The effects of the European bank mergers and acquisitions on bank value and risk

Transcription

1 The effects of the European bank mergers and acquisitions on bank value and risk Study for large cross-border bank M&As in Europe ANR : Name : S tanislav Tinev Topic : Mergers and Acquisitions Study Program: MSc Finance Supervisor : Olivier De Jonghe

2 Table of Contents Management Summary... 2 Preface... 3 Chapter 1 Introduction Research relevance Current state of literature Research method Structure Data... 8 Chapter 2 Literature review and hypothesis development Bank value creation/destruction Change in bank total risk Change in bank s systematic risk Chapter 3 Research Method Methodology and sample Event studies Total risk Systematic risk Chapter 4 - Results Abnormal and cumulative abnormal returns Abnormal and cumulative abnormal returns for the whole sample Abnormal and cumulative abnormal returns before and during the crisis Changes in Total Risk Total Risk for the whole sample ( ) Total Risk before and during the last financial crisis Changes in Systematic risk Change in Systematic risk for the whole sample ( ) Changes in Systematic risk before and during the last financial crisis Chapter 5 Conclusion Major findings Managerial recommendations Limitations Future research Appendix Reference

3 Management summary Given the fact that recent years Cross-border bank Mergers and Acquisitions (M&A) increased and continue to take place after the last financial crisis, this research is addressing some important issues. The findings from the analysis would provide valuable insights to both academic and managers interested in the overall risk and value change due to engaging in a bank merger. The research is using the total relative risk and systematic risk calculations in order to provide valuable findings over the general risk a bank is taking. Therefore, the presented findings would give a complete picture over the shift in risk due to engaging in a Cross-border bank Merger or Acquisition. Furthermore, this thesis would observe the change in shareholders value in terms of return. In order to achieve this, a data over a 12 year horizon ( ), including the announcement dates of Cross-border bank M&As is used. The study is conducted by applying event studies and the consecutive calculation of abnormal and cumulative abnormal returns multiple times over different time periods (CARs). Moreover, these findings are compared with the sample of domestic bank M&As which is used as a benchmark for different periods of time pre and throughout the last financial crisis, as well as for the whole sample period. The results based on the empirical methods yield not always significant, but interesting findings regarding the differences of the cross-border flag, as well as the period of the merger announcement. Summing up the main results, this research is able to provide the following conclusions. First, there is a clear difference between the acquirer performance after engaging in cross-border and in domestic bank mergers - the return, total, and systematic risks are much more intact on domestic level, which is not the case for most of the values discovered on cross-border level. Second, the findings also present a shift in the value and systematic risk change attributable to the last financial crisis. While the cross-border bank M&As before the crisis were associated with increase in overall return and systematic risk (related to the home bank index) and decrease of total risk, the throughout crisis deals turn out to be different. They presented negative change for all the variables of interest return, total, and systematic risk (related to home and host bank indexes). Thus, we can suggest that the change was due to the crisis itself and the regulations developed in that period, which inline the findings much more to the existing theories. Therefore, engaging in a cross-border bank M&A can decrease the risk due to the diversification effect, which is at the expense of a decrease in shareholders value. 2

4 Preface: This thesis is the concluding element of the one-year Master degree program in Finance at Tilburg University. The analysis provided herein was conducted under the supervision of Mr. Olivier De Jonghe in the department of Finance, between June 2014 and November This work is the outcome of my interest in Mergers and Acquisitions in the banking sector, resulting from the education provided to me by. It elaborated upon my research skills and my knowledge on the topic, and helped me connect the gained theoretical insights with practical issues. I would like to express my deepest gratitude to Mr. Olivier De Jonghe for his guidance, advice and constructive criticism throughout the writing process. Stanislav Tinev November

5 Chapter 1 Introduction 1.1 Research Relevance In the recent years, a significant number of European cross-border bank mergers and acquisitions (M&As) appear to take place. In the current research, we are interested in observing banks M&As and provide more clear understanding of their outcome in terms of change in potential risk, as well as value of bidder and target banks. A common assumption in this research area is that cross-border mergers have a potential for reducing banks risk of insolvency due to diversification. However, there are two existing reasons why bank M&As might in fact increase their risk by diversifying abroad. The first one is the acquirer s incentive for higher risk taking, which can be seen from the expansion of its operations in a different country by acquiring a risky bank. If this risky investment pays off, the bank has the benefit of holding the positive outcome, while if the target fails the acquirer is threatened by insolvency, which can result in regulatory bail by domestic or host regulator. The second reason is connected to the monitoring costs for consumer loan and the cost structure of the target bank. Since, expanding operations in new markets is highly related to potentially new regulatory problems, high monitoring cost can also increase the risk of insolvency. In order to address these issues and provide more clarity on the matter, the current research will focus on whether mergers create or destroy shareholders value for both bidder and target banks. It is important to determine whether mergers add or destroy value to banks shareholders, since the information available in the financial literature is outdated and most of the time contradicting instead of confirming certain findings. Thus, the main question of focus is weather European cross-border mergers are beneficial, and if so what is their effect over the acquirer and the target? More specifically, the analysis will focus on cross-border M&As. Noticeably, the existing literature on cross-country bank mergers and acquisitions seems to be limited in the sense that its main investigations and findings are based primary on a domestic level. The share of cross-border bank mergers and acquisitions in particular is relatively small related to other industries. A possible explanation given by Buch, DeLong (2004), suggests that the infrequency of cross-border bank M&As is due to their limited success. Since M&As are considered the best way for a company to enter a new market, a possible drawback can be that cross-border bank mergers might create institutions that cannot compete successfully in the host markets. A clarification about this phenomenon is provided by Berger et. al. (2001), who suggest that the possible problems arising from the cross-border bank M&As are the 4

