Do Wealth Creating Mergers and Acquisitions Really Hurt Acquirer Shareholders? * School of Banking and Finance, Australian School of Business, UNSW

Transcription

1 Do Wealth Creating Mergers and Acquisitions Really Hurt Acquirer Shareholders? * Ron Masulis, Peter L. Swan and Brett Tobiansky, School of Banking and Finance, Australian School of Business, UNSW Draft: June 16, 2011 ABSTRACT We examine the expected economic benefits of mergers and acquisitions. We conclude that both signaling and revelation biases are responsible for the commonly reported finding that on average takeovers are harmful to acquirer shareholder wealth. After accounting for these two biases that lead to a price fall on announcement of 18.9% ($563.9 million), we demonstrate that acquirers generally benefit from takeovers with an average 81% share of the economic gains from the transaction. By studying bids that fail for exogenous reasons, which are largely free of signaling and revelation biases, we confirm the neoclassical view that takeovers are positive NPV projects for a typical acquirer, which produce a sizeable return on capital of 21% to the acquirer ($626.6 million) and 21.2% ($772.2 million) to the combined acquirer-target. This conclusion is based on two important findings. First, on a failed acquisition announcement, the combined acquirer and target value on average falls, where both target and acquirer suffer significant negative abnormal returns. Second, acquirers share in a significant portion of the economic benefits of a successful acquisition, reflected in a significantly positive relationship between acquirer and target stock returns utilizing a 60-day initial bid announcement window and a 100-day period following the termination announcement. Over the same window, exogenously failed cash bidders significantly underperform successful cash bidders by 10.7% and exogenously failed stock bidders significantly underperform successful stock bidders by a further 15.5% making a total differential of 26.2%. Moreover, in the long term, stock-funded targets typically only receive half the premium of cash targets. Key Words: M&A, takeover bids, acquisition benefits, acquirer gains, acquisition synergies, failed bids. JEL Codes: G34, G14 * We thank Ken Ahern and Emir Hrnjic for constructive comments. Also, participants at Financial Intermediation Research Society (FIRS) Conference, Sydney Eugene Chua, Nick Orlic, and Ewe Helmes provided assistance earlier on. r.masulis@unsw.edu.au. Contact author. Department of Banking and Finance, ASB, UNSW Sydney NSW 2052 Australia. Tel: +61 (0) peter.swan@unsw.edu.au. brett@unsw.edu.au. 1

2 Do Wealth Creating Mergers and Acquisitions Really Hurt Acquirer Shareholders? Draft: June 16, 2011 ABSTRACT We examine the expected economic benefits of mergers and acquisitions. We conclude that both signaling and revelation biases are responsible for the commonly reported finding that on average takeovers are harmful to acquirer shareholder wealth. After accounting for these two biases that lead to a price fall on announcement of 18.9% ($563.9 million), we demonstrate that acquirers generally benefit from takeovers with an average 81% share of the economic gains from the transaction. By studying bids that fail for exogenous reasons, which are largely free of signaling and revelation biases, we confirm the neoclassical view that takeovers are positive NPV projects for a typical acquirer, which produce a sizeable return on capital of 21% to the acquirer ($626.6 million) and 21.2% ($772.2 million) to the combined acquirer-target. This conclusion is based on two important findings. First, on a failed acquisition announcement, the combined acquirer and target value on average falls, where both target and acquirer suffer significant negative abnormal returns. Second, acquirers share in a significant portion of the economic benefits of a successful acquisition, reflected in a significantly positive relationship between acquirer and target stock returns utilizing a 60-day initial bid announcement window and a 100-day period following the termination announcement. Over the same window, exogenously failed cash bidders significantly underperform successful cash bidders by 10.7% and exogenously failed stock bidders significantly underperform successful stock bidders by a further 15.5% making a total differential of 26.2%. Moreover, in the long term, stock-funded targets typically only receive half the premium of cash targets. Key Words: M&A, takeover bids, acquisition benefits, acquirer gains, acquisition synergies, failed bids. JEL Codes: G34, G14 2

