Is merger & acquisition activity value creating or destructive?

Transcription

1 Is merger & acquisition activity value creating or destructive? An empirical study of acquiring-firm returns during the sixth merger wave Master thesis Tilburg School of Economics and Management Student name: Administrationnumber: Supervisor: Date: Robbert van Wingerden dr. D.A. Hollanders

2 Abstract This report looks at the value generated or destroyed to acquiring public firm shareholders around mergers & acquisitions in the U.S. for the periods (merger wave) and the period (not a merger wave). Abnormal return is estimated by using the market model, where abnormal return is defined as the difference between the forecasted return and realized return. Cumulative abnormal return is then used to find the total loss or profit from merger activity. This is done for both periods, as in which subsequently regressions between cumulative abnormal return and deal & firm characteristics are estimated to find significant relations. The results suggest that merger announcements are significant positive for acquiring (public) firms during a merger wave ( ), which is not the case when there is no merger wave present ( ). Acquiring firms perform worse when they undertake merger and acquisition activity at the end of a merger wave ( ). The question why acquiring firms perform in the way they do around merger announcements cannot be answered entirely in this study. It is probably due to overvaluation of firms and the economic situation present. The economic situation has a direct influence on stock prices and therefore public firms. If an acquiring firm targets a public firm in an economic situation that is about to change (end of the merger wave) this probably has consequences for the results of the acquirer. During the end of the last merger wave there was overvaluation present, which probably caused firms to perform worse when engaging in merger and acquisition activity. The results show also that paying with cash, a higher net debt level and high transaction value in relation to the total assets of a firm has a positive effect on the performance of an acquiring firm around merger announcements, where cash/total assets, same industry, equity in payment and private target does not show a significant relation with cumulative abnormal returns. For the period , the ratio of net debt/assets(market) only shows a significant negative relation in relation to cumulative abnormal returns. No other notable features are present for that period( ), which suggest when one is not in a merger wave the investigated deal and firm characteristics cannot explain cumulative abnormal returns. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 2

3 Table of contents I. Introduction 4. II. Current state of literature 5. III. Main research question 12. IV. Methodology & data collection 13. V. Descriptive statistics 17. VI. Deal and firm characteristics 24. VII. Conclusion 32. VIII. Bibliography 33. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 3

4 I. Introduction A decade ago, America Online merged with Time Warner in a deal valued at a stunning $350 billion. It was then, and is now, the largest merger in American business history. The Internet, it was believed, was soon to vaporize mainstream media business models on the spot. America Online s frothy stock price made it worth twice as much as Time Warner s with less than half the cash flow. When the deal was announced on Jan. 10, 2000, Stephen M. Case, a co-founder of AOL, said, This is a historic moment in which new media has truly come of age. His counterpart at Time Warner, the philosopher chief executive Gerald M. Levin, who was fond of quoting the Bible and Camus, said the Internet had begun to create unprecedented and instantaneous access to every form of media and to unleash immense possibilities for economic growth, human understanding and creative expression. The trail of despair in subsequent years included countless job losses, the decimation of retirement accounts, investigations by the Securities and Exchange Commissionand the Justice Department, and countless executive upheavals. Today, the combined values of the companies, which have been separated, is about oneseventh of their worth on the day of the merger. To call the transaction the worst in history, as it is now taught in business schools, does not begin to tell the story of how some of the brightest minds in technology and media collaborated to produce a deal now regarded by many as a colossal mistake. Box 1- source Ney York Times, January In the box above one of the possible outcomes of a merger is shown. Mergers & acquisitions (from now on M&A) is an interesting topic for researchers, governments and the general public. The topic M&A raises a lot of questions, whether mergers are profitable for the firms involved is one of them. This study will begin with a theoretical framework where the relevant theories behind M&A activity and important literature will be discussed, which can be found in section 2. It contains the main findings from previous studies done on M&A, which will be compared with each other. The conclusions and findings from section 2 will be the foundation on which the main research question is based, which will be given in section 3. Section 4 will show how the data is collected and more details will be given in how the sample is constructed and in which way it will be used. Also the methodology will be discussed here. Section 5 will contain the descriptive statistics and significance levels, and section 6 the deal & firm characteristics and the regression models estimated used to answer the main research question and sub questions. Event study methodology will be used just as in the study of Moeller et al. (2005), meaning in this case that the abnormal return around M&A activity will be analyzed for the acquiring firm, to see whether it is positive or negative and significant. The regression models estimated will give the opportunity to relate the short term results of deals with deal and firm characteristics. The regression models estimates possible effects and consequences of those deals. This will be done for the last merger wave ( ), the years and the period , and will be compared with the study of Moeller et al. (2005) and each other. Also endogeneity and its role will be discussed in this section. When the results are found and analyzed, conclusions and recommendations for future research will be given in section 7. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 4