6 efficiency barriers, such as cultural and lingual differences, as well as geographical distance. Furthermore, the adverse regulatory and supervisory structures can also stress-out the cross-border activity and therefore offset some of the gains from cross-border bank interaction (Berger et. al. (2001)). Having these considerations in mind, we believe that the current analysis will provide valuable insights about crossborder mergers and acquisitions in particular. However, some of the results of the empirical studies for domestic bank mergers are also found to be intriguing and relevant for the current research and thus will be further discussed in the following chapters. The research will concentrate and examine three common measures for value and risk change in total risk, systematic risk and abnormal returns of recent bank M&As in Europe. Those measures will be investigated respectively by studying the variation in volatility, regression β beta, abnormal and cumulative abnormal returns with the help of event studies, which is introduced in a later stage of this research. By the means of this study more light will be shed оn the mixed findings in the existing literature about European M&As. Further explanation of the problem, as well as existing literature is provided in the next section. From managerial point of view, this research will help decision makers and bank executives to make better decisions concerning cross-border M&As, mainly by shaping their expectations on whether target/acquirer risk and value will change and what will be the true effect of a future cross-border expansion. From academic point of view we expect this research to provide significant results and accurately define the consequences of bank M&As, which is anticipated to clear out the confusion and define the consequences of European cross-border mergers. 1.2 Current state of literature and research questions The European banking sector has been studied from numerous points of view, such as analysis on the efficiency effect in the European banking (Vennet, 1996), competition investigation (Mayer, 1990), etc. In fact, the European banking sector deviates in many respects from the one in the United States, which makes it interesting to further investigate how the European market reacts to merger announcement. The most important differences considered in this paper are the economic, cultural, and legal dissimilarities between the two markers, which according to Rad,Beek (1999), are likely to have stronger positive consequence on the diversification effect of cross-border mergers and acquisitions. Another difference between Europe and the US is the ability of European banks to engage in wider range of activities. For example, the commercial banks in United States are forbidden to engage in investment banking, which is not the case in Europe and thus many banks engage in both commercial and investment banking activities. 5

7 With regards to M&As, currently the financial literature in the banking industry seems to provide an abundance of research, but there is also some confusion. The general results in terms of economic value added and increased shareholders wealth are far from settled. Thus, the information and results of banks mergers and their positive effects on shareholders value are widely challenged and their significance is seen as questionable. Since the 1980s, many researchers argued that bank mergers resulted in efficiency gains, such as value creation to shareholders. The market responded positively on the bank M&A news which led to a raise of their stock price. For example, research by Tourani, Van Beek (1999), based on EU bank mergers found that target bank shareholders experience significant positive abnormal returns. In the same time, a research conducted by Amihud, DeLong & Saunders (2002) points out both positive and negative reactions to shareholders value and total as well as systematic risk. Similar findings are also provided in the context of cross-border M&As by Amihud et al. (2002), who stipulated that international mergers of financial institutions, neither increase, nor decrease risk in banking. On the other hand, some earlier researches on bank M&As in 1980s s, such as Berger, Demsetz, Strahan (2009), reported negative or no gains, instead of added value to the shareholders. These findings were further supported by other researches such as the one of Rhoades (1998), thus shedding skepticism over the belief of positive outcome of bank M&As. Because of this skepticism and continuous disagreement between researches, bankers, and the existing empirical studies in this field, this research aims to present a clear view and further assessment of the European bank mergers and acquisitions during the period of January 1 st 2002 until January 1 st 2014 (12 year period). This period of time was selected to capture M&A deals before and during the last Financial Crisis. Since the overall aim is not only to capture the change in risk and value after a merger, but also to divide the sample period in two different parts and further exploit the differences between them. The main logic behind this division is to capture the effect of bank M&A on value and risk in the presence of economic growth during the period of and further collate it with the one of the most recent financial crisis, which took place in the period With the use of the information from the existing literature and the already mentioned methods for conducting this empirical study, we strive to find the answers of the following research questions: Main research question: - How do the European cross-border bank mergers and acquisitions affect the banks value and risk? 6

