Finances

Merger Theory, Stock Returns and Deal Drivers – The Impact of International Bank M&As on Targets, Bidders and Peers May 30th 2008 Abstract Our paper empirically analyzes the capital market perception of deal drivers of bank M&As. Using event study methodology we investigate abnormal returns of targets, bidders, and their five most likely transaction peers upon takeover and closing or termination announcements. Comparing acquisition probability, pre-emptive merger, collusive merger, as well as economies of scale and scope hypothesis we derive which theory explains the observed share price reactions best. Based on a sample of 691 bank M&As in North America and Europe in the period from 1995 to 2008 we find surprising results: Although our descriptive statistics and corresponding significance levels are in-line with previous literature, the collusive merger hypothesis offers the highest explanatory power for international bank M&As. This is a new empirical result since the collusion theory has been rejected by researchers so far. However, due to the obvious challenge to realize economies of scale within the banking sector, collusion might be a good opportunity for banks to achieve safe merger gains. Especially in the context of the ongoing scientific discussion and mixed empirical results about the existence of scale economies in the banking industry our collusion story seems plausible. Moreover, our results offer potential for future research investigating the empirical fit of the collusion theory via new methodologies. Keywords: M&A, Banks, Peer Returns, Event Study, Imperfect Competition JEL-Classification: G34, G14, G20, L13 1. Introduction The analysis of past M&A transactions and the respective market reactions to those transactions suggests that rationales behind mergers and acquisitions are often ambiguous. In theory, the economically most desirable motives are synergies which improve profits and enhance firm value. Upon announcement of a transaction, these synergies are most of the time mentioned as the primary underlying reasoning driving the deal. Taking this into consideration, short-term market reactions should reflect an appreciation of these beneficial synergies by raising stock prices of the involved corporations. However, empirical results show otherwise: in many transactions value is only created for the transaction target whereas the bidders’ and the combined entities’ values are destroyed, at least when measured by the short-term share price reaction of the entities. This divergence in assumption and empirical result leaves the question open whether or not it is the market that does not appreciate the benefits resulting for an M&A transaction, whether the market does not see the benefits or whether or not there have been no benefits in the first place. To help explain this divergence, past research on Mergers and Acquisitions has brought forward different theories which are used to explain the occurrence of and rational behind Mergers and Acquisitions as well as the reaction of transaction parties and markets to them. Four of these theories will be analyzed in the paper at hand: the Acquisition Probability Hypothesis, the Collusive Merger Theory, the Pre-emptive Merger Theory and a fourth theory we call the Economies of Scale and Scope Hypothesis. In direct comparison, all four theories investigate the reasons for and consequences of Mergers and Acquisitions on different levels, yet they share a common implication: each theory tries to provide explanation on why a company might or might not engage in a merger or acquisition. The Acquisition Probability Hypothesis gives an explanation of why corporate transactions tend to bundle and explains the dynamics behind it. The theory behind Collusive Mergers analyzes why market participants benefit from possible collusive behavior resulting from a merger and hence states why mergers should take place. The Pre-emptive Merger Theory tries to show why M&A transactions are conducted even though the participating bidder is aware that the transaction destroys corporate value. The implication here is that a company should consider a merger or acquisition in order to preempt a competitor from buying or merging with a potential target company, regardless of any synergies or economic benefits. The background of the Economies of Scale and Scope Hypothesis is that cost synergies realized through Mergers and Acquisitions attract more customers through lower product prices. As Switching Costs exist, customers can be locked in over longer periods of time and be imposed with higher prices through Switching Costs which results in higher profits. -1- The implications of these four theories have been tested in various studies over the past years. Most of the tests were conducted by looking either at the transaction parties’ stock returns upon announcement and/or cancellation of the deal or by analyzing the stock reaction of the transaction parties’ peers. However, no study has directly compared the theories and tested them against each other to find out whether or not there are implications that one of the theories might be better able to explain M&A transactions than another. As the theories give differing reasons for M&A transactions, we regard it as highly necessary to find out which of the implications prevails in explaining the reasoning behind corporate transactions. Thus, the objective of this paper is to investigate these four theories and provide an explanation on which theory can best explain the occurrence of Mergers and Acquisitions. In order to test the theories against each other, we use a method which allows us to make them directly comparable. We apply the event study method to analyze the stock returns of the two parties involved in an M&A transaction and the stock returns of the two parties’ peers, thus taking into account the reaction of all companies directly swayed by the deal. We thereby look at two events: the announcement of the transaction and its closing or cancellation. We investigate whether or not the parties and their peers have either positive or negative abnormal returns at the two events. The four theories suggest different patterns of abnormal returns for the involved companies and peers at the events; we will thus be able to determine which theory is mostly able to describe the reasoning behind M&A transactions by analyzing which theory’s pattern of abnormal returns has the highest match with the empirically determined abnormal returns. We derive the expectations concerning the stock return patterns by making assumptions on how the respective rationales behind the theories influence the transaction parties and their peers: would a deal closure – following the rational of the Collusive Merger Theory e.g. – have positive or negative implications for the parties and their peers' Our approach is unique because we analyze the different rationales behind the four theories, apply these rationales to a rectified stock return pattern in order to make the theories comparable and finally measure how many transactions follow a given pattern. We will develop these theories and the resulting hypotheses in subsequent chapters of the paper at hand in more detail. The hypotheses will be tested empirically using a data set of 691 M&A transactions in the period from 1995 to 2008, consisting of 648 completed transactions and 43 cancelled transactions. The data set consist of M&A deals in North America and Europe. Both regions will be analyzed separately, i.e. no intercontinental M&A transactions are considered. The transaction sample we analyze is limited to M&A transactions among -2- banks.1 Our paper is structured as follows: Section 2 reviews relevant empirical and theoretical M&A literature. Section 3 explains the essential M&A theories and derives our model. In section 4 we outline our data set and provide descriptive statistics. Section 5 highlights our research methodology and test statistics. Our empirical results including robustness checks and a paper discussion are presented in section 6, while section 7 finally concludes our paper. 