3 1. INTRODUCTION Many studies document that acquirers systematically destroy shareholder wealth in mergers and acquisitions (M&As), while targets benefit at their expense. 5 Moeller, Schlingemann, and Stulz (2005) find that shareholders of acquiring firms over the period lose 12 cents per dollar on takeover announcement and an incredible $240 billion in dollar terms. In their survey, Betton, Eckbo, and Thorburn (2008) find a significantly negative abnormal return of percent for large stock bidders of public targets, which shrinks to -0.3 percent for large cash bidders of public targets. Small stock bidders approximately break-even, while small cash bidders gain 3.06 percent. Thus, if one sets a zero NPV as the floor return to the bidder in an auction market, only small bidders, not large bidders where most of the investment is, satisfy this rationality criterion. Bayazitova, Kahl, and Valkanov (2011) find that these large acquirers (mega-mergers) account for 43% of all merger outlays. Apparent value-destroying deals representing large negative NPV projects do not represent some minor backwater in which irrationality flourishes. Strangely, in view of its wealth-destructive properties, M&A activity plays a significant role in the global economy, constituting $2.1 trillion in the US in just one year alone, or 15% of GDP (see Bao and Edmans (2011)). The virtually universal finding that acquiring shareholders do not gain from M&A activity features in practically every major textbook (see, for example, Ross, Westerfield, and Jordan (2008, p. 835) or Copeland, Weston, and Shastri (2005, p. 778)). The overall synergistic benefits made up of the net change in the value of the bidder and target on takeover announcement are marginally positive, where acquirer shareholders are distinct losers, while target shareholders are clear winners. These virtually universal findings stand in stark contrast to what may be termed the neoclassical theory of M&A (see Ahern and Weston (2007)) which asserts the profit motive of the acquirer will naturally drive the ownership of assets to their highest value use. It follows from this motivation that the initiator (acquirer and its shareholders) will benefit from such wealth-enhancing transactions, rather than suffer losses. Thus, if there are changes in technology due (say) to innovation, regulatory changes, or shifts in demand for goods and 5 For an early example of a study showing that acquirer value falls at the time of takeover announcement see Dodd (1980). Betton, Eckbo, and Thorburn (2008, Table 6) summarize 16 relatively recent large-sample studies of acquirer returns. Most report sizeable takeover samples in which the bidder s share price reaction is negative. 3

4 services such that the existing ownership of assets is no longer optimal, M&A activity should occur to redistribute the ownership and control of assets from a target to a bidder so as to enhance the overall value of the merged firms, resulting in improved utilization of the combined firm s assets. In this paper and consistent with neoclassical theory, we show that there is no foundation for the belief that, even where target shareholders benefit, this typically comes at a considerable cost to acquirer shareholders. To the contrary, we demonstrate that not only are the net economic benefits of M&A large, but they are typically shared between acquirer and target. In fact, an acquirer on average gains the lion s share from a typical acquisition, capturing a 81% share or US$626.6 million of the associated total economic benefits, which average US$772.2 million. These gains are not obvious to researchers since making a bid also releases bad news about a bidder s value, which reduces its equity capitalization by $563.9 million or 18.9%. While this loss in value is offset by the prospective benefits of a typical M&A bid, the net effect implies little change in a bidder s equity value. Most recently, behavioral financial economists have focused on the use of the acquirer s own stock, rather than cash as the means of payment for the target (e.g., Shleifer and Vishny (2003) and Dong, Hirshleifer, Richardson, and Teoh (2006)). It is argued that managers of acquirers may use their firms relatively overvalued stock to benefit their own shareholders at the expense of target shareholders, with no combined economic gain to bidders and targets, but potentially huge losses to society as a whole. We show the findings, that acquirer shareholders and, particularly, large acquirer shareholders, do not gain from takeovers and may be worse off, is due to a fundamental revelation bias in the standard event study methodology used to identify synergistic gains and, to assess whether acquirer shareholders are victims of manager-shareholder agency problems. First, acquirer managers rewarded for short-term performance often instigate a signaling bias by deliberately timing the good news of the bid announcement to coincide with the release of bad news such as unmet earnings targets (see Bhagat, Dong, Hirshleifer, and Noah (2005)). This common practice, documented in the Wall Street Journal, 1998, p. C1, confounds the information about an acquisition s value to a bidder that is capitalized into the bidder s stock price on the acquisition announcement, and results in a downward revaluation of a bidder s stand-alone value. 4

5 Second, the bid itself typically releases bad news about an acquirer, even when there is no motive to distort the market s perception of company value. For example, it may reveal serious empire-building tendencies of incumbent management, or management perception that the bidder has run out of profitable internal growth opportunities (see Fuller, Netter, and Stegemoller (2002), and Hietala, Kaplan, and Robinson (2003)). Shleifer and Vishny s (2003) market timing explanation for stock bids when the stock is relatively overvalued falls into the same category. Third, the acquirer may already have a long history of generally successful bids, such that a new acquisition announcement is no surprise to the market, which leads to a close to zero market reaction even to a beneficial acquisition. Serial acquirers such as General Electric, Cisco, GlaxoSmithKline and Capital One for example are possible examples in this activity. 6 Fourth, for reasons relating to conflicts between existing and new shareholders put forward by Myers and Majluf (1984), stock financed acquisitions represent new equity issues that by their very nature can represent bad news about the stock s true value since stock may be preferred to retained earnings and debt when it is overvalued. Finally, the initial bidder announcement return could be downward-biased because of the likelihood of failure, perhaps because of the subsequent entry of competing bidders. Bhagat, Dong, Hirshleifer, and Noah (2005) introduce the probability scaling method (PSM) that uses post-bid data to factor in the probability of a successful bid and the intervention method (IM) to take account of the likelihood of competing bids. They show that both of these methods increase the perceived value of the takeover and reduce the likelihood of finding that the bidder is overpaying for the target. Hence, release of information that is extraneous from the critical question of whether an acquisition will create bidder shareholder value poses a severe problem from the perspective of traditional M&A event study methodology. The PSM and IM methods tackle only limited dimensions of this complex problem. Our study builds on the well-known contribution by Shleifer and Vishny (2003) that put forward a conditionality explanation for stock acquisition offers as opposed to cash bids. Acquiring shareholders may gain and target shareholders correspondingly lose from the resulting swap of the firm s own equity for the target s relatively underpriced equity. Hence, the issuance of stock bids is endogenous in this arbitrage operation and is conditional on management discovering that their stock is overpriced relative to the targets. This mechanism 6 The point made here about expectations and share price reaction is quite distinct from the considerable empirical literature on whether serial acquirers are successful or not. 5