5 II. Literature review Mergers & Acquisitions is a topic that attracts lots of attention, in both economic research and in the public news. The consequences of M&A activity for the involved parties are extensively studied in the economic literature present today. In the first part of this section the possible motives to undertake M&A activity are discussed. In the second part the possible deal and firm characteristics that affect wealth creation will be discussed. Thirdly, the effect on acquiring firms will be discussed, to conclude with the merger trends. Motives for M&A activity The modern finance theory (Manne, 1965), states that shareholder wealth maximization can be reached by firms that invest when the sum of the present values of the present cash flows exceeds the initial investment. Translating this to M&A, a firm has to undertake a merger if there is added value from the bidders perspective. In other words, the added value exceeds the transaction costs and the acquisition premium. The added value consists of operating synergies for both the bidder and the target (Berkovitch & Narayan, 1993). Bruner (2004) states that: true synergies create value for shareholders by harvesting benefits from mergers that they would be unable to gain on their own. So the added value consists of the synergy benefits they would be unable to gain on their own. The operating synergies can be divided in cost reduction trough economies of scale (Porter, 1985), revenue enhancements and a possible source for new tangible, intangible and human resources and capabilities (Simmonds, 1990). Another reason for M&A are financial synergies. Financial synergies lower the cost of capital, thus they create value even when operations of merged firms do not have any profit from it. This is in contrast with the findings of Miller & Modigliani (1958), who state that in an efficient market without taxes, informational asymmetries, and default costs the market value of a company does not depend on its capital structure, thus a financial synergy cannot be found. The capital structure of a firm does matter however, when the assumptions of Miller & Modigliani (1958) are not entirely correct. For instance, Lewellen (1971) explains that a merger or acquisition between two or more firms with not perfectly correlated cash flows reduce the joint probability of financial distress and thereby increases the level of debt capacity for the combined firm. This means that coinsurance can be a possible motivation to engage in a merger. In line with this, Seth (1990) found that synergy gains from coinsurance are likely to be lower for target and bidders facing the same market or demand because of a high correlation between their earnings stream. More empirical evidence that coinsurance can be a possible motivation to engage in a merger, is given by Higgins & Schall (1975) and Kim & McConnell (1976). Master thesis Robbert van Wingerden Mergers & Acquisitions Page 5

6 Besides synergies, the hubris of a certain manager can be a reason why firms engage in M&A activity. The hubris hypothesis formulated by Roll (1986) claims that managers systematically make errors in evaluating possible M&A opportunities due to excessive self-confidence, which leads to too much optimism. This leads to higher valuations for the target companies, compared to the true market value. Rational managers would not have made the valuations that are too high. Managerial motives can be important motives to engage in M&A activity, as they try to maximize their own utility and perhaps practice empire building (Trautwein,1990 and Zalewski, 2001) instead of maximizing shareholder wealth. Managerial hubris can be seen as an agency problem due to the separation of control (managers) and ownership (shareholders), which have both different interests (Alchian & Demsetz, 1972 and Jensen & Merckling, 1976). These three motives are the main reasons why firms engage in M&A activity. Where operational and financial synergy motives influence the abnormal return of both the bidder and target positively, the hubris hypothesis has a positive effect on the target, but negative on the bidder. Therefore, it seems likely that targets have positive abnormal returns around M&A announcements. The return for bidders however can be whether positive, given the operational and financial synergy motives to engage in M&A activity, or negative, given the hubris hypothesis. In the past numerous studies looked at the consequences of M&A for acquiring firms. First of all it is interesting to see why firms engage in M&A activity. As discussed earlier, this is because of financial or operational synergies and the hubris hypothesis. Trautwein (1990) states there are seven theories present that explain the motivation why firms engage in M&A activity. These seven theories are given in the following picture: Picture 1: theories of merger motives, source: Trautwein (1990) Master thesis Robbert van Wingerden Mergers & Acquisitions Page 6

