Finance and International Business

Published on February 2017 | Categories: Documents | Downloads: 27 | Comments: 0 | Views: 363
of 89
Download PDF   Embed   Report

Comments

Content

Finance and International Business 2011/2012

Master Thesis Author: André van Bragt Student ID: 282796 Supervisor: Robert Ormrod

Master Thesis

The effects of rumors of mergers and acquisitions on shareholder wealth – evidence from the European pharmaceutical and biotechnological industry

Aarhus School of Business 2011/2012

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Executive summary
In this research paper the author investigates whether or not there are abnormal returns related to rumors of mergers and acquisitions within the pharmaceutical and biotechnological industry in Europe. The test sample consists of 63 rumors from 2001 to 2012. The sample will be split to analyze the abnormal returns on a total level; comparing domestic against cross-border and comparing inter-industry against cross-industry. After the introduction a brief section with the history and trends from the recent years, 1970 to the early 2000’s, will be presented to give the reader some general knowledge before investigating the problem at hand. Here it is also explained that a large share of the recent pharmaceutical acquisitions have taken place in Europe. Afterwards a literature review will follow including general literature on abnormal returns as well as literature specific on the pharmaceutical and biotechnological industry, domestic and cross border and literature related to firm profitability, size and method of payment. The methods used in this research paper are an event study and a number of linear regression analyses. There are 8 different t-tests that are being used in the event study of which both parametric and nonparametric tests. For the event study an estimation period of 250 days is used with an event window of 3 days (i.e. +/-1 day of the event). Furthermore, there are 3 regression models that are used for analyzing the abnormal returns. The methods will be described in depth in the methodology section and the way they are used to answer the research questions. In the data description the characteristics of the sample are described. Here it is concluded that the sample comes close to a normal distribution and is assumed to be normally distributed, but it is not a perfect normal distribution. This is the main reason that a range of tests has to be used in order to capture significant abnormal returns if they are present. In the analysis it is concluded that there are positive abnormal returns following a rumor of a merger or acquisition in the pharmaceutical or biotechnological industry in Europe. Furthermore, in most of the samples it is found that the return on the event day is higher and is significant in more tests than the other days, indicating that because the event is a rumor, the market reacts different than when an announcement or completion of a merger or acquisition was taken as event.

Page | 1

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Table of Contents
Introduction .......................................................................................................................................... 4 Problem statement ........................................................................................................................... 5 Structure of the paper ...................................................................................................................... 6 Delimitation ...................................................................................................................................... 7 Assumptions ..................................................................................................................................... 8 History .................................................................................................................................................. 9 Literature review ................................................................................................................................ 11 Mergers and acquisitions in the pharmaceutical and biotechnological industry ............................ 14 Domestic & cross-border, profitability and size .............................................................................. 16 Literature summary ........................................................................................................................ 17 Methodology part I – choice of method ............................................................................................. 18 Choice of method ........................................................................................................................... 18 Regression analysis ..................................................................................................................... 18 Event study ................................................................................................................................. 20 Qualitative interviews ................................................................................................................. 21 Discussion and choice of method................................................................................................ 22 Methodology part II - the tests and models in the analysis ................................................................ 24 Event study ..................................................................................................................................... 24 Abnormal return ......................................................................................................................... 24 T-tests ......................................................................................................................................... 26 Ordinary least squares regression analysis ..................................................................................... 36 Data .................................................................................................................................................... 41 Data extraction ............................................................................................................................... 41 Data filtering ................................................................................................................................... 42 Data description.............................................................................................................................. 45 Analysis ............................................................................................................................................... 49 Abnormal returns analysis .............................................................................................................. 49 Event study results and analysis ..................................................................................................... 51 Total sample................................................................................................................................ 53 Domestic and cross-border samples ........................................................................................... 54 Industry samples ......................................................................................................................... 55 Event study summary ................................................................................................................. 59 OLS Regression results and analysis ................................................................................................ 59

Page | 2

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth Regression analysis summary ..................................................................................................... 65 Problem statement and research questions answered .................................................................. 66 Discussion ........................................................................................................................................... 69 Conclusion .......................................................................................................................................... 72 References .......................................................................................................................................... 74 Appendix ............................................................................................................................................. 81

Page | 3

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Introduction
Every firm that is listed on the stock exchange needs to fulfill different criteria to be successful. The most important one of these is, arguably, that the firm has to maximize the returns for its shareholders. Shareholders can see this return in the form of dividends or by an increase in share price. Dividends can be directly influenced by the firm’s decisions to either invest or distribute excess capital and stock price reflects the markets perception of the firm and is indirectly influenced by a firm’s decisions and performance. Both the firm and the market can see how the decisions of the firm affect its stock prices. A popular analysis to detect the impact of these decisions on the stock price is an event study. In an event study particularly the changes in a firm’s stock price in the days around an event are compared to price changes on normal days. The difference between the expected return on a normal day and the actual return from the days around the event is referred to as the abnormal return. While these decisions can be monitored in the short and long run, the focus in this paper is whether there are abnormal returns following the rumors of a merger or acquisition. For that to hold, it has to be assumed that markets are efficient and that the market can predict the long term effects of these decisions or changes in the short run. This theory is also known as “efficient markets” (Fama 1970), and will be explained further in the assumption section of this research paper. The events that will be focused on in this research paper are mergers and acquisitions. However, because the motives for mergers and acquisitions vary between countries and industries, the author chooses to focus on the pharmaceutical and biotechnological industry within Europe, the motivations behind this will be further explained in the delimitation section of the research paper. This way it was possible to do a more in debt analysis of the event, the industry, the deal and geographical implications linked to it. The main area of research will be to investigate whether or not there are abnormal returns in the pharmaceutical and/or biotechnological industry in Europe. Further areas of research are domestic deals compared to cross-border deals and inter-industry deals compared to crossindustry deals (e.g. pharmaceutical company acquiring a biotechnological company). Last the influence that various company and country characteristics have on the abnormal return will be investigated. There are a number of reasons in the pharmaceutical and biotechnological industry to engage in mergers and acquisitions, but results from previous event studies in these industries
Page | 4

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

have experienced different outcomes regarding the abnormal returns (Kirchhoff 2011; Higgins 2006; Hassan 2007; Andrade 2001). Also the drivers have been different behind the various mergers and acquisitions. Some research articles have identified research & development, growth, patents, market control and others as motivations behind mergers and acquisitions (Danzon 2007; Mittra 2007; Kipp 2008; Kirchhoff 2011). However, to answer the question whether or not there exist any abnormal returns in the pharmaceutical and/or biotechnological industry, there has not been one single outcome across the various event studies. Some studies found that acquiring firms had negative abnormal returns (Soresco, 2003), some no abnormal returns (Campa, 2004), and others positive abnormal returns (Higgins, 2006). However, it is expected that if there would be any abnormal returns, it would most likely be among firms that had alliances or worked together before the merger or acquisition, are research and development driven or between firms that engage in cross-industry mergers or acquisitions, considering this was the outcome of other related studies (Kirchhoff 2011; Higgins 2006, Andrade 2001). It also was shown among studies that troubled firms which engage in mergers and acquisitions as a way out are mostly not successful (Danzon 2007; Mittra 2007). As for the acquired firm, most studies seem to witness that an acquired firm does experience a positive significant abnormal return (Kirchhoff 2011). This can most likely be explained by the fact that the target firms can benefit from the acquirer. However, while the focus of this study is on the firm acquiring, the target firms will also be analyzed to determine whether there are any characteristics of the target firm that reflect positively or negatively on the acquiring firm.

Problem statement
As mentioned in the introduction, the main focus of this research paper is, whether or not there exist any abnormal returns related to mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe, whether or not there are differences between domestic and cross-border deals, whether or not the differences between inter and crossindustry deals and whether there are other dependent factors related to abnormal returns. As well as, whether or not they can be detected at the date of the rumor. The main reason to focus on rumors rather than announcements or completed mergers is because there has not been much research done in this area. However, there are a great deal of noise traders (i.e.
Page | 5

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

traders that trade on rumors and buzz) on the market that trade on noise as if it is information (Wang, 2010), which would therefore make it interesting to investigate the effects of a rumor instead of an annoucement. To condense this, the main problem statement of this paper is: Are there significant abnormal returns related to rumors of mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe? Additionally, to go further in debt with this, the following research questions will be used: 1. Are there significant differences, in abnormal return following a rumor, between domestic and cross-border mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe? 2. Are there significant differences, in abnormal return following a rumor, between inter- and cross-industry mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe? 3. Are there characteristics of the acquiring or target firms that have a significant influence on abnormal returns following a rumor of a merger or acquisitions in the pharmaceutical and biotechnological industry in Europe? Considering that different studies have found different outcomes, the research questions can assist in determining what influence different firm and deal characteristics have on the observed abnormal returns following rumors of mergers and acquisitions.

Structure of the paper
The paper will be structured based on other event study papers (Kirchhoff 2011, Higgins 2005, Hassan 2007). Firstly, after the introduction, a brief history of the development of the pharmaceutical and biotechnological industry will follow. This is done to give the readers some background knowledge and a better chance of comparing the results with past industry trends. Secondly, a literature review will follow where findings from past studies will be given. This will help connect the findings of this paper with previous studies and will show if there are differences between the outcome of this paper and past research. In the literature

Page | 6

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

review general literature on event studies will be discussed as well as literature specifically related to the research in this paper. Thirdly, the methodology for the research will be explained. This section is divided into two parts. In the first part different methods from different studies are discussed and the author weighs the pros and cons against each other to choose the right method(s) to use in this research paper. In the second part the methods chosen will be described in more detailed and the author will explain how these methods will be used to answer the research questions and the problem statement. Fourthly, the data used for the research will be discussed in a data section. Here the statistical aspects of the sample are discussed as well as the sample composition. In this section it is also explained how the data was retrieved and filtered and which decisions and assumptions were made in the process. Fifthly, in the analysis section, the results from the research will be presented; the data will be analyzed and linked to previous findings. Results of domestic and cross-border deals will be compared, inter and cross-industry deals will be compared and the regression analyses of specific factors will be analyzed in order to determine the impact of industry and company specific variables. The focus in this section will lay in identifying statistically significant results and explaining them. Sixthly, the results from the analysis will be discussed in this section with focus on any unexpected results. Also the implications of the delimitations will be discussed and what could be done for further research in this area. Finally, the paper will be concluded with a summary of the paper where the answers on the research questions and the problem statement will be presented.

Delimitation
As was mentioned in the introduction, the focus of this paper is on mergers and acquisitions in the biotechnological and pharmaceutical industry in Europe. The reason for delimitating the research to a specific industry and geographical area is to be able to do a more in debt analysis with focus on one industry and location specific traits and patterns. Also, the author found that industries with a high level of innovation and research &

Page | 7

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

development are more likely to have abnormal returns, therefore such an industry was chosen for this research (Kohers 2000; Kohers 2004; Higgins 2006). Furthermore, the author focuses on the effect of the rumors of the merger or acquisition, the short term effect, assuming that the market can predict the long term impact in the days around the event (Fama 1970). Moreover, the author will perform 5 parametric tests, 3 non-parametric tests to analyze the impact of the rumor on the stocks return and 3 regression analyses with selected variables, these tests are used in other studies (Bartholdy, 2007; Kirchhoff, 2011) and are expected to be capable of identifying abnormal return if there is. There are more possible tests to use, but the author wanted to focus on the results of the 8 test chosen, to be able to go more in debt with the findings. Last, the data in the research will be from January 2001 to March 2012 to reflect the current market and to be able to add more value with the findings of this paper. Kirchhoff et al. (2011) also states in his research paper that more research should be done where the recent mergers and acquisitions are incorporated.

Assumptions
The first assumption that will be made is that event study methodology holds, which implies that markets are efficient. The theory behind efficient markets has been described in Fama et al. (1970), and there it is assumed that the market has all the information available, and can therefore predict the full effect that the merger or acquisition has in the long run. While in theory markets can be efficient, in practice it is hardly ever the case that the market has all information available and everyone in the market can equally well predict the long term effects of the event. The second assumption is regarding the decision on the event window and the estimation period. Picking the right window and period is crucial and should be chosen with consideration. For this paper a 3 days event window and a 250 days estimation period is chosen, this is also used in other recent event studies (Bartholdy 2007; Kirchhoff 2011) and is expected to be adequate. Furthermore, it is assumed that the sample used in the study is following a normal distribution. The characteristics of the sample are discussed and compared to a normal distribution and the differences between a normal distribution and the sample are analyzed. However, this will be further elaborated on under the methodology section.
Page | 8

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

History
In the following section the history and trends of the pharmaceutical and biotechnological industry will be presented. The main motivation for doing this is to be able to link past trends with potential findings in the research and for the reader to get an understanding of what has happened in this industry in the past decades. A trend that has been seen throughout the last decades is that the larger firms in the industry have been dictating the course for the smaller ones (Mittra, 2007). To limit how far back the historical analysis of the pharmaceutical and biotechnological industry will go, the author decided to limit it to 1970s till the early 2000s. The main reason for starting in the 1970s is because this is where the economic expansion follow the 2nd world war, really came to a highpoint creating a perfect environment for mergers and acquisitions. The 1970s was a period where economies were growing and in the pharmaceutical industry it was not any different. In this industry, the firms that were focusing on innovation grew the most in that decade and most of the firms that had a large focus on innovation are currently among some of the biggest (Achilladelis, 2001). Most of these had a track record of being known for their innovativeness such as Bayer, Roche, Hoechst and others. It was also in the end of this decade that firms between themselves started partnering up or at least looking after alliances to strengthen their position. However, the majority of these groups of alliances only started to occur in the early 1980s (Roijakker, 2005). Because of the innovation of the 70s, in the 80s pharmaceutical companies grew steadily due to their increase in patents. A popular approach at this point was the shotgun approach, where companies would try to experiment with a large number of compounds hoping one would have success (Nigro, 2011). It was also in the 1980s that biotechnology started to take off and reached high growth due to intensive research and partner network expansion. When the biotechnological industry started to grow, pharmaceutical firms started to notice the potential that this industry could offer (Roijakker, 2005). With this pharmaceutical companies could expand their product pipeline in new dimensions than was possible if allying with a pharmaceutical company. While there can be argued for the changes that biotechnology introduced to pharmaceutical firms, the key figures of importance in the 1990s were economies of scale and scope. These became more and more important throughout the 90s as R&D spending and innovativeness increased with the economies of scale (Nigro, 2011). The 1980s was a period
Page | 9

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

in which a wave of M&A took place in most industries. This trend only reached the pharmaceutical industry in the 1990s. This was possibly the result of the growing importance of scale and scope. However, while in the 90s pharmaceutical companies grew, their productiveness declined. This was partly the result of sharper rules by the Food and Drug Administration (FDA), making it harder for generic products to hit the market (Nigro, 2011). In this period the alliances and partnerships between pharmaceutical companies and biotechnological companies grew even more and included a number of the big players such as Merck, Lilly, Glaxo and others (Roijakker, 2005). Most firms had a chance to test the waters in earlier years to finally go for mergers and acquisitions with known partners. In Nigro et al. 2011’s paper arguments are made for the complementary assets between the pharmaceutical and biotechnological industry to be the main reason for the cross-industry interest. Biotechnology brings in new methods in different stages of the discovery, development and manufacturing to the table to support weaknesses in pharmaceutical firms. Around 2000 it was mainly in Europe that pharmaceutical companies were involved in mergers and acquisition. All big players in the pharmaceutical industry were part of one or more mergers or acquisitions in between 1997-2006 (Mittra, 2007). This is also one of the reasons the author choose to focus on the European market when choosing the sample for the study. A larger sample of mergers and acquisitions would make it easier to analyze the effects. What also characterized the years after 2000 was that a great deal of pharmaceutical companies were facing future expirations of patents, but had hoarded up cash from their blockbuster drugs, while biotechnology had a lot of potential but no cash to develop this (Roijakker 2005; Malik 2009). This made them the perfect match for each other and could possibly hold abnormal returns if mergers and acquisitions were driven by mutual strengths and complementary assets. To sum up the above, firms in the pharmaceutical industry have experienced growth over the last decade, mainly due to high levels of innovation and mergers and acquisitions. The main geographical area where these mergers and acquisitions took place was in Europe. Aside from that, the pharmaceutical industry has been increasingly exploring alliances, partnerships, mergers and acquisitions in the biotechnological industry. From first glace there appear to be valid reasons for these two industries to enter in mergers and acquisitions, but the results from hard data in this research paper will show whether or not these mergers and acquisitions have positive abnormal returns for the shareholders.

Page | 10

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Literature review
In the following section the existing literature will be discussed. This will include literature related to abnormal returns and event studies in general; domestic as opposed to cross-border mergers and acquisitions; mergers and acquisitions within the pharmaceutical and biotechnological industry; returns to the target and the acquirer and other characteristics that are related to abnormal returns following rumors or announcements of mergers and acquisitions. Considering that the amount of literature on mergers and acquisitions is extensive, there will only be a brief and partial survey of the literature with focus on the literature related to the problem statement. However, the author considered it relevant to include both generic literature as well as industry specific literature to give a better overview of the research done and because there is only a limited amount of literature that focuses on the pharmaceutical and biotechnological industry.