2. Literature Review Empirical M&A Literature Empirical research on background, conduct, and outcome of M&A transactions emerged in the late 1970’s and early 1980’s. Seminal research using event study methodology includes the work of Dodd and Ruback (1977), Dodd (1980), and Asquith (1983), who analyze abnormal stock returns of targets and bidders on takeover announcement and deal closing or cancellation. Bradley et al. (1983) focus on abnormal returns of targets and bidders of unsuccessful tender offers. Similar studies are Davidson et al. (1989) and Croci (2006) who investigate stock returns of firms involved in cancelled M&As. All authors conclude that a takeover bid results in positive abnormal returns for targets and slightly negative abnormal returns for bidders. Moreover, although a deal cancellation is bad news for targets in the short run, they are able to retain higher share prices in the long run (Bradley et al., 1983). Further research focuses on cancelled M&A transactions, investigating determinants and consequences of deal terminations. Holl et al. (1997) find that the positive abnormal returns of targets are driven by the relation between bidder and target.2 In more detail the relationship between stock ownership, board composition, and stock returns following failed takeover bids are analyzed by Davidson III et al. (2002). Safiedinne and Titman (1999) as well as Jandik and Makhija (2005) focus on financing decisions of target firms after unsuccessful M&A transactions by investigating their change in leverage. Long-term effects of takeover bids are analyzed by Hviid and Prendergast (1993) as well as Dassiou and Holl (1996). Both studies show that a failed M&A can increase profitability of the target but decrease profitability of the bidder. Two other highly covered research areas are determinants and success factors of M&A transactions. Epstein (2005) identifies several success factors for takeovers, while Cole et al. (2005) investigate valuation effects of bidders. Research with highest relevance for our paper are studies focused on bank M&A. Houston Companies are differentiated by SIC Codes; for further details see Part 4. The relatedness regards industrial relatedness, i.e. whether or not the merger is horizontal or vertical Abnormal target returns of both horizontal and vertical transactions are positive, whereas the latter transactions yield higher returns than the former. 2 1 -3- and Ryngaert (1994) analyze merger gains of target and bidder banks and identify deal characteristics which capital markets perceive as value enhancing. Although they only find slightly positive and statistically insignificant gains, they are able to identify value increasing deal characteristics, such as bidder profitability or merger synergies. Pilloff and Santomero (1997) provide a detailed literature overview on different types of economic merger gains. More recent papers on bank M&A include Beitel et al. (2004) and Lorenz and Schiereck (2007). Beitel et al. (2004) analyze drivers of abnormal target and bidder returns in European bank M&As and identify 13 variables explaining excess returns. Lorenz and Schiereck (2007) test abnormal target and bidder returns in cancelled bank M&As supporting the findings of Dodd and Ruback (1977), Dodd (1980), and Asquith (1983): Failed bidders experience negative value impacts, while targets profit from a sustainably positive revaluation. Merger Theory Literature The theory of collusive mergers arose from a strand of literature analyzing returns of rival firms of acquisition targets at takeover announcement. A series of analyses, e.g. Eckbo (1983 and 1985) or Mitchell and Mulherin (1996), finds that rivals show positive abnormal returns upon merger announcements. These results have been interpreted as follows: Since M&As eliminate competitors and thus reduce the number of market participants, collusion among the remaining firms is facilitated. Hence, positive abnormal rival returns are the consequence of anticompetitive takeover effects. The collusion theory was first tested by Eckbo (1983). Analyzing different horizontal mergers on their effects on competition, he rejects the theory as no anticompetitive effects can be observed. Stillman (1983), Eckbo and Wier (1987), as well as James and Wier (1987) support these results. Especially relevant for our paper are the findings of James and Wier (1987) who focus on bank M&As. Even though the collusion theory has been rejected by the papers above, we feel the need to reexamine the hypothesis by using different methodology. All mentioned papers try to measure the anticompetitive effects of M&As. Our analysis, however, disregards the actual existence of collusion and solely relies on capital market perception of reduced competition through M&As. As a result of the rejection of the collusion theory Song and Walkling (2000) introduced the acquisition probability hypothesis. They conclude that positive abnormal returns of target rivals are driven by an increased takeover probability. Consistently, the acquisition probability hypothesis states that any unexpected M&A transaction signals potential of further takeovers and thus triggers subsequent M&A activities. Carrying this logic to the extremes an increased takeover probability can actually result in a merger wave. By testing a sample of unregulated industry acquisition targets and their rivals Song and Walkling (2000) confirm their hypothesis. Another paper supporting their findings is Otchere and Ip (2006). Although the acquisition probability hypothesis has been confirmed by several papers we regard it as necessary -4- to test the hypothesis in direct comparison to the collusive merger theory. The pre-emptive merger theory addresses the question whether a firm should participate in M&As to prevent its rivals from taking over a desired target and realizing merger synergies which could be harmful to its competitive position. Deneckere and Davidson (1985), Kwoka (1989), Ziss (2001), and Brito (2003) analyze this issue and find mixed results. Brito concludes that firms might engage in M&As to protect their competitive position even if the takeover does not promise any direct benefits. Hence, although the takeover itself is disadvantageous it is still the better alternative than not acquiring the target since this would economically be even worse. Fridolfsson and Stennek (2005) draw a direct link between M&A related profit reductions and share price increases. Consistently, this adverse capital market reaction can force firms into pre-emptive M&As. So in contrast to other hypotheses the pre-emptive merger theory is able to economically explain value diminishing takeovers. Since the financial services industry is characterized by a high level of product standardization the existence of switching costs plays an important role for competition. Switching costs are defined as the costs customers bear when changing a supplier (Porter, 1980). Basically, three issues of switching costs have been subject to extensive research: Jones et al. (2002) as well as Sharma and Patterson (2000) analyze determinants of switching costs. De Ruyter et al. (1997) and Caruana (2003) investigate implications of switching costs for consumer behavior. Klemperer (1987 and 1988) as well as Farrell and Shapiro (1988) focus on anticompetitive effects of switching costs and their impact on welfare. However, most relevant for our paper is the research by Fornell (1992) as well as Gruen und Fergusson (1994), who find that switching costs are positively correlated with customer loyalty. Moreover, Gremler and Brown (1996) show that this relationship is even stronger for service suppliers like banks. Another important finding for our model stems from Klemperer (1987, 1995), who concludes that switching costs create lock-in effects and thus result in imperfect competition. 