6 is a major contributor to the revelation bias that occurs on a bid announcement as it signals that the offer s acquisition currency, bidder stock, is relatively overvalued. Stock price falls on announcement of a stock financed bid, not because it is a bad bid, lacking synergies and likely to destroy shareholder wealth, but because this is a strong signal that the stock is currently relatively overvalued, much like the signal released by a seasoned equity offer announcement. 7 Here, decisions to issue equity by a stock offer are conditional on some value-destruction occurring within the firm itself that reveals itself to the market along with the offer to the target or, less plausibly, that target shares are undervalued. The most recent contribution to this literature is a study by Savor and Lu (2009) who restrict their analysis to acquirers only. They find that withdraw stock bids suffer a more severe value reduction than those that are not withdrawn. They interpret this finding to be due to the unsuccessful bidders inability to swap their overvalued equity for fairly valued target equity and conclude that, indeed, stock acquirers realize sizable benefits from successful bids. In our analysis, we include the market responses of matched targets following bid failures to show that value reduction occurs because failed bidders are unable to realize the synergistic benefits associated with their bids, with both bidders and targets losing value as a consequence. Both Shleifer and Vishny (2003) and Savor and Lu (2009) assume a complete absence of synergistic gain, but since neither investigates the price changes in the associated targets, their argument remains only partially tested. Since firms with relatively overvalued equity are likely to have insufficient internal growth prospects, it makes sense for such firms to concentrate on external growth prospects achieved via synergistic gains. Our contribution is related to the fact that failed bids are subject to a much weaker revelation bias relative to new bids, and this is especially true for stock bids. The bidder stock price following the initial bid already reflects the endogenous valuation error of announcing a stock bid due to perceived relative overvaluation, managerial empire-building incentives, evidence of weak internal growth prospects, and all other revelation problems. By contrast, the induced change in a stock bidder s value following an offer withdrawal represents a much cleaner experiment than the market reaction to the initial bid 7 Seasoned equity issues by overvalued but mature firms are rare but nonetheless have statistically significant effects in terms of explaining the probability of an SEO (see DeAngelo, DeAngelo, and Stulz (2010)). This rarity could be explained in part by the likelihood that investors see through the ploy and thus underprice the offering. Conditionality problems also plague event study analysis of stock splits, as well as stock-based takeover announcements and seasoned equity issues. Stock splits generally occur only following stock price runups. Hence, an event study of stock splits will show considerable value-enhancement whereas in reality none has occurred (see Brown, Goetzmann, and Ross (1995)). 6

7 announcement. This is especially the case where a bid has failed for exogenous reasons unrelated to the existing value of the bidder or the target. Unlike Savor and Lu (2009), by matching our bids with their targets, we show that both bidder and target are subject to large losses in value on bid withdrawal due to exogenous failure, consistent with both being subject to loss of synergistic benefits. The more the target falls, the more the bidder falls, and vice versa, indicating that the bidder as well as the target suffers a loss of synergistic benefits. Thus our fundamental insight is that when failed bidders are matched with their associated failed targets and the reason for deal failure is exogenously triggered, then it is possible to impute the share of gains received by the bidder when the deal succeeds. We find that studying failed bids yields a great deal more insight than conventional approaches concerned largely with successful bids. With failed bids, one can observe the subsequent long-term history of both the bidder and target, whereas with successful bids one cannot generally observe either individual firm post-acquisition, as they are combined entities. Thus, failed bids are a huge and largely untapped mine of new information about the fundamental causes and consequences of mergers and acquisitions. In this study, we overcome the problems in the existing empirical literature by devising a new approach to assessing the economic benefits of the M&A offers by using successful and failed takeover deals in our sample of both bidders and targets. Moreover, as a means of increasing our analysis of matched bidders and targets when the bid fails, we compile a much larger database of mergers and acquisitions over four countries: Australia, Canada, the United Kingdom and the United States, compared with conventional studies that focus on a single country. 8 Using our new database, we take a novel approach to analyzing the gains and distribution of the synergistic benefits in mergers and acquisitions between bidder and target. Following Savor and Lu s (2009) approach of studying bidders in M&A offers that are cancelled for exogenous reasons, we extend this concept to analyze target firms and a matched sample, where requisite financial information about both an acquirer and a target of an offer is available for our offer sample. Utilizing this large sample, we incorporated new variables into our cross-sectional regression models, which include indicators of failed offers and exogenously failed offers and acquirer and target reactions to their counterparty s abnormal announcement returns. 8 We are not the first to consider acquisitions outside of the United States. For example, Netter, Stegemoller, and Wintoki (2011) examine a comprehensive set of mergers and acquisitions. 7