Acquirers, targets and the combined entity
In the first part of the literature review the returns to the acquiring firm will be discussed. Among the different studies it appears that there is no clear conclusive outcome on the shareholder returns for the acquiring company. A number of researchers have found that the acquiring company experienced negative abnormal returns (Eckbo, 2000; Kohers, 2000; Houston, 1994). Some researchers allocate the negative abnormal returns due to the winners curse (Malik, 2009). The winners curse “refers to a situation in competitive bidding when the high bidder, by virtue of being the high bidder, has very likely overestimated the value of the item being bid on” (Berk, 2011 chapter 23). However, Samuelson et al. (1985) claims that if a firm is well informed or better informed than the market, it can overcome the winners curse. This also reflects the results in which firms with prior alliances or relationships experience higher abnormal returns (Higgins, 2006). Other researchers find that the returns for the acquirer are not significantly different from zero. Rad at al. (1999) studied the effects of announcements of mergers and acquisitions in Europe and found no significant increase or decrease of shareholder wealth for the acquirer. Ismail et al. (2005) found abnormal returns not significantly different from zero for his cross-border sample, but did found slightly positive returns for his domestic sample. However, it should be noted that in his research he had lower target returns than in most other studies. Goergen et al. (2003) found that in their sample domestic acquisitions were more costly for the acquirer than cross-border. Moreover,
Page | 11

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

the results showed that domestic acquisitions had abnormal returns not significantly different from 0, while the cross-border acquisitions had an abnormal return of 3.09%. Kiymaz et al. (2004) also showed positive abnormal returns for acquirers, mainly for those acquirers that had a target in a developing market. The positive abnormal returns of the acquirer are also found in other studies (Higgins, 2006; Fuller, 2002). Therefore it is difficult to conclude on whether a mergers or acquisitions are value creating or destroying. It appears to be related to different characteristics of the deal that results in positive or negative abnormal returns. Moeller et al. (2005) investigated acquisitions from 1980 to 2001. The results showed that in the acquisitions from 1990 to 1997 had been profitable for the acquirer but that the acquisitions from 1998 to 2001 destroyed all gains made from earlier acquisitions. Moreover, it was found that acquiring firms remain having positive abnormal returns for each acquisition announcement until an acquisition is showed to have negative shareholder returns, after which no significantly positive deals have been captured. The outcome of the research can possibly be summed in the following quote: “it is possible that the acquisition demonstrates to investors that the acquiring firm’s strategy of growing through acquisitions is no longer sustainable and will not create as much value as they believed previously” (Moeller, 2005 P.758). As can be derived from the above quote, an important part of the acquisition process is also the signal it sends to the market. Acquisitions paid with cash have better returns than acquisitions paid with stocks. This is due to the fact, that if a firm pays with stocks the market perceives it as the stocks being overvalued. This is also seen in a study from Andrade et al (2001), where the results indicate that deals which are financed with cash are experiencing better returns that those that include shares. The same was found in the study of Travos et al. (1987), who investigated mergers and acquisitions with focus on the financing of the deal, acquisitions that are financed with cash are better perceived than those with stock. Malik et al. (2009) therefore also argues for cash rich pharmaceutical firms to acquire cash poor biotech firms. Kirchhoff et al. (2011) makes the same conclusion, but adds that the stock performance of the acquirer also should be factored in. If the acquirer’s stock was already over performing when the acquisition was announced, it would indicate to the market that it was dealing with a firm that was well managed and having a superior strategic position. So far the returns to the acquirer have been discussed, but the target is important to consider as well. Opposite to the acquiring firms, the target firms always experience

Page | 12

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

significant gains. These results have been found across different studies (Eckbo, 2000; Higgins, 2006; Andrade, 2001). It could be argued that the winner’s curse which is used for explaining negative returns for acquirers can also be used to explain positive results for targets, because if the acquirer overpaid, the target received more than what it was worth. It is unanimous that takeovers are value creating for the target and the effect seen at the announcement when comparing to the pre-announcement share price (Renneboog, 2006). Sowlay et al. (2009) discusses the situation when the target is underperforming, this will result in a situation where the acquirer potentially can optimize the performance of the target, which is value creating for the target and in some cases also for the acquirer. Beitel et al. (2004) states that successful acquirers can be identified from the targets they choose. Choosing well managed targets that have a high market to book ratio often lead to positive returns for both the target and acquirer, while poorly managed targets that have to be optimized are often only value enhancing for the target. The returns to the acquirer and target have been discussed, in which the acquirer seemed to be experiencing negative or slightly positive returns and the target significant positive returns. A great deal of research has also been done with regard to the combined entity. However, the results are not unexpected. Because the acquirer has returns around zero and the target significantly positive returns, the combined entity is also experiencing positive returns (Andrade, 2001). Unfortunately this is mainly due to the return that the target experienced. The same was also found by Kirchhoff et al. (2011) and Higgins et al. (2006), that the combined entity has significant positive returns. Renneboog et al. (2006) mentions in his publication: “as the target shareholders earn large positive abnormal returns and the bidder shareholders do not lose on average, takeovers are expected to increase the combined market value of the merging firms’ assets” (Renneboog, 2006 P.27). This was also reflected in the findings of other studies (Becher, 2000), that even though the acquirer’s shareholder wealth had declined the combined entities’ shareholder wealth had increase, indicating synergies which are value creating of which the upside mainly went to the target. Conversely, Ravenscraft et al. (2000) found that the combined entities had returns around zero. The sample consisted of mergers from the pharmaceutical industry between 1985 and 1996, in which the target had abnormal returns of 13.3% and the acquirer negative abnormal returns of -2.1%, which resulting in the combined firm having returns of zero.

Page | 13

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Mergers and acquisitions in the pharmaceutical and biotechnological industry
In the literature review up to this point the general effects of mergers and acquisitions have been discussed. In the following section the results from the pharmaceutical and biotechnological industry will be presented. Multiple studies have found that a strong drive behind mergers and acquisitions in these industries is expiring patents and gaps in the product pipelines (Danzon, 2003; Higgins, 2006). However, these types of deals that are motivated by internal weaknesses do not seem to create value for the shareholders and are in most cases even destroying shareholder value (Vernon, 2005; Mittra, 2007; Danzon, 2003). Only in some cases mergers and acquisitions that are driven by these motives can increase value, the two main scenarios are if the acquiring firm has more knowledge about the target due to previous alliances (Higgins, 2006) and scenarios where the target is cash poor but very innovative and possibly cross-industry (Kirchhoff, 2011). Sorescu et al. (2003), which investigated the short-term returns in the pharmaceutical industry with a 3 day event window found significant results in his sample. The abnormal returns were significantly negative for the acquirer. However, Sorescu et al. (2003) also used larger event windows with up to 2 years after the merger or acquisition and found that the results were not significantly different. This would indicate that the market was efficient in predicting the long term value of the deal at the time of the announcement. Contradictory to this, Higgins et al. (2006) found that in a 3 day event window acquirers had positive cumulative abnormal returns. However, the positive abnormal returns were mainly present with acquisitions were the acquirer had a past alliance or was well informed before the acquisition. All the above studies had used a 3 day event window, indicating that this is a justifiable choice for using as event window in the study at hand. As mentioned earlier in this section, Kirchhoff et al. (2011) found that cross-industry, pharmaceutical and biotechnological, mergers and acquisitions had positive results. Arora et al. (1990) found similar results, where pharmaceutical firms and biotechnological firms will complement each other indicating the potential for synergies. In another study (Malik, 2009) also evaluated these cross-industry mergers and acquisitions between pharmaceutical and biotechnological firms. The pharmaceutical firms are often cash rich with expiring patents and gaps in their product pipelines, while biotechnological firms are cash poor but with great level of innovation and potential, making them a perfect match and with potential for positive
Page | 14

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

abnormal returns. However, Eckbo et al. (1992) found that intra industry mergers had positive results in the event window surround in announcement. Also, the benefit of having pre- merger or acquisition partnerships or alliances (Higgins, 2006) has a positive relationship to shareholder returns in the biotechnological and pharmaceutical industry. This relationship has also been discussed in other research papers (Baum, 2000) and has also been identified as one of the factors that is important for achieving positive abnormal returns, as the earlier partnership can assist the firm in truly assessing the target. Also Malik et al. (2009) found that earlier alliances between the acquirer and the target have had positive results on shareholder value, it was stated that it was a way to “test the waters to establish whether the companies are a good strategic fit” (Malik, 2009 P.819). Nicholsen et al. (2002) examined mergers between pharmaceutical and biotechnological firms, focusing on the need to merge between the two entities from a R&D perspective, investigating expiring patents and gaps in the product pipeline. It was found that firms that have this as main driver will remain to have low growth in R&D and have not been able to create shareholder value from the merger. These results are also found in research from Danzon et al. (2007). Rothaermel et al. (2001) conducted similar research, but instead of focusing on the synergies of past relationships, the focus was put on similar product development. The results indicated that alliances between firms which work with similar products are more successful than firms which have none related products. Other researchers investigated the relationship of firm size in the pharmaceutical and biotechnological industry. Cockburn et al. (1996) found that innovation is positively correlated to firm size, mainly due to benefits and spillover between the different products. Another study shows that the relatedness between the industry of the acquirer and the target is positively correlated with expected returns (Campa, 2004). This was a study in which European mergers and acquisitions were analyzed, but returns were not significantly different from zero. In the same study was also discussed that firms that are diversified can be acquired at a discount compared to specialized firms, although the results from that study proven otherwise. However, in another study of European mergers and acquisitions, it was found that diversifying firms had higher performance than their industry peers (Renneboog, 2006). This shows that the results from the same research area have not been exclusive in the outcome.

Page | 15

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Domestic & cross-border, profitability and size
One of the research questions is regarding the performance of domestic compared to cross-border acquisitions, therefore relevant literature on this topic should be discussed. Mangold et al. (2008) finds that due to larger competition in domestic markets acquirers often pay more than in cross-border mergers and acquisitions. However, domestic acquisitions are often less challenging than cross-border. So it depends on which advantage outweighs the other. Goergen et al. (2003) and Renneboog et al. (2006) finds that the difficulties in a crossborder acquisition, such as cultural and regulatory differences, are greater than cost advantage and market expansions, resulting in lower returns at the date of the announcement of a merger or acquisition. Gugler et al. (2003) analyzed post announcement performance of national and international mergers and acquisitions. It was found that also in the period after the announcement acquiring firms in cross-border acquisitions had significant worse returns than the domestic acquisitions. In a study regarding shareholder wealth created in the pharmaceutical industry (Hassan, 2007), it is stated that most of the available literature supports that cross-border acquisitions are outperformed by domestic acquisitions. There also have been a number of studies that studies profitability and size. Some studies found that acquisitions in which the target is poorly managed, there is potential for turning it around and gaining positive abnormal returns (Ekkayokka, 2009; Hernando, 2009). Beitel et al. (2004) finds the opposite and states that well managed target will have better returns. The results have not been exclusive with regard to how well managed and profitable the target is. Moeller et al. (2004) did a study with focus on firm size related to abnormal returns in acquisitions. In the sample of acquisitions from 1980 to 2001 in the US, he found that acquisitions by smaller firms perform better than acquisitions by larger firms. The motivation behind the poorer performance of the large firms is mainly due to premium pricing and empire building motives by the acquiring firm’s management. Conversely, Cockburn et al (1996) found opposite results when looking at firm size performance. It was found that the relation between firm size and productivity in the pharmaceutical industry was positively related. In the same research paper it was argued that firm size in the pharmaceutical industry was becoming increasingly important. Rothermael et al. (2004) also found that firm size and post-merger performance was positively correlated. However, in other studies it was found that there is no significant relation between the relative target size and performance (Haushalter, 2002; Powell, 2005).

Page | 16

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Literature summary
To summarize the above, there have not been many exclusive outcomes between studies. The abnormal returns for the acquirer have been around zero with some studies finding slightly positive or negative returns. The only exclusive outcome throughout all studies is that the target experiences significant positive abnormal returns. The combined entity has in most cases increased value compared to each of the separate entities, indicating that synergies can be achieved or that knowledge and other advantages can be shared. The signal to the market is an important part of the deal, which is an important reason for deals financed by cash to outperform deals financed by stock. In the literature regarding the pharmaceutical and biotechnological industry it has been found that acquirers that have had ties with the target before the announcement are experiencing positive abnormal returns. Also, research & development, expiring patents and gaps in the product pipeline appear to be some of the main drivers behind mergers and acquisitions in this industry. However, most of these deals are value destroying as the motive behind the deal is a weakness of the acquiring firm, rather than a strength that can be used in the target firm. Furthermore, in most cases cross-industry mergers and acquisitions had better results than the intra-industry deals. In most domestic and cross-border studies it is found that domestic mergers and acquisitions outperform the cross-border mergers and acquisitions both at the announcement period and the period after. There has been research done with regards to firm size but no exclusive outcome has resulted from it.

Page | 17

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Methodology part I – choice of method
In the following section the author will start with briefly explaining the different methods and the processes that could be chosen to find the answer on the problem statement and the research questions. Afterwards the author will explain the reasons for his choices regarding the chosen methods and processes. Last, the author will explain the methods more in depth and with focus on the problem at hand and how this methodology is applied to answer the main problem statement and the research questions.

Choice of method
When the author was faced with the choice of how to analyze the effects of mergers and acquisitions in the pharmaceutical and biotechnological industry, there were three main methods that have been used in other articles. Firstly, Danzel et al. (2007), Hess et al. (2010) and Higgins et al. (2006) have used a regression analysis with various accounting data and/or ratios for analyzing the value creation in the pharmaceutical industry and biotechnological industry. Secondly, researchers such as Kirchhoff et al. (2011), Hassan et al. (2007) and Andrade et al. (2001) have used an event study for analyzing the effects of mergers and acquisitions in the pharmaceutical industry and biotechnological industry. Thirdly, Mittra et al. (2007) and Kipp et al. (2008) have used qualitative interviews for analyzing the effects of mergers and acquisitions in the pharmaceutical industry and biotechnological industry. In the following three sections the author will explain the methodology of each and will discuss the advantages and disadvantages of using each specific analysis for answering the main problem statement and the research questions. The author will conclude on the three sections with a summary, which will be followed by a more in debt methodology description of the methods that the author chose and how they will be used to answer the research questions and the main problem statement.

Regression analysis

Under this section linear regression analysis methodology is explained together with the advantages and disadvantages of using a multiple regression analysis to analyze the

Page | 18

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

effects of mergers and acquisitions. In a linear regression ‘a dependent variable’s magnitude to measures for a group of explanatory variables working together’ is analyzed (Colwell, 2009 P.255). When preforming a linear regression analysis the basis for the analysis is a model in a number of dependent variables are used. The most standard formula for a regression analysis can be described as: Y = +  1X1 + X2 + … nXn +  (Colwell, 2009) Y has in theory a linear relationship to each of the variables and each explanatory variable X1 to Xn has a significant relationship to Y (Colwell, 2009). The in the model represents a value that is independent of the variables and can be seen as an intercept value. The  1 to n in the model represent the variables contribution to the models value (Colwell, 2009). Aside from a standard linear multivariable regression analysis, variables can be used in logarithmic, powered, cross-multiplied and other mathematical forms. Now that the reader is introduced to the simplified methodology of the regression analysis the advantages and disadvantages will be presented. The first advantage of using a regression analysis for analyzing the effects of mergers and acquisitions is that, compared to an event study, the company does not have to be listed to compare results. Moreover, a regression analysis can work and is sometimes even preferred if insignificant variables are part of the model (Tu, 1996). This gives the researcher a greater deal of flexibility in the making the model for the analysis. The second advantage is that there are a wide range of programs designed for conducting regression analyses (Shelton, 1987), making it less demanding to perform the analysis. The third advantage is that a regression analysis is more objective than an interview or other assessment procedures (Dilmore, 1972). This is mainly because it draws on hard data, rather than a subjective judgment. A regression analysis is especially useful to put a weight behind every variable and which variables have the largest influence (Dilmore, 1972). The first disadvantage of using a regression analysis is that the data has to satisfy a number of criteria, better known as assumptions, to be considered valid. These assumptions state that the data has to be 1) random, 2) independent, 3) normally distributed, 4) have a constant variance and have a 5) residual mean equal to zero (Shelton, 1987). If the assumptions are not satisfied the results from using the data may be invalid or ambiguous.
Page | 19

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Selecting the right variables is critical in order to be able to satisfy the different assumptions (Dilmore, 1972). This adds more difficulty to finding a data sample that can be considered for the analysis. The second disadvantage is that a large data sample is required to be able to draw proper conclusions (Dilmore, 1972). If the dataset is too small the results are not significant. All in all there are important pros and cons to take into consideration when deciding whether to use a regression analysis for the data analysis in this research paper.

Event study

In this section event methodology will be explained and the advantages and disadvantages of applying an event study to answer the research questions will be discussed. When conducting an event study the first step is to define the event and to identify the estimation period or range (MacKinlay, 1997). After those have been identified, the tests have to be chosen which will have the greatest chance of successfully measuring the implications of the event. This depends on how closely the sample follows a normal distribution. When conducting an event study it is also important to assume that markets are efficient (MacKinlay, 1997), meaning that all information is available on the market and that the market is efficient enough to predict what the implications are of an event in the long run on the day it is known to the market (Fama, 1970). When testing for abnormal returns the data of the estimation period is regressed against market data to reduce the chance of abnormal returns being mistaken for general changes in the market. After having analyzed the relation between the market and the stock it is possible to calculate the expected return on a given day. Any return that differs from this can potentially be abnormal. This can be seen in the following formula where the observed return is deducted from the expected return. Ait = E[rit] – rit (Bartholdy, 2007) After having performed the different tests it is possible to see whether or not a specific event results in positive or negative abnormal returns and whether they are material. However, this form of analyzing data also has a number of advantages and disadvantages. These will be discussed in the following sections. The first advantage is that an event study is considered highly useful for measuring the economic impact of specific events while at the same time being fairly easy to
Page | 20

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

conduct (MacKinlay, 1997). The second advantage is that when preforming an event study the researcher has various tests to his disposal to be able to capture possible abnormal returns even if the data is not perfectly normally distributed (Bartholdy, 2007). Considering that in reality data hardly ever mirrors a perfect normal distribution, this is an important advantage to take into consideration. The third advantage is that “regardless of the institutional setting” the researcher is able to determine whether there are abnormal returns with great accuracy (Bartholdy, 2007 P.227). While there are pros to using an event study there are also cons. The first disadvantage is that event studies are somewhat limited to listed firms so that the changes of different securities can be measured over the estimation period (MacKinlay, 1997). The second disadvantage is that when conducting an event study, markets have to be assumed to be efficient. This does not reflect the real market place and can therefore only be an assumption rather than a condition (Fama, 1970).

Qualitative interviews

In the following section the methodology of qualitative interviews is explained as well as the advantages and disadvantages. The basis of an interview is an interviewer asking an interviewee questions. These questions have to be designed to answer the problem that the interviewer has. The answers that come from an interview are different than can be gained from a statistical study. It depends greatly on the motives of the researcher whether he wants to use a statistical study or interviews for data collection. Yeung et al. (1995) believes that “qualitative information from an interview give the researcher a more realistic ‘feel’ of the world that cannot be experienced with ‘cold’ statistics” (Yeung, 1995 P.325). Talja et al. (1999) describes a similar comparison referring to interviews as “humanistic approach” (Talja, 1999 P.460) while referring to statistical data as a “hard approach” (Talja, 1999 P.460). It is therefore important to look at the motives of the researcher, what the aim is of the study and what the researcher wants to capture in his study to decide whether or not to use qualitative interviews as data source. Some of the advantages of doing a qualitative interview are that it is considered easy to make a random sample and conduct interviews with independent candidates (Mitchell, 2007). This mainly due to that potentially by telephone no one is out of reach and that information that is not available yet, can be obtained during the interview. Another
Page | 21

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

advantage is that it is possible that more or different information is received that was anticipated on forehand (Yeung, 1995). Because an interview has humanistic factor involved in it, aside from that the interviewer can attempt to guide the interview, the answers and information received can never be completely predicted. This also leads us to the next advantage, which is that in the terms of structure and design the qualitative interview allows for more flexibility (Pole, 2002). This is mainly due to that the interviewer is not restricted by specific tests or other parameters when designing and conducting the interviews. Aside from the advantages, there are also a number of disadvantages to using qualitative interviews as the main source of data. Some of the disadvantages are that, unless an interview is face to face, there will likely not be enough depth in the information received (Mitchell, 2007). This can appear to be problematic if the candidates for the interview are not focused in a specific area, which would therefore be a problem for this research. Another disadvantage is that there is a chance that the answers will be biased. This can occur due to that the interviewee adjusts his answers so they are more “acceptable”, “desirable” or to try to “impress” rather than giving the real answer (Mitchell, 2007; Pole, 2002; Talja, 1999). The reason lies in the humanistic factor that is introduced in the interview because the data is ‘soft’. Last when doing an interview in a larger geographical area the response rate tends to be lower and questions and answers can be misinterpreted or differently interpreted due to the different backgrounds (Yeung, 1995; Talja, 1995). The different interpretation does not necessarily need to be a disadvantage, but it does introduce complexity and could introduce confusion. All in all there are some advantages and disadvantages to using a qualitative interview for collecting the data for the analysis. However, in the end it is more a choice of preference and matching the data collection method to the aim of the study, rather than saying that a qualitative interview is better or worse than the other methods discussed.