8 Secondary research questions: - Do bank mergers create value for the shareholders of the bidding/target bank in different economic conditions? - How do bank mergers and acquisitions influence the acquirers bank total risk in different economic conditions? - What is the impact of bank mergers and acquisitions over acquirer s bank systematic risk in different economic conditions? 1.3 Research method In order to achieve the purposes of this research, the method of event studies will be used. The focus in this method is going to be on three key components. The first is the total risk, usually measured by the volatility of host and domestic country. Second is the systematic risk, calculated by the change in beta coefficient of acquirer stock return. Third is the abnormal return, which is the return to equity investors (shareholders) using world, domestic, and host indexes for target. The estimation window will be set to 220 days, which consists of -250 to -30 days prior the merger announcement date in order to avoid any leakage of information. In sum, a few CARs windows will be calculated in order to capture the percentage return before and after the announcement. 1.4 Structure This research will begin with introduction and a deeper overview of past research data and findings (Section II). As described in this academic paper, there are plenty of disagreements between the overall findings in this area that will be further discussed and compared. In the end of Section II, the research questions and the hypothesis will be developed. Section III will be dedicated to the Methodology, which is explained briefly in the next paragraph. Section IV will deal with empirical studies of the data and hypothesis testing, such as event studies that will be developed for the computations of abnormal returns and their respective cumulative abnormal returns. Furthermore, multiple regression analysis for calculating the systematic bank risk measured by beta will be conducted too. Done by regressing the acquirers bank returns on the European bank index, home country bank index and host country bank index, the following results will be summarized in tables. Section V will deal with conclusion of the results found in the previous section and further recommendations. 7

9 1.5 Data In this research, the appropriate data of the DataStream database will be used with the help of some external sources, such as SDC Platinum, in order to determine the right number of bank M&As. Only the banks that are part of this database will be included in this research. A list of the European banks that merged or had been acquired by another bank in different country (cross-border) for 1 st of January st of January 2014 is created with the help of SDC Platinum. Therefore, another dataset will be created for the same time period but for the bank deals that occurred in a Domestic level and will be used as a benchmark. Furthermore, as mentioned above, the main focus will be on banks daily stock returns for domestic and host bank returns, as well as European, home and host indexes for the corresponding data frame based on DataStream. 8

10 Chapter 2 Literature review and hypothesis development In this part of the master thesis, in-depth literature exploration will be provided, as well as hypothesis development and methodology insights. The literature examination will be selective and critical, helping for further defining the hypothesis. Thus, explanations and current view of the reasons why cross-border bank mergers and acquisitions take place, as well as an alternative sight on their effect over value and risk will be provided. 2.1 Bank value creation/destruction based on share price The main goal behind recent European M&As seems to be expanding acquirers bank size in order to sustain competitiveness in the highly intense European market. However, the pre-crisis bank mergers and acquisitions in recent years seem to focus on few important aspects, the main of which is creation of synergies, as well as management related benefits. In fact, for some banks merger can be seen as less costly method for replacing lacking management. There is already an extensive body of literature that covers the pre and post-merger effects on shareholders value, such as the works mentioned in the previous section Tourani et. al.(1999), Amihud et al. (2002), DeLong et al.(2002), etc. However, the purpose of a merger is not always concentrated on creating value for its shareholders. According to management-utility hypothesis, managers interests do not always align with the shareholders one. In fact, if the management utility depends on the firm size, managers are willing to engage in M&A activity regardless of their wealth creation to stockholders, which can be found as a possible explanation for unsuccessful mergers. Furthermore, Roll (1986) argues that managers usually overestimate the future outcomes of an acquisition, since they tend to be overconfident and think they are better of managing new acquisitions than their direct competitors. Additionally, regarding the risks of a bank M&A, the acquirer has a tendency to focus on more risky cross-border acquisitions. John et al. (2000) discussed the incentives of banks that exceed the level of risk exposed beyond the optimal one in the host country. They find it is often done, if the capital requirements, net deposit insurance, and bank closure are properly priced. Following this logic, it can be argued that the management decision to engage in merger or acquisition of another bank can actually jeopardize the value of acquiring bank, instead of improving it. Thus, our first hypothesis is the following: 9