3. Motives for Bank M&As We develop a model predicting certain and individual abnormal equity returns of bidders, targets, and peers for different M&A motives. Given public information, rational investors, and efficient capital markets, different transaction motives must results in different share price reactions because they imply divergent economic effects. Thus, our empirical results will show which M&A motives are predominant in the financial services industry. Going forward we will first provide an overview of our theoretical model by explaining different M&A motives before we finally wrap-up our hypotheses in table 1. Our model stresses the four hypotheses of “Acquisition Probability Hypothesis”, “Pre-emptive Merger Theory”, -5- “Collusive Merger Hypothesis”, and “Economies of Scale and Scope Hypothesis” as possible transaction motives which are detailed out in the following paragraphs. 3.1 Acquisition Probability Hypothesis The first theory we address is the acquisition probability hypothesis. Explaining M&A transactions by firms following a market trend, this hypothesis provides the theoretical background for merger waves. Since the theory assumes a lack of economic rationale for M&As, firms pursue takeovers for the simple reason of good market timing. The hypothesis therefore suggests that the transaction is not in the best economic interest of the firm und thus destroys value for the combined entity. However, perusing a transaction without any economic benefits indicates that management decisions are driven by either outside pressure or personal utility maximization. Besides simple market or peer pressure such managerial M&A motives include e.g. herding3, hubris4, and empire building5. Hence, in our paper we subsume all transactions driven by these effects under the acquisition probability hypothesis, since their effects yield the same economic results for the transaction parties. Looking at share price reactions according to the acquisition probability hypothesis we expect the following abnormal returns: At M&A announcement date (event #1) abnormal target returns should be positive because of the offered takeover premium. Bidder returns, on the other hand, should be negative due to market’s perception that the transaction is definitely not in the best interest of the firm. Consistently, for the transaction peers, which are defined as the five most probable targets and the five most likely bidders of the respective takeover, we expect exactly the same share price reactions at M&A announcement. Peer targets should show positive abnormal returns, while we anticipate negative abnormal returns for peer bidders as takeover probability increases. Thus, according to our acquisition probability hypothesis any actual takeover boosts M&A probability disregarding transaction specific facts and economic fundamentals. Our second event is determined by the announcement date of the outcome of the respective transaction. If the takeover is successfully completed we are talking about a closed deal (event #2a), whereas a termination results in a failed deal (event #2b). Dependent on the actual transaction outcome we expect divergent share price reactions of target, bidder, and peers. For closed deals we anticipate the same sign of abnormal returns for all parties as Herding explains M&A transactions as a mass phenomenon such as merger waves. Here an initial takeover triggers follow-up transactions, because managers are not able to resist the market pressure of “falling behind” by neglecting a transaction. However, economic consequences are completely disregarded. 4 Hubris blames overconfidence as driver of M&A transactions. Here the firms’ management is overconfident in respect of identifying and quantifying future economic pay-offs resulting from M&As. This transaction motive is not about disregarding economic prospects but rather about managerial misjudgment of their future materialization. 5 Empire building relies on management’s aim to maximize its own utility by spending free cash-flows to expand power. Managers promote M&A transactions to increase their influence in terms of firm size and related compensation structures. Consequently, the predominant driver of takeovers is management’s forwardness. 3 -6- before. So, targets should exhibit positive and bidders negative abnormal returns at closing announcement. Consequently, peer targets should reveal positive abnormal returns, while peer bidders should show negative abnormal returns, as takeover probability in the market increases. If the deal, however, is terminated we would anticipate exactly the opposite share price reactions for the transaction parties. In this case we expect targets to exhibit negative abnormal returns, whereas bidders should reveal positive abnormal returns. Reasoning thereby suggests that even though the takeover premium is lost, at least a value-deteriorating M&A transaction was abandoned. Hence, peer targets and bidders should show the same share price reactions as their transaction counterparts, since the overall probability of takeovers decreases. 3.2 Pre-emptive Merger Hypothesis Second, we want to highlight the theory of pre-emptive mergers as a possible transaction motive. A pre-emptive merger is characterized by the fact that the bidder intends to acquire a specific target to prevent it from being taken over by one of its competitors. Consistently, preemptive mergers are not driven by the idea of value creation but rather considered to limit possible losses due to competitive disadvantages. In other words, a pre-emptive merger compared to the respective target being taken over by a main competitor clearly is the lesser of two evils. Hence, it is still better to sacrifice some money to secure your own competitiveness than to suffer substantial and sustainable future losses. However, this implies that preemptive mergers as such are value-diminishing transactions.6 Within the framework of our pre-emptive merger hypothesis we anticipate the following abnormal returns: At M&A announcement (event #1) target shares should show positive abnormal returns due to the offered takeover premium. However, abnormal bidder returns should be negative since the transaction is motivated by the intention to reduce potential future losses and thus provides a negative outlook. Peer targets should be characterized by positive abnormal returns because after the most desirable target is acquired they might be in focus of forthcoming transactions themselves. Consistently, we expect peer bidders to show negative abnormal returns since their preferred target has been taken over by a direct competitor and hence promising synergies are forgone. Given the deal’s closing (event #2a) we predict exactly the same share price reactions as for the M&A announcement date. Thus, targets should show positive and bidders negative ab- Assume the Spanish Banco Santander Central Hispano SA was interested in acquiring Commerzbank AG to enter the German retail banking market. Given this scenario, Deutsche Bank AG might want to takeover Commerzbank itself to protect its domestic market position. However, in a competitive auction, Deutsche Bank would have to pay a rather high price for Commerzbank shares. Even though the premium might considerably exceed potential merger synergies and hence destroy value, it is eventually less harmful than the negative impact of increased competition. So, Deutsche Bank’s acquisition of Commerzbank is the better of two poor alternatives. 