Discussion and choice of method

It is an important point to consider that a regression analysis is required for calculating abnormal returns of a stock over a period of time, which is a requirement for conducting an event study. In a linear regression analysis it is possible to identify more specific factors to give a more detailed explanation to results from an event study. They therefore work complementary to each other. Furthermore, it is important to consider the
Page | 22

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

preference of the researcher. As discussed under the qualitative interview section, the main choice to go for either an interview or a statistical analysis is whether the researcher wants to use a “humanistic” or “hard” approach. The advantage to the humanistic approach is being able to include soft data, data that cannot be expressed in numbers. The advantage to the hard approach is that it involves hard data that is based on facts and numbers. Another point to take into consideration is the availability of data. For interviews the researcher has to be able to reach the enough experts to gather information for the analysis. In the event study the researcher is limited to the stock listed firms that have easy accessible records. The regression analysis requires a sample that fulfills certain criteria in order to be considered suitable for testing. Last, the form of the data available. Considering that in hard statistical studies the researcher often faces not normally distributed samples and in humanistic studies data that can be interpreted in different ways. Taking everything into consideration regarding the advantages and disadvantages of the different research methods, the author decided to base the research in this paper on an event study with complementary regression analyses. The reasons for choosing these two methods is that they complement each other; they both use hard numbers and statistics; they are objective and unbiased; and when working with these methods is the rules and boundaries are easier to outline and identify. Moreover, event studies and regression analyses work even when data is not normally distributed due to the various t-tests in the event study and the different variables in the regression analyses that can aid in creating the perfect combination of tests. In the following section the author will discuss how the chosen methods will give a more in depth explanation of the methods and how they will be used to answer the problem statement.

Page | 23

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Methodology part II - the tests and models in the analysis
In the previous sections the author explained that an event study with multiple tests will be used to answer on the first and second research question and that regression analyses will be used to answer on the third research question. The regression analysis will also work complementary on the results of the even study and is required for calculating the abnormal return of the different stocks needed for the event study. The following sections will go more in debt with the methodology behind abnormal returns, the different parametric and nonparametric t-tests and the regression analyses and how they will be used to answer on the research questions and the main problem statement.

Event study
When preforming an event study the first task is to identify the event that has to be studied (MacKinlay, 1997). In this case that is the rumor of a merger or acquisition within the European pharmaceutical or biotechnological industry. The event window in this study will be 3 days, 1 day before and after the day of the rumor. This has also been mentioned under the assumptions in the introduction, mainly because it is a choice that is left to the researcher and various event studies have different event windows ranging from 1 to 251 days (Kirchhoff, 2011; Hassan, 2007). However, because other recent event studies also used an event window of 3 days (Bartholdy, 2007; Andrade, 2007; Higgins, 2006), this is assumed to be a correct choice for the current study. Also, as stated before, it is important to assume that markets are efficient in order to justify that the effect of a merger or acquisition can be measured from the moment the market is informed about this. Furthermore, abnormal returns are also an important point in event studies. This will be explained in the following section.

Abnormal return

Before conducting the different t-tests and the regression analysis it is important to calculate the abnormal return for each stock per merger or acquisition. Abnormal return is the difference between the expected return and the actual return. Abnormal return can both be positive or negative. The expected return can be calculated by using the following formula: E(Rit) = αi + βiRmt (Bartholdy, 2007)
Page | 24

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Where: E(Rit): the expected return of stock i on day t αi: the interception term βi: the slope coefficient derived from the market return Rmt: the return on market(m) index on day t The coefficients αi and βi will be calculated using a linear regression on the share price and the market return over the 250 day estimation window. The motivation for using the market to regress against and as input variable is that it adjusts for market wide fluctuations. Afterwards the abnormal returns for the estimation period and the even window have to be calculated. This will be done with the following formula: Ait = E[rit] – rit = -249, -248, ..., -2, -1, 0, +1 (Bartholdy, 2007) Where: Ait: the abnormal return for stock i on day t. E[rit]: the expected return for stock i on day t. rit: the actual return for stock i on day t. The cumulated abnormal return (CAR) for the event window is calculated by summing the daily returns (Āt) using the following formula: CAR = Ā+1 + Ā 0 + Ā +1 (Bartholdy, 2007) Last, the thickness of the trade is also an important factor to take into consideration. Thickness of trade is basically how often the stocks are traded and can be divided into 3 categories. Thinly traded stocks are stocks that have been traded less than 40% of the time in the estimation period. Medium traded stocks are those that have been traded between 40% and 80% of the time. Last, thickly traded stocks are the stocks that have been traded 80% or more of the estimation time (Bartholdy, 2007). The return on a trading day can be calculated using the following formula:

Page | 25

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

(Maynes, 1993) Where: Rt: the return on the trading day Pt: price of the stock on day t Pt-1: price of the stock on day t The returns that have a 0 illustrate the trading days where no trade took place. By summing up the days with a value the thickness of the trade can be calculated. According to Maynes et al. (1993) event study methodology may not hold and results might turn out to be insignificant for most parametric t-tests, if stocks are not frequently enough traded. Therefore the author will include only thickly traded stocks, traded 80% or more, to avoid having the problem of event study methodology not holding or the results not having significant explanatory power.

T-tests

To test whether or not the daily abnormal returns and the cumulated abnormal returns can be considered significant, parametric and nonparametric tests are used. According to the publication “Conducting Event Studies on a Small Stock Exchange” (Bartholdy, 2007) it is essential to use a range of tests and to make a conclusion based on all of them which would enhance the strength of the explanation and make it less situational. It is also stated that usage of both parametric and nonparametric tests will give a better result, especially since nonparametric tests perform better when the sample is not following a normal distribution. In the following sections the parametric tests and nonparametric tests used in the research will be explained.

Parametric tests

Parametric tests perform best when they are used on samples that follow a normal distribution. For this research paper, the author has chosen 5 different parametric tests each with different test conditions. The tests chosen have been used in other research papers
Page | 26

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

(Bartholdy, 2007). The 5 tests chosen are listed below together with the respective articles they relate to. The tests are numbered for the reason to easily distinguish between the various tests when they are discussed and explained.      T1: T-test with cross sectional dependence (Brown and Warner, 1985 and Patell, 1976) T2: T-test with cross sectional independence (Brown and Warner, 1985 and Patell, 1976) T3: T-test with standardized excess return cross sectional independence (Sallinger, 1992) T4: T-test adjusted for event induced variance cross-sectional (Boehmer, Musumeci and Poulsen, 1991) T5: T-test adjusting for event induced variance standardized cross-sectional method (Boehmer, Musumeci and Poulsen, 1991) The T1, the t-test with cross sectional dependence, is conducted by dividing the average event-period residual by its contemporaneous cross-sectional standard error (Brown, 1985). In the T1 test, the variance of the average is the average of the individual variances. The sample is assumed to be normally distributed when performing the T1 test. However, this is not a requirement that has to be fulfilled if the sample size is large enough (Bartholdy, 2007). Moreover, since a battery of tests is performed the chances of successfully detecting abnormal returns, if they are present, is increased. To calculate the mean or average abnormal return on day t (Āt), the formula below is used:

(Brown, 1985) Where: N: The number of firms in the sample Ajt: Security j’s abnormal return on day t

Page | 27

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

After having calculated the abnormal return’s mean for the different day’s the mean of the overall sample ( ) has to be calculated. This is done with the following formula:

(Brown, 1985) Where: T: Number of days in the estimation period After having calculated the mean of the days in the estimation period and the overall mean, it is possible to calculate the standard deviation of the average excess return over the estimation period (S(Ā)). This is done by the following formula:

(Brown, 1985) After having made the above calculations it is possible to test for abnormal return on any day, in this case +1, 0, -1 of the event. This is done by dividing the mean abnormal return of a given day by the standard deviation of the average excess return.

(Brown, 1985) To test for significance of the CAR over the entire event window the following formula is used:

(Brown, 1985)

Page | 28

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

The T2, the t-test with cross sectional independence, is assumes that abnormal retursn across stocks are independent. This would indicate that, because the variables are independent, they can be summed, considering that it is assumed that each abnormal return will have the same variance. Therefore the variance in T2 is calculated as the sum of variances of the abnormal return of the individual stocks (Brown, 1985). In the T2 test, if stocks are not thickly traded it would require adjustments for the thinly traded stock (Bartholdy, 2007). However, as indicated before, only thickly traded stocks were chosen for this research. The T1 and T2 test are very similar and the main difference is the cross sectional independence in the T2 test. This also makes the formula for testing for significant returns in the T2 test easier than in the T1. This is done with the following formula:

(Brown, 1985) Where: Ait: Abnormal return for stock i on day t T: Number of days The t-statistics for CAR is calculated in the same way as for T1 using the standard deviation which is the denominator of Tt2 by using the formula from Brown et al. (1985). This is a method that is used in various t-test for cumulative abnormal returns. The T3, the t-test with standardized excess returns cross sectional independence, the abnormal returns are scaled by their individual performance and these are afterwards summed to produce the test statistic (Sallinger, 1992; Bartholdy, 2007). By adjusting the abnormal returns the test has increased success on samples with a short estimation period or long event window. It is also assumed that the abnormal returns are independent. For producing the test statistic first the standardized abnormal returns (Asjt) for security i have to be calculated. This is done by using the formula below:

Page | 29

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

(Bartholdy, 2007) Where: T: Number of days Ait: Abnormal return for stock i on day t Ajt: Adjusted abnormal return j on day t of the test day The denominator is the standard deviations for each and the nominator being the adjusted abnormal returns for day t. Since the standard deviation of a standardized variable equals 1, the cross sectional variance of the average standardized excess returns is given by:

(Bartholdy, 2007) Where: N: The number of observations The test statistic for event day t is then given by taking the average of the standardized abnormal returns and dividing this by one divided by the square root of the number of returns. However, this can be simplified to the sum of standardized abnormal returns divided by the square root of the number of returns. This can be seen in the formula below:

(Bartholdy, 2007)

Page | 30

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

In or to calculate the Cumulative Average Residuals (CAR) using a window of three days the following formula is used:

(Bartholdy, 2007) The test statistics for CAR is given by:

(Bartholdy, 2007) The method is in essence the same as used in Brown et al. (1985). The sum of test statistics divided by the square root of the number of abnormal returns to be tested (3 in this case). As mentioned after the explanation of the T1 and T2 the standard method for calculating the test statistic for the CAR. In the T4, the t-test adjusted for event induced variance cross-sectional, it is assumed that security residuals are uncorrelated and that event-induced variance is insignificant. It adjusts the out of sample prediction and will have a higher standard deviation than the estimation period. The adjustment allows for heteroskedastic event day residuals (Boehmer, 1991). In the other t-tests the variance is estimated over the estimation period, in this test only the event window is used.

(Boehmer, 1991) The test statistic for the cumulative abnormal returns is calculated by summing the tvalues for the 3 days and dividing them by square root of 3, the method as used by Bartholdy et al. (2007) and Brown et al. (1985).

Page | 31

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

In the T5, the t-test adjusting for event induced variance standardized crosssectional method, the residuals are first standardized and then the variance is estimated during the event window, to counter for the temporary increase in systematic risk and uncertainty regarding the effects of the event, which increases the variance of the returns around the event (Boehmer, 1991). In the following formula the abnormal returns are adjusted for event induced variance:

(Boehmer, 1991) The standardized abnormal returns are then used in the formula below that will produce the test statistics. By dividing the variances all the abnormal returns will have the same distribution and making this test therefore more successful that some of the other tests (Bartholdy, 2007).

(Boehmer, 1991) The test statistic for the cumulative abnormal returns is calculated by summing the tvalues for the 3 days and dividing them by square root of the number of days in the event window, the method as used by Bartholdy et al. (2007) and Brown et al. (1985). In this section the 5 parametric tests that are used in this paper are described and discussed. Each of the tests has its unique properties in order to adjust or to test various samples and measure existing abnormal returns. When performing parametric tests it is assumed that the same is normally distributed. However, in practice a sample is hardly ever a perfect normal distribution, but the closer it comes to a normal distribution the better the tests perform. Therefore a range of tests should be used in an event study including both

Page | 32

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

parametric as non-parametric tests. In the following section the author will discuss and describe the non-parametric tests chosen for this research.

Non-parametric tests

Non-parametric tests should perform better than parametric tests when used on a sample that is normally distributed. Considering that finding a perfectly normally distributed sample from real data is difficult, it is important to include non-parametric tests in order to increase the chance of measuring significant abnormal returns if they are present (Bartholdy, 2007). In this research paper 3 non-parametric tests were chosen, these have also been used by Bartholdy et al. (2007). In non-parametric tests the abnormal returns are exchanged for other variables to conduct the tests on. This way it is possible to capture abnormal returns even if the sample is not normally distributed. The tests with the respective articles they come from are listed below.    T6: Rank test (Corrado, 1989; Corrado, 1992) T7: Sign test (Corrado and Zivney, 1992) T8: Generalized sign test (Cowan, 1992 and Cowan and Sergeant, 1996) In T6, the ranked t-test, each of the abnormal returns is ranked with relation to the other abnormal returns for that specific stock. By ranking the abnormal returns it also introduces the possibility to adjust for missing values.

(Corrado, 1992) Where: Kit: K(Ait) = rank(Ait), t = -249, ….,+1 Ti; Number of observations for stock i The sum of the number of observations under the T6 test is the number of observations divided by two. This is due to that each of the observations is scaled and divided by the total.

Page | 33

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

(Corrado, 1992) Considering that each of the abnormal returns is ranked and divided by the number of observations for each stock the mean for each of the stocks will be ½. This can be seen in the formula below.

(Corrado, 1992) After having calculated the mean it is possible to continue with the T6 test and calculating the test statistics. This formula is similar to some of the parametric tests. The denominator is the standard deviation and the nominator the sum of excess returns divided by the square root of the number of observations. The formula for the T6 test is shown below.

(Corrado, 1992) To calculate the test statistic for the cumulative abnormal return the formula below can be used which is similar to the formula from Brown et al. (1985). The S(U) denotes the standard deviation.

(Corrado, 1992)

Page | 34

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

In the T7, the sign t-test, the abnormal returns are replaced with other values, this are -1’s, 0’s and 1’s. The values are assigned according to whether they are higher lower or equal to the median of abnormal returns of the stock. Higher returns are assigned a 1, lower a -1 and returns equal to the median a 0. In this test the abnormal returns excess return is assumed to have a median around zero (Corrado 1992). The formulas for the sign test are very similar to the rank test. This is mainly due to that the same academic is responsible for them. Also in this formula the mean is a ½, because the median is middle number in a range of numbers and since there should be equally many numbers below and above the median and the values only are 1, 0 and -1, the mean is a ½. The formula for calculating the test statistic is show below:

(Corrado, 1992) Where: Git: Git = Gsign (Ait-median (Ait)), t= -249,…,+1 Gjt: Values for the test day N: Number of observations The Cumulative abnormal returns test statistic can be calculated with the formula below, which is similar to the Brown et al. (1985) and the rank test (Corrado, 1992) CAR formula.

(Corrado, 1992) The last test is the T8, the generalized sign t-test. In this test the abnormal return values are replaced with 1’s and 0’s depending on whether or not the abnormal return is positive or negative. Every abnormal return that is positive is assigned a 1 and the remaining abnormal returns are assigned a 0. This would indicate that the general sign test analyzes
Page | 35

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

whether the stocks have more often positive abnormal return in the event window than the rest of the estimation period. The generalized sign test is more properly interpreted as a test of the median CAR (Cowan, 1992). The formula for calculating the t-statistic is:

(Cowan, 1992) Where: N: Number of observations Wt; Number of positive Ait in the event window

, Where φ = 1 if Ait > 0 and φ = 0 otherwise For calculating the CAR the formula from Brown et al. (1985), Corrado et al. (1992) or the other articles can be used.

Ordinary least squares regression analysis
In the following section the ordinary least squares (OLS) regression analyses and models, which will be used in this research paper, will be discussed and described. The basis for the regression analyses are the models that will be used and the variables that will be included in the model. Choosing the right variables is essential in order to increase the robustness and the quality of the regression analysis. It is also important to have the variables fulfill the assumptions of randomness, independence, normality, having a constant variance and having a residual mean equal to zero (Shelton, 1987). Moreover, as mentioned in the regression analysis section under the choice of method, the form in which these variables can be included can vary (e.g. log, squared, etc.). Another type of variable used in regression analyses is a dummy variable. This is a variable that is assigned a 1 or a 0 depending on whether it fulfills the criteria set for the dummy variable. Dummy variables are very useful when comparing different groups or classes that do not hold a numeric value. Although it should be noted that dummy variables also are useful for when

Page | 36

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

applied on numeric data in which the data is divided in groups, rather than it being used in its unaltered continuous form (Tsou, 2010). However, aside from using normal and dummy variables in the regression analysis, including control variables increases the quality of the regression analysis. Control variables are used in order to capture more variables that explain the abnormal return (Elliott, 2011). An important point to note that distinguishes a control variable from the other variables is that they are not related to the theories or problems that the research is attempting to answer (Spector, 2011). According to Elliot et al. (2011) regression analyses with control variables have improved results when compared to regression analyses without control variables. Therefore a number of control variables will be included in the model. Below the main model is shown that will be used for the ordinary least squares regression analysis. The model includes dummy variables, normal variables and control variables. However, to have a greater chance at finding significant variables, multiple linear regression analyses will be performed with fewer variables includes.

Model1: β4(EBITDAi)

CARi = β0 + β1(Domestici) + β2(Sharesi) + β3(Profit_after_tax) + + β5(Acquirer_assetsi) + β6(Deal_valuei) + β7(Target_ROAi) +

β8(Pharm_Pharmi) + β9(Pharm_Bioi) + β10(Bio_Pharmi) + β11(Bio_Bioi) + β12(Sizei) + ei

Domestic is a dummy variable that will be indicating whether it was regarding a merger or acquisition that is domestic or cross-border. The domestic variables should help in answering whether domestic or cross-border deals have the best performance with regard to shareholder value. Studies have found cross-border mergers and acquisitions to be more value enhancing than domestic deals (Goergen, 2003). The variable shares is used for indicating whether the deal is rumored to be financed by shares or by cash or cash equivalents. It is a dummy variable where 1 indicates that the deal is financed by shares. As discussed in the literature review, it is important to send the right message to the market when a merger or acquisition takes place. Deals that are financed by shares show to the market that the stock is overvalued and that can lead to a drop in stock price (Andrade, 2001; Travos, 1987). This variable will show how the market reacts to rumors regarding mergers or acquisitions financed by shares in the pharmaceutical and biotechnological industry.