11 H1: Cross-border bank mergers and acquisitions have negative impact on shareholder s value. This might be due to their different nature related to the domestic M&As. This research expects to prove that the main focus of cross-border deals is on reducing risk through diversification and expansion in new markets, even though this does not always create value for the shareholders and in some cases even destroys it. 2.2 Change in bank total risk A major implication of the modern portfolio theory is that the return variance of a portfolio consisting of financial assets can be reduced by diversification. The application of this theory in the banking industry suggests that diversification can reduce the probability of failure. An empirical study by Demsetz, Strahan, (1997) proves this and further confirms the overall believe in the banking literature, that diversification and size go hand in hand. Furthermore, they investigat the portfolio theory and state that in practice bank holding companies diversification will decrease its firm specific risk but leave its systematic risk unaffected (Demsetz, Strahan, (1997)). Moreover, Penas,Unal (2004) point out two complexities which are specific to the bank mergers that can further be important for the change of risk and bondholder gains. First, they consider the fact that as a result of the merger, the combined bank might become too-big-to-fail in the eyes of the federal deposit insurer. Becoming too-big-to-fail allows to exploit the value of the implicit guarantees by letting all uninsured liabilities to actually have an insurance coverage. Second, unlike other industry firms, banks are in fact subject to capital requirements set by regulators, which simply banns shareholders from increasing the leverage and thus compensate for the merger. According to Penas,Unal (2004), shareholders can loose and bondholders can gain, since they appropriate some of the M&A gains. Hence, bondholders of both acquirer and target banks realize significant positive risk and bank mergers are considered as default-risk reducing events. Therefore, combining those recent finding, we set out a second hypothesis as follows: H2: Cross-border bank mergers in Europe have positive impact (decrease) over acquirers total risk (volatility). Confirmation of the hypothesis might be due to diversification phenomena that reduce the return variance or due to the different nature and applied regulations to the financial firms and in this case bank holding companies. 10

12 2.3 Change in bank systematic risk A study of US multinational corporations conducted by Agmon and Lessard (1977) concerned with the beta change after a merger with respect to the home and host country indexes was considered in this research. The outcome of their study can be simply explained in the following way, showing that the return beta of multinational companies is lower with respect to the home market and the beta of the world index is higher. Moreover, after a domestic bank acquisition on a foreign bank, there is change in the income shares coming from foreign and domestic markets. The shares of income that are received from foreign markets are increased, while the shares of the income from the home market are declined. Thus, according to Amihud et al. (2002), who studied this dependence, it is expected that after a merger takes place, the acquirer s return will show a weaker covariance with the home market banking index and a stronger covariance with foreign market banking indexes. Thus, the expectations are that cross-border mergers will increase acquirers beta with respect to the host and world indexes and decrease it with respect to the home market one. Hence, our third hypothesis is as follows: H3: The betas of the acquirer will decline with respect to the home market indexes and will increase with respect to the host and European indexes. This change is expected to be due to shift in the returns between foreign and domestic markets (increase of the returns percentage from foreign markets and decrease of the return percentage from domestic one). 11

13 Chapter 3 Research method 3.1 Methodology and sample Most of the research done for bank mergers and acquisitions is grounded on cross-section analysis and statistical evidence. The reason for this lies in the large number of available data on mergers that took place over time. Most of the criticism on this methodology was developed by Geroski (1988) and Hannan, Wolken (1989), who argued that cross-section analysis does not capture accurately the firm and industry specific characteristics. Taking this into account, this research is focused on three major investigations that are believed to be the most accurate regarding M&As nature. In order to investigate the potential shareholders value creation or destruction, there are two different types of available research possibilities first, based on accounting measures and second, based on event studies of stock returns in finance. This study will focus on the financial measures, such as investigation of stock returns around the announcement date of bank merger or acquisition. More specifically, it consists of time-series analysis of different M&A deals that took place over the selected data horizon. The information about the M&A deals that took place over the selected horizon (1/1/2002 1/1/2014) is carefully selected with the help of SDC Platinum. The data of occurred M&A deals was selected such that the target and the acquirer are both publicly traded commercial banks with available stock price history in Data-stream. The restriction Target and acquirer headquarters in a different country was added as a string for allowing only cross-border deals in the dataset. Moreover, only bidders that have at least 51% of the targets shares were included in order to capture only M&A deals with acquirer majority share after the transaction. The final list of the deals available includes 29 cross-border commercial bank mergers. Moreover, the study also includes Domestic mergers with the same type of criteria and time frame which are further used as a benchmark for the Cross-border model. In this investigation a rather narrow definition of banks is used, since it includes only European commercial banks (bank-holding companies). On the other hand, the definition of Europe is not limited to the European Union, but also includes all Eastern and Western European countries and the Russian federation. Furthermore, this empirical study is divided into two parts - deals before and deals after the September, 2008 financial crisis. Consequently, it not only challenges the basic view of the changes in risk and value before and after M&A, but also captures weather the nature of the financial crisis had some 12