6 -7- normal returns. While target returns are driven by the final realization of takeover premiums, bidder returns are the consequence of deteriorating competitive perspectives. Peer targets again are expected to exhibit positive abnormal returns as their takeover probability increases. Moreover, peer bidders should reveal negative abnormal returns because they miss out a valuable opportunity. However, in the case of deal cancellation (event #2b) anticipated outcome and underlying storyline are twisted. Now, we expect target as well as bidder to consistently show negative abnormal returns. The reasoning therefore is that target shareholders loose offer premiums. Bidders, on the other hand, miss their opportunity of a pre-emptive merger. Thus, the thread of a direct competitor acquiring the respective target emerges, which is the worst case scenario. Consequently, peer targets should show negative abnormal returns as their chance of becoming an actual future takeover target fades. At the same time, peer bidders are anticipated to exhibit positive abnormal returns since the chance of acquiring their preferred target increases due to the failed pre-emptive merger. 3.3 Collusive Merger Hypothesis As third M&A motive we introduce the theory of collusive mergers. This hypothesis originally goes back to competition theory arguing that the probability of collusive behavior increases as the number of market participants decreases. A rather pragmatic reasoning therefore is that the fewer parties sit at the bargaining table the easier it is to reach the agreement to collude. Based on this logic, striving for market power is a desirable transaction motive since every takeover reduces the number of players and hence narrows competition. So, within this framework the predominant intention of a bidder is to acquire one of his direct competitors and thereby enforce collusion. Thus, added value in this scenario is solely created by extracting consumer surplus whereas operating synergies play no mentionable role. However, collusion seems to be a primarily legal competition issue and thus could successfully be mitigated by effective market regulation. Especially when looking at the banking industry it might be argued that due to the relatively high regulation of the financial services industry this hypothesis does not seem too realistic. Nevertheless, we include the collusive merger hypothesis in our analyses because it cannot be neglected ex-ante. Talking about share price reactions we expect the following abnormal returns: At M&A announcement (event #1) target and bidder should consistently show positive abnormal returns. While target shareholders profit from the offered takeover premium, bidders benefit from their increased market power after the acquisition. Furthermore, also peer targets and peer bidders are anticipated to reveal positive abnormal returns, since the chance of collusion in the respective market is facilitated. So, within the framework of our collusion hypothesis all market participants gain from M&As because a lower number of players will decrease -8- competition and finally boost future profits. If the deal is successfully closed (event #2a) we predict exactly the same abnormal returns as for the transaction announcement date. Again target and bidder as well as peer targets, and peer bidders should consistently exhibit positive abnormal returns. For target shareholders takeover premiums are secured, while for bidders and all other market participants the chance of a profitable collusion strategy emerges. Thus, ultimately all players benefit from decreasing competition and the opportunity to extract a bigger share of consumer surplus in future. On the other hand, we expect a withdrawn deal (event #2b) to consequently result in the opposite outcome. In this case, target, bidder as well as peer targets and peer bidders should persistently show negative abnormal returns. Here target shareholders loose offered takeover premiums. Moreover, bidders miss the opportunity to increase their market power and hence to extract more consumer surplus. Due to the failed deal the number of players stays constant and thus there is no chance of facilitating collusion within the market. 3.4 Economies of Scale and Scope Hypothesis The fourth and final theory we look at is the economies of scale and scope hypothesis. It explains M&A transactions motivated by the intention to realize merger synergies boosting future cash flows and thus enhancing firm value. Such synergies include operating as well as financial synergies either due to an increased firm size (scale) – or as a result of firm specific combination advantages (scope). So this hypothesis summarizes e.g. revenue increases through cross- and/or up-selling, cost reductions due to efficiency gains as well as benefits from new opportunities of financial engineering, tax savings or cash slack. However, our paper focuses on cost synergies since, according to the relevant literature, this is the predominant form of synergies in bank M&As.7 Nevertheless, the following considerations for the expected share price reactions to takeover and termination announcements qualitatively also hold for any other synergy type. We assume the global financial services industry to be characterized by more or less homogeneous goods, the existence of switching costs and thus, the persistence of imperfect competition. Hence, we introduce a combination auf Bertrand’s price competition and Klemperer’s switching cost model to illustrate the impact of cost synergies on banks’ future cash flows.8 Within our model we interpret competition in the financial services sector basically as a two period game. In period one banks set prices and compete for clients while in period two due to the existence of switching costs they are able to exploit their customer base by extracting consumer rent without losing a substantial number of clients. Given this scenario, any M&A transaction resulting in cost synergies is highly beneficial for participating banks 7 8 See e.g. Cornett, Tehranian (1992). See Klemperer (1987a, 1987b, 1995). -9- since lower costs allow for lower prices and thus a larger market share in period one. Consequently, a bigger customer base boosts profits and cash flows in period two and hence, increases firm value right at the takeover announcement. In terms of share price reactions our synergy hypothesis predicts the following abnormal returns: Target and bidder should consistently be characterized by positive abnormal returns at M&A announcement (event #1). On the one hand target shareholders are offered a takeover premium, while on the other hand bidder shareholders anticipate positive merger synergies boosting future cash flows. By contrast peer targets and bidders are expected to show negative abnormal returns, since due to the synergies of the merging banks their own competitive position is deteriorating. So, while any M&A transaction motivated by the realization of synergies is positive for participating banks it has a negative impact on the future operating and financial performance of their competitors. Given the deal’s closing (event #2a) we expect positive abnormal returns for target and bidder, whereas peer targets’ and bidders’ abnormal returns should be negative. The reasoning is exactly as above, while target shareholders realize the takeover premium, bidder shareholders will benefit from future synergies. Consistently, these M&A synergies will enable merging banks to reduce costs, increase market share and thus, boost future cash flows, causing harm to all competitors. However, should the attempted merger fail (event #2b), we expect the cancellation announcement to result in negative abnormal returns for target and bidder. In this case target shareholders lose the offered takeover premium and bidder shareholders forgo value enhancing merger synergies. Consequently, peer targets as well as bidders should show positive abnormal returns upon deal termination. Since the threat of a deteriorating competitive position will not materialize their market share and earnings perspectives are secured. Table 1 summarizes the anticipated abnormal returns of target, bidder and peers to takeover and termination announcements according to our four M&A hypotheses: - 10 - Table 1: Expected Abnormal Returns upon M&A Announcements Sign of expected abnormal return given respective type of event, party, and hypothesis. Takeover Announcement (event #1) Target Bidder Combined Target Bidder Entity Peers Peers Acquisition Probability Pre-emptive Mergers Collusive Mergers Economies of Scale and Scope + + + + Target + + + + + + + - + Bidder Peers Deal Closing (event #2a) Bidder Combined Target Entity Peers Acquisition Probability Pre-emptive Mergers Collusive Mergers Economies of Scale and Scope + + + + Target + + + + + + + - + Bidder Peers Deal Cancellation (event #2b) Bidder Combined Target Entity Peers Acquisition Probability Pre-emptive Mergers Collusive Mergers Economies of Scale and Scope - + - + - + + + + SOURCE: Own illustration. 