7 Shortly explained, the efficiency theory takes financial, operational or managerial synergies as main motives to engage in M&A activity. Monopoly theory can be explained as M&A activity that is planned and executed to gather more market power. The raider theory can be interpreted as wealth transfers to the shareholders of the acquiring firm. The valuation theory argues that firms engage in M&A activity because the acquiring firm has better information of the value of a certain target than the market. The empire building theory argues that mergers are done by managers who try to maximize their own utility instead the value for the shareholders of the firm. The process theory states that M&A activity comes forth out of the strategic decision process. The disturbance theory states that because of economic disturbances the non-owners of assets place a higher value on the asset then their owners and vice versa, which results in a merger wave. The seven theories stated by Trautwein (1990) are more extensive than the theories given above, and give a better idea why firms engage in M&A activity. Possible deal and firm characteristics that affect wealth creation The question remains however if a merger does create wealth for the shareholders of the bidding firm. Datta, Pinches and Narayanan (1992) found the following factors that affect the wealth creation; Regulatory changes, number of bidders, bidders approach, mode of payment and type of acquisition. They found several factors that influence the returns of the bidding firm, given in the picture below. Picture 2: Influence of factors on wealth creation according to Datta, Pinches and Narayanan (1992), source: Datta, Pinches and Narayanan (1992) Master thesis Robbert van Wingerden Mergers & Acquisitions Page 7

8 The picture shown above gives information on how certain factors of a deal can influence the wealth creation for the acquiring firm. How the acquiring firm pays for the acquisition can differ, as it can pay with cash, stock or a combination of both. Jensen (1986) argues that acquisitions financed with cash (or debt) will have a larger positive effect on the results of the acquiring firm then when financed with stock. Besides that they also argue that firms with a large amount of cash, or a high cash flow, are probably more inclined to use cash as mode of payment. This implies that acquiring firm characteristics matter when a firm engages in M&A activity. Myers & Majluf (1984) argue that the acquiring firm holds superior information about the value of the targeted firm. When the targeted firm is overvalued, stock will be used as mode of payment. When the targeted firm is undervalued, cash or debt will be used as mode of payment. Moreover, Hansen (1987) suggests that when acquiring firms are uncertain about the value of the targeted firm, they are more inclined to use stock as mode of payment. This because target shareholders then share part of the risk, may the acquiring firm pay more than the appropriate value. Looking at empirical work done, Travlos (1987) shows that there are significant differences between cash and stock payment when acquiring a firm. Acquiring firms that use stock offers, experience a significant negative effect of -1.47%. Acquiring firms that use cash as mode of payment however, experience an insignificant positive effect of 0.24%. Therefore it seems likely that cash as mode of payment has a positive effect for acquiring firms. Moeller, Schlingemann and Stulz (2004) show that small firms perform significantly better than large firms when they make a M&A announcement. The abnormal return for small firms is larger than the abnormal return for large firms. Large firms experience more shareholder wealth losses when they engage in M&A activity. This holds when they control for both firm and deal characteristics. Large firms are more likely to use stock as mode of payment, which could explain this. This means that certain deal and firm characteristics do matter when an acquiring firm is engaging in M&A activity. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 8

9 Effect on acquiring firms Whether the effect is positive or negative in total when controlling for firm and deal characteristics for acquiring firms, remains a question as several studies are done about the effect on the short term results of acquiring firms. Mulherin and Boone (2010) for instance found a small insignificant negative return to shareholders of the acquiring firm of -0,37%. Franks et al (1991), Kuiper et al (2003) Jensen & Ruback (1983) and Andrade, Mitchell & Stafford (2001) find also supporting evidence that bidders have a negative return around M&A activity. There are however, also several studies that show small positive returns to acquirer shareholders. For example, Bradley et al (1988), who show evidence of a significant positive return of 0,97% in a US sample in the time period of 161 tender offers. Dod & Ruback (1977) and Bradley & Sundaram (2004) also found evidence that supports this suggestion. Moeller et al (2005) found that the returns for acquiring-firms are positive until 1997, but that the losses form 1998 trough 2001 wiped out all the gains made earlier. This implies that it matters in what stage of a merger wave a takeover occurs, as the merger wave they researched was from 1993 to They find that the losses are from relatively few acquisitions announcements that have shareholder wealth losses of $1 billion dollar or more. The firms experiencing a large loss deal, all have a high valuation but is in their study not sufficient enough to explain the large loss deals. Looking at the numerous studies done, not a decisive answer can be given of the effect of M&A activity is positive or negative for acquiring-firms. Below merger trends will be discussed, to get a better idea of what a merger wave really is. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 9