Page | 37

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Including the variables Profit_after_tax, EBITDA (Earnings Before Interest Tax Depreciation and Amortization) and Target_ROA will help determining the relation between performance of the target and expected abnormal returns for the acquirer. Each of these variables is included as a normal variable. In some studies it has been shown that performance variables, such as ROA (Return On Assets) and profit, have a negative relation to expected abnormal returns, mainly because the opportunity of turning a poorly managed target around is not existing (Ekkayokka, 2009; Hernando, 2009). Therefore it will be interesting to test whether this is also the case for the sample at hand or whether the results differ. Acquirer_assets is variable representing the assets of the acquirer. This variable is used as a control variable. Deal_value is also used as control variable. This variable represents what the rumored deal would cost. The deal value should have a direct relationship to the observed return, whether negative or positive, which makes it a perfect choice to use as control variable. These are used as control variable because the assets of the acquirer and the deal value are not the focus of the research but are expected to have a relationship to the expected abnormal returns of the deal. Pharm_pharm, pharm_bio, bio_pharm and bio_bio are included as dummy variables to identify the performance of the different mergers and acquisitions with regard to the target and acquirers industry. These are important variables to include to analyze what type of deal is perceived to increase the shareholder value the most. Including these variables in the regression analysis will assist in answering on the research question regarding which type of merger or acquisition, when comparing industries, is perceived by the market to have the highest return. As mentioned under the history section, it is expected that the dummy variable for pharmaceutical firm acquiring a biotechnological firm will have the highest positive relation to the abnormal return. This is because the potential of synergies and covering each other’s weaknesses is greatest here (Roijakker 2005; Malik 2009). Size is a normal variable. This variable is a ratio of how much bigger the acquirer is compared to the target. This variable should help identify the importance of size and if relatively larger firms are more successful in absorbing their targets. This is expected to have a positive relation to the abnormal returns and has been researched in the pharmaceutical industry (Kirchhoff, 2011) as well as other industries (Hernando, 2009; Ekkayokka, 2009). Various studies have seen that relative size has a significant influence on abnormal returns;

Page | 38

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

however the coefficients are varying between different studies (Moeller, 2004). Therefore it is seemed like a suitable variable to include in testing which other variables have an influence on the expected abnormal returns. As stated earlier in the section, aside from the main model for the regression analysis, multiple other analyses will be performed with fewer variables to be able to capture more significant variables and to focus on specific elements. The use of multiple regression analyses has also been seen in various other studies (Higgins, 2006; Kirchhoff, 2011; Hess, 2010). Below the different models can be seen, which include the same but fewer variables than were used in the main model.

Model2: CARi = β0 + β1(Domestici) + β2(Pharm_Pharmi) + β3(Pharm_Bioi) + β4(Bio_Pharmi) + β5(Bio_Bioi) + β6(Acquirer_assetsi) + ei

In model 2 the focus is on the variables that are related to the samples in the event study together with the acquirer’s assets as control variable. The main reason for including this smaller model is to analyze how much influence each of the variables has which were included as sample in the event study, without the influence of other test variables. Furthermore, because the sample included rumors, most of the variables were not complete for each of the firms, making the test sample smaller for model 1 and therefore the results less significant. However, because the tests variables in model 2 are all known for each firm, there are no firms excluded for this, leading to a model that potentially has more explanatory power than a regression analysis with fewer firms included. It is also interesting to see how the explanatory power of model 2 is, because it are these variables that were used for creating sub samples in the event study, thus the results will work complementary on the results of the event analysis if the explanatory power is high.

Model3:

CARi

=

β0

+

β1(Sharesi)

+

β2(Profit_after_tax)

+

β3(EBITDAi)

+

β4(Acquirer_assetsi) + β5(Deal_valuei) + β6(Target_ROAi) + ei

Page | 39

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

In model 3 the focus is on the variables that are not related to the samples in the event study. The results of this test will help answer on the third research question, which firm or deal characteristics have influence on abnormal returns following a rumor. Doing a regression analysis without the domestic variable and the industry variables included can show how the weight of the above variables relates to the cumulative abnormal returns. Also the explanatory power will show how much of the CAR can be explained by the above variables. In the previous sections the author described the methods, models and test that will be used to answer on the main problem statement and the research questions. The methods chosen are an event study and a regression analysis. The event study will help answer on whether there are abnormal returns, if they are significant and in which sub-samples they are related to. For testing whether abnormal returns exist in the sample 5 parametric tests and 3 non-parametric tests will be used in order to increase the chance for success when the sample distribution does not mirror a normal distribution. The regression analysis will put weight behind specific variables in order to identify which variables have the greatest impact on return and whether a combination of different factors is the key to achieving positive abnormal returns.

Page | 40

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Data
In this section of the research paper, the author will describe how and from where the data for the research was extracted, how and why it was filtered, and a description of data final data. The empirical analysis is based on a sample of 63 M&A rumors over the period 01/01/2001-01/03/2012. But how the author composed the sample will be described in the following sections.

Data extraction
The data was extracted from Zephyr, one of the most comprehensive databases for information about various deals (Zephyr, 2012), including mergers and acquisitions. The following criteria were used: 1. Listed/Unlisted/Delisted companies: listed acquirer. Having a listed acquirer is required in order to be able to measure abnormal returns in the stock price. 2. Deal type: Acquisition, Merger. These are the two types of deals that are being analyzed. 3. Percentage of stake: Percentage of initial stake (max: 50 %); Percentage of final stake (min: 50 %). The author wanted to focus on deals were the initial stake in the company was under 50% and the final share was over 50%. This would mean that the acquiring firm would go from no stake or a minority stake in the target firm to a majority stake in the target firm. 4. World regions: Balkan States, Baltic States, Eastern Europe, Scandinavia, Western Europe, European Union enlarged (27) ( Acquiror AND Target ). This criterion was chosen to limit the geographical region to Europe. This is a criterion that both had to restraint the acquirer and the target firm, considering that the scope of the paper only would focus on European mergers and acquisitions. 5. Zephus classification: Biotechnology, Pharmaceuticals and Life Sciences ( Acquiror AND Target ). This was chosen to be able to have the right industry class in the sample. The undesired firms are filtered out at a later stage. 6. Time period: on and after 01/01/2001 and up to and including 01/03/2012 (rumoured, completed, announced, include updated deals, include new deals). The data had to be as recent as possible to be able to add value and new information to the already

Page | 41

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

existing research. Therefore the author chose to use data from 01/01/2001 till 01/03/2012. Under these search criteria 210 deals (mergers and acquisitions) were found. These qualified under the above mentioned conditions. However, it was still required to further filter these down to have a more optimal sample. The next step is to put stock data behind the 210 deals found. This is done by doing a data extraction from Datastream, one of the largest databases with historical financial data (Reuter, 2012). During this extraction both the acquirers company data as well as market data was extracted. The market index used was the Morgan Stanley Composite Index (MSCI) of the acquirer’s country. This index is composed of the 50 largest stocks on the market and is considered to be able to represent the entire countries market; in addition it is also adjusted for stock splits, dividend payments and other irregularities (Morgan, 2012). The reason for using the acquirer’s countries index rather than the industry index, is that if the index consist of the 50 largest firms most of the largest firms engaging in M&A activity would have a material share in the index, which would mean that if the firm’s stock is regressed against the index it is partially regressed against itself.

Data filtering
As for the range of the data extraction, a window of 1 day after the rumor date and 1 year before the rumor date was used. This was after the extraction reduced to 249 trading days before the event and 1 day after the event day. Afterwards the stock price changes for each day were calculated leaving 248 days before the event and 1 day after. This would indicate an estimation period of 250 days and an event window of 3 days. After the stock data associated with the 210 deals was extracted the thickness of the trade was calculated. As mentioned in the methodology section, firms with a trade percentage of 80% and higher are considered thickly traded. All other deals were filtered out to leave a more optimal sample. Furthermore, deals where the either market index or a firm’s stock information was incomplete were also filtered out. This reduced the sample of 210 deals to 127 deals. The next step was to filter the data for deals that had a merger or acquisition within the 250 working days estimation period. This was done by extracting another set of mergers and acquisitions from Zephyr to cross check with. Explanation will be given for the changes in the extraction criteria. The following criteria were used:

Page | 42

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

1. Listed/Unlisted/Delisted companies: listed acquiror. 2. Deal type: Acquisition, Merger 3. Percentage of stake: Percentage of initial stake (max: 50 %); Percentage of final stake (min: 50 %) 4. World regions: Balkan States, Baltic States, Eastern Europe, Scandinavia, Western Europe, European Union enlarged (27) ( Acquiror ). In the new data extract only the acquirer had to be bound by the industry classification. This was done in order to filter out deals that had a merger or acquisition in the estimation period but outside the originally specified geographic region. 5. Zephus classification: Biotechnology, Pharmaceuticals and Life Sciences ( Acquiror ). For the same motive as with the alteration of the geographical parameter only the acquirer had to be bound by the industry classification. This was done in order to filter out deals that had a merger or acquisition in the estimation period but outside of the industry. 6. Time period: on and after 01/01/2000 and up to and including 01/03/2012 (rumoured, completed, announced, include updated deals, include new deals). An additional year was added to be able to cross check in the estimation period before the actual merger or acquisition rumor. After the data extraction was performed and the deals were cross-checked, the sample of 127 was further reduced to 67 deals. Removing the deals with a merger or acquisition in the estimation period would improve the chance of measuring abnormal returns in the event window if they would exist. However, it could also indicate a self-inflicted bias in the sample. Because filtering out these deals might exclude bigger player that regularly make acquisitions or other firms with other properties which could be important to measure. The last step in filtering out deals from the sample was the industry classification, considering that the zephyr search also included life sciences. The following US SIC (standard industry classification) codes were used in the sample:
Table 1: Acquiror SIC codes

Acquiror US SIC code 283 2833

Acquiror US SIC description Drugs Medicinal chemicals and botanical products manufacturing

Industry acquiror Count Pharmaceutical 1 Pharmaceutical 5

Page | 43

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Pharmaceutical 2834 3841 5169 8731 87314953
Source: CD, 2012 Table 2: Target SIC codes

Pharmaceutical preparations manufacturing Pharmaceutical Surgical and medical instruments and apparatus Chemicals and allied products, not elsewhere Pharmaceutical classified wholesale dealing in Biotechnological Commercial physical and biological research Commercial physical and biological research Biotechnological Refuse systems

41 1 2 12 1

Target US SIC code 283 2833 2834 28343841 28347389 28348731 2844 8071 8731
Source: CD, 2012

Target US SIC description Drugs Medicinal chemicals and botanical products manufacturing Pharmaceutical preparations manufacturing Pharmaceutical preparations manufacturing Surgical and medical instruments and apparatus Pharmaceutical preparations manufacturing Business services, not elsewhere classified Pharmaceutical preparations manufacturing Commercial physical and biological research Perfumes, cosmetics and other toilet preparations manufacturing Medical laboratoires Commercial physical and biological research

Industry target Pharmaceutical Pharmaceutical

Count 1 1

Pharmaceutical 42 Pharmaceutical 1 Pharmaceutical 1 Pharmaceutical 1 Pharmaceutical Biotechnological Biotechnological 1 1 14

Table 3: Removed SIC codes

US SIC code US SIC description 2076 Vegetable oil mills, except corn, cottonseed and soybean 2899 Chemicals and chemical preparations, not elsewhere specified manufacturing
Source: CD, 2012

After having classified and filtered the deals for the usable SIC codes the sample was further reduced from 67 to 63 deals. Using codes starting with 283 as classification for pharmaceutical companies combined with 2844, 3841 and 5169 and codes starting with 8731 and 8071 as biotechnological companies, as an appropriate classification (Danzon, 2007; Higgins, 2006). The reason for including codes 2844, 3841 and 5169 is because the secondary codes are starting with 283. These are the final 63 deals that compose the total sample.

Page | 44

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Data description
In the following section the properties of the data sample will be presented and discussed. However, a more detailed outline can be found in the appendix with more specifics than directly presented and discussed in the main paper (e.g. target and acquire country count). In this section the same symbols will be used in the various formulas and will only be specified the first time they are used. Considering that having normalized data will increase the success of the test, it is therefore important to assess whether the data in the sample is normally distributed. A sample that follows a normal distribution is bell shaped, but in reality it is rare to come by a distribution that perfectly follows a normal distribution. Therefore there are different statistics that can describe the data and that will assist in determining whether the data is normally distributed, 4 of them are mainly used. These 4 are the average daily return (mean), the standard deviation, the coefficient of skewness and the coefficient of kurtosis (Boutahar, 2009). The first statistic, the average daily return or mean, will show what kind of µ the data sample has and shows what the average expected daily return is. The mean is calculated by summing up the observations and dividing the sum by the number of observations. E(xt ) = μ (Keller, 2005) The second statistic, the standard deviation, which is denoted as σ, is a measure that indicates how far away the different observations in the sample lay from the mean or µ (Keller, 2005). The greater the standard deviation is, the greater the spread away from the mean. The standard deviation is calculated as the square root of the variance. σ2 = variance(xt) = E(xt - µ)2 σ = variance(xt)½ (Elton, 2011) Where: xt: the observation µ: the mean

Page | 45

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

The third statistic, the coefficient of skewness which is denoted by an S, is a measurement of symmetry. If the coefficient of skewness is positive it would indicate that the distribution of the sample is right skewed having a longer tail on the right and if it is negative it would indicate that it is left skewed having a longer tail on the left. The coefficient of skewness is calculated with the following formula:

(Boutahar, 2009) The fourth statistic, the coefficient of kurtosis which is denoted by a K, is a measurement of flatness. A high coefficient of kurtosis indicates that the distribution is peaked with a steeper decline from the mean than in a normal distribution and a low value would indicate that it is less peaked with a slower decline. The coefficient of kurtosis is calculated with the following formula:

(Boutahar, 2009) After having introduced the different descriptive statistics, the following step is to analyze the descriptive statistics from the sample. The average daily stock return is -0,00028; indicating that on average the return declines. This could be related to that the sample includes deals from January 2001 to March 2012, of which the later deals were done during the financial crisis, where general stock declines were observed (Longstaff, 2010). Furthermore, the sample has a standard deviation of 0,02811; indicating that if the sample was normally distributed 68% of the observations lay between -0,02839 – 0,02783 and that 95% lay between -0,05650 – 0,05594. The coefficient of skewness of the sample is -1,89448; signifying that the sample is left skewed and that the left tail is longer than the right tail. This also indicates that the sample is not perfectly normally distributed. Last, the sample has a coefficient of kurtosis of 86,45831; signifying that is highly peaked with a very steep decline. Another indication of that the sample is not perfectly normally distributed. In cases where the sample is not perfectly normally distributed non-parametric tests should perform better than
Page | 46

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

parametric tests (Bartholdy, 2007), however performing both will yield an even greater chance of success. In the table below a summary of the descriptive statistics of the sample can be found.
Table 4: Descriptive statistics of the stock returns

Number of stocks Average daily return Standard deviation Coefficient of skewness Coefficient of kurtosis

63 -0,00028 0,02811 -1,89448 86,45831

Aside from the statistical properties there are other characteristics of the sample, such as the composition that can be analyzed. The total sample consists of acquirers from 15 countries and targets from 17 countries. The countries with the most deals from the acquirer’s perspective are Great Brittan with 12 and Germany with 9 mergers and acquisitions and from the target’s perspective are Great Brittan with 18 and France with 9 mergers and acquisitions. This could possibly indicate that there are industrial trends in these countries or it might be related to the size of the country. To answer the first research question, whether there significant differences in abnormal return between domestic and cross-border mergers and acquisitions, it is important to look at the how the sample is split with regard to that. The sample consists of 28 domestic deals and 35 cross-border deals. The countries with most domestic deals are Great Brittan with 9 deals and France, Switzerland and Germany with each 4 deals. As for the countries representing the cross-border sample, the main acquiring countries are Belgium with 7 deals and Germany with 5 deals and the main target countries are Great Brittan with 9 deals and France with 5 deals. Furthermore, to answer the second research question, whether there significant differences in abnormal return between inter-industry and cross-industry mergers and acquisitions, how the sample is composed regard to industry classification. Both the acquirers and targets are divided into pharmaceutical or biotechnological firms creating 4 categories. The pharmaceutical firms represent 40 of the deals as acquirer and 48 as target and the biotechnological firms 13 of the deals as acquirer and 15 as target. Of these deals 47 are interindustry and 16 are cross-industry. This could indicate that companies are mainly involved in mergers and acquisitions within their own industry. Also, the majority of the sample consists
Page | 47

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

of pharmaceutical firms. As mentioned earlier, because often pharmaceutical companies are cash in assets but are facing expiring patents, they are the perfect choice to engage in acquisitions and biotechnological companies that often have a great potential but lack the resources would be a perfect target (Roijakker 2005; Malik 2009). The sample only includes 9 deals in which a pharmaceutical company is rumored to acquire a biotechnological company, which is less than the 41 deals were pharmaceutical companies acquire another pharmaceutical company. However, the number of deals can only be used as indication if there are possible trends, the test will show whether one type of deal outperforms the other. More specifics regarding the acquirers and targets can be found in the appendix. To sum up the above, the initial sample of 210 deals was reduced by excluding deals that did not have tickly traded stocks, lacked data, had a merger or acquisition rumor in the estimation period or did not match the industry characteristics. This resulted in a sample of 63 deals. The descriptive statistics of the sample indicated that the sample is not perfectly normally distributed; it is highly peaked and is skewed to the right. Therefore it is critical to include non-parametric test in order to increase the chance of measuring abnormal returns if they are present.

Page | 48

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Analysis
In this section the results from the various tests will be presented, analyzed and possibly linked to existing literature. All of the information from the tests is on the attached CD, with the most important data added in the text in the form of tables. Considering that most of the references made are to the calculations and the data, the references in the analysis section will only be specifically mentioned if it is to other literature, it can therefore be considered that when a specific reference is missing in the analysis section it is regarding the data on the attached CD (CD, 2012). Also, considering that not all the data is equally important to have in the text, the reader can always consult the appendix and CD for details on the calculations and miscellaneous statistics. The analysis will be split into two parts. In the first part the results of the event study and the different t-tests will be presented and analyzed. In the second part the results of the regression analysis are presented and discussed. However, before starting on both analyses the properties of the abnormal returns will be presented and discussed.

