory intervention (i.e.,

even if only indicated in the SDC deal synopsis), target news, or disagreement on management

terms. Target size is the target’s market value of equity in 2010 $bn. Relative deal size is

the transaction value over the acquirer’s market value of equity. Offer premium is normalized

by the target’s market capitalization at one month prior to the announcement of the bid, and

truncated between 0 and 2. We include indicator variables for whether the bid was hostile or a

leveraged buyout. KZ index is the four-variable version in Lamont et al. (2001). All non-deal-

related variables (e.g., q ratios) are measured at the end of the year prior to the unsuccessful

deal’s announcement, and all q and KZ variables are winsorized at the 1st and 99th percentiles.

Industry fixed effects are based on one-digit SIC codes. A constant term is always included

in the absence of fixed effects. Robust standard errors are in parentheses. */**/*** denote

significance at the 10%/5%/1% level, respectively.

Acquirer CAR (B-25, F+25)

Cash ∈ [0, 1] 0.132** 0.138** 0.056 0.156*(0.06) (0.07) (0.14) (0.09)

Log(target size) -0.002(0.02)

Log(relative deal size) 0.016(0.02)

Offer premium -0.148(0.12)

Hostile -0.004(0.08)

LBO 0.077(0.18)

q of acquirer 0.006(0.04)

q of target -0.020(0.03)

KZ of acquirer 0.028(0.05)

KZ of target 0.018(0.03)

Cash × Premium 0.189(0.15)

Cash × q of acquirer -0.054(0.06)

Cash × KZ of acquirer 0.007(0.06)

Industry & year FE N Y Y N

Sample All All All CN 249 249 217 82