8 Our central line of reasoning is as follows. One would assume that a rational economic motivation for a bidder to enter into a merger and acquisition transaction would be the ability to invest in a positive net present value investment and thus, return excess abnormal returns to their long-term shareholders. This view is in line with the findings of Schlingemann (1994), Cooney and Kalay (1993), and Lang et al. (1991). Hence, we seek to investigate whether mergers and acquisitions are positive NPV investments for acquirers using either cash or stock financing and yield genuine economic gains and do not simply transfer wealth (theft!) from target shareholders in the case of stock bids. Our research question follows from the above reasoning, which examines whether the pursuit of synergistic economic benefits serves as the primary motive for acquiring firms participating in merger and acquisitions. We find support for our first hypothesis that (HI) on the announcement of a failure of a merger, the combined value of the acquirer and target fall. Having shown that synergistic benefits do exist for the combined firm, we then evaluate our second hypothesis that (HII) acquirers gain a significant share of the synergistic benefits created on announcement of a merger or acquisition. We conduct our analysis in two stages. First, using our entire sample of 2,963 acquirers and 4,606 targets, we conduct an investigation into the abnormal returns around: (i) the initial announcement of a merger or acquisition, and (ii) the announcement date that the merger and acquisition either succeeds or fails with certainty. Specifically we find a mean stock price run-up and announcement return for acquirers of two percent and 1 percent respectively, representing an approximate net gain to acquirers on announcement of one percent. This result is consistent with evidence from Ahern and Sosyura (2011) suggesting that stock acquirers release a string of positive announcements in the lead up to a merger or acquisition announcement, thereby boosting their share price. In their sample, merger talks begin on average 64.5 days prior to the public announcement. Consistent with the findings of Andrade, Stafford and Mitchell (2001), we find that acquirer and target abnormal returns over a fiveday window are 1.05 percent and percent, respectively. Hence, our sample replicates the standard finding using conventional methodology that acquisitions are destructive of acquirer value, thus earning (apparent) negative synergies. We find that the combined firm announcement returns appreciate on the initial offer announcement (2.67 percent) and fall on the announcement of an offer failure ( 1.82 percent). Thus, this evidence yields strong support for the existence of overall synergistic gains in our sample, consistent with Hypothesis I and the findings of many prior empirical studies. 8

9 The second stage of our analysis utilizes a series of cross-sectional regression models to investigate the creation, and subsequent loss of synergistic benefits. Using an innovative method of examining synergistic benefits, we split our analysis into three subsections. The first subsection analyses the creation of synergistic benefits using an event study of an initial announcement of a merger or acquisition. We improve on existing methodology by extending the event window to include a 60-day period preceding the initial announcement to capture any price run-up, potentially due to proprietary information leaks, and define the dependent variable to be equal to this run-up plus the five-day window around the offer announcement. We find that for long-horizon models of target abnormal returns, the market predicts offer failures and exogenously failed bids, where targets of these bids earn significantly lower returns leading up to and on the initial offer announcement. Additionally, we find a significant positive relationship between acquirer and target returns, hence finding leading support for Hypotheses I and II. How can we rule out reverse causation - that the deal failed because the target found a way to trigger deal failure after it correctly anticipates poor future acquirer performance? This might occur when the target obtains information about acquirer overvaluation from its due diligence investigation. We rule out this otherwise plausible story by also incorporating the target s reaction to the announcement of deal failure when examining bidder abnormal returns and vice versa when examining target abnormal returns around the same announcement. Were the reverse causality story true, both the bidder and target would decline in value together due to release of bad news about the bidder until the endogenous collapse of the bid. The bidder would then continue to fall in value as it could no longer receive the benefit of the relatively underpriced target, but the target s stock would shoot up in value as it is now free of the yoke of being tied to the overvalued bidder. We do not find evidence consistent with this scenario. In fact, we find quite the opposite. The values of both the target and bidder continue to fall pari passu together for the next 100 days, consistent with a mutual loss of synergistic benefits. As noted by Fuller, Netter and Stegemoller (2002) and Hietala, Kaplan and Robinson (2003), information is released on the M&A offer announcement. The takeover announcement reveals information about the stand-alone value of the bidder and target in addition to any potential synergies arising from the combination of the two firms, and distribution of these gains between target and acquirer. Due to this complication in measuring synergistic gains on the initial announcement of an acquisition, we study the loss of synergistic benefits on the announcement of a failure of an acquisition. We find evidence in 9