10 Merger trends Looking at the past it seems that M&A s occur in waves and that within these waves there is a clustering by industry. In the past several waves have occurred, which ended most of the times with a strong turmoil in the financial market (Martynova & Renneboog, 2008). The following picture shows the more recent merger waves. In total six great merger waves occurred in the US history. Graph 1 most recent merger waves up to 2010, source: Martynova & Renneboog, 2008 The last wave includes merger and acquisition announcements between 2003 and Numerous aspects of sixth merger wave include private equity, leveraged buyout and shareholder activism. Waves are driven by economic, regulatory and technological shocks like the dot.com bubble. Often they arise after a financial recession when there is a rapid expansion of credit, caused by growing capital markets and growing stock markets. When a merger wave develops itself further, managerial hubris also has it influence. The results of M&A in the last part of a merger wave often perform poorer than in the beginning. Martynova & Renneboog (2008) expect that the heterogeneity in the forces driving mergers may explain the varying patterns and profitability. The question arises if we are currently (2014) in the beginning of the next big merger wave, since the BRIC- countries are performing well and the economy as a whole shows signs of improvement. The period however cannot be classified as a merger wave (Dolfsma & McCarthy, 2012). The graph pictured below proves the suggestion of Dolfsma & McCarthy (2012) that the period cannot be classified as a merger wave. Deal value is used in the graph, instead of number of transactions. This is done because of the sample restrictions, because when number of transactions is used a different graph will be displayed due to the sample restrictions. So in the recent past one merger wave occurred, the period The other period, from 2009 to 2013, cannot be classified as a merger wave. These two most recent periods will be further investigated in the report. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 10

11 Total deal value p/y Total deal value M&A done by public US companies ($ millions) $ 800,000 $ 700,000 $ 600,000 $ 500,000 $ 400,000 $ 300,000 Deal value ($ millions) $ 200,000 $ 100,000 $ year Graph 2 Total deal value of deals done in the period To conclude, the wealth-effect of M&A activity on acquiring firms on the short term is not yet fully clear. Recent studies found both negative and positive results for acquiring firms, and the explanation when deal and firm characteristics where used did also differ. Moreover, also the stage of which a merger wave is in seems to be important. To get a decisive answer further research is needed, and all previously named factors have to be accounted for. So, besides the period , the period will be investigated also during this study, to check if results for acquiring firms can be explained differently when there is no merger wave present. Also the assumption of Moeller et al. (2005), that acquiring firms perform worse in the last stage of a merger wave, will be looked at. The deal and firm characteristics of the firms at the time of the M&A will also be investigated per period and in total for any notable features in relation to cumulative abnormal returns. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 11

12 III. Main research question From the literature review an important question arises for both researchers as firms: Are M&A creating value for acquiring firms? This is still an ongoing debate among academic researchers about the effects of M&A activity as shown above. The different suggestions presented above are the main reasons for this thesis. The goal is to look if returns for an acquiring-firm are value creating or destroying around M&A activity for the last merger wave. This will be compared with the period , which is not a merger wave. It is interesting to see whether the suggestion formulated by Moeller et al (2005), that M&A in the last stage of a merger wave firms perform worse, is justified for the period The period is as previously discussed not classified as a merger wave, but is interesting to investigate and to see what the characteristics of M&A s are in that period, as the M&A undertaken in that period are not part of a wave. In this way the results found for each period will be compared for similarities and differences. The following research question will be investigated in this thesis: - Are M&A value creating or destructive for acquiring-firm returns in the US market? To make sure we find a solid answer to this question the following sub questions are formulated: a) What are the average and cumulative abnormal returns for acquiring-firms for the different periods? b) Are there large aggregate dollar losses present and when to they occur? c) What is the statistical significance of the different (cumulative and average) abnormal returns? d) Can firm and deal characteristics explain the cumulative abnormal returns? The formulated sub questions are in the spirit of the study Moeller et al (2005) did, but now for different periods, so that their implication that firms that undertake M&A perform worse in the last stage of a merger wave, can be validated for the period Also the period will be investigated for any notable features compared to the period Furthermore, they will help answering the main research question. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 12