Abnormal returns analysis
In this section the abnormal returns are presented and discussed. In the table below an overview and the descriptive statistics of the abnormal returns can be seen. This is an overview of the abnormal returns of the total sample.
Table 5: Descriptive statistics of the abnormal stock returns

Number of stocks Average daily abnormal return Standard deviation Coefficient of skewness Coefficient of kurtosis

63 0,00000 0,02620 -2,17628 112,61872

The average daily abnormal return is rounded to 0 with a precision of 17 decimals. This could also be expected, considering that the abnormal return is estimated per day based on how much higher that specific day’s return is compared to the total data range. This would imply that by taking the average, all abnormal returns should net out against each other, bearing in mind that the total data sample should be normal for there to be abnormal returns on specific days. The standard deviation is 0,026 and with a mean of 0 it would indicate that 68% of the abnormal returns lay between -0,026 and 0,026 and 95% lay between -0,054 and
Page | 49

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

0,054. This shows that most of the observations are very close to 0 which is not unlikely, considering that abnormal returns are returns that are not expected and should therefore not be likely to have many significant abnormal returns; otherwise they would be normal rather than abnormal. The coefficient of skewness is -2,176 indicating that the sample is left skewed and that the left tail is longer than the right tail. That is because the median is higher than the mean, therefore having more observations left from the mean than right. Considering that the median is higher than the mean and the mean of the sample is 0, there are more observations that are positive than negative. However, because the mean is 0 it indicates that the in the sample the chance of positive abnormal returns is higher than negative abnormal returns, but the positive abnormal returns on average are lower than the negative abnormal returns. The coefficient of kurtosis is 112,619 indicating that the data is very peaked. This should not be a surprise because in most days there should not be any significant abnormal returns, which resulting in that most of the observations are around zero. The abnormal returns for each of the days of the event window and the cumulative abnormal returns for each of the samples are presented below.
Table 6: Abnormal return overview

Sample Total Domestic Cross-border Pharmaceutical Biotechnological Pharmaceutical-Pharmaceutical Pharmaceutical-Biotechnological

-1

0

+1 0,55%

CAR 1,07%

-0,27% 0,80% -0,64% 0,92% 0,02% 0,71%

-0,08% 0,19% 1,05% 0,54% 1,78% 1,57% -0,83% 1,82%

-0,04% 1,07%

-1,17% -0,22% 0,56% 0,29% 0,64% 0,88%

-1,56% 2,99% 0,04%

-1,02% 0,41% 1,15% 1,93% -3,19%

Biotechnological-Biotechnological 0,74% Biotechnological-Pharmaceutical

-2,80% -0,44% 0,06%

As can be seen from the table, all samples except for the biotechnologicalbiotechnological samples have both higher abnormal returns on the day of the rumor and the day after the rumor compared to the day before the rumor. This could be explained with by the fact that, unlike an announcement or a completion of a merger or acquisition, a rumor is less predictable decreasing the chance of the market anticipating the circulation of a rumor
Page | 50

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

before it happens. This is also in line with some of the other studies (Higgins, 2006) where the abnormal returns are more significant in the period at or after the announcement than the period up to the announcement. It can also be seen that the cross-border sample has a higher abnormal return than the domestic sample. In the literature review it is stated that from first glance cross-border acquisitions look more profitable than domestic, even though the long run returns show otherwise. This is often because there are complications associated with cross-border acquisitions such as cultural or regulatory difficulties (Renneboog, 2006) which outweigh the benefits. It also appears that the inter-industry samples (pharmaceutical-pharmaceutical and biotechnological-biotechnological) have higher abnormal returns than the cross-industry samples (pharmaceutical- biotechnological and biotechnological-pharmaceutical). Eckbo et al. 1992 found similar results, where inter-industry mergers and acquisitions had higher abnormal returns than those cross-industries. Campa et al 2004 argues that firms that are operating in the related industries have better results than those from unrelated industries. Depending on how related industries are interpreted, it can be considered different interindustry mergers in which the firms have different product areas to be related or to consider pharmaceutical and biotechnological industry to be related. Contrariwise, most studies found that cross-industry mergers and acquisitions should perform better (Arora, 1990; Kirchhoff, 2011; Malik, 2009), which is not the case here. From the table can also be seen that the pharmaceutical acquirer samples is having higher abnormal returns than the biotechnological acquirer sample. This is not unexpected, considering that in the literature review and in the history section it was stated that other studies (Malik, 2009; Kirchhoff, 2011) found that biotechnological firms often were very innovative but cash poor and that pharmaceutical firms often were cash rich but were missing the innovation to close the gaps in the product pipe line. Therefore the market could perceive mergers and acquisitions with a pharmaceutical firm as acquirer more value enhancing. However, the tests will show whether or not any of these abnormal returns is significant and at which significance level. This will follow in the subsequent sections of the analysis.

Event study results and analysis
In this section the results of the different t-tests will be presented and discussed, furthermore an overview will be presented of the results of the t-test with regard to the
Page | 51

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

cumulative abnormal returns (CAR), in the appendix the overview for each of the days in the event window can be found. The results of the t-tests should help with answering the first and second research question, whether there are significant abnormal returns in the abnormal return following a rumor, between domestic and cross-border mergers and acquisitions and between inter-industry and cross-industry mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe. Answering on these questions will help in finding the answer to the main problem statement. In the table below the results of each of the tests for each of the samples are presented. Each of the probability values have been calculated adjusting for the size of each of the sub samples (CD, 2012).
Table 7: Cumulative abnormal return (CAR) test statistics and probability values

Sample Total

Test T-stat P-value T-stat

T1 1,78 8,08% 0,18 85,70% 2,68 1,15% 2,70 0,96% -0,45 66,47% 3,08 0,38% 0,23 82,51% 0,84 46,06% -1,14 31,67%

T2 2,10 4,00% 2,17 4,01% 1,96 5,85% 2,48 1,68% 1,42 18,51% 2,19 3,46% 2,82 3,02% 1,32 27,75% 1,46 21,88%

T3 3,34 0,15% 0,87 39,12% 3,70 0,08% 3,60 0,08% 0,29 77,56% 3,67 0,07% 0,65 53,99% 0,78 49,20% -0,32 76,25%

T4 1,17 24,77% -0,35 73,15% 2,35 2,51% 1,67 10,08% -0,48 64,04% 2,69 1,07% -0,60 56,98% 1,22 30,84% -1,46 21,80%

T5 2,37 2,09% 0,37 71,78% 2,72 1,04% 2,52 1,51% 0,02 98,20% 2,52 1,59% -0,09 93,21% 0,88 44,15% -0,62 56,78%

T6 -1,78 8,05% 0,03 97,51% -2,61 1,36% -1,86 6,87% -0,25 80,71% -1,97 5,64% -0,17 87,21% -1,14 33,59% 0,77 48,35%

T7 1,25 21,58% -0,54 59,38% 2,40 2,23% 1,05 30,04% 0,80 44,43% 1,07 29,31% 0,18 86,23% 0,87 44,86% 0,22 83,75%

T8 2,01

CAR 1,07%

4,91% 0,75 46,09% 2,03 1,78% 5,12% 2,13 3,88% 0,25 -0,83% 80,64% 2,38 2,23% -0,07 0,41% 94,58% 0,69 53,82% -0,30 -3,19% 77,95% 1,93% 1,82% 1,57% 0,19%

Domestic P-value T-stat Cross-border P-value T-stat Pharmaceutical P-value Biotechnological P-value PharmaceuticalPharmaceutical PharmaceuticalBiotechnological BiotechnologicalBiotechnological BiotechnologicalPharmaceutical T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat

When looking closer at the test statistics for each of the samples, it can be seen that both the parametric as nonparametric tests are producing test statistics that are significant and that they vary between samples. It is for that reason that it was critical to include a battery of tests (Bartholdy, 2007) in order to capture any abnormal returns if they are present, because each sample is unique and different. Therefore including adjustments or having different
Page | 52

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

methods (i.e. different tests) is needed in order to capture results in these different samples. There is no single test that outperforms the others. In the following sections each of the samples will be analyzed and discussed with focus on answering the research questions.

Total sample

The first statistics that will be analyzed and discussed are those related to the total sample, considering that this will help answer on the main problem statement, whether or not there are any abnormal returns following a rumor. The abnormal returns for each of the days in the event window -1, 0 and +1 are -0,27%; 0,80% and 0,55% respectively, with a cumulative abnormal return of 1,07%. The CAR for the total sample is significant at 5% in 4 tests and at 10% in 6 tests. That is an indication that the CAR of 1,07% is significantly different from 0 and that the CAR are positive would indicate that the rumor about a merger or acquisitions within the European pharmaceutical and biotechnological industry is creating significant value. These results are in line with the findings of other studies (Higgins, 2006; Kirchhoff, 2011), even though most studies find that mergers and acquisitions are not creating value (Danzon, 2003) or are value destroying (Sorescu, 2003). However, it should be noted that in these other studies the focus was not on the rumors but on announcements and completions of mergers and acquisitions. When looking at each of the days in the event window for the total sample (Appendix; CD, 2012), it can be seen that the abnormal returns on event day -1 are not significant at 10% level in any of the test; event day 0 has abnormal returns that are significant at a 10% level in 7 of the tests and at a 5% level in 4 of the test and event day +1 has abnormal returns that are significant at a 10% level in 2 of the tests and at 5% in one of the tests. This confirms the earlier statement which was in line with Higgins et al. (2006), that in most studies abnormal returns only occur on the day of or the days after the announcement, in this case the rumor. Furthermore, it is mainly the event day that has significant abnormal returns and contributes most to making the CAR in the event window significant. It could possibly be explained by the fact that the event is a rumor, which at the time it circulated the market caused the market to react. But a rumor is hard to anticipate resulting in the -1 day in the event window not to have significant abnormal returns and after the rumor hit the market there is no consequence for the firm’s returns aside from a possible lag that caused the +1 day in the event window to also have significant returns. However, considering that the total sample consists of

Page | 53

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

domestics, cross-border, inter-industry and cross-industry mergers and acquisitions, it is important to analyze each of the sub samples. This is critical, because the abnormal returns could possibly be attributed to a smaller sub sample, which possibly had a significant influence on the total sample. Therefore in the following sections the different sub-samples will be discussed, starting with the domestic and cross-border samples and finishing with the industry samples.

Domestic and cross-border samples

The second set of statistics that will be analyzed and discussed are the statistics regarding the domestic and cross-border samples. This will help answer on the first research question, whether or not there are differences in abnormal returns between the domestic and cross-border rumors of mergers and acquisitions. The first sample that will be looked at is the domestic sample. The domestic samples has abnormal returns in the event window of -0,64% on day -1; 0,92% on day 0 and -0,08 on event day +1, this gives an cumulative abnormal returns of 0,19%. The cumulative abnormal return of the domestic sample is not significant at a 10% level in all the tests. This would indicate that over the 3 days in the event window, the abnormal returns are not significantly different from 0. Also, when the abnormal returns for each day in the event window are tested on, it does not produce test statistics that are significant at a 10% level. Therefore we can state that based on these results, the abnormal returns are not statistically different from 0. The following sample that will be analyzed is the cross border sample. The cross border sample has abnormal returns of 0,02%; 0,71% and 1,05% respectively on event days 1, 0 and +1. The cumulative abnormal return is 1,78%; which is significant at a 10% level in all the tests and in 6 of the tests on a 5% level. This is an indication that the rumors of crossborder mergers and acquisitions within the pharmaceutical and biotechnological industry having a significantly positive influence on shareholder wealth. This can also be seen when analyzing each of the days in the event window. The abnormal return on event day -1 is not significant in any tests at a 10% level. This is in line with the results of the total sample, which was explained that the anticipation of a rumor is smaller than the anticipation of an announcement or an actual merger or acquisition; therefore the results of the days before the rumor are less significant than in other studies. However, event day 0 is significant at a 10% level in 6 of the tests and at a 5% level in 3 of the tests. Furthermore, event day +1 had

Page | 54

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

significant abnormal returns at a 10% level in 5 of the tests and at a 5% level in 4 of the tests. Thus event day 0 and +1 were significant in some of the test, while the cumulative abnormal returns were significant in all of the tests and in most at a 5% level. This is explained by Higgins et al. (2006), where it is stated that by increasing the event window the significance will increase marginally as well and going for 1 event day to an event window of 3 days is an increase of the number of days were the abnormal returns are tested on. This results in better performance in the tests. The abnormal return results in the domestic and cross-border sample are in line with what was discussed under the abnormal return section of the analysis, when purely the abnormal returns were discussed without looking at significance levels, in which the CAR for the cross-border sample was stated to be twice as high as for the domestic sample. However, it does conflict with most of the literature discussed, because domestic acquisitions should perform better than cross-border acquisitions, mainly because there are difficulties cultural and regulatory difficulties that arise with cross-border mergers and acquisitions (Hassan, 2007; Renneboog, 2006). Mangold et al. (2008) found that domestic acquisitions often are more expensive than cross-border, therefore when a rumor hit the market with a possible domestic merger or acquisition, the first reaction of the market could be negative if the market only looked at the short term gains, while ignoring the long run difficulties. While the above results can help answer on the first research question, performing the OLS regression and analyzing the outcome will be complementary to the results of the event study. Therefore the first research question will be answered after presenting and analyzing the results from the regression analysis.

Industry samples

The third set of statistics that will be analyzed and discussed are the statistics regarding the industry samples. This is done in order to find the answer on the second research question, whether or not there are differences in abnormal returns between the interindustry and cross-industry rumors of mergers and acquisitions. The first two samples that will be analyzed are the pharmaceutical acquirer sample and the biotechnological acquirer sample, after which each of the smaller sub samples will be analyzed were both the acquirer and the target are known.

Page | 55

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

The pharmaceutical acquirer sample has an abnormal return of -0,04% on event day -1; 1,07% on event day 0 and 0,54% on event day +1, which adds up to an cumulative abnormal return of 1,57%. The CAR of 1,57% is significant at a 10% level in 6 of the tests and in 5 of the tests at a 5% level. This is an indication that there are significant positive returns following a rumor when the acquirer is a pharmaceutical firm. This can be explained with that pharmaceutical firms often are cash rich, therefore the market might find it more justifiable to use funds to acquire another firm (Kirchhoff, 2011; Malik, 2009). When looking at each of the event days the results seem to be similar with the other samples. Event day -1 is not significant at a 10% level in any of the tests, which can be explained in the same way as stated earlier, that it is not likely a rumor will be predicted. The abnormal return of event day 0 is significant at a 10% level in every test and is significant at a 5% level in 3 of the tests. However, event day +1 is only significant at a 10% level in 2 tests of which only 1 remains significant at a 5% level. Showing the quick but short lasting effect of the abnormal returns after the rumor circulated the market, resulting in event day 0 being significant in more of the tests than day -1 and +1. The sample with the biotechnological acquirer was, as discussed before, one of the only samples in which the CAR was negative. However, even though the CAR is negative it is not significant in any of the tests indicating that the CAR are not statistically different from 0, which would be an indication that a rumor of a merger or acquisition with a biotechnological firm as acquirer does not significantly have any effect on shareholder wealth. The results for each of the days -1, 0 and 1 is respectively -1,17%; -0,22% and 0,56%. These are, just as the CAR for this sample, all insignificant at a 10% level. This strengthens the argument that there are no significant results in this sample. This would indicate that the effect, from a rumor of a merger or acquisition circulating the market, is stronger when it is regarding a merger or acquisition were a pharmaceutical firm is the acquirer than were a biotechnological firm is the acquirer, based on the amount of tests in which the results were significant. Furthermore, the biotechnological acquirer sample had results that were not significantly different from 0 and the pharmaceutical acquirer sample had results that were significantly positive. However, the results could also have been affected by the type of target that was acquired in each of the samples. Therefore it is important to perform an analysis of each of the sub samples in which both the acquirer and target is identified, in order to find out which combination has the best and worst significant returns.

Page | 56

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

The first sample with both the acquirer and the target specified is the pharmaceuticalpharmaceutical samples. This is an inter-industry sample where both the target and the acquirer is a pharmaceutical firm. The abnormal returns in this sample were progressive, 0,29% on day -1; 0,64% on the event day and 0,88% on day +1 leading to a CAR of 1,82%. The CAR of 1,82% is significant at a 10% level in 7 of the tests and at a 5% level in 6 of the tests. The event day -1 abnormal return is significant at 10% in 1 test, which is so far the first sample with a significant event day -1 abnormal return. However, because it is only significant in 1 of the tests and only at a 10% level, it is questionable whether this abnormal return truly is abnormal. The event day abnormal return is significant at 10% in 3 of the tests and only in 1 of the tests at 5%. Event day +1 is significant at 10% in 5 of the tests and 5% in 2 of the tests. What is noteworthy here is that the 5 tests in which the abnormal returns was significant were parametric tests. However, the CAR is significant in 2 out of the 3 nonparametric tests and had better results in the tests than each of the days. This is a similar case as when the cross-border sample was discussed and analyzed and explained with that a larger window (i.e. going from 1 day to 3 days) can increase the significance level marginally (Higgins, 2006). Furthermore, the positive significant CAR results in this sample can be explained with the acquirer being a pharmaceutical firm. This was discussed earlier in the section and it was concluded that the cash richness of pharmaceutical firms make them good acquirers and that the market is more positive towards rumors with a pharmaceutical firm as acquirer. Eckbo et al. (1992) found supporting results, that inter industry mergers and acquisitions had better performance than cross-industry. However, this cannot be concluded based on the first sample. The biotechnological-biotechnological sample is the other interindustry sample which is important to make a conclusion here, and the cross-industry samples have to be analyzed in order to compare the results. The second sample with both the acquirer and the target specified is the pharmaceutical-biotechnological sample, which is a cross-industry sample. According to the literature this sample is expected to have the best positive returns due to the complementary properties of the pharmaceutical firm as acquirer and the biotechnological firm as target (Malik, 2009). When looking at the CAR in this sample of 0,41% this does not seem to be the case. The CAR is also only significant in 1 of the tests at 5%. However, as was the case in some of the other samples, looking at each of the days in the event window might give more insight. Event day -1 has an abnormal return of -1,56% and is insignificant in every test, signifying that it is not statistically different from the other returns in the sample and the

Page | 57

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

return on day -1 not being abnormal. The return on the event day of 2,99% is the highest abnormal return of any day in any of the samples, which could be an indication that the market is very positive when a rumor of a merger or acquisition with a pharmaceutical firm as acquirer and biotechnological firm as target hits the market. However, it is only significant in 3 of the tests at 10% and in 2 of the tests at 5%. Indicating that while the results are high, they are abnormal in some of the tests compared to the other days in the sample. On event day +1 the abnormal return is -1,02, which is just as day -1 in this sample, insignificant in all of the tests. This is not unexpected, as in most of the other samples the event day was the day that was significant in most of the tests, considering that the event is a rumor. The third sample with both the acquirer and the target specified is the biotechnological-biotechnological sample, which is the second inter-industry sample. This sample has a cumulative abnormal return of 1,93% which is the highest positive abnormal return of any sample. But while being the highest positive abnormal return, it is insignificant in every test. This could possibly be explained by the fact that the sample is relatively small compared to the total sample, requiring higher abnormal returns to be considered significant. Also each of the days in the event window of this sample are insignificant in every test at a 10% level. Considering that each of the abnormal returns was positive (0,74%; 0,04%; 1,15%), which would have resulted in the CAR being more significant than any of the days as was seen in the other tests, it is not unexpected that the event days are not significant. This strengthens the argument that the average returns of this sample are not significant. The results from this sample are not in line with the results seen from the first inter-industry sample. The fourth sample with both the acquirer and the target specified is the biotechnological- pharmaceutical sample, which is the second cross-industry sample. This sample has an abnormal return of -3,19% and is the sample with the lowest abnormal return. However, even though it is the lowest abnormal return and the absolute value of the return the highest of any sample, it is not significant at a 10% level in any of the tests. This could possibly be related to that this sample is relatively small compared to the total sample and some of the less specific samples. The event day’s abnormal return (-0,44%) in this sample and the event day +1’s abnormal return (0,06%) are not significant at a 10% level in any of the tests. However, the abnormal return of event day -1 of -2,80% is significant at 10% in 2 of the tests. Considering that in the other samples the event day -1 seemed to be less related

Page | 58

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

to the event than the other tests, it could be possible that the significant abnormal return on event day -1 is, while significant, not related to the event.