10 favor of negative abnormal returns to acquirers, targets, and the hypothetical combined firm on the announcement of failed bids. A third form of analysis involves investigating the relationship of the expected synergistic benefits found in a long-horizon study of acquisition wealth creation, and the subsequent losses of these synergistic benefits. Measuring abnormal returns commencing 60 days prior the initial announcement until 100 days post the date participants know whether the bid is successful or fails, we find strong evidence for the existence of synergistic benefits for acquirers, supporting Hypothesis II. Specifically, we find that failed cash bidders significantly underperform successful cash bidders by approximately 10.7 percent and failed stock bidders underperform successful stock bidders by a further 15.5 percent in bids that fail for exogenous reasons. Additionally, we find that failed target stock prices in stock bids fall back to their pre-bid levels, which is the reverse of the predictions made by Savor and Lu (2009). We also find that after 100 days following deal outcome successful stock bidders do only about half as well (relative return of -13.3%) as successful cash bidders. This substantially subtracts from the target s premium and synergistic share. According to the market timing hypothesis, acquirers purchasing targets use relatively overvalued equity as currency. One would expect that on the news of bid failure, long-term target shareholders would react positively, which is the exact opposite of our findings. We agree that bidders with overvalued stock use their shares as currency. Our point is that the market is not entirely fooled. The terms on which targets accept bidder stock reflects to some extent the bidder overvaluation and collective synergistic gains so that the target price can only fall on deal failure, not rise as proponents of no synergistic gains predict. Additionally, we find strong positive correlations between acquirer and target announcement return, suggesting that there is a strong complementarity and sharing of synergistic gains between acquirer and target shareholders. This complementarity is inconsistent with the Roll (1986) hubris hypothesis, which predicts a more positive bidder stock return (with a larger target price fall) following a bid failure announcement reflecting a reversal of a bidder s stock price discount for its expected overpayment for the target. It is also inconsistent with the Shleifer and Vishny (2003) and Savor and Lu (2009) theft explanation as the bidder price falls and target must rise in response to deal failure and the unwinding of the theft of the target s relatively undervalued equity, according to this explanation. Our study makes at least two valuable contributions to the literature. First, we propose a new approach to examining the synergistic benefits in merger and acquisitions. Our evidence suggests that looking at failed bids provides an additional opportunity to more clearly identify 10

11 the gains in mergers and acquisition by studying the unexpected loss of these expected gains. Second, we show the possibility of improvements to current methodology by including the acquirer price run-up as a dependent variable when studying acquirer and target returns in M&A. This price run-up is significantly negatively correlated with the unexpected losses incurred if the bid subsequently fails typically many months hence. Remarkably, it is not only correlated with the fall in the price of the bidder, but also with the target with the impact on the target only slightly smaller in magnitude. This is to be expected if there is sharing of the synergistic gains, with on-going private negotiations occurring between bidder and target during the run-up period. We structure the study as follows. Section 2 reviews existing literature and develop a contextual basis for our key hypotheses. Hypotheses are further constructed in Section 3. Section 4 describes our data sources, acquisition sample and empirical methods. Section 5 contains the empirical work and discusses the results, Section 6 presents our estimates as to how synergistic gains are shared and Section 7 concludes. 2. LITERATURE REVIEW Real synergy involves combining assets in the form of positive net present value (NPV) projects, creating positive excess (abnormal) returns to an acquirer s long-term shareholders. By contrast, Roll (1986) posits the hubris hypothesis that at least some managers over-bid for targets due to errors in valuing synergies, leading to no overall value improvement. These inefficient investments can also be due to agency problems such as empire-motivated managers not acting in shareholder interests. Jennings and Mazzeo (1991) conclude that bidders learn nothing from the share price reaction to a bid. Thus, a negative share price reaction does not make deal completion less likely. Luo (2005) disagrees with the earlier finding, utilizing a larger deal set and different methodology and set of tests. We find that the pairwise correlation coefficient between the cumulative abnormal returns (CARs) of the bidder over the five-day window is uncorrelated with bid outcome in agreement with Jennings and Mazzeo (1991). When we add in a full set of controls, the market is able to predict failure of the sample inclusive of endogenously failed bids but not exogenously failed bids. Failure significance disappears even for endogenous sample inclusive of the run-up period after maintaining the full set of controls. Hence, the full market reaction to the bid does not appear to predict bid failure, suggesting that the presumed inside information is not affected by the reaction of outside investors. This 11