13 IV. Methodology & data collection Event study methodology is often used to measure the impact of a certain event on the value of a certain firm, a good example of such an event is a merger announcement, but also dividend or corporate earnings announcements are good examples. Besides that, macroeconomic news can also be such an event. The two major reasons to perform an event study are to see whether the market efficiently incorporates new information and to examine the impact of the event on the wealth of the firm s security holder. An event study can be executed at many different ways, but most of the time the same applies. First the event of interest and the timing of the event are identified, then a benchmark for normal stock return behavior is specified and next the abnormal returns are calculated an analyzed around the event date. When performing an event study on stock returns there can be an efficient or inefficient market present. When performing an event study, it is appropriate to make an assumption of market efficiency (McWilliams & Siegel, 1997). The efficient market hypothesis (EMH) will be discussed next. The efficient market hypothesis is defined as a market in which prices always fully reflect available and relevant information (Fama, 1998), a market where prices will only change if (new) information arrives (Fama, 1965) and a market where information cannot be predicted. Fama (1970) states that there are three versions of the EMH possible, where the difference is defined by the term all information available : the weak, semi-strong, and strong form of the hypothesis. The weak form of the EMH suggests that stock prices reflect all information from the past. This implies that stock prices are not dependent of each other, and follow a random walk. Firm specific events and systematic effects are not accounted for in this form of the EMH, and is therefore a weak test. The semi-strong form states that stock prices not only reflect past information, but also all other published information. Compared to the weak form, it does control for firm specific events. The disadvantage of the semistrong form is that it is possible that benchmarks are not perfect, because noise and disarranging effects can influence the results. This can be minimized though by using large datasets and relevant benchmarks (Bruner 2004). The strong form of EMH states that stock prices not only reflect public information but any (public of private) information that might be relevant. This means that no inefficiencies exist, so no abnormal return can be earned. Given these different forms of EMH, the semi-strong form is most likely to be used for event studies, as found in the literature. Despite the disagreement of the validity of EHM (La Porta et al, 1977 for example), the assumption is made that one can measure the effect of an unexpected merger announcement by analyzing the difference between the actual return around the event, and the expected return when the event had not taken place. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 13

14 The model used to determine the abnormal return is the market model. There are more statistical models available to determine the abnormal return, like the mean adjusted return model, the market adjusted model and a multi-factor model (Fama and French, 1993). These statistical models will not be discussed further, because they are not relevant for answering the main research question and therefore do not serve the goal of this paper. The market model is used in most of the previous studies to examine the effects of a M&A announcement and also contains some of the other methods, and is therefore used in this study also. Fama et al. (1969) argue that for any given stock i the following OLS (Ordinary Least Square) regression can be applied: (1) Where is the return of stock i, and is the return of the benchmark (the expected return on the stock) at time t. and are firm specific variables to be estimated by an OLS regression (1) using data found in the estimation period, which will be defined below. is the random zero-mean error term. The abnormal return for a given stock i is defined as the difference between the observed return and the expected predicted return(benchmark): (2) This equation shows that the forecasted error of the market model is defined as the abnormal return, since this coefficient represents the part of the (abnormal) stock return that results from the M&A announcement. To determine the total firm effect of stock movement for the complete event window (- 2,+2) during merger announcements, the cumulative abnormal return (CAR) is calculated by adding up all abnormal returns for the event window: (3) To calculate the abnormal returns, an estimation window, event date and event window have to be formulated. In figure one, T1-T2 shows the estimation window, t1-t2 the event window and t=0 is the event itself (the merger announcement). The estimation window is this study will consist of the period -210, -10 days prior to the event. In this window the expected return will be estimated. The length of this window is based on earlier studies of Peterson (1989) and Armitage (1995). They state that when performing a study which has to deal with daily information, an estimation window of days is appropriate to assess the parameters of the statistical pricing model well. Besides that, in line with MacKinley (1997), the event window and the estimation window are separated, to prevent them from influencing each other. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 14

15 Figure 1: estimation and event window The event date in this study is defined as the official announcement date of the M&A deal, as suggested by Dodd & Ruback (1977). This definition is important because it prevents misidentification, which could alter the results of the event study, as stated by Brown & Warner (1980). If there would be a perfect efficient market present, choosing the event window as the event date itself would be adequate. However, the event window is usually larger. This is because of two reasons, namely information leaking (Keown & Pinkerton, 1981) and information availability (Masulis, 1980). Information leaking occurs at a significant level sometimes even up to twelve days before an announcement. Besides that, it is not clear when the announcement is made, during trading hours or when the stock market is closed. Considering these two factors, the assumption can be made that abnormal returns can be found on both sides of the event day. This means the strong form of the efficient market hypothesis can be rejected and a larger event window is appropriate. Therefore, an event window of five days (-2,+2) is more suitable for this event study. In order to evaluate performance of M&A for acquiring firms in the US we need to create a sample. The data needed is collected from the Securities Data Company s (SDC) U.S. Mergers and Acquisition Database. Also the sample has to meet the following criteria which are almost similar as in the study of Moeller et al. (2005): 1) The announcement date is in the period; 2) The acquirer controls less than 50% of the shares of the target at the announcement date and obtains 100% of the target shares is the target is a public or private firm; 3) The deal value is equal or greater than $ 1 million; 4) The target is a U.S. public firm, private firm, subsidiary, division, or branch; 5) Data on the acquirer is available from CRSP (Center for Research in Security Prices) and COMPUSTAT; and 6) The deal is successfully completed in less than 1,000 days. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 15