4. Data Set Based on Thomson One Banker data our total sample contains 691 intra-industry M&A transactions of public banks in Europe and North America in the period from 1995 to 2008. We classify firms as banks by looking at their SIC codes and include companies with industry codes from 6000 to 6289 plus SIC 6712.9 These industry codes define banks as financial intermediaries in the sense of savings and loan banking, investment banking, and brokerage services. Thus, we derive a homogeneous group of banks suitable for our model and M&A theories. Insurance as well as real estate companies, also listed in the SIC 6000 section, are excluded as they might distort the comparability of our results, since inter-industry M&As may be characterized by different transaction motives and varying economic effects. 6021: National commercial banks; 6022: State commercial banks; 6029: Commercial banks; 6035: Savings institutions (federally chartered); 6036: Savings institutions (not federally chartered); 6099: Depository banks; 6111: Credit agencies; 6141: Personal credit institutions; 6159: Business credit institution; 6162: Mortgage banks; 6163: Loan brokers; 6211: Security brokers; 6282: Investment advisors; 6712: Bank holding companies. 9 - 11 - The countries of our data set include Austria, Belgium, Croatia, Czech Republic, Denmark, Finland, France, Germany, Greece, Republic of Ireland, Italy, Luxembourg, The Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and UK for Europe and Canada as well as the USA for North America. Moreover, we exclude all intercontinental M&A transactions where one transaction party is incorporated in Europe and the other in North America. This geographical segmentation is useful for two reasons: First, it ensures the quality of our peer selection, which would be distorted, if we included North American peers for European banks and vice versa. Second, this procedure allows us to use the North American deals as control sample for a robustness check of the results of our European transactions Our final data set consists of a total of 691 bank M&As, of which 648 transactions are closed and 43 are cancelled respectively. Out of these 691 transactions 87.1 percent are conducted by the ultimate parent.10 Altogether 94.4 percent of the deals are defined as “small”, with a total deal volume of less than 10 billion US-Dollars. Furthermore, the average profitability of targets and bidders is approximated by their cost/income ratio, while their size is measured by total assets, market capitalization, and enterprise value. The average cost/income ratio is 73.5 percent for targets and 71.9 percent for bidders. Average total assets yield 56 billion US-Dollars for bidders and 49 billion US-Dollars for targets, with an average market capitalization of 6 billion US-Dollars for bidders and 2 billion US-Dollars for targets. The results for average enterprise value are 19 billion US-Dollars for bidders and 17 billion US-Dollars for targets.11 5. Research Methodology We apply event study methodology to investigate the abnormal returns of targets, bidders, and their five most likely transaction peers at the takeover announcement and deal closing or cancellation date. To validate our results we conduct an event study using the index model12, the constant mean return model, and the market model13. The estimation period for the constant mean return and the market model is fixed to 250 trading days in the time period from 300 to -50 days prior to the takeover announcement. Altogether we analyze three different events: For all our sample deals we identify the takeover announcement date as event #1. Whereas for closed deals the date effective is defined as event #2a, for cancelled deals the We control for whether or not the transactions are conducted by the ultimate parent or a subsidiary. This is not only crucial for our regional differentiation of the North American and European market but also for our peer selection. A selection of peers for a regional subsidiary would distort the view that the transaction is actually conducted by the ultimate parent. Exceptions, in which a specific subsidiary conducts a domestic transaction – in which the peers should be selected for the subsidiary and not for the ultimate parents – are not accounted for. 11 For further descriptive statistics please refer to Appendix (A), Table 2. 12 The Dow Jones STOXX 600 Banks and the DJ STOXX Americas 600 Banks is used as benchmark index. 13 10 with a risk-free rate of 4.5 percent and a market risk premium of 5.5 percent. - 12 - withdrawal date equals event #2b. As event windows we investigate the symmetric [-10;+10], [-3;+3], and [-1;+1] windows for every of these three events to provide a further robustness check of our findings. We calculate the cumulative abnormal returns (CARs) for the respecttive event windows. To test for significant abnormal returns (ARs) we finally apply standard mean and median tests by using the t-Test and the Wilcoxon-Signed-Rank-Test respectively. As we conduct our event study not only for the actual transaction parties but also for their five most likely peers, we introduce a set of four key variables to assure a sound peer selection. This procedure is of crucial importance, since we claim the selected five peer targets and five peer bidders to be the ten most likely firms which would have taken part in the respective deal instead of the actual transaction parties. Consequently, we determine the transaction peers by the following four variables in order to maximize this likelihood: SIC code, market capitalization, sales region, and profitability. First, the bidder’s or target’s four digit primary SIC code must exactly match the primary SIC of the peers. This criterion is implemented to account for operating differences between banks. Second, the peer’s market capitalization must be within a range of 25 percent of the transactions party’s market capitalization for acquirers and 50 percent range for targets. These values are chosen to reflect differences in size between bidders and targets. Third, the sales region is determined by the country, in which the bidders’ and targets’ headquarters are located; the peers are expected to be located in the same country. The country in which the respective firm is incorporated is used as a proxy for the geographic focus of its business activities. Thus, this selection variable helps assure that actual transaction parties and their peers basically have the same sales region. Fourth, the profitability proxy is based on empirical evidence: prior studies found that targets tend to be the least profitable companies within their peers group, whereas typical bidders are the most profitable among their peers.14 Hence, we select target peers by choosing the least profitable companies matching the other criteria and the bidder peers by selecting the most profitable peer companies. A list of all stock listed banks in the US, Canada and Europe is filtered with these variables in order to find the transaction parties’ peers. 6. Empirical Results The initial research question the paper at hand set out to answer is: which of the analyzed theories, i.e., Acquisition Probability Hypothesis, Collusive Merger Hypothesis, Pre-Emptive Merger Theory and Economies of Scale and Scope Hypothesis, explains best the abnormal stock returns measured upon announcement, cancellation and closing of an M&A transaction' The following part will introduce the results of the event study tests and explain which See Hannan, Pilloff (2006); Hernando et al. (2007); Altunbas, Marqués (2008); Pasiouras et al. (2007); Lanine, Vennet (2007). 