12 evidence is consistent with our finding that the market reaction to the bid announcement does not say a great deal about likely synergistic gains accruing to the bidder. We do find, however, a significant negative correlation between the target announcement CAR and bid failure that disappears when a full set of controls is included. Once we take account of the target s price run-up and thus the information leakage, the market appears able to predict deal failure for targets, but not for bidders. This is not so surprising when one considers that the target event is much cleaner as it is not so contaminated by the revelation of bad news about the bidder. Schlingemann (1994) conducts a study of 623 cash takeovers during the period He concludes that retained-earning sourced cash transactions positively relate to bidder announcement abnormal returns. He attributes this finding to firms using internally generated cash to invest in positive NPV investments (Cooney and Kalay (1993)). 9 The results of the study are also consistent with Lang et al. (1991) who hypotheses that managers at times have incentives to waste excess free cash flows in wealth destroying investments. Martynova and Renneboog (2009) conclude that transactions financed by internally generated funds underperform those financed with debt. Lang et al. (1991) also identifies a relationship between Tobin s q and investment policy such that that high q firms, compared to low q firms, are more likely to have positive NPV projects. Hence, low q firms should on average pay out their excess cash, rather than invest in poor acquisition prospects. Grinblatt and Titman (2002) suggest, the stock returns of the bidder at the time of the announcement of the bid may tell us more about how the market is reassessing the bidder s business than it does about the value of the acquisition. 10 Similarly, Hietala, Kaplan and Robinson (2003) note that the announcement of takeovers reveals information about the stand-alone value of the bidder and target, any potential synergies arising from the combination, and the distribution of gains between the target and acquirer. Additionally, they state that it is often not possible to infer the respective synergies, overpayment, and distribution of gains simply from the change in the market price of the bidder and target. Our model hopes to fill this gap by including a failed sample where the bid has failed for exogenous reasons. Thus, an unexpected offer failure provides valuable additional evidence about the value of an offer captured in the forgone economic gains of the offer. 9 Cooney and Kalay (1993) adopt the innovative methodology of including negative NPV projects into the financing decision, shows that the new issue of equity to finance projects can signal an exceptional valuable project, where the market may react positively to the news. This is contrary to the prediction by Myers and Majluf (1984) where they believe that the market will never react positively to new issue of equity. 10 Grinblatt and Titman (2002, p. 708). 12

13 A negative share price reaction to offer withdrawal announcements for targets should indicate the size of the loss of synergistic benefits (Bradley, Desai, and Kim (1983) and Samuelson and Rosenthal (1986)). Bradley et al. (1983) find that after an offer withdrawal the target s share fall back over an extended time-period toward its previous pre-offer levels, as it becomes clear that no subsequent acquisition bid is likely. Comparatively, those targets that receive subsequent offers experience additional increases in share price. Bradley s (1980) study of 100 percent cash offers shows that bidders experience severe negative reactions to offer failure, compared to positive reactions for successful offers. Savor and Lu (2009) incorporate failed bid announcements into their sample. They interpret the larger reduction in share price for failed stock offers, compared with successful, as an inability to swap overvalued equity for more fairly-valued target equity. A more plausible explanation is the inability of failed bidders to attain anticipated synergistic benefits. Savor and Lu also point out that failed cash bids do not suffer such a loss in value and continue to perform relatively well compared to failed stock bids. We agree but our explanation is entirely different: stock bids by their nature, with high relative Tobin s q and high-priced equity indicative of bidder relative managerial advantage, achieve far higher synergistic gain. Hence, the loss on exogenous deal failure is also far higher. Their supposition of no synergistic gain for stock bids predicts the exact opposite of their finding: their failed stock bids should do much better than failed cash bids as the former suffer no loss of embodied synergistic gain by assumption. An important part of Savor and Lu s approach is their solution to the endogeneity problem that arises from the inclusion of failed bids. Bidders with overvalued stock have greater incentives to propose stock as the mode of payment rather than cash. This complicates matters as the bid may fail due to target shareholders discovering the acquirer s overvaluation. Savor and Lu solve this endogeneity issue by attempting to create an exogenous failed subsample where mergers fail for exogenous reasons outside of the control of the bidder or target, and thus are unrelated to the acquirer s or target s valuation. M&A researchers rarely use the technique of failed acquirers in their analysis. A natural extension of the Savor and Lu approach is to focus on target reactions to bid failure. If the market-timing theory is correct and no synergistic benefits exist, then rational target shareholders should expect that a bidder s stock to be relatively overvalued. If this happens systematically over time then in anticipation of future acquirer price falls, the price of the target may fall below the notional cash value of the equivalent acquirer stock even though the acquisition seems very likely to go ahead. Thus, upon the announcement of the 13