16 To investigate if M&A are value creating or destroying an event study with abnormal returns around the announcement date (-2,+2 days) is used, based on the findings of Keown & Pinkerton (1981) and Masulis (1980). The abnormal returns are estimated using standard event study methods. CRSP equally weighted index return is used to find the market model abnormal returns, where the parameters are estimated over (-210,-10) event window relative to the announcement day, also based on the findings of Keown & Pinkerton (1981) and Masulis (1980). In this way the assumption that large loss deals occur in the end of a merger wave can be validated. Variables such as type of deal (hostile, friendly) and method of payment (cash or equity) and other deal and firm characteristics are all included in the model, to explain the loss or gain on a M&A deal. To prevent M&A events from influencing each other, events(m&a) done by the same firm (bidder) in the given time period are excluded from the sample, which is different compared to the study performed by Moeller et al. (2005). In this way each firm has only one M&A event in the given time period to prevent them from influencing the returns (MacKinley, 1997), and thus the reliability of the event study becomes higher. Furthermore, in order to minimize the problem of non-normal return distributions of thinly traded stocks, as stated by Maynes & Rumsey (1993) and Cowan & Sergeant (1996), a stock has to be traded at least 200 days of the total estimation period to be in the sample. Also if there was no stock information available from CRSP for a given M&A deal, this deal was deleted from the sample. Using all these parameters named above, 372 US M&A deals are found in the SDC database and CRSP for the period The test statistics used to see whether found abnormal returns around the M&A announcements are significant or not will be described below. It is important to see if the average abnormal returns around the event dates are significantly different from zero. A standard t test will be used to investigate this, using four different assumptions for the found observations, in order to ensure a reliable outcome of the test: Normally distributed Independent Constant variances (homoscedasticity) Expected value of abnormal return is zero To estimate the p value for the CAR for the different periods different linear regressions are estimated to find the robust standard error. The advantage of using robust standard errors, which is not the case with a standard OLS regression, is that you can adjust for heteroscedasticity. This is common when investigating stock returns and time-series data. With an OLS regression the sample errors often have an equal variance and are also uncorrelated. Stock return and time series data do not always have the same variance, and you have to correct for this. This means that a correction for the standard errors is made, as you adjust for heteroscedasticity. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 16

17 V. Descriptive statistics The sample of 372 US M&A announcements during the period , as stated earlier, is divided into two groups in order to answer the main research question. The period (merger wave) is one period, where the period (not a merger wave) is the other one. Below in table 1, you can find the summary statistics for the (realized) return, predicted return, abnormal return (AR) and Cumulative Abnormal Return (CAR) during the event window sorted per year. After that, in table 2 the summary statistics of the same stock returns in the event window are given by period. Table 1 - summary statistics of stock returns per year for all 372 M&A deals present in the sample during the event window year statistic return predicted return AR CAR(-2,+2) year statistic return predicted return AR CAR(-2,+2) 2003 mean mean min min max max sd sd p p year statistic return predicted return AR CAR(-2,+2) year statistic return predicted return AR CAR(-2,+2) 2004 mean mean min min max max sd sd p p year statistic return predicted return AR CAR(-2,+2) year statistic return predicted return AR CAR(-2,+2) 2005 mean mean min min max max sd sd p p year statistic return predicted return AR CAR(-2,+2) year statistic return predicted return AR CAR(-2,+2) 2006 mean mean min min max max sd sd p p year statistic return predicted return AR CAR(-2,+2) year statistic return predicted return AR CAR(-2,+2) 2007 mean mean min min max max sd sd p p year statistic return predicted return AR CAR(-2,+2) year statistic return predicted return AR CAR(-2,+2) 2008 mean Total mean min min max max sd sd p p Table 1 shows the summary statistics mean, minimum, maximum, standard deviation and median per year for the stock returns named above. Noteworthy are the years 2007 and The mean of the AR and CAR is negative for 2007, the mean of the return remains still positive. For 2008 the mean of AR and CAR is positive, the mean return however, is negative for this year. Also the standard deviation and minimums are relatively large for both years, comparing it to the other years. As presented in the Master thesis Robbert van Wingerden Mergers & Acquisitions Page 17