14 - 13 - pattern fits best the assumed theory. Going forward, the second part will discuss the methodology, results and possible limitations of our analysis. 6.1 Target, Bidder and Peer Returns in cancelled Bank M&As Our results yield that of the investigated 691 transactions, 66 transactions (9,6 percent of the total transactions) followed the pattern of the Collusive Merger Hypothesis, 15 transactions (2,2 percent) followed the Acquisition Probability Hypothesis pattern, 14 transactions (2 percent) matched the Pre-Emptive Merger Theory pattern and 13 transactions (1,9 percent) matched the Economies of Scale and Scope Hypothesis pattern. 15,6 percent of all measured transactions thus follow our presumed patterns. We assume that a pattern, in order to match one of our theoretic abnormal return expectations, must at least have 7 of the possible 8 abnormal return results in the investigated event windows. We make this assumption due to random economic reasons which might influence either a transaction party or its peers in singular special circumstances. As the scope of the paper cannot include these exceptional cases, we account for them by accepting patterns with one missing match.15 Figure 1 plots the frequencies of all patterns in our sample. Altogether there are 512 different theoretical patterns that assemble as follows: Four different stock prices (acquirer, acquirer peers, target and target peers) can either move up or down in two different states in time (announcement and closing or withdrawal). This results in 28 = 256 different patterns. As the end of transaction is either the date of closing or date of withdrawal we have to multiply these 265 patterns by 2 and finaly derive 2x28 = 512 different stock price patterns upon M&A announcements. To every pattern we assign a unique numerical code. This code is generated by a binary 8 digit number, where each digit takes the value of zero if the stock moves down and a value of one if it moves up. The digits are: 1 = acquirer peers at deal announcement, 2 = acquirer at deal announcement, 3 = acquirer peers at the end of the transaction, 4 = acquirer at the end of the transaction, 5 = target peers at deal announcement, 6 = target at deal announcement, 7 = target peers at the end of the transaction, and 8 = target at the end of the transaction. We do this for closed deals as well as for withdrawn deals and sort the two code samples together in a numeric order. Therefore, the leftmost value in figure 1 shows transactions that have only negative movements in all stock prices, whereas the rightmost value shows transactions with only positive movements. If we assume a purely random distribution of joint abnormal returns each single pattern would have a probability of 0,195 percent to occur and the sample of stock return patterns should be equally distributed. However, our results suggests otherwise: the fact that four specific 15 A „total match“, i.e. 8 out of 8 abnormal returns, is tested 25 times for the Collusive Merger Hypothesis (3,50%), 3 times for the Economies of Scale and Scope Hypothesis (0,42%), and 2 times each for the Acquisition Probability Hypothesis and the Pre-Emptive Merger Theory (0,28%). - 14 - patterns occur with a summarized frequency of over 15 percent indicates that these patterns do not occur randomly. Hence the market – in our case manifested by the transactions parties and their peers – seem to clearly react to the transactions. As one pattern occurs with considerably higher frequency than others, the capital market reactions strongly suggest the existence of a dominating theory underlying this pattern in M&A transactions. These results are even more outstanding since no other pattern, bar two, occurs as often as the investigated patterns based on the four theories. We can thus clearly state that the market has certain perceptions of an M&A deal, reacts accordingly, and that the perception mostly followed is represented by the Collusive Merger Hypothesis. Analyzing the CARs itself, we find that for the +/- 3 day event window upon announcement the transaction targets have an average positive CAR of +4,9 percent and the bidders a slightly negative CAR of -0,22 percent. Upon closing, we find slightly negative average CARs for the targets (-0,12 percent) and slightly positive CARs for the bidder (+0,17 percent). Similar results are delivered by the cancellation event window: the targets show slightly negative average CARs of -0,059 percent, the bidders again slightly positive average CARs of +0,53 percent. Looking at the peers, we find that upon announcement, bidder peers have an average CAR of +0,09 percent, targets’ peers of +0,19 percent. Upon deal closing, bidders’ peers have an average positive abnormal CAR of +0,22 percent, the targets’ peers an average CAR of -0,1 percent. Finally, upon cancellation, average bidder peers’ CARs are slightly negative with -1,06 percent, average target peers’ CARs are slightly positive with +0,15 percent. As explained in section 5, we tested for significance using the t-Test and the WilcoxonSigned-Rank test. Looking at the Wilcoxon-Signed-Rank test z-value, we find highly positive results for target and bidder CARs upon announcement with z-values of 7,56 for targets and -3,2 for bidders. Closing and cancellation event CARs are insignificant for the transaction parties; for the peers we find the CARs to be insignificant upon announcement for both bidders and targets. However, upon closing, bidder peers’ CARs are significant even though target peers’ CARs are insignificant.16 As robustness check we subdivide our sample by geography to test whether our findings are driven by country effects. Therefore, we split our data set into the subsamples North America and Europe. The rational behind this geographical analysis is that the North American and European financial services industry are characterized by different banking systems, varying market consolidation, and divergent regulation. Thus, these differences might impact the results of our event study. However, if we compare the two subsamples we find that our results in terms of sign and significance levels of abnormal returns qualitatively hold for both re- 16 For more detailed results please refer to Appendix (B), Table 3. - 15 - gions.17 Thus, our findings suggest that capital market reactions to bank M&As are at least qualitatively the same in North America and Europe. These geographically consistent share price reactions of targets, bidders, and peers to takeover and closing or cancellation announcements indicate the fit of our model. Thus, our robustness check emphasizes the explanatory power of our research approach and validates our empirical results. Further robustness checks validate our sample additionally: testing for different event windows and using three models to estimate the expected returns and CARs, we find matching results for all event windows and estimation models. 