14 impact over the deals. In fact, in theoretical aspect, the investigation aims to capture the possible differences in the M&A deals when the markets are performing well and when they are underperforming as in the case of financial crisis. A variance of the outcome in terms of shareholders value and change in risk might be due to overall underperformance of host/home/world markets and forced mergers by government authorities. 3.2 Event studies First, since this empirical study is concentrated on capturing abnormal returns and change in shareholders value around M&A announcement dates, an event study is conducted. Its purpose is to capture prior and post event stock price reactions to M&A announcement for any possible leakage of information or slowed response on the stock prices. As pointed out in Chapter 1, the estimation window is selected to be from 250 working days to 30 working days prior to the M&As announcement, which takes place at t=0 by definition. Therefore, the event window is set to be [-30, +20] and the following CARs are estimated based on that. The abnormal returns (AR) were calculated during the already selected period. The ARs were defined as the actual returns minus the predicted returns with the help of the Capital Asset Pricing Model (CAPM) as shown in the following equation: ARi, t = Ri, t (αi + βeurope, t RBeurope, t + βhome, t RBhome, t + βhost, t RBhost, t) Where RB Europe,t represents the return on the market index (European index) on day t, RB d,t the domestic (home) index on day t, R h,t the host index on day t and R i,t stands for the return of stock i on day t. And α i and β i are the market model parameters estimated for the stock i. Further, The Cumulative Abnormal Returns (CARs) were calculated over several periods in order to measure the merger influence related to the periods surrounding it. Where CAR is calculated as: +20 CARi = ARi, t Total risk Second, the study also tends to capture any possible change in total bank risk, where the appropriate methods will be deployed, such as stock price volatility. A common measure of total risk is the variance of a bank s stock return. The total relative risk is computed as, dividing the acquirer bank return variance to the variances of home, host, and European indexes. 13

15 TRRi, k = Var(Ri) Var(RBk) Where,TRRi,k = total relative risk of acquirer j, with respect to the European, home, host bank indexes. R i is the daily return on acquirer j and RB k is the return on bank index k (Europe, home, host). The change of the total relative risk is calculated as follows: TRRi, k = TRRi, k (after) TRRi, k (before) Where, before refers to 250 toward 10 days before the merger and after to +10 towards +250 days after the merger announcement was made. 3.4 Systematic Risk The systematic risk is measured by the β (beta) coefficient, also known as regression β. Thus, the acquirers systematic risk is captured from a regression with independent variables for the European performance index as well as host and home indexes, while removing the effect of European bank index (EBI) from them. In order to remove the mentioned effect, home, and host indexes are regressed on the EBI, the respective residuals are obtained from the regression. Furthermore, dummy variables denoted with Db (before) and Da (after) the merger announcement are added, where Db = 0 and Da = 1. Thus, the change in beta is directly obtained as follows: βi, k = βi, k (after) βi, k (before) =θik Where, Δ βjk is the change in beta coefficient of acquirer j on the bank index k. The following regression is used in order to determine the θik, which represent the change in βjk : Ri, t = αi + αidt + βeurope, i RBeurope, t + θi RBeurope, t Dt + βhome, t RBhome, t + θi RBhome, t Dt + βhost, t RBhost, t + θi RBhost Dt 14

16 Chapter 4 Results Introduction of the data This chapter summarizes the results reached by applying the empirical methods mentioned in Chapter 3. Therefore, a deep analysis of the empirical results will be conducted, thus giving some insights about the hypothesis already developed in Chapter 2. We begin with introduction of the collected and used data. After a careful consideration and exclusion of banks from the dataset which were either not listed on DataStream or did not have the complete set of data for the estimation window, we ended up with the following values. Full sample Cross-border M&As 58 Acquirer sample Cross-border M&As 29 Full sample Domestic M&As 108 Acquirer sample Domestic M&As Abnormal and cumulative abnormal returns In this section the results generated by Stata with the help of the methods mentioned above are presented. The time frame (estimation window) for Cumulative Abnormal Returns (CAR) is carefully set from -250 days to -30 days before the announcement date and calculated by the formulas stated in Chapter 3.1. Therefore, the market reactions of both acquirers/targets, as well as only the acquiring banks to the M&A announcement are presented in the following tables for several periods before and after the information was publicly announced Abnormal and cumulative abnormal returns for the whole sample First, a table with results containing Cumulative Abnormal Returns and their respective t-values for both bidders and targets is presented. CARs represent the returns in specified time diapason, where they vary across the announcement date which is at t=0 by definition. 15