Event study summary

The abnormal returns of the total sample and the sub samples have been analyzed and discussed. Because each of the sub samples is smaller than the total sample, some of the sub samples had high abnormal returns but were not significant. This can be explained by the higher required t-statistic in order to have significant returns as adjustment for the smaller sample. Furthermore, it could be seen in most of the samples that the abnormal returns were on the event day. This was possibly due to the nature of the event, since a rumor gives a different market reaction than an announcement or completed merger or acquisition. A rumor appears to have no anticipatory effect and the lasting effects are also smaller than with an announcement. Also the results in each of the samples were not always following the majority of the literature. There were positive abnormal returns in the total sample which have been seen to a less degree than negative or none existing abnormal returns. Also the cross-border sample had better results than the domestic sample, even though the literature would suggest that the domestic sample should have better results. This was possibly explained by looking at the nature of the event, being a rumor, the market might focus on the short term gains while ignoring the long term complications. The industry acquirer samples had significant returns that did not come as a surprise, with the pharmaceutical sample having positive significant return. Furthermore, if purely focusing on event day, then the abnormal return of the pharmaceutical-biotechnological sample had the highest abnormal return. This was also expected based on the existing literature. However, before any final conclusions can be made, the results of the regression analyses have to be analyzed and discussed.

OLS Regression results and analysis
In the following section the results of the regression analyses will be discussed, analyzed and linked to the existing literature. The analyses from this section will complement the results of the event study for the first and second research question and are used for answering the third research question. The calculations and detailed statistics of the regression analyses can be found on the attached CD (2012). However, there is an overview of the most important results presented in the text, which can be seen below. In the table the variables are listed with their attribution to the cumulative abnormal return together with an
Page | 59

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

indication of the significance level. Furthermore the explanatory power, R 2, and the sample size, N. The reason for variations in the sample size (N) in the different models is that some of the variables were incomplete for some of the deals, which resulted into having to exclude those from that model. As was discussed in the methodology section, 3 models have been made which include the same variables in order to perform different tests. The models will be analyzed in their numeric order.
Table 8: Regression analyses overview

Variable Intersection variable (β0) Domestic Shares Target_EBITDA (€000) Target ROA Acq_Assets (€000) Deal_Value (€000) Size Pharm-Pharm Pharm-Bio Bio-Pharm Bio-Bio R2 N
** Denotes significance at a 5% level * Denotes significance at a 10% level Source: CD, 2012

Model 1 -0,1799** -0,02688 0,08389**

Model 2 -0,01527** -0,02578**

Model 3 0,01566**

0,04246 -2,138E-06 6,354E-07 0,01333 -6,150E-10
*

Target_Profit after tax (€000) -2,981E-06** 6,739E-07 0,08647** 2,856E-08* -2,131E-09 -6,502E-4 0,2139** 0,3134** 0** 0,2280** 95,75% 17 0,0503** 0,0261 0,0308 0** 19,61% 60

1,714E-08 -1,374E-09 -1,415E-4

25,39% 17

Model 1 is the model that has all 12 variables included of which of which 6 dummy variables, 4 linear variables and 2 linear control variables. The model has 9 variables which are significant at a 10% level of which 7 are also significant at a 5% level. This is the model that has the highest explanatory power, over 95%, which would indicate that more than 95% of the CAR can be explained with the variables that have been used in this model. The higher the explanatory power the better the model. However, due to the small sample size that was used for this model it can be questioned whether the explanatory power truly is over 95%, as
Page | 60

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

the large amount of variables and the small amount of deals included in the sample potentially could have caused abnormal returns to be attributed to variables that they were not meant for. However, since this is only a speculation and no results prove this, it will be assumed that each of the variables is explaining that what it is meant for. The variables will be analyzed in the order they are listed in the overview. The intersection variable is a static variable which is independent of the input of the other variables, it only varies between samples as the samples composition and the variables in the sample vary. The first variable to analyze is the domestic dummy variable. The attribution of having a domestic deal is -0,027% to the CAR. The statistic is not significant at a 10% level, nevertheless, the negative sign in front of the percentage indicates that while not significant domestic factor has a negative attribution to shareholder wealth following a rumor. As discussed in the event study section of the analysis, this could possibly be explained with the short term benefits of cross-border mergers and acquisitions, such as lower price, while the long term effects are ignored (Mangold, 2008). The second variable is the shares dummy variable. As was discussed in the literature review a deal which is financed with shares should show to the market that the acquirer’s firm’s shares are overvalued (Andrade, 2001). However, in this model the variable is significantly positively influencing the CAR. This is not in line with the majority of the literature. It could possibly be explained with the fact that the event is a rumor. Therefore the uncertainty about the way of payment might have a different influence on the market perception, than when an actual announcement would be made that an acquisition is taking place which is paid with shares. The next three variables that will be analyzed are the target’s profit; the target’s EBITDA and the target ROA, as this are all indications of the target’s performance. The target’s profit is significantly negatively influencing the CAR. While the literature differs on whether or not a target that has a high profit is a good target for acquisitions, the negative influence can be explained. The argument here is that a profitable firm gives less potential for the acquirer, because if the firm was less profitable, the acquirer could turn the firm around and make it profitable, creating shareholder wealth in the process. The target EBITDA has a positive but not significant influence on shareholder wealth following a rumor. This could be related to that the EBITDA is an indication of the amount of revenue a firm has but is not necessarily an indication of how profitable a firm is. Therefore the EBITDA can be seen as

Page | 61

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

the market share the firm is acquiring and might therefore be positively influencing shareholder wealth, considering that the acquiring firm will have more market control in the amount of the added EBITDA. The target ROA is the net profit divided by the assets of the firm. This variable is significantly positive at a 5% level. This is not in line with the literature discussed (Ekkayokka, 2009; Hernando, 2009), considering that a firm with a high ROA has less potential for the acquirer to optimize the target firm. The acquirer’s assets as well as the deal value are significant in the analysis at a 5% level. While these results are significant, the variables were included as control variables and will therefore not be analyzed as deeply as the other variables. As can be seen from the acquirer’s assets, this is significantly positively influencing the shareholder wealth. That would be an indication that a large acquirer is more value creating than a small acquirer, which can be explained by the argument that it is easier for a large acquirer to absorb a target firm than for a small acquirer (Kirchhoff, 2011; Hernando; 2009). However, the author would argue that it is relative size that is important here for analyzing this, which is why the size variable has been included and this variable as control variable. The deal value is significantly negatively affecting the abnormal return. This could be explained by the argument that the more is paid or will be paid for the target, the less upside there is left for the acquirer. Therefore it should not come as a surprise that this variable has a negative influence on the abnormal return and is one of the reasons that it has been chosen as control variable. The relative size of the acquirer compared to the target is negatively affecting the CAR, but is insignificant. Moeller et al. (2004) found that acquisitions by smaller firms often have better results than by larger firms, that could explain why the sign in front of the size variable (-6,5E-4) is negative. Another noteworthy statistic is the biotechnological-pharmaceutical variable. This variable has a contribution of 0 but is highly significant even at a level of below 1%. This is mainly due to the 4 industry dummy variables. Considering that there are 4 groups of acquirer-target combinations, it was only required to be 3 variables to divide these in these groups (e.g. just as there is only 1 variable to divide deals in domestic or cross-border). Therefore the attributions of these industry variables have to be seen relative to each other, as it is certain that having a bio-pharm deal has an influence on the CAR whether or not it is significant. As a result of this the bio-pharm variable is 100% correlated with the intersection

Page | 62

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

variable and the model does not suffer any loss in explanatory power from excluding one of the industry variables as the model only requires 3 to remain functional. However, removing 1 of the industry statistics would make the intersection variable insignificant. The attribution of each of the industry variables is: 0,21 for the pharm-pharm sample; 0,31 for the pharm-bio sample; 0 for the bio-pharm sample and 0,23 for the bio-bio sample. That would indicate that, when comparing these variables relative to each other, the pharm-bio sample adds 0,31% more to the CAR than the bio-pharm sample. That is in line with the literature in which it is argued that biotechnology firms are a perfect target for a pharmaceutical firm (Malik, 2009; Kirchhoff, 2011), and that can be seen here as the contribution is higher than the other variables with each of the industry variables being significant. Furthermore, as was seen from the results in the event study, the deals which had a biotechnological firm as acquirer and a pharmaceutical firm as target had the lowest abnormal returns (-3,19%), which explains why the contribution of the bio-pharm sample is relatively the smallest compared to the other industry samples. The results of model 1 have been discussed up to this point; the next model to discuss is model 2. Some of the main differences between model 1 and 2 is the sample size, which is related to having more complete variable data but fewer variables. Furthermore, the explanatory power of the model is greatly lower, only 19,61% is explained in model 2 compared to 95,75% of model 1. Moreover, even though less of the same variables have been used the results are different on some of the variables. In model 2 the domestic variable has similar results as in the first model, a coefficient of -0,026 indicating that there is a negative relation between domestic rumored deals and shareholder wealth. However, in this model the domestic variable is significant at a 5% level. That could be explained by the fact that there are fewer variables, so there is more weight put on the variables in the model, making some of the insignificant variables significant. But with the lower explanatory power of the model, the significance in the second model indicates that it is only significant when it explains a smaller part of the car. The control variable, the acquirer’s assets, has a negative relation to shareholder wealth in this model, compared to the first model, and in this model the variable is insignificant. This could possibly because the variable only is significant in combination with one or more of the variables that have been excluded in this model.

Page | 63

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

The last variables to discuss in this sample are the industry variables. Similarly to the first model, there is 1 of the variables which has a coefficient of 0 which is highly significant. This did not come as a surprise after having analyzed the variables in the first model. However, aside from the similarity with the coefficient of 0, the variables are not all significant in this model and the factor of contribution to the CAR is not the same as in the first model. In the first model the contribution of the different variables was 0,21%; 0,31%; 0% and 0,23% for the respective samples of pharm-pharm, pharm-bio, bio-pharm and biobio. In the second model the variables have coefficients of 0,050%; 0,026%; 0,030% and 0% for the respective samples pharm-pharm, pharm-bio, bio-pharm and bio-bio. That shows that the relative relationship of the variables has changed since the order of lowest to highest contribution is not the same. This, in combination with the low explanatory power, could be an indication that the variables that have been left out in this model were important for making an accurate model for predicting CARs. Model 3 has some of the same variables as model 1, but with the focus on those variables that did not have a sample dedicated to them in the event study. This is done to analyze if the CAR could be explained by other factors than those used in the event study. However, model 3 has no significant variables and the explanatory power is only 25,93% compared to model 1’s 95,25% explanatory power. That would indicate that no strong grounded conclusions can be made purely based on the results from the third model. However, it is still possible to analyze the results and compare them with the results from the first model, as they do work complementary. Each of the results in the third model has the same sign in front of the coefficient. This is an indication that the relation to the CAR is the same as in the first model. Furthermore, the magnitude of the coefficient is within the same precision. Aside from these findings it is not value adding to analyze each variable in details as none of them is significant and they are similar to the ones in the first model.

Page | 64

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth Regression analysis summary

The different regression analyses have produced a number of significant and insignificant variables which will complement the results of the event study and will help in answering on the three research questions. It was similar in the different models that the domestic variable had a negative relationship to the CAR, which was in line with the findings from the event study. Furthermore the shares variable had a positive relationship to the CAR. This is not in line with the literature and could possibly be attributed to the nature of the event, it being a rumor, and therefore the method of payment is not fixed and therefore not as influential as in an announcement. The profitability variables; target profit, target EBITDA and target ROA; had mixed influences on the CAR. Profit had a negative influence on the CAR while EBITDA and ROA had a positive influence. This could possibly be explained by the argument that a high profit will leave little space for optimizing the target, while high EBITDA and ROA are factors of market share and efficiency without necessarily being profitable. Relative size had a negative relation to the CAR, indicating that firms that are closer to each other’s size are more value enhancing than when the acquirer is greatly larger than the target. This was explained by some of the literature. However, the variable was insignificant and it is therefore difficult to make grounded conclusions based on the results. Last, the industry variables had mixed results in the different modes. However, the results from the first model had the highest significance levels and the highest explanatory power; the high explanatory power might be related to the few observations with a large set of variables, which could lead to variables explaining more or other things than what they were created for. Therefore the results of that analysis of each test will be used with keeping a critical eye on the explanatory power and variables when formulating an answer on the research questions.

Page | 65

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Problem statement and research questions answered
Now that the results of the event study and the regression analyses have been calculated, analyzed and discussed; it is possible to answer the research questions and in turn the main problem statement.

1. Are there significant differences, in abnormal return following a rumor, between domestic and cross-border mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe?

The results from the event study indicated that the rumors regarding domestic mergers and acquisitions were not significant. Furthermore, the regression analyses showed that having a domestic rumor would negatively influence the CAR and that variable was significant in one of the models. The cross-border sample had different results in the event study. The rumors regarding cross-border mergers and acquisitions had significantly positive influence at 10% and 5% levels in the different tests. Therefore the answer on the first research question would be that there are significant differences between the rumors of mergers and acquisitions in the pharmaceutical and biotechnological industry. The domestic rumors are not significant and indicated not to be different from 0, while the cross-border rumors are significantly positive influencing shareholder wealth.

2. Are there significant differences, in abnormal return following a rumor, between inter- and cross-industry mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe?

In the event study the tests for the rumors of inter- and cross-industry mergers and acquisitions had mixed results. It appears that it is the specific combination of the acquirer and the target that is more important and a more accurate statement than that either inter- or cross-industry performs better. However, the inter-industry samples did have the highest returns, but only the rumors where a pharmaceutical company was both the target and the acquirer had most significant returns when looking at the CAR. However, when looking at the event day, which seems to be more accurate based on the sub-conclusions made after each
Page | 66

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

sample analysis, the pharmaceutical-biotechnological sample had the highest 1 day positive returns and of the event day returns the sample with the most significant return among the industry samples. The high significant return of this sample was also predicted in the literature review. Furthermore, the regression analysis had mixed results regarding the relation of the acquirer-target combination. The pharmaceutical-biotechnological variable was significant and had the highest influence of the industry variables in the first model, while the other cross-industry variable biotechnological-pharmaceutical had the lowest influence. The results of the inter-industry variables were similar. There it depends on what is looked at when answering the second research question, if the focus is on the event day and the regression analysis the pharmaceutical-biotechnological cross-industry combination is most significant. While if the focus is on the CAR the inter-industry samples had the highest abnormal return of which the pharmaceutical-pharmaceutical sample significant in most tests. Therefore the conclusion is that the differences in abnormal returns in the inter and crossindustry rumors is more related to the combination of target and acquirer and the scope that range of days that is being looked at that purely saying that inter or cross-industry is better. However, the samples and variables with the pharmaceutical firm as acquirer had overall the best results.

3. Are there characteristics of the acquiring or target firms that have a significant influence on abnormal returns following a rumor of a merger or acquisitions in the pharmaceutical and biotechnological industry in Europe?

The characteristics that have been looked at in this research paper which were not specified under the other research questions are method of payment, target profit, target EBITDA, target ROA and relative size. Payments made with shares had a positive influence on the abnormal return and was significant in the first models. The profit of the target had a negative influence on the abnormal return and was significant in the first model. The EBITDA of the target had a positive influence on the abnormal return but was in none of the model significant. Indicating that EBITDA by itself among these variables did not had a significant influence on the abnormal return. Target ROA had a positive influence on the abnormal return and was significant in the first model. The relative size had a negative influence on the abnormal return, but was insignificant in each of the models. It appears that

Page | 67

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

these variables have influence on the abnormal return. However, when these variables were combined in a model the explanatory power was 25,39%, indicating that while they have influence on the abnormal return it is only to a limited extend and not each of the variable has been significant. Therefore, the answer on the third research question is that there are characteristics that influence the abnormal returns following a rumor; mainly the method of payment, the target’s profit and the target’s ROA had a significant influence on the abnormal return.

Now that each of the research questions has been answered, the main problem statement will be answered. Are there significant abnormal returns related to rumors of mergers and acquisitions in the pharmaceutical and biotechnological industry in Europe?

Based on the results from the analyses and the answers of the research questions it is possible to answer on the main problem statement. There are significant positive abnormal returns following a rumor of a merger or acquisition in the pharmaceutical and biotechnological industry. It is not that each rumor is followed by significant positive returns, but in general they are present. The market is most positive on rumors regarding mergers or acquisitions that are cross-border or that have a pharmaceutical firm as acquirer. Also the method of payment, the profit of the target and the target’s ROA are influential for the abnormal returns. Thus, based on the results of this research paper, if a rumor start circulating the market regarding a merger or acquisition in the pharmaceutical or biotechnological industry, than it will increase shareholder wealth if it is regarding a merger or acquisition which is cross-border, has a pharmaceutical firm as acquirer, is paid with shares, has a low profit target and/or has a target with a high ROA.

Page | 68

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Discussion
In this section the implications of the assumptions and limitations will be discussed. Furthermore, the results that differed from the expectations and any abnormal findings will be discussed as well. In this research paper a number of assumptions were made, the first assumption was that markets are efficient. The theory behind this is that as soon as the market has information, it is rational and efficient enough to predict what the implications are of the information. Thus when an event happens the market can predict at the moment of the event what the long run consequences are. While this theory holds in a perfect world, in the real market not every person acts rational and not all the information is equally available for everyone. Therefore markets will never be completely efficient. However, markets are efficient enough to assume them to be efficient, which is why it is assumed that this theory holds. Furthermore, the theory that had been used for announcements and completed mergers and acquisitions was assumed to hold for rumors as well, considering that the market would receive information and reacted on that. However, the effect of a rumor differs from an announcement, even though there are noise traders on the market. Therefore the results did differ and were more focused around the event day. Next are the assumptions of choosing the right event window and estimation period. The event window of 3 days and the estimation period of 250 days were assumed to give the best results considering that these had been used in other recent event studies. However, considering that each study is different and each sample is different it is difficult to determine the right event window and estimation period. Therefore, if another study of this kind would be conducted, changing the event window or the estimation period could produce different results which possibly are less significant, more significant or completely different from the results in this study. However, for the purpose of the study the chosen event window and estimation period are assumed to be the most appropriate and to have given the best results. Another assumption that was made in the paper was that the sample was normally distributed. In order to justify the use of parametric tests in the event study, the sample has to be normally distributed or assumed to be normally distributed. Therefore the sample was analyzed and the differences between the sample and a perfectly distributed sample were mentioned. It was stated that there were a number of differences. However, in the real world it is difficult to find perfectly distributed samples. Therefore the sample is assumed to be normally distributed, even though this can affect the results of the test.