6.2 Discussion Finally, we discuss our research approach to address potential limitations of our paper. Our analysis tests the explanatory power of four established M&A theories by investigating abnormal returns of targets, bidders, and their most likely transaction peers. Therefore, we compare acquisition probability, pre-emptive merger, collusive merger, as well as economies of scale and scope hypothesis. However, in reality there clearly exist more than these four deal drivers. Further M&A motives include, but are not limited to, corporate strategy such as expansion e.g. in terms of increase in market share or entry in new markets, geographical or industrial diversification, mis-valuation, and – especially relevant for banks – financial distress. Although it might be argued that our analysis is not comprehensive in this respect we provide a sound comparison of the four most established M&A theories. As a matter of fact, our results show a pattern which occurs with the second highest frequency (48 times) next to collusion. This pattern, with its specific CAR reactions of bidders, targets and peers, could hint at financial distress as an M&A motive. Investigating the actual M&A deals which resulted in the pattern, we find that the majority of those deals actually involved a target bank in financial distress. We can thus say that of all transactions motives which were not covered in the paper at hand, there is only one reason which actually occurred in our pattern results. To our knowledge, this is the first paper to jointly test these four theories and evaluate their actual explanatory power for bank M&As. Hence, we contribute in two ways to the existing literature and ongoing academic discussion we offer the first empirical comparison of all four hypotheses ever and introduce standardized event study methodology to jointly test the theories. At least for the collusive merger theory, this is a new research approach which also assures the overall comparability of our findings. Moreover, we derive surprising empirical results: Even though the descriptive statistics and corresponding significance test are in-line with previous M&A literature, the collusive merger hypothesis offers the highest explanatory power for international bank M&As. This second finding seems astonishing, since previous studies reject the fit of this hypothesis. However, 17 The statistical results for the two subsamples North America and Europe are not reported in the paper. - 16 - our results offer promising potential for future research opportunities investigating the explanatory power and empirical fit of the collusion theory via new methodology. 7. Conclusion In our paper we empirically analyze the capital market’s perception of drivers of bank M&As. Applying event study methodology we investigate abnormal returns of targets, bidders, and their five most likely transaction peers upon the three events takeover announcement (event #1) and deal closing (event #2a) or cancellation (event #2b) announcements. Comparing the four M&A theories of acquisition probability, pre-emptive merger, collusive merger, as well as economies of scale and scope hypothesis we derive which theory can explain best the observed share price reactions. Based on a sample of 691 bank M&As in North America and Europe in the period from 1995 to 2008 we find new and surprising results: The collusive merger hypothesis offers the most exact prediction for international bank M&As, opposed to the acquisition probability, preemptive merger, and economies of scale and scope hypothesis which fail to forecast the empirical return patterns. This is a new empirical result since the collusion theory has so far been rejected by past empirical research. However, our descriptive statistics and corresponding significance levels are completely in-line with previous M&A literature. Moreover, all our results are subject to three robustness checks: First, we geographically split our sample in the two subpanels North America and Europe to control for county fixed effects. Second, to validate the computed abnormal returns we apply the index, the constant mean return, and the market model for our event study. Third, we investigate the three symmetric event windows of [-10;+10], [-3;+3], and [-1;+1] days around the event to assure the stability of the observed share price reactions. As a result all our key findings qualitatively hold. Although our findings contradict previous empirical studies using different methodology, they are not implausible. As recent empirical evidence shows it is a real hard challenge to realize economies of scale within the banking industry. Thus, collusion might be a good opportunity for banks to achieve relatively safe merger gains. So, if a bank is able to reduce the number of market participants by conducting a takeover and thereby facilitating collusion, the aim of profitable growth finally seems possible. Finally, our collusion story also seems plausible in the context of the lasting scientific discussion about the existence of scale economies in the banking sector. Given our findings and the empirical context of banks in the area of conflict between growth and profitability, collusion might be a driving force for M&A transactions within the financial services industry. Since this question goes far beyond the scope of our paper this issue is left open for future research. - 17 - References Altunbas, Y., Marqués, D. (2008): Mergers and acquisitions and bank performance in Europe: The role of strategic similarities, in: Journal of Economics and Business, 60 (08), pp. 204-222. Asquith, P. (1983): Merger bids, uncertainty, and stockholder returns, in: Journal of Financial Economics, 11(1-4), pp. 51-83. Beitel, P.,Schiereck, D., Wahrenburg, M. (2004): Explaining M&A Success in European Banks, in: European Financial Management, 10(1), pp. 109-139. Bradley, M., Desai, A. Kim, E. (1983): The rationale behind interfirm tender offers: Information or synergy', in: Journal of Financial Economics, 11(1-4), pp. 183-206. Brito, D. (2003): Preemptive mergers under spatial competition, in: International Journal of Industrial Organization, 21(10), pp. 1601-1622. Cole, R. A., Fatemi, A., Vu, J. (2005): Do mergers create or destroy value' Evidence from unsuccessful mergers, in: MPRA Paper, pp. 1-24. Cornett, M. M., Tehranian, H. (1992): Changes in corporate performance associated with bank acquisitions, in: Journal of Financial Economics, 31 (92), pp. 211-234. Croci, E. (2006): Stock Price Performance of Target Firms in Unsuccessful Mergers and Acquisitions, Working Paper, July 2006. Dassiou, X, Holl, P. (1996): Merger failure and merger profitability: An alternative to the Hviid and Prendergast model, in: Applied Economics Letters, 3(4), pp. 271-273. Davidson, W. N.,,Dutia, D., Cheng, L. (1989): A re-examination of the market reaction to failed mergers, in: The Journal of Finance, 44(4), pp. 1077-1088. Davidson, W. N., Rosenstein, S., Sundaram, S. (2002): An empirical analysis of cancelled mergers, board composition and ownership structure, in: Applied Financial Economics, 12(7), pp. 485-491. De Ruyter, K., Wetzels, M. (1998): On the relationship between perceived service quality, service loyalty and switching costs, in: International Journal of Service Industry Management, 9(5), pp. 436-453. Deneckere, R., Davidson, C. (1985): Incentives to form coalitions with Bertrand competition, in: RAND Journal of Economics, 16 (4), pp. 473-486. Dodd, P. (1980): Merger proposals, management discretion and stockholder wealth, in: Journal of Financial Economics, 8(2), pp. 105-137. Dodd, P., Ruback, R. (1977): Tender offers and stockholder returns: An empirical analysis, in: Journal of Financial Economics, 5(3), pp. 351-373. Eckbo, B. E. (1983): Horizontal mergers, collusion, and stockholder wealth, in: Journal of Financial Economics, 11(1-4), pp. 241-273. Eckbo, B. E. (1985): Mergers and the market concentration doctrine: Evidence from the capital market, in: The Journal of Business, 58(3), pp. 325-349. - 18 - Eckbo, B. E., Wier, P. (1985): Antimerger policy under the Hart-Scott-Rodino Act: a reexamination of the market power hypothesis, in: Journal of Law & Economics, 28(1), pp. 119-149. Epstein, M. J. (2005): The determinants and evaluation of merger success, in: Business Horizons, 48 (05), pp. 37-46. Farrell, J., Shapiro, C. (1988): Dynamic competition with switching costs, in: RAND Journal of Economics, 19(1), pp. 123-137. Fornell, C. (1992): A national customer satisfaction barometer: the Swedish experience, in: Journal of Marketing, 56, pp. 6-21. Fridolfsson, S.