17 Table 1. Cross-border and Domestic deals CARs. Full sample Cross-border M&As Full sample Domestic M&As Return (%) t-value Return(%) t-value CAR [-30, -1] -1.76% % CAR [-20, -1] -0.85% % 0.84 CAR [-10, -1] 0.13% % 1.00 AR [0] 5.60% % 2.16 CAR [+1, +10] -0.41% % 1.16 CAR [+1, +20] -1.61% % 0.31 CAR[-20, +5] 4.45% % 2.25 CAR[-10, +10] 5.31% % 2.78 # of observations Looking at the percentage return available in column 1 for Cross-border deals, it is obvious that 20 to 30 days before the announcement was made the return was negative of between -1.8% and -0.8%, which, however, is not supported by the large t-value and thus not significant. Almost the same is observed in the coefficients and the t-values in the second column for Domestic M&As for the time prior the M&As. Looking a little further at the time of the announcement (AR 0), the Cumulative Abnormal Returns are quite large and positive, also significant with respectively 5.6% for cross-border deals and 2.9% for domestic one. This speaks of little to no knowledge and leakage of information before the news for the upcoming deals close to the real announcement dates. Furthermore, it is interesting to see that those abnormal returns for both types of deals during the announcement date were significant and yield a large increase in the share price with respect to the market movements. Consequently, looking at the post event CARs, it is clear that there is a difference between the deals based on different Cross-border flag. In other words, the Cumulative Abnormal Return for cross-border deals decline and turns out to be negative, where the one for the domestic deals stray positive for a short time after the announcement. Therefore, in order to capture this change combined, a CAR calculations on [-20, +5] and [-10, +10] days around the announcement date are presented in the last two rows of Table 1. It is evident for both types of deals that an M&A announcement increases the value to the shareholders. This can be concluded based on the positive coefficients that yield around 5%-6% positive change in the returns with quite strong t-values for significance for both types of deals. These numbers, based on targets and acquirers pre and post announcement performance yield a large enough evidence to conclude that a merger announcement for either cross-border or domestic deal results in a positive gain in shareholders value, which is in contradiction to our already developed hypothesis earlier in Chapter 2. However, this data is based on both target and acquirer stock and cannot be seen as a conclusive to answer the previously developed research questions. Thus, the following table, presenting the change in revenue for the acquirers will be set as a true performance measure. 16

18 Table 2 Acquirers CARs for Cross-border and Domestic M&As. Acquirers Cross-border M&As Acquirers Domestic M&As Return (%) t-value Return(%) t-value CAR [-30, -1] -3.14% % CAR [-20, -1] -2.46% % 0.06 CAR [-10, -1] 0.12% % 1.13 AR [0] 0.50% % 0.81 CAR [+1, +10] 0.01% % CAR [+1, +20] -0.09% % CAR[-20, +5] -1.93% % CAR[-10, +10] 0.64% % 1.06 # of observations Table 2 presents the different CARs for acquirers engaged in both cross-border and domestic deals. However, in respect to the previous findings following bidders and targets, the findings do not seem to be in line to such an extent. Thus, it can be noted that the acquirers involved in Cross-border deals have significant underperformance in the periods before the mergers announcement is made. It is obvious that going from 30 days before the event to the announcement date itself, the CARs slowly levels and even turns positive at day 0. However, observing the shorter period of [-20, +5] presents a negative impact of the announcement over the shareholders value. The t-value shows it is not statistically significant but it is large enough to take serious the already developed hypothesis in Chapter 2, that the banks value will be destroyed rather than enlarged. Moreover, combining those findings with the ones available in column 2 for domestic M&As, it is obvious that they do not align. In fact, the domestic deals show insignificant results with quite small change in return, which seems to be normal for deals that are conducted inland. Furthermore, in order to find out whether there is significant difference between pre and post crisis cross-border performance, the next table presents the corresponding values of the acquirers CARs for both cross-border and domestic M&As in the following periods ( ) as before crisis period and ( ) as during and post crisis one. 17