Page | 69

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

In this research paper a number of limitations were made. The focus of the paper was on the pharmaceutical and biotechnological industry. This was chosen in order to go further in depth with the results as there have been many event studies that focused on the total market. Also the pharmaceutical and biotechnological market had the potential for abnormal returns, as this was seen in other studies and discussed in the literature review. Aside from the industry limitation, the geographical limitation was on Europe. This limitation was also chosen to limit the sample, in order to go further in depth with the information at hand. Also in the history it was stated that a great deal of the mergers and acquisitions in the chosen industry took place in Europe, which suggested that this was a suitable geographic region. Furthermore, because of the low trade barriers in Europe due to the European Union, the results would differ from other areas. Another limitation was to focus on firms that did not have a merger or acquisition rumor in the estimation period. This was done in order to compare the results of the event window with an estimation period that would not be filled with other events that could influence the outcome. This could have introduced a self-inflicted bias, as it might have filtered the sample from specific types of mergers and acquisitions. However, considering that there were significant abnormal returns found in the sample, the limitation is considered justifiable. Otherwise, if the sample had more events in the estimation period, the event would occur more frequent and the abnormal returns of the event would become more normal. The last limitations that were introduced were in terms of the variables chosen for the regression analyses. There are a large variety of variables used in different studies; however the author chose to focus on a selected group of variables to be able to analyze them and the results more thoroughly. Otherwise, by including too many variables it would be at the cost of the depth of the analysis per variable. Aside from the assumptions and the limitations, any noted abnormalities are discussed in this section. The first item that will be discussed is the high explanatory power of regression model 1. As stated in the analysis, it is possible that due to the large amount of variables and the small sample, the variables were able to explain a large part of the abnormal returns, because they acted as factors rather than variables the should represent. However, it is difficult to determine if this is the case, although the unusually high explanatory power

Page | 70

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

would indicate that something is off. Otherwise it would mean that it is a superior model, which seems more unlikely than that the model acts different than expected. Furthermore, in the different tests the cross-border related rumors had better results than the domestic related rumors. This is contradicting most of the theory available on mergers and acquisitions. This result could possibly be explained by one or two factors. The first argument is that the event is a rumor and a cross-border merger or acquisition is more often cheaper thus the short term pay of seems better than a domestic merger or acquisition. However, in the long run the complications of a cross-border merger or acquisition outweigh the advantage of the lower price. The second argument is that the sample only includes European firms. Considering that the trade boundaries within the European Union are smaller than other non-European countries (Hernando 2009; Ekkayokka 2009), the difficulties of a cross-border merger might not be as great in Europe as between other countries outside of Europe. The last notable unexpected result was the positive relationship between share payment and the abnormal return. In all of the studies the payment of an acquisition with shares had a negative influence on the abnormal return. In this study it had a positive relationship in each regression model it was included in. There are two possible explanations for this. The first argument for explaining it is that because the event is a rumor, the payment of the deal is not fixed; therefore it might not influence the return negatively until it is announced. The other possibility is that, because it was used as dummy variable, the binary selection could also be applied to other characteristics that actually have a positive influence on the abnormal return. At any rate, it is contradictory to the bulk of the literature and it seems unlikely that payments with shares are value increasing based on the fundamentals behind it. In this section the different assumptions, limitations and extraordinary findings have been discussed. Each of these topics discussed have their implications on how the results of the study turned out. It are decisions and assumptions that make a study unique and it depends on where the focus of the study is put what assumptions and limitations are important to include in the study.

Page | 71

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Conclusion
In this research paper the effects of rumors of mergers and acquisitions in the European pharmaceutical and biotechnological industry were analyzed with regard to shareholder wealth. The event of this paper were rumors of 2001 to 2012 and the nature of the event combined with the most recent data, makes sure that the results of this paper are value adding to the available literature. The main problem statement was whether or not these rumors resulted in significant abnormal returns and whether or not these returns were positive or negative. This was done by answering on the 3 research questions. The first research question was whether there are any significant differences when analyzing rumors regarding domestic or cross-border mergers and acquisitions. The results of the study indicated that the domestic sample had returns that were not significantly different from 0 and that the crossborder sample had returns that were significantly positive at a significance level of 5%. The second research question was whether there are any significant differences when comparing rumors regarding inter- and cross-industry mergers and acquisitions. The results showed that there were differences, but that they were more related to the combination of acquirer and target, considering that the tests of the different inter- and cross-industry samples did not share similar results. It appeared that the rumors regarding a merger or acquisition with a pharmaceutical firm as acquirer had significantly positive results and that the combination of a pharmaceutical firm as acquirer and a biotechnological firm as target had the highest positive results when looking at the event day. Furthermore, the rumors with the biotechnological firm as acquirer did not appear to be significantly different from 0 even though the abnormal return was lower in some cases or negative compared to the other samples. The third research question was regarding other characteristics that were not covered in the first and second research questions and that did not have a specific sample in the event study. The results of the regression analyses showed that rumors regarding mergers and acquisitions that were financed with shares, had a target with low profit or a target with a high return on assets had a significantly positive effect on the abnormal return. What was also seen among the different samples and tests in the event study; was that the day of the event was more significant when compared to the other days. This could be related to the nature of the event, considering that most studies focus on announcement and completed mergers and acquisitions. Therefore the effect of the rumor was insignificant on
Page | 72

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

the day before the announcement, as it is harder to anticipate a rumor than an announcement. Also the day after the event was only significant in some cases, which are also explained with that a rumor should not have long lasting effects as the outcome is still uncertain. After the research questions were answered it was possible to formulate the answer on the main problem statement. Are there significant abnormal returns following a rumor of a European merger or acquisition within the pharmaceutical and biotechnological industry? The answer to this question is that there are significant results and that they are positive. However, these positive significant results are mostly related to mergers and acquisitions that are cross-border, have a pharmaceutical firm as acquirer, have a target with low profit or have a target with a high return on assets.

Page | 73

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

References
Achilladelis, B. and Antonakis, N. (2001), “The dynamics of technological innovation: the case of the pharmaceutical industry”, Research Policy 30, pp. 535-588 Elsevier. Andrade, M.; Mitchell, M. and Stafford, E. (2001), “New Evidence and Perspectives on Mergers”, Journal of Economic Perspectives, Vol. 15, No. 2, pp. 103-120. Arora A and Gambardella A (1990), “Complementary and External Linkages: The Strategies of the Large Firms in Biotechnology”, Journal of Industrial Economics, Vol. 38, No. 4, pp. 361-379. Bartholdy, J.; Olsen, D. and Peare, P. (2007), “Conducting Event Studies on a Small Stock Exchange”, The European Journal of Finance, Vol. 13, No. 3, pp. 227–252. Baum, J.A.C.; Calabrese, T. and Silverman, B.S. (2000), “Don’t Go It Alone: Alliance Network Composition and Startups’ Performance in Canadian Biotechnology”, Strategic Management Journal, Vol. 21, No. 3, pp. 267-294. Becher, D. (2000), “The valuation effects of bank mergers”, Journal of Corporate Finance, Vol. 6, pp. 189–214. Beitel, P.; Schiereck, D. and Wahrenburg, M. (2004), “Explaining M&A success in European banks”, European Financial Management, Vol. 10, pp. 109–39. Berk, J. and DeMarzo, P. (2011), “Corporate Finance”, 2nd edition, Pearson International Edition. Boehmer, E.; Musumeci, J. and Poulsen, A. (1991), “Event-study methodology under conditions of event-induced variance”, Journal of Financial Economics, 30(2), pp. 253– 272. Boutahar, B. (2009), “Behaviour of skewness, kurtosis and normality tests in long memory data”, Statistical Methods and Applications, pp. 1-23. Brown, S.J. and Warner, J.B. (1985), “Using daily stock returns: The case of event studies of event-induced variance”, Journal of Financial Economics, 14(1), pp. 3–31. Campa, J.M. (2004), “Shareholder Value Creation in European M&As”, European Financial Management, Vol. 10, No. 1, 2004, 47–81

Page | 74

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

CD (2012) is used as reference when referring to the attached CD with the data collection, description of the data, calculations and results of the different tests on it. Cockburn, I. and Henderson, R. (1996), “Scale, scope, and spillovers: the determinants of research productivity in drug discovery”, Rand Journal of Economics, Vol. 27, pp. 32– 60. Colwell, P.F.; Heller J.A. and Trefzger, J.W. (2009), “Expert Testimony: Regression Analysis and Other Systematic Methodologies”, The Appraisal Journal, Summer 2009, pp. 254-262. Corrado, C.J. (1989), “A nonparametric test for abnormal security-price performance in event studies”, Journal of Financial Economics, Vol. 23, No. 2, pp. 385–395. Corrado, C.J. and Zivney, T.L. (1992), “The specification and power of the sign test in event study hypothesis tests using daily stock returns”, Journal of Financial and Quantitative Analysis, 27(3), pp. 465–478. Cowan, A. (1992), “Nonparametric event study tests”, Review of Quantitative Finance and Accounting, Vol. 2 No. 4, pp. 343–358. Cowan, A. and Sergeant, A. (1996), “Trading frequency and event study test specification”, Journal of Banking & Finance, Vol. 20, No. 10, pp. 1731–1757. Danzon P.M.; Epstein, A.J. and Nicholson, S. (2003), “Mergers and Acquisitions in the Pharmaceutical and Biotech Industries”, NBER Working Papers, National Bureau of Economic Research. Danzon, P.M.; Epstein, A. and Nicholson, S. (2007), “Mergers and Acquisitions in the Pharmaceutical and Biotech Industries”, Managerial and Decision Economics 28: 307– 328 Published online in Wiley InterScience. Demirbag, M.; Ng, C.K. and Tatoglu, E. (2007), “Performance of Mergers and Acquisitions in the Pharmaceutical Industry: A comparative Perspective”, Multinational Business Review, Vol. 15, No. 2, pp. 41-61. Dilmore, G. (1972), “Notes and comments: Multiple Regression Analysis as an Approach to Value”, Appraisal Institute Magazine, Vol. 2, pp. 459-461.

Page | 75

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Eckbo, E. (1992), “Mergers and the value of antitrust deterrence”, Journal of Finance, Vol. 47, pp. 1005–1030. Eckbo, E., Thorburn, K. (2000), “Gains to bidder firms revisited: domestic and foreign acquisitions in Canada”, Journal of Financial and Quantitative Analysis, Vol. 35, pp. 1– 25. Ekkayokka, M.; Holmes, P. and Paudyal, K. (2009), “The Euro and the Changing Face of European Banking: Evidence from Mergers and Acquisitions”, European Financial Management, Vol. 15, No. 2, pp.451–476. Elliot, G. (2011), “A control function approach for testing the usefulness of trending variables in forecast models and linear regression”, Journal of Econometrics, Vol. 164, pp. 79–91. Elton, E; Gruber, M.; Brown, S. and Goetzmann, W. (2011), “Modern Portfolio Theory and Investment Analysis”, 8th ed., Wiley 2011. Fama, E. (1970), “Efficient capital markets: A review of theory and empirical work”, Journal of Finance, Vol. 25, No. 2, pp.383-417. Fuller, K; Netter, J. and Stegemoller, M. (2002), “What Do Returns to Acquiring Firms Tell Us? Evidence from Firms That Make Many Acquisitions”, The Journal of Finance, Vol. 57, No. 4, pp. 1763–1793. Goergen, M. and Renneboog, L. (2003) “Shareholder Wealth Effects of European Domestic and Cross border Takeover Bids”, European Financial Management Journal, Vol.10, pp. 9-45. Gugler, K., Mueller, D.C., Yurtoglu, B. and Zulehner, C. (2003), “The effects of mergers: an international comparison”, International Journal of Industrial Organization, Vol. 21 No. 5, pp. 625-53. Hassan, M.; Patro, D.K.;Tuckman, H. and Wang, X. (2007), “Do mergers and acquisitions create shareholder wealth in the pharmaceutical industry?”, International Journal of Pharmaceutical and Healthcare Marketing, Vol. 1 No. 1, pp. 58-78. Haushalter, G.D.; Heron, A.R.; Lie, E. (2002), “Price uncertainty and corporate value”, Journal of Corporate Finance, Vol. 8, pp. 271 – 286.

Page | 76

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Hernando, I.; Nieto, M. and Wall, L. (2009), “Determinants of domestic and cross-border bank acquisitions in the European Union”, Journal of Banking & Finance, Vol. 33, pp. 1022–1032. Higgins, M.J. and Rodriguez, D. (2006), “The Outsourcing of R&D Through Acquisitions in the Pharmaceutical Industry”, Journal of Financial Economics, Vol. 80, No. 2, pp. 351383. Houston, J. and Ryngaert, M. (1994), “The overall gains from large bank mergers”, Journal of Banking and Finance, Vol. 18, pp. 1155–76. Houston, J.; James, C. and Ryngaert, M. (2001), “Where do merger gains come from? Bank mergers from the perspective of insiders and outsiders”, Journal of Financial Economics, Vol. 60, pp. 285-331. Ismail, A. and Davidson, I. (2005), “Further analysis of mergers and shareholder wealth effects in European banking”, Applied Financial Economics, Vol. 15, pp. 13–30. Keller, G. (2005), “Statistics for Management and Economics”, 7th ed., Thomson. Kipp, O and Leiding, P. (2008), “Marketing Strategy Effects of mergers & acquisitions on the day-to-day work lives of general practitioners”, Journal of Medical Marketing, Vol. 8, pp. 211–219. Kirchhoff, M. and Schiereck, D. (2011), “Determinants of M&A Success in the Pharmaceutical and Biotechnological Industry”, Journal of Business Strategy. Kiymaz, H. (2001), “Cross-border acquisitions of US financial institutions: Impact of macroeconomic factors”, Journal of Banking & Finance 28, pp. 1413–1439. Kohers, N. and Kohers, T. (2000), “The value creation potential of high-tech mergers”, Financial Analysts Journal, Vol. 56, pp. 40–48. Kohers, N. and Kohers, T. (2004), “Information sensitivity of high tech industries: evidence from merger announcements”, Applied Financial Economics Applied Financial Economics, Vol. 14, pp. 525–536. Longstaff, F.A. (2010), “The subprime credit crisis and contagion in financial markets”,Journal of Financial Economics, Vol. 97, pp. 436–450.

Page | 77

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

MacKinlay, A.C. (1997), “Event Studies in Economics and Finance”, Journal of Economic Literature Vol . 35, pp. 13-39. Malik, N.N. (2009), “Biotech acquisitions by big pharma: why and what is next”, Drug Discovery Today, Vol. 14, No. 17/18 , pp. 818-821. Mangold, N. and Lippok, K. (2008), “The effect of cross-border mergers and acquisitions on shareholder wealth: Evidence from Germany”, Journal of international business and economics, Vol. 8, No. 3. Maynes, E. and Rumsey, J. (1993), “Conducting event studies with thinly traded stocks”, Journal of Banking and Finance 17, pp. 145–157. Mitchell, M.L. and Jolley, J.M. (2007), “Research Design Explained”, 7th Edition. Wadsworth Cengage Learning. Mittra, J. (2007), “Life Science Innovation and the Restructuring of the Pharmaceutical Industry: Merger, Acquisition and Strategic Alliance Behaviour of Large Firms”, Technology Analysis & Strategic Management, Vol. 19, No. 3, pp. 279–301. Moeller, S.B.; Schlingemann, S.P. and Stulz, R.M. (2005), “Wealth Destruction on a Massive Scale? A Study of Acquiring-Firm Returns in the Recent Merger Wave”, The Journal of Finance, Vol. LX, No. 2, pp. 757-782. Morgan (2012), “http://www.msci.com/products/indices/”, Extracted 05/04/2012. Nicholson, S, Danzon, P.M. and McCullough, J. (2005), “Biotech-Pharmaceutical Alliances as a Signal of Asset and Firm Quality”, Journal of Business, Vol. 78, No. 4, pp. 14331464. Nigro, G.L.; Perrone, G. and Chiapparrone, S. (2011), “Governance forms drivers in biopharmaceutical inter-firm relationships”, Int. J. Production Economics. Patell, J. M. (1976), “Corporate forecasts of earnings per share and stock price behavior: Empirical tests”, Journal of Accounting Research, Vol. 14, No. 2, pp. 246–276. Pole, J. and Lampard, R. (2002), “Practical social investigation: qualitative and quantitative methods in social research”, Pearsons Eduction Limited.

Page | 78

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Powell , R.G. and Stark, A.W. (2005), “Does operating performance increase post-takeover for UK takeovers? A comparison of performance measures and benchmarks”, Journal of Corporate Finance, Vol. 11, pp. 293–317. Rad, A.T. and Van Beek, L. (1999), “Market valuation of European bank mergers”, European Management Journal, Vol. 17, pp. 532–540. Ravenscraft, D.J.; Long, W.F. (2000), “Paths to creating value in pharmaceutical mergers. In Mergers and Productivity”, Conference Report Series. University of Chicago Press: Chicago. Renneboog, L. (2006) “Forthcoming in Advances in Corporate Finance and Asset Pricing”, (ed.), Amsterdam: Elsevier, March 2006. Reuters (2012), “http://thomsonreuters.com/products_services/financial/financial_products/az/datastream/”, Extracted 05/04/2012. Roijakker, N.; Hagedoorn, J. (2005), “Inter-firm R&D partnering in pharmaceutical biotechnology since 1975: Trends, patterns, and networks”, Research Policy, Vol. 35, pp. 431–446. Rothaermel, F. (2001), “Complementary assets, strategic alliances, and the incumbent’s advantage: an empirical study of industry and firm effects in the biopharmaceutical industry”, Research Policy 30, pp. 1235–1251. Salinger, M. (1992), “Standard errors in event studies, Journal of Financial and Quantitative Analysis” Vol 27, pp. 39–53. Shelton, F.A. (1987), “Using regression analysis: A guided tour”, Compu. Educ., Vol. 11, No. 2, pp. 95-111. Sorescu, A.B.; Chandy, R.K. and Prabhu, J.C. (2003), “Sources and Financial Consequences of Radical Innovation: Insights from Pharmaceuticals”, Journal of Marketing, Vol. 67, No. 4, pp. 82-102. Spector, P.E. and Brannick, M.T. (2011), “Methodological Urban Legends: The Misuse of Statistical Control Variables”, Organizational Research Methods, Vol. 14, No. 2, pp. 287-305.