18 paragraph Merger trends, the last merger wave ended in This may imply that the suggestion of Moeller et al. (2005), which is that firms that undertake M&A perform worse in the last stage of a merger wave, is correct. For the years 2009, 2010, 2011, 2012 and 2013 no trend can be found. The mean of the different stock returns deviates from positive to negative for different years, which underlines the assumption that there is no merger wave present in this period, as stated by Dolfsma & McCarthy (2012). Looking at the total for all years, M&A activity tends to be positive since there is a positive AR and CAR. Also the median is positive for CAR, which means the middle of the distribution of all CAR for all given years is above zero. The median is lower than the mean, which means that there may be some outliers present. Table 2 - summary statistics of stock returns during the event window per period for all 372 M&A deals present in the sample period statistic return predicted return AR CAR(-2,+2) period statistic return predicted return AR CAR(-2,+2) mean mean min min max max sd sd p p Table 2 shows the summary statistics of the above named stock returns for the two periods. Comparing the two periods, remarkable is the much higher mean predicted return for the period This leads to a lower mean AR, and eventually to a lower mean CAR for this period. The cause of this can be found in the year 2008, where the predicted return is much lower relatively to all other years. The mean of the return itself is almost equal. Also the minimum and maximum are much larger for the period This can be due to the fact that this period is a merger wave, where companies often perform well in the beginning, but worse at the end, causing more outliers. For the other period, which cannot be classified as a merger wave as stated earlier, this is not the case which means less outliers. The standard deviation of both periods compared does not show any notable characteristics. The median of CAR though is much higher for the period for CAR. This means that the distribution of CAR for the period is shifted more to the right than for the other period. On the next page table 3 is shown, which gives the distribution of aggregate transaction values, aggregate dollar returns and cumulative abnormal returns by announcement year and period. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 18

19 Table 3 - Full sample distribution of aggregate transaction values, aggregate dollar returns and cumulative abnormal returns by announcement year and period. This table shows the sample of successful acquisitions by publicly listed U.S. firms who performed one merger in the period The acquisitions shown are from the SDC Merger and Acquisition Database of U.S. targets that are subsidiaries, public of private firms. Column N shows the number of observations. The column aggregate transaction value (in million dollars) is the total value of consideration paid by the acquirer, excluding fees and expenses. The column aggregate dollar return (in million dollars) is the sum of the acquisition dollar returns (change in market capitalization from day -2 to day +1, day is relative to the event day) divided by the sum of market capitalizations two days before the event day (merger announcement). The column CAR(-2,+2) represents the average cumulative abnormal return measured over the five days of the event window using the market model. Bidder Year N Aggregate Transaction Value ($ millions) Aggregate Dollar Return ($ millions) CAR (-2,+2) The table above supports the suggestion that firms that undertake M&A perform worse in the last stage of a merger wave of Moeller et al. (2005), since the CAR in 2007 and aggregate dollar return in 2008 are both negative. These negative returns occur at the end of the last merger wave, and are in line with the suggestion of Moeller et al. (2005). The difference with the previous tables is that the market capitalization over time of the different acquiring firms is considered, where the change is shown in aggregate dollar return. As shown above, M&A activity at the end of a merger wave could destroy firm value. The question however is if this due to deal and/or firm characteristics will be discussed in the next section. The statistical significance of the abnormal returns will be discussed first however. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 19

20 0 5 Percent The statistical significance of abnormal returns In this paragraph the significance of both the average abnormal returns as cumulative abnormal returns will be discussed. First the average abnormal returns(aar) for the different periods will be looked at using a t-test. Second, the p-value for the CAR across all companies will be estimated for the different periods and then looked at if it is significant or not. t-test To test whether the average abnormal returns are significant, the following test is done: Where N is the number of days in the event window, and SD(AR) is the standard deviation of the abnormal returns for the event window. CAR is the sum of the abnormal returns for the event window, also called cumulative abnormal returns. This means that a test is performed on the average abnormal returns, since CAR is divided by N. This gives the following results, presented below in different histograms sorted per period. First the entire period of the sample, is given in histogram 1. Next to that, histogram 2 is given which represents the merger wave. Histogram 3 shows the end of the merger wave, which consists of the years Histogram 4 shows the period that cannot be classified as a merger wave, the period Percent 1. Histogram significance CAR t-test 4. Histogram significance CAR Percent 0 5 Percent Histogram significance CAR t-test 3. Histogram significance CAR t-test t-test Master thesis Robbert van Wingerden Mergers & Acquisitions Page 20