19 4.1.2 Abnormal and cumulative abnormal returns before and during the crisis Table 3 Acquirers CARs for both periods (Cross-border and Domestic deals) Acquirers Cross-border M&As ( ) Acquirers Cross-border M&As ( ) Return (%) t-value Return (%) t-value CAR [-30, -1] -3.64% % CAR [-20, -1] -2.37% % CAR [-10, -1] -0.52% % 1.23 AR [0] 1.06% % CAR [+1, +10] 0.92% % CAR [+1, +20] 0.23% % CAR[-20, +5] -0.66% % CAR[-10, +10] 1.46% % # of observations 20 9 Acquirers Domestic M&As ( ) Acquirers Cross-border M&As ( ) Return (%) t-value Return (%) t-value CAR [-30, -1] -0.19% % CAR [-20, -1] -0.19% % 0.18 CAR [-10, -1] 0.14% % 1.19 AR [0] -0.07% % 0.86 CAR [+1, +10] 0.18% % CAR [+1, +20] 0.82% % CAR[-20, +5] -0.76% % 0.24 CAR[-10, +10] 0.25% % 1.11 # of observations Combining the findings in a single table, it is obvious that there are some slight differences between the deals related to their cross-border flag. Generally speaking, the findings are mostly insignificant at any level of CAR for both type of M&As. However, using the t-value, we can distinguish that for the period before the crisis the results for cross-border mergers are mostly significant. There is also a tendency for its changes from quite large and significant returns 30 days before the announcement (-3.64%) to increase in return and repaying positive 1% change at the time of the announcement (which is also statistically significant based on the t-value presented). Thus, combining those results in the calculation of the CAR for [-10, +10], it is obvious that engaging in cross-border M&A yields a positive increase in shareholders value, where for the period after the crisis the results turn out negative for all periods before and after the event combined. Moreover, as seen in Table 2, the domestic M&As again do not present any significant results for all the periods the CARs were calculated. A simple proposition for explanation of the results can be the nature of the deals. Since those deals were conducted inland and do not include any outside post factors, the banks engaging in domestic M&As do not capture the diversification effect which is present in cross-border deals. 18

20 Therefore, we cannot make a general conclusion, since the number of deals in the later period of cross-border M&A is limited to nine, but we can formulate that the crisis reflected on (decreased) the overall returns for cross-border M&As. This might be a result due to the raising uncertainty that seems to be true only for those banks that were involved in cross-border deals. Nevertheless, based on all three tables presented above (Table 1,2,3), it can be concluded that from -30 to 10 days prior to the announcement cross-border bank M&As showed significant underperformance (negative return based on CARs) that leveled and turned positive at the time of the announcement and after it. The results on domestic bank M&As used as a benchmark presented much more insignificant and inconclusive outcomes. A table with the acquirers performance (CAR -10,+10) engaged in cross-border and domestic mergers and acquisitions including name, country and announcement date of the deal are available in the appendix. 4.2 Changes in Total Risk As explained briefly in the previous chapter, the total risk for both domestic and cross-border mergers in this research is measured by the difference of the Total Relative Risk (TRR). The TRR is computed as the variance of the bank is divided by the variance of the European, home, and host market indexes. Or in other words, the change in TRR represents the difference between the Total Relative Risk after and Total Relative Risk before a merger is announced. As already stated, the qualification before stands for [-250,-10] days before the merger took place and after for [+10,+250] after the merger was already announced Total Risk for the whole sample ( ) In the following tables the results for change in TRR for both cross-border and domestic mergers full samples, as well as only the acquirer side are presented, related to the different market indexes. The null hypothesis is that the mean is positive (µ > 0) which, if rejected, will lead to confirmation of the already developed hypothesis. It is likely that due to the different factors stated in Chapter 2, such as diversification, the banks variance decreases as a result of a merger. Furthermore, we are mostly interested in the results covering the change of risk related to the home bank index and European one, since they should better explain and align with the banks change in total risk. 19

21 Table 4. ΔTRR for both Domestic and Cross-border Mergers ΔTRR - Cross-border Mergers Observations Mean Std. error Std.dev 95% Confidance Interval t-value ΔTRReurope full sample ΔTRRhome full sample ΔTRRhost full sample ΔTRReurope Acquirer ΔTRRhome Acquirer ΔTRRhost Acquirer ΔTRR - Domestic Mergers Benchmark Observations Mean Std. error Std.dev 95% Confidance Interval t-value ΔTRReurope full sample ΔTRRhome full sample ΔTRReurope Acquirer ΔTRRhome Acquirer Beginning with the change in TRR for the full sample cross-border mergers, it can be noted that the mean is quite high and positive related to all market indexes used. However, they are all not significant at 5% level, but still significant at a 10%, which shows that this positive effect is most likely to occur for both targets and acquirers. Nevertheless, we cannot reject the null hypothesis that the mean is strictly positive, since the values in the lower bound of the confidence interval are negative for all fullsample Cross-border ΔTRR. Furthermore, looking at the acquirer side of the cross-border mergers, it can be noted that there is positive mean for the acquirer related to the European index and negative means related to both host and home indexes with a change of -14.8% and -9% respectively, based on the coefficients. Thus, the results are again not significant at a 5% and 10% level and all of them include positive values in the rejection interval, which in practice cannot be taken as enough evidence neither to reject nor confirm our null hypothesis. However, looking at the results for the change in Total Relative Risk for the full sample Cross-border mergers, it is obvious that it is most likely to have a positive impact over the total risk (volatility) for the banks that are part of mergers occurring between different countries. But the results show a difference between this and the sample looking only at the acquirer side of the deal. There is again no evidence to confirm our hypothesis of a negative mean, but it still seems that it is more likely to have a negative rather than positive impact over the total risk, which is in line with the already developed hypothesis, especially for the change in TRR related to the acquirers home bank index. Thus, the hypothesis of the presence of diversification effect over the acquirers involved in cross-border mergers can be observed to some extent. Next, the findings related to the domestic sample, which are used as a benchmark in this research and present much larger ΔTRR coefficients will be compared. The only difference in calculating the change in Total Relative Risk for the sample of domestic mergers is that the variance is compared only 20