Page | 79

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Sowlay, M (2009), “The current M & A environment and its strategic implications for emerging biotherapeutics companies”, Macmillan Publishers Ltd. 1462-8732 Journal of Commercial Biotechnology, Vol. 16, No. 2, pp. 109–119, www.palgrave-

journals.com/jcb/ . Talja, S. (1999), “Analyzing Qualitative Interview Data: The Discourse Analytic Method”, Library & Information Science Research, Vol. 21, No. 4, pp. 459–477. Travlos, N.G. (1987), “Corporate takeover bids, methods of payment, and bidding firms’ stock returns”, Journal of Finance, Vol. 42 No. 4, pp. 943-63. Tu, J.V. (1996), “Advantages and Disadvantages of Using Artificial Neural Networks versus Logistic Regression for Predicting Medical Outcomes”, J Clin Epidemid, Vol. 49, No. 11, pp. 1225-1231. Uddin, M. and Boateng, A. (2009), “An analysis of short-run performance of cross-border mergers and acquisitions Evidence from the UK acquiring firms”, Review of Accounting and Finance, Vol. 8 No. 4, pp. 431-453. Vernon, J. (2005), “Examining the link between price regulation and pharmaceutical R&D investment”, Health Economics, Vol. 14, pp. 1–17. Wang, A.F. (2010), “Information arbitrage with speculative noise trading”, Journal of Banking & Finance, Vol. 34, pp. 304-313. Yeung, H.W.C. (1995), “Qualitative Personal Interviews in International Business Research: Some Lessons from a Study of Hong Kong Transnational Corporations”, International Business Review, Vol. 4, No. 3, pp. 313-333, Elsevier Science Ltd. Zephyr (2012), “http://www.bvdinfo.com/Products/Economic-and-M-A/M-A/ZEPHYR.aspx”,

Extracted 05/04/2012.

Page | 80

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Appendix
Table 9: Acquiror country specification

Acquiror country code Acquiror name Count FR BOIRON SA 1 IPSEN SA 1 SANOFI-SYNTHELABO SA 2 STALLERGÈNES SA 1 VIRBAC SA 2 AT SANOCHEMIA PHARMAZEUTIKA AG 1 BE IBA ION BEAM APPLICATIONS SA 1 OMEGA PHARMA NV 2 TIGENIX NV 1 UCB SA 3 BG SOFARMA AD 2 CH ACINO HOLDING AG 2 ACTELION LTD 2 ARPIDA AG 1 NOVARTIS AG 1 SANTHERA PHARMACEUTICALS HOLDING AG 1 DE BAYER AG 2 EVOTEC OAI AG 1 MEDIGENE AG 1 MORPHOSYS AG 1 NOVEMBER AG 1 PLASMASELECT AG 1 STADA ARZNEIMITTEL AG 1 WILEX AG 1 DK ALK-ABELLÓ A/S 1 H LUNDBECK A/S 1 TOPOTARGET A/S 1 ES LABORATORIOS ALMIRALL SA 1 GB AMARIN CORPORATION PLC 1 ASTRAZENECA PLC 1 BTG PLC 1 DECHRA PHARMACEUTICALS PLC 2 GLAXOSMITHKLINE PLC 5 PROTEOME SCIENCES PLC 1 SHIRE LTD 1 HU EGIS GYOGYSZERGYAR RT 1 RICHTER GEDEON VEGYÉSZETI GYÁR NYILVÁNOSAN MUKÖDO RT 3 IT NEWRON PHARMACEUTICALS SPA 1 RECORDATI SPA 3 NL QIAGEN NV 1 RO ANTIBIOTICE SA 1 RU FARMSINTEZ OAO 1 PROTEK OAO 1 SE MEDA AB 1 OREXO AB 1

Page | 81

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth
Table 10: Target country specifics

Target country code Target name FR AVENTIS SA BIO VETO TESTS SA CIS BIO INTERNATIONAL SA'S RADIOPHARMACEUTICALS DIVISION ELAIAPHARM SA IPSOGEN SAS LABORATOIRES DOLISOS SA ORPHAN EUROPE SARL TROPHOS SA AT ALVETRA & WERFFT AG BG BULGARSKA ROZA SEVTOPOLIS AD UNIFARM AD CH ADIPRA AG APOXIS SA AXOVAN AG MERFARM AG'S INTERNATIONAL BUSINESS SPIRIG PHARMA AG'S GENERICS BUSINESS TLT MEDICAL LTD CZ HERBACOS-BOFARMA SRO DE BAYER AG EVOTEC NEUROSCIENCES GMBH GRÜNENTHAL GMBH'S ORAL CONTRACEPTIVES BUSINESS HEIDELBERG PHARMA AG IBL GESELLSCHAFT FÜR IMMUNCHEMIE UND IMMUNBIOLOGIE MBH PHYLOS GMBH SCHERING AG SCHWARZ PHARMA AG DK H LUNDBECK A/S PHARMALETT A/S VETXX HOLDING A/S ES CELLERIX SA INTERNATIONAL PHARMACEUTICAL IMMUNOLOGY SA'S SPAIN AND PORTUGALBASED IMMUNOTHERAPEUTIC ACTIVITIES GB AVIDEX LTD CELLTECH GROUP PLC CUTIVATE EMGEL GENITRIX LTD GLAXOSMITHKLINE PLC'S CERTAIN OVER-THE-COUNTER PHARMACEUTICAL BRANDS HUNTER-FLEMING LTD KUDOS PHARMACEUTICALS LTD LAGAP PHARMACEUTICALS LTD LAXDALE LTD LECTUS THERAPEUTICS LTD'S KEY PHARMACEUTICAL ASSETS OXISTAT PHARMAKODEX LTD PROTHERICS PLC SEROTEC LTD SHIRE PLC Page | 82

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth SHIRE PLC'S NON-CORE PRODUCT PORTFOLIO TEMOVATE MEDESTEA INTERNAZIONALE SRL NEWRON PHARMACEUTICALS SPA PIERREL MEDICAL CARE SPA ARTU BIOLOGICALS EUROPE BV ZAKLADY FARMACEUTYCZNE POLPHARMA SA INSTITUTUL CANTACUZINO ANVILAB OOO KHIMIKO-FARMATSEVTICHESKII KOMBINAT AKRIKHIN OAO SERDIX LTD SINBIO OOO ANTULA HEALTHCARE AB DR F FRIK ILAÇ SAN VE TIC AS VERNALIS PHARMACEUTICALS INC.

IT

NL PL RO RU

SE TR US

Table 11: Domestic deal specifics

Acquiror FR country code AT BG CH

Acquiror name BOIRON SA SANOFI-SYNTHELABO SA VIRBAC SA SANOCHEMIA SOFARMA AD PHARMAZEUTIKA AG ACINO HOLDING AG ACTELION LTD ARPIDA AG BAYER AG EVOTEC OAI AG NOVEMBER AG WILEX AG AMARIN CORPORATION ASTRAZENECA PLC PLC BTG PLC DECHRA GLAXOSMITHKLINE PLC PHARMACEUTICALSPLC

DE

GB

RO RU SE

SHIRE LTD ANTIBIOTICE SA FARMSINTEZ OAO PROTEK OAO MEDA AB

Target name LABORATOIRES DOLISOS SA AVENTIS SA BIO VETO TESTS SA ALVETRA & WERFFT AG BULGARSKA ROZA SEVTOPOLIS AD UNIFARM AD ADIPRA AG MERFARM AG'S INTERNATIONAL BUSINESS AXOVAN AG TLT MEDICAL LTD SCHERING AG EVOTEC NEUROSCIENCES GMBH IBL GESELLSCHAFT FÜR IMMUNCHEMIE UND HEIDELBERG PHARMA IMMUNBIOLOGIE MBHAG LAXDALE LTD KUDOS PHARMACEUTICALS LTD PROTHERICS PLC GENITRIX LTD CUTIVATE EMGEL OXISTAT TEMOVATE SHIRE PLC INSTITUTUL CANTACUZINO SINBIO OOO ANVILAB OOO ANTULA HEALTHCARE AB

Page | 83

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth
Table 12: Event day +1 test statistics and probability values

Sample Total Domestic Cross-border Phamaceutical Biotechnological PhamaceuticalPhamaceutical PhamaceuticalBiotechnological BiotechnologicalBiotechnological BiotechnologicalPhamaceutical

Test T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value

T1 1,56 12,34% -0,14 89,20% 2,74 0,99% 1,61 11,30% 0,53 60,99% 2,59 1,33% -0,99 35,49% 0,87 43,27% 0,03 97,38%

T2 1,29 20,17% 1,32 19,86% 1,25 22,08% 1,60 11,63% 0,73 47,81% 1,69 9,81% 1,42 19,94% 0,77 48,17% 0,72 50,65%

T3 2,78 0,73% 1,03 31,07% 2,80 0,85% 2,55 1,39% 1,11 29,14% 2,67 1,09% 0,31 76,74% 1,03 36,32% 0,56 59,91%

T4 1,28 20,39% -0,11 91,35% 2,35 2,48% 1,09 28,14% 0,70 49,71% 1,77 8,48% -0,67 52,45% 1,19 30,03% 0,04 96,67%

T5 1,92 6,00% 0,59 56,35% 2,44 2,01% 1,70 9,62% 0,87 40,14% 1,71 9,53% 0,25 81,09% 1,19 30,06% 0,35 74,11%

T6 -1,02 31,30% 0,38 70,75% -1,86 7,17% -0,86 39,47% -0,61 55,41% -0,91 36,84% -0,07 94,64% -0,95 39,40% 0,07 94,43%

T7 0,11 91,41% -1,25 22,33% 1,44 15,92% -0,24 81,00% 0,83 42,55% -0,13 89,57% -0,31 76,30% 0,75 49,34% 0,38 72,02%

T8 0,07 94,64% -1,08 29,03% 1,06 29,83% 0,10 92,38% -0,04 96,88% 0,33 74,04% -0,49 64,18% 0,13 90,47% -0,17 86,95%

Day +1 0,55% -0,08% 1,05% 0,54% 0,56% 0,88% -1,02% 1,15% 0,06%

Table 13: Event day 0 test statistics and probability values

Sample Total Domestic Cross-border Phamaceutical Biotechnological PhamaceuticalPhamaceutical PhamaceuticalBiotechnological BiotechnologicalBiotechnological BiotechnologicalPhamaceutical

Test T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value

T1 2,30 2,50% 1,50 14,64% 1,85 7,36% 3,18 0,26% -0,20 84,29% 1,89 6,64% 2,90 2,00% 0,03 97,50% -0,27 79,42%

T2 1,47 14,77% 1,65 11,05% 1,15 25,98% 1,80 7,76% 0,87 40,08% 1,29 20,37% 2,45 3,99% 0,89 41,35% 0,86 42,20%

T3 2,77 0,75% 0,98 33,35% 2,83 0,77% 2,82 0,70% 0,57 58,23% 2,22 3,23% 1,91 9,32% 0,11 91,33% 0,66 53,12%

T4 1,68 9,85% 0,97 33,96% 1,66 10,52% 1,95 5,66% -0,22 82,63% 1,68 10,04% 1,20 26,55% 0,03 97,38% -0,29 78,20%

T5 1,99 5,15% 0,95 35,07% 1,74 9,10% 1,95 5,70% 0,47 64,77% 1,51 13,87% 1,37 20,81% 0,08 93,99% 0,62 55,78%

T6 -1,98 5,23% -0,76 45,33% -2,10 4,36% -1,96 5,54% -0,53 60,59% -1,52 13,61% -1,61 14,64% -0,19 85,86% -0,57 59,01%

T7 2,06 4,37% 1,25 22,29% 1,76 8,73% 1,93 5,88% 0,83 42,40% 1,45 15,46% 1,57 15,55% 0,00 100,00% 1,14 29,88%

T8 2,59 1,20% 1,57 12,89% 2,07 4,60% 2,36 2,23% 1,07 30,49% 1,90 6,52% 1,52 16,78% 0,13 90,35% 1,34 22,77%

Day 0 0,80% 0,92% 0,71% 1,07% -0,22% 0,64% 2,99% 0,04% -0,44%

Page | 84

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth
Table 14: Event day -1 test statistics and probability values

Sample Total Domestic Cross-border Phamaceutical Biotechnological PhamaceuticalPhamaceutical PhamaceuticalBiotechnological BiotechnologicalBiotechnological BiotechnologicalPhamaceutical

Test T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value T-stat P-value

T1 -0,78 43,66% -1,04 30,59% 0,06 95,59% -0,12 90,50% -1,10 29,30% 0,86 39,59% -1,51 16,59% 0,56 59,71% -1,74 12,50%

T2 0,88 38,29% 0,78 44,15% 1,00 32,21% 0,89 37,59% 0,86 40,56% 0,81 42,34% 1,02 33,22% 0,63 55,45% 0,95 37,52%

T3 0,24 81,27% -0,51 61,58% 0,77 44,47% 0,86 39,29% -1,17 26,45% 1,46 15,16% -1,09 30,51% 0,21 83,91% -1,79 11,73%

T4 -0,94 35,10% -1,46 15,45% 0,06 95,64% -0,14 88,67% -1,31 21,21% 1,20 23,64% -1,57 15,12% 0,90 40,46% -2,28 5,63%

T5 0,21 83,65% -0,90 37,49% 0,53 59,95% 0,72 47,44% -1,30 21,57% 1,15 25,64% -1,77 11,02% 0,27 79,96% -2,05 7,99%

T6 -0,08 93,39% 0,44 66,62% -0,56 57,62% -0,41 68,67% 0,71 49,28% -0,98 33,44% 1,39 19,87% -0,84 43,40% 1,83 10,97%

T7 0,00 100,00% -0,94 35,74% 0,96 34,35% 0,12 90,43% -0,28 78,70% 0,53 60,06% -0,94 37,14% 0,75 48,00% -1,14 29,28%

T8 0,82 41,32% 0,81 42,45% 0,38 70,61% 1,23 22,53% -0,60 56,15% 1,90 6,50% -1,15 27,84% 0,95 38,11% -1,69 13,50%

Day -1 -0,27% -0,64% 0,02% -0,04% -1,17% 0,29% -1,56% 0,74% -2,80%

Page | 85

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth

Table 15: Model 1 regression statistics

Multiple R R Square Adjusted R Square Standard Error Observations

0,97851719 0,957495891 0,663986851 0,022114969 17

Table 16: Model 1 Anova

df Regression Residual Total 12 5 17

SS 0,055086943 0,002445359 0,057532302

MS 0,0045906 0,0004891

F 10,23960775

Significance F 0,0187806

Table 17: Model 1 variable statistics

Coefficients Intercept Domestic Shares Target_Profit after tax (€000) Target_EBITDA (€000) Acq_Assets (€000) Deal_Value (€000) Target ROA pharm-pharm pharm-bio bio-pharm bio-bio Size -1,80E-1 -2,69E-2 8,39E-2 -2,98E-6 6,74E-7 2,86E-8 -2,13E-9 8,65E-2 2,14E-1 3,13E-1 0,00E+0 2,28E-1 -6,50E-4

Standard Error 3,71E-2 1,92E-2 1,79E-2 1,14E-6 3,80E-7 1,14E-8 8,97E-10 3,11E-2 2,92E-2 9,57E-2 0,00E+0 4,13E-2 4,50E-4

t Stat -4,85E+0 -1,40E+0 4,70E+0 -2,62E+0 1,77E+0 2,50E+0 -2,38E+0 2,78E+0 7,33E+0 3,28E+0 6,55E+4 5,52E+0 -1,44E+0

P-value 4,67E-3 2,21E-1 5,35E-3 4,71E-2 1,36E-1 5,42E-2 6,35E-2 3,89E-2 7,40E-4 2,21E-2 7,23E-73 3,77E-5 2,08E-1

Lower 95% -2,75E-1 -7,63E-2 3,80E-2 -5,91E-6 -3,02E-7 -7,57E-10 -4,44E-9 6,49E-3 1,39E-1 6,74E-2 0,00E+0 1,22E-1 -1,81E-3

Upper 95% -8,46E-2 2,25E-2 1,30E-1 -5,69E-8 1,65E-6 5,79E-8 1,75E-10 1,66E-1 2,89E-1 5,59E-1 0,00E+0 3,34E-1 5,07E-4

Page | 86

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth
Table 18: Model 2 regression statistics

Multiple R R Square Adjusted R Square Standard Error Observations

0,4427811 0,1960551 0,1030973 0,0432025 60

Table 19: Model 2 Anova

df Regression Residual Total 6 54 60

SS 0,024579 0,100788 0,125367

MS 0,0040965 0,0018665

F 2,633757

Significance F 0,0261147

Table 20: Model 2 variable statistics

Coefficients Intercept Domestic Acq_Assets (€000) pharm-pharm pharm-bio bio-pharm bio-bio -1,53E-2 -2,58E-2 -6,15E-10 5,03E-2 2,61E-2 3,08E-2 0,00E+0

Standard Error 1,77E-2 1,16E-2 5,84E-10 1,94E-2 2,44E-2 2,41E-2 0,00E+0

t Stat -8,61E-1 -2,21E+0 -1,05E+0 2,59E+0 1,07E+0 1,28E+0 6,55E+4

P-value 3,93E-1 3,10E-2 2,97E-1 1,24E-2 2,89E-1 2,07E-1 7,23E-73

Lower 95% -5,08E-2 -4,91E-2 -1,79E-9 1,13E-2 -2,28E-2 -1,75E-2 0,00E+0

Upper 95% 2,03E-2 -2,44E-3 5,56E-10 8,93E-2 7,51E-2 7,91E-2 0,00E+0

Page | 87

André van Bragt 282796 The effects of rumors of mergers and acquisitions on shareholder wealth
Table 21: Model 3 regression statistics

Multiple R R Square Adjusted R Square Standard Error Observations

0,503873614 0,253888619 -0,326420234 0,069061571 17

Table 22: Model 3 Anova

df Regression Residual Total 7 9 16

SS 0,014606797 0,042925505 0,057532302

MS 0,0020867 0,0047695

F 0,437506

Significance F 0,8557525

Table 23: Model 3 variable statistics

Coefficients Intercept Shares Target_Profit after tax (€000) Target_EBITDA (€000) Acq_Assets (€000) Deal_Value (€000) Target ROA Size 1,57E-2 4,25E-2 -2,14E-6 6,35E-7 1,71E-8 -1,37E-9 1,33E-2 -1,41E-4

Standard Error 3,04E-2 4,37E-2 2,91E-6 9,89E-7 3,07E-8 2,69E-9 6,04E-2 4,29E-4

t Stat 5,16E-1 9,71E-1 -7,34E-1 6,42E-1 5,58E-1 -5,11E-1 2,21E-1 -3,30E-1

P-value 6,19E-1 3,57E-1 4,82E-1 5,37E-1 5,91E-1 6,22E-1 8,30E-1 7,49E-1

Lower 95% -5,30E-2 -5,64E-2 -8,73E-6 -1,60E-6 -5,24E-8 -7,45E-9 -1,23E-1 -1,11E-3

Upper 95% 8,44E-2 1,41E-1 4,45E-6 2,87E-6 8,67E-8 4,71E-9 1,50E-1 8,30E-4

Page | 88

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close