21 The histograms above show the significance of the AAR over the different periods. It is notable that all of the histograms show a high density close to zero, meaning that the different AAR are not significant. The 10% threshold of significance, or is only met by a part of the AAR. These histograms do not provide a definite answer, therefore the p-value per period will be estimated using robust standard errors to determine whether the different CAR are significant or not. p-value cumulative abnormal returns across all companies To estimate the p-value for the CAR for the period the following linear regression is estimated to find the robust standard error. The advantage of using robust standard errors, which is not the case with a standard OLS regression, is that you can adjust for heteroscedasticity. This is common when investigating stock returns and time-series data. With an OLS regression the sample errors often have an equal variance and are also uncorrelated. Stock return and time series data do not always have the same variance, and you have to correct for this. This means that a correction for the standard errors is made, as you adjust for heteroscedasticity. Period (full sample period) N coefficient Robust Standard Error t P> t 95% Confidence Interval As you can see in the table above the CAR for the period is significantly different from zero as the p value is below This means that a merger announcement has positive consequences for an acquiring firm, given the value above zero. Also the values in the 95% confidence interval show this, as the value of CAR is 95% significant between and Besides this, the t- value is above the 5% threshold of 1.96, meaning that CAR is significantly different from zero for the period Master thesis Robbert van Wingerden Mergers & Acquisitions Page 21

22 The periods , and will also be looked at separately, to check whether the cumulative abnormal returns are significant for a period that is a merger wave, a period at the last stage of a merger wave and one that is not a merger wave. The regression estimated is the same as given above, only then per period. First the period , the period of the merger wave will be given. Period (merger wave) N Coefficient Robust Standard Error t P> t 95% Confidence Interval As you can see in the table above the CAR for the period is significantly different from zero as the p value is below This means that a merger announcement has positive consequences for an acquiring firm, given the value above zero. Also the values in the 95% confidence interval show this, as the value of CAR is 95% significant between and Besides this, the t- value is above the 5% threshold of 1.96, meaning that CAR is significantly different from zero for the period Period (end of merger wave) N Coefficient Robust Standard Error t P> t 95% Confidence Interval Above you see the regression estimated for the years As you can see in the table above the CAR are not significantly different from zero as the p value is above Besides this, the t-value is below the 5% threshold of 1.96, meaning that CAR is not significantly different from zero for the years , what suggest that in the last one or two years of a merger wave the CAR is not significantly positive or negative for acquiring firms. The non-significant results can also be due to the small sample, since N=58. Period (no merger wave) N Coefficient Robust Standard Error t P> t 95% Confidence Interval As you can see in the table above the CAR for the period is not significantly different from zero as the p value is above Besides this, the t-value is below the 5% threshold of 1.96, meaning that CAR is not significantly different from zero for the period This suggests that in a period that cannot be classified as a merger wave, the CAR is not significantly positive or negative for acquiring firms. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 22

23 VI. Deal and firm characteristics In this section the deal and firm characteristics will be given at the event day. The deal and firm characteristics data are both from the SDC merger database. In this way the characteristics can be examined per year and per period for any notable features. Deal characteristics Table 4 - deal characteristics for the period Transaction value ($ million) is the total value paid by the acquirer, excluding fees and expenses. The days in between is measured as the difference between the date announced and the date effective. Cash and equity in payment represent the amount of cash of equity used in the payment, found in the SDC database. The same goes for pure cash deals and hostile deals. Same industry deals involve targets with a 2-digit SIC code identical to the one of the bidder. Private, public or subsidiary target is also found in the SDC database. Column (1) denotes the averages of the period for the different characteristics. The same applies to table 5. Deal characteristics for period (1) Transaction value (TV) TV/Assets (market) TV/Equity (market) Days in between Cash in payment(%) Equity in payment(%) pure cash deal 27.9% 33.3% 37.8% 36.4% 37.1% 43.5% 36.0% Hostile deal 0% 0% 0% 0% 0% 4% 0.7% Same industry 62.8% 61.9% 51.1% 54.5% 45.7% 73.9% 58.3% Private target 39.5% 38.1% 40.0% 50.0% 45.7% 39.1% 42.1% Public target 25.6% 14.3% 22.2% 18.2% 40.0% 26.1% 24.4% Subsidiary target 32.6% 45.2% 33.3% 31.8% 14.3% 34.8% 32.0% Table 4 shows the deals characteristics for acquiring firms from period The characteristic cash in payment is notable because the level of cash used in payment rises during the period, and peeks in 2007 and This means that at the end of the merger wave, more M&A are paid by cash. This is probably due to the fact that the economy during this period was doing well, resulting in cash surpluses by firms. As a consequence the level of equity used in the payment drops during this period. Also the characteristic pure cash deal shows this, as the level rises during the period. Furthermore the level of public targets is interesting. In the year 2007, the level of public takeovers is considerably higher than in the other years. Also in the year 2008 this level is higher. The characteristic same industry shows that for the year 2007, more than half of the M&A done by acquirers where targets not in the same industry. The year 2008 however, which is the last year of the merger wave, shows that 73.9% of the target firms where in the same industry as the acquirer. Master thesis Robbert van Wingerden Mergers & Acquisitions Page 23