14 stock bid failure, the target share price could go up if the target s price has significantly fallen for this reason. Alternatively, it may drop slightly representing only a fraction of the notional value of the bid. In contrast, this study postulates that synergistic gains exist. Hence, upon the offer failure announcement, we expect to see a fall in the combined value of the acquirer and target equity capitalization due to a loss of jointly shared synergistic gains. More specifically, we do not expect to see a target s share price rise because shareholders are no longer to receive burdensome stock that may be rapidly falling in value. Instead, we expect the bidder stock price to fall in a complementary fashion to the fall in target price, due to a shared loss in synergistic benefit. Ahern (2011) finds that in terms of dollar gains, targets do not do a geat deal better than acquirers that tend to be much larger. Moreover, in vertical acquisitions a target s relative scarcity and product market dependence help explain its share of total merger gains. Cai, Song, and Walkling (2011) present evidence to show that some bids come as more of a surprise to the market than others. In particular, the first bid in an industry has both greater surprise and a higher return than subsequent bids. They conclude that, after accounting for anticipation, bidding activity generates wealth. Malmendier, Opp, and Saidi (2011) find that targets react much more favorably to cash bids than do stock bids on a deal failure announcement utilizing a relatively short 25-day window. They suppose that cash targets are initially undervalued so that by paying in cash rather that stock, the bidder captures the entire gain. We find no difference in the way stock and cash bids react negatively to deal failure with a much longer event window post the deal outcome news. It is true that we find stock financed targets return to their pre-bid levels within the 100-day window following a deal failure announcement and cash financed targets do not. Hence, we believe that failed targets of cash bids remain in play for longer, even though Malmendier, Opp, and Saidi (2011) find no evidence that they do subsequently become targets. Dimopoulos and Sacchetto (2011) structurally estimate preemptive bidding and target resistance based on an extension of Fishman s (1988) theoetical model. They show that only rarely is there a second bidder, meaning that most initial bids are preemptive. Their simulations imply that the initial bidder s valuation of the target is 97% of the pre-bid value and potential rivals far lower at only 58%. This means that prospective synergistic gains are likely to be the monopoly of just one bidder and the main barrier to the bidder extracting most of the gains is target resistance. Our findings are the first to justify the high initial bidder valuations that are otherwise unexplained by their modelling. Furthermore, in their framework, deals fail endogenously because synergsitic gains are insufficient to overcome 14

15 target resistance. Thus, endogenously failed deals yield fewer synergistic benefits than exogenously failed deals, which is precisely what we find with our sample of endogenously and exogenously failed deals. 3. Hypothesis Development and Construction 3.a Hypothesis Development As outlined above, despite a comprehensive body of research documenting the performance of firms around acquisition announcements and the motives behind these takeovers, there are serious limitations with these methodologies due to revelation and other biases. We will analyze the short and long-term impacts of M&A offer failures. From this analysis, we seek to fill some important gaps in our understanding of the motives for mergers and acquisitions and in the process, to improve on the interpretation of M&A announcement effects. We first assess whether the pursuit of synergistic benefits is a plausible motive for bidders to make takeover offers. Hypothesis I: In our analysis, we include failed and exogenously failed takeover bids. In doing so, we overcome the revelation bias issue posed by Hietala, Kaplan and Robinson (2003), where the announcement of a takeover reveals information about the stand-alone value of the bidder and target and their businesses, in addition to any potential synergies arising from the combination of the two firms. Based on the existing literature and empirical evidence, we propose the following hypothesis: HI: On the announcement of the failure of a merger and acquisition offer, the combined value of the acquirer and target will fall. We foresee that this result will prove robust to a variety of samples, methodological approaches, and estimation techniques, revealing that on an M&A offer failure there is a decline in the combined acquirer and target equity capitalization, and hence a loss in synergistic benefits. 15

16 Hypothesis II: We hypothesis that: HII: Acquirer shareholders gain a significant share of the synergistic benefits created on the announcement of a merger and acquisition and, similarly, lose a significant amount on failure of the bid. To assess the validity of this proposition, we employ both event study analysis and cross sectional regression models to analyze the returns around initial bid announcements and failure announcements. We seek to assess whether bidders gain a significant share of the expected synergistic benefits on announcement of an M&A offer. To test this proposition, we examine whether on the announcement of a bid failure, bidder stocks lose their expected portion of the synergistic benefits created at the bid announcement. 4. DATA AND METHODOLOGY 4.a Sample Construction This study focuses on initial merger and acquisition offers and possible later withdrawals announcements. We study merger and acquisition offers for public targets from four major Anglo developed economies: Australia, Canada, the United Kingdom and the United States, mainly to increase sample size particularly for the failed offer sample. Our study is the first to combine data on matched acquirers and targets from a number of Anglo countries with quite similar competitive tender offer rules. The core data used in this study represents merger and acquisition characteristics, which comes from the Securities Data Corporation s (SDC) Platinum Global and US Mergers and Acquisitions database. We examine initial bids announced between January 1, 1985 and December 31, 2009 and obtain the following information on from the SDC Platinum database: (i) the identities of the parties involved in the transaction, (ii) whether the deal was consummated, (iii) the deal s mode of payment, (iv) any toehold the acquirer or target held in each other prior to the offer, (v) the initial offer announcement date, (vi) the announcement date of offer consummation or withdrawal, (vii) offer characteristics, (viii) the industry and nation of the acquiring and target firms. 16