-O. Stennek, J. (2005): Why mergers reduce profits and raise share prices-A theory of preemptive mergers, in: Journal of the European Economic Association, 3(5), pp. 1083-1104. Gremler, D. D.,Brown, S. W. (1996): Service loyalty: Its nature, importance, and implications, Working Paper 1996. Hannan, T. H., Pilloff, S. (2006): Acquisition Targets and Motives in the Banking Industry, in: Finance and Economics Discussion Series, Federal Reserve Board, Washington, D.C. Hernando, I., Nieto, M. J., Wall, L. (2007): Determinants of national and cross border bank acquisitions in the European Union, Working Paper 2007. Holl, P., Dassiou, X., Kyriazis, D. (1997): Testing for asymmetric information effects in failed mergers, in: International Journal of the Economics of Business, 4(2), pp. 155-172. Houston, J. F., Ryngaert, M. D. (1994): The overall gains from large bank mergers, in: Journal of Banking & Finance, 18(6), pp. 1155-1176. Hviid, M., Prendergast, C. (1993): Merger Failure and Merger Profitability, in: The Journal of Industrial Economics, 41(4), pp. 371-386. Jandik, T., Makhija, A. K. (2005): Debt, debt structure and corporate performance after unsuccessful takeovers: evidence from targets that remain independent, in: Journal of Corporate Finance, 11(5), pp. 882-914. Jones, M. A., Mothersbaugh, D. L., Beatty, S. E. (2002): Why customers stay: measuring the underlying dimensions of services switching costs and managing their differential strategic outcomes, in: Journal of Business Research, 55(6), pp. 441-450. Klemperer, P. (1987a): The competitiveness of markets with switching costs, in: RAND Journal of Economics, 18(1), pp. 138-150. Klemperer, P. (1987b): Markets with consumer switching costs, in: The Quarterly Journal of Economics, 102(2), pp. 375-394. Klemperer, P. (1995): Competition when consumers have switching costs: An overview with applications to industrial organization, macroeconomics, and international trade, in: Review of Economic Studies, 62(4), pp. 515-539. Kwoka, J. E. (1989): The private profitability of horizontal mergers with non-Cournot and Maverick behavior, in: International Journal of Industrial Organization, 7(3), pp. 403-411. - 19 - Lanine, G, Vennet, R. V. (2007): Microeconomic determinants of acquisitions of Eastern European banks by Western European banks, in: Economics of Transition, 15(2), pp. 285-308. Lorenz, J.-T., Schiereck, D. (2007): Completed versus cancelled banking M&A transactions in Europe, Working Paper 2007. Mitchell, M. L. and J. H. Mulherin (1996). "The impact of industry shocks on takeover and restructuring activity." Journal of Financial Economics 41(2): 193-229. Otchere, I., Ip, E. (2006): Intra-industry effects of completed and cancelled cross border acquisitions in Australia: A test of the acquisition probability hypothesis, in: Pacific-Basin Finance Journal, 14(2), pp. 209-230. Pasiouras, F., Tanna, S., Gaganis, C. (2007): What drives acquisitions in the EU banking industry' The role of bank regulation and supervision framework, bank specific and market specific factors, in: Coventry University Working Paper Series, 3/2007. Pilloff, S. J., Santomero, A. M. (1996): The value effects of bank mergers and acquisitions, Working Paper 1996. Porter, M. E. (1980): Competitive Strategy, New York: Free Press 1980. Safieddine, A., Titman, S. (1999): Leverage and corporate performance: Evidence from unsuccessful takeovers, in: The Journal of Finance, 54(2), pp. 547-580. Sharma, N. Patterson, P. G. (2000): Switching costs, alternative attractiveness and experience as moderators of relationship commitment in professional, consumer services, in: International Journal of Service Industry Management, 11(5), pp. 470-490. Song, M. H., Walkling, R. A. (2000): Abnormal returns to rivals of acquisition targets: A test of the acquisition probability hypothesis, in: Journal of Financial Economics, 55(2), pp. 143-171. Stillman, R. (1983): Examining antitrust policy towards horizontal mergers, in: Journal of Financial Economics, 11(1-4), pp. 225-240. Ziss, S. (2001): Horizontal mergers and delegation, in: International Journal of Industrial Organization, 19(3-4), pp. 471-492. - 20 - Appendix (A) Descriptive Statistics Table 2: Descriptive Statistics This table shows selected descriptive statistics of our main sample. We list market value, enterprise value, the cost to income ratio as well as the total assets for both transaction companies, i.e., the acquirer and the target. The bottom of table 1 shows some propotions of how many transactions fall inside certain categories. All applicable values are reported in billion USD. All Acquirer Market Value mean median s.d. N 6.736,0 277,9 27.132,6 1291 19.884,3 126,7 74.928,8 1203 71,9% 67,4% 47,5% 155 Target 2.332,5 67,6 10.988,0 1297 17.061,3 336,4 47.065,2 409 73,5% 76,2% 60,5% 68 49.595,7 2.934,5 128.263,9 432 North America Acquirer Target 3.623,8 219,1 19.373,8 980 5.761,6 92,1 35.489,0 915 64,2% 57,3% 25,1% 51 22.017,3 1.133,6 118.043,0 951 961,4 50,0 4.725,8 980 3.459,3 90,5 14.319,8 261 71,9% 66,5% 26,4% 12 14.142,0 902,4 45.527,5 277 Europe Acquirer Target 16.542,8 2.323,8 41.844,8 311 64.753,4 8.011,2 129.794,1 288 75,7% 71,5% 55,0% 104 6.571,0 422,8 20.053,2 317 41.048,6 11.225,3 69.840,0 148 73,9% 76,7% 65,8% 56 Enterprise Value mean median s.d. N Cost-to-Income mean median s.d. N Total Assets mean 56.047,6 median 1.544,3 s.d. 184.845,1 N 1252 163.565,6 112.954,8 30.751,6 44.858,1 288.244,4 189.807,8 301 155 13,2% 84,0% 81,8% 51,9% 318 Deals Withdrawn Transaction100 mio. Transaction Entity Peer Wilc. signed rank test H0: median=0 All Date Transaction Entity Announcement Closing withdrawal Announcement Closing withdrawal Announcement Closing withdrawal Announcement Closing withdrawal n 691 648 43 691 648 43 215 196 19 215 196 19 Acquirer z median -0,652 -0,086 -1,224 0,035 0,204 -0,092 -2,140 0,516 2,089 0,204 0,323 0,455 -3,20 *** -0,40 0,42 0,67 1,77 * -1,16 -5,94 *** 2,14 ** 1,01 0,71 0,77 -0,32 Target z median 1,178 -0,175 -0,940 0,034 0,030 0,163 3,365 -0,132 1,618 0,327 -0,191 0,163 7,56 *** -0,63 -0,36 0,85 0,11 -0,28 7,31 *** 0,29 0,32 1,67 * -0,59 0,64 Peer Tvalue>100 mio. Transaction Entity Peer The asterix *, ** and *** mark the significance on the 10, 5 and 1 percent level - 22 - (C) Frequency Distribution of observed Share Price Patterns Figure 1: Frequency of stock price movement patterns This figure shows the frequencies of all possible stock price reaction. Each dot represents one pattern of target, bidder, and transacton peers to takeover, closing and termination announcements respectively. Altogether there are 512 different patterns that assemble as follows: Four different stock prices (acquirer, acquirer peers, target and target peers) can either move up or down in two different states in time (announcement and closing or with8 drawal). This results in 2 = 256 different patterns. As the end of transaction is either the date of closing or date of 8 withdrawal we have to multiply these 265 patterns by 2 and finaly derive 2x2 = 512 different stock price patterns upon M&A announcements. To every pattern we assign a unique numerical code. This code is generated by a binary 8 digit number, where each digit takes the value of zero if the stock moves down and a value of one if it moves up. The digits are: 1 = acquirer peers at deal announcement, 2 = acquirer at deal announcement, 3 = acquirer peers at the end of the transaction, 4 = acquirer at the end of the transaction, 5 = target peers at deal announcement, 6 = target at deal announcement, 7 = target peers at the end of the transaction, and 8 = target at the end of the transaction. We do this for closed deals as well as for withdrawn deals and sort the two code samples together in a numeric order. Therefore, the leftmost value in this figure shows transactions that have only negative movements in all stock prices, whereas the rightmost value shows transactions with only positive movements. 30 25 20 15 10 5 0 - 23 - (D) Frequency Distribution of M&A theories Figure 2: Frequency of M&A theories associated to stock price movement patterns This figure shows the corresponding frequencies of M&A theories associated to the empirical stock price movement patterns reported in figure 1. 30 25 20 1 5 1 0 5 0 Financial Distress APH Collusion Synergy Pre-emptive - 24 -