22 with European and home indexes, since there is no host country. Again, for the full sample domestic mergers, e.g. both acquirer and target included, the results suggest positive impact over the change in TRR. However, this time the change in the Total Relative Risk related to the country specific bank indexes is strictly positive at a 5% level with a huge positive change of a 110%. In the same time, observing the acquirers side of the domestic deals, it can be noticed again that there is largely negative impact over the TRR, which is in line with the findings of the cross-border sample. Therefore, this issue has to be addressed, since these findings suggest that the decreasing risk might be due to the merger itself and not to the diversification phenomena. Systematizing the entire set of conclusions made above, it is clear that there is no significant difference between the nature and change in total risk between merges at domestic and cross-border levels for targets and acquirers as a whole, as well as only for the acquirers part of the sample. Consequently, it cannot be concluded that the change (most likely decrease) in the total risk of the banks engaged in a merger is due to the cross-border flag and not the merger itself. As a next step, the following paragraph is dividing the observation samples in two parts, pre and post the last financial crisis, in order to capture any significant change in the total risk reaction to a merger Total Risk before and during the last financial crisis The first group in Table 5 is for the time period and will be discussed simultaneously with Table 4 and Table 6 from this chapter. Table 5 Full sample Cross-border and Domestic mergers for the years ΔTRR - Cross-border Mergers Observations Mean Std. error Std.dev 95% Confidance Interval t-value ΔTRReurope full sample ΔTRRhome full sample ΔTRRhost full sample ΔTRReurope Acquirer ΔTRRhome Acquirer ΔTRRhost Acquirer ΔTRR - Domestic Mergers Benchmark Observations Mean Std. error Std.dev 95% Confidance Interval t-value ΔTRReurope full sample ΔTRRhome full sample ΔTRReurope Acquirer ΔTRRhome Acquirer A first look at Table 5 shows significantly more consistent results. It is clear that there are positive means for all changes in TRR related to the three banking indexes used for the cross-border mergers (in rows 1,2 and 3). In addition to that, it can be concluded that for the time period the mean of ΔTRR related to the European banking index is strongly positive at the 5% level. Since the null 21

23 hypothesis of an increase in the risk is confirmed, this is in contradiction to the overall expectation of a decrease in the risk when engaging in a cross-border deal. Looking at the acquirer side of the Crossborder mergers, we cannot be as confident of the true outcome since the t-value shows that the results are not statistically significant. Yet, the results yield negative values of the change in TRR related to the variances of the banks with the variances of the home and host indexes, which is still in line with the expectation of a decrease in risk due to diversification. Furthermore, the domestic mergers data also provides statistically significant results for a positive effect of mergers on the total bank risk. For the full sample of both targets and acquirers, we can confirm that the mean of the change in Total Relative Risk is significantly larger than zero, thus confirming the null hypothesis. Still, the results for the acquirer side of the mergers are not significant but quite eloquent. Thus, in the years before the last financial crisis an existing difference between the change in total risk engaging in cross-border and domestic deals can be found. Consequently, in order to be able to distinguish the differences between before and during the crisis samples, Table 6 is presented with the results covering the crisis period ( ). Table 6 - Full sample Cross-border and Domestic Mergers for the years ΔTRR - Cross-border Mergers Observations Mean Std. error Std.dev 95% Confidance Interval t-value ΔTRReurope full sample ΔTRRhome full sample ΔTRRhost full sample ΔTRReurope Acquirer ΔTRRhome Acquirer ΔTRRhost Acquirer ΔTRR - Domestic Mergers Benchmark Observations Mean Std. error Std.dev 95% Confidance Interval t-value ΔTRReurope full sample ΔTRRhome full sample ΔTRReurope Acquirer ΔTRRhome Acquirer The results provided in Table 6 are striking related to the previously observed ones. At first glance, it can be noticed that the means for all of the cross-border deals are lower than zero. Furthermore, they are not significant and we cannot confirm our hypothesis, but it becomes clear enough to show the difference between pre and post crisis mergers performance and expectations. The same tends to be the main finding over the domestic benchmark used in this research. However, the acquirer side of the domestic deals changed dramatically. In fact, it rejects the null hypothesis of a strictly positive impact of mergers on acquirers total risk with a huge mean of change for the total risk in respect to the European banking index which is significant at a 5% level. These findings provide interesting insights 22