17 Where possible stock returns, firm size and accounting data, are obtained from the CRSP/Compustat merged database. Due to the limited coverage in the CRSP/Compustat merged database of international stocks, we used Datastream to collect stock returns, firm size and accounting data. Since Datastream has limited financial statement coverage, we also relied on Aspect Huntley, ORBIS and SDC Platinum to obtain the missing accounting information. The M&A sample criteria are as follows. (i) The target is a public listed firm that is incorporated in Australia, Canada, the United Kingdom or the United States. (ii) The acquirer is a publicly listed firm. (iii) The deal can be clearly classified as successfully completed or a failure. (iv) The bidder seeks to acquire more than 50 percent of the target firm shares in order to gain control and holds less than 50 percent of its shares beforehand. (v) The deal value must be greater than one million dollars. 11 (vi) The method of payment used in the bid must be solely cash or stock. (vii) The firm s stock is actively traded and its stock price and market value must be readily available from either CRSP or Datastream. (viii) The firm s annual financial statement information must be available from CRSP/Compustat Merged Database, Datastream, Aspect Huntley s FinAnalysis, ORBIS or SDC Platinum. (ix) Deal value must represent five percent or more of a bidder s equity capitalization to insure that the deal has a material impact on a bidder s stock price. The final sample includes 3,147 acquirers and 4,793 targets firms. Table 1 reports the reduction in the sample as we impose additional sample criteria. We start with raw data collected from SDC Platinum and in the end obtain our final sample shown in Table 1, Panel A. We finally require the bidder and target firms to have stock price and accounting data available, which yields our final dataset described in Table1, Panel B. <<Insert Table1 about here>> 11 Deal value is defined in US$ as the consideration paid by the acquirer for the target, excluding fees and expenses. 17

18 The four-country selection allows us to analyze the motives for cross-border acquisitions and to analyze merger and acquisition activity outside the US, as well as to substantially expand our sample size. Along with the usual sample criteria, we required that bidders seek more than 50 percent of target shares since at this level of ownership an acquirer will typically have full control over the target. We also find that setting an absolute cut-off on the method of payment of either 100 percent cash or 100 percent stock, does not significantly reduce the sample size. 12 The final sample consists of 2,963 acquirers, 4,606 targets and 1,941 deals where data is available for both acquirer and target. Figure 1 describes the time-series distribution of the sample and the US dollar amount of stock bids relative to cash bids in each year. The figure first shows that that we have more targets than acquirers with the required data and even fewer deals where we have data for both bidder and target. Second, we identify the frequencies of stock and cash financed deals in the sample. Stock based deals become more popular than cash financed deals in the 1990s, which corresponds to the 1990s merger and acquisition boom. The merger boom of the second half of the 1990s is noteworthy for the largely stock financed transactions. Table 2 presents the sample distribution of completed and failed bids, by country and by stock and cash financed deals. <<Insert Figure 1 and Table 2 about here>> According to Shleifer and Vishny s (2003) market-timing hypothesis, overvalued firms have a relatively greater incentive to make stock acquisitions. Hence, an acquiring firm may use mispricing of their stock to swap their overvalued equity for relatively less overvalued target equity. This is very similar to the adverse selection in the Myers and Majluf (1986) model of stock offers. Thus, the acquirer is using its overvalued stock to purchase a target s assets at a discount. This hypothesis corresponds to the evidence presented in Figure 1, where we see that the stock market and merger booms of the 1990s corresponds to an increasing use of stock over cash as an acquisition currency. An announcement of an equity-finance merger and acquisition may signal that the acquiring firm is mispriced and overvalued. Accordingly, one expects on this negative signal of overvaluation that the acquirer s stock should fall in value. Hence, if synergies exist from combining two firms, then on the offer announcement they will co-mingle with the negative signal. This makes it difficult to isolate the synergy 12 A table of bids lost by imposing the 100 percent cash or stock requirement is available on request. 18

19 effect when studying the distribution of returns on the announcements of M&A deals. Thus, we construct a sample of failed bids caused by exogenous factors, where the bids fail for reasons extraneous to the valuation or decisions of the acquirer or target. 4.b Exogenous Failed Bids: Sample Construction We follow a method similar to Savor and Lu (2009) to construct a sample of exogenously induced merger and acquisition bid failures, shown in Table 3. As postulated in the markettiming theory, there is a positive relationship between a firm s overvaluation and the probability of a stock-based acquisition when bids fail due to target shareholder recognition that a bidder s equity is overvalued. Empirically, on the initial announcements of stock-based deals, bidder returns significantly underperform cash-based deals. <<Insert Table 3 about here>> To account properly for loss of expected synergies, we must first be sure that expected synergies are associated with the acquisition announcement. We know that a variety of important pieces of information are released with the initial bid announcement. The announcement reveals information about the stand-alone values of the bidder and target, the expected synergies arising from the combination, as well as the distribution of the expected synergies between target and bidder shareholders. Thus, we must exclude bids where the offer is doomed to fail from the beginning, or failed because of any revelations concerning bidder or target valuation. Thus, we construct an exogenously caused offer failure sample, which reduces our failed bid sample by more than 50 percent Our methodology differs slightly from Savor and Lu (2009) because we include target firms and a matched sample (where we have both acquirer and target firms in our sample). Therefore, we determine what is exogenous in context of the three different sample categories. The goal of the exogenous sample is to keep only deals that fail unexpectedly and for reasons that are outside of the two firms control, which in a semi-strong form efficient market means that new information was not previously priced by the market. Ideally, we want to study bids that trigger market capitalization of a bid s expected synergies on its initial announcement date, which is subsequently lost on the bid s announced failure. It is critical that the withdrawal news does not co-mingle with signals concerning revision in the bidder s 19