VALUE AND COST OF LOCAL INFORMATION:
A STUDY IN CROSS-BORDER MERGERS AND
ACQUISITIONS
University of Amsterdam Amsterdam Business School
MSc Business Economics, Finance track Master’s Thesis
July 7th 2016
Student: Martijn Vink Student number: 10211063 Supervisor: Tolga Caskurlu
S
TATEMENT OF ORIGINALITYThis document is written by Student Martijn Vink who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
A
BSTRACTThis study investigates the value and cost of having ‘local’ information in a cross-border merger or acquisition. ‘Local’ information is defined as having a board member or C-suite executive native to the target’s country. The underlying assumption is that he or she possesses the necessary information to break value decreasing cultural and financial barriers. This study finds that the geographical distance between the acquirer and target is an important determinant of the value: having ‘local’ information has a negative effect on the acquirer’s announcement return when the acquirer and target are relatively close to each other. As the distance between the two parties increases, ‘local’ information becomes more valuable. There seems to be a threshold of approximately 13 to 15 thousand kilometers from whereon ‘local’ information becomes favorable when involved in a cross-border deal. This study also shows that ‘locals’ receive a relatively higher compensation increase following a deal than ‘non-locals’ when controlling for tenure.
T
ABLE OFC
ONTENTS1.
I
NTRODUCTION... 5
2.
L
ITERATURE REVIEW... 8
2.1.MOTIVES OF CROSS-BORDER MERGERS AND ACQUISITIONS ... 8
2.2.CROSS-BORDER MERGERS AND ACQUISITION PERFORMANCE ... 8
2.3.CROSS-BORDER CULTURAL DIFFERENCES ... 9
2.4.CROSS-BORDER FINANCIAL DIFFERENCES ... 11
2.5.COST OF ‘LOCAL’ INFORMATION ... 11
3.
R
ESEARCH QUESTION AND HYPOTHESES... 12
4.
M
ETHODOLOGY... 13
4.1.EVENT STUDY METHODOLOGY ... 13
4.2.APPLICATION OF EVENT STUDY METHODOLOGY ... 14
4.3.ORDINARY LEAST SQUARES (OLS) ESTIMATIONS ... 16
4.4.TESTING FOR SUBSAMPLE DIFFERENCES ... 19
4.5.MEASURING DISTANCE ... 20
4.6.ENDOGENEITY ISSUES ... 20
5.
D
ATA AND DESCRIPTIVE STATISTICS... 21
5.1.DATA AND DATASET CONSTRUCTION ... 21
5.2.DESCRIPTIVE STATISTICS ... 23
6.
R
ESULTS... 28
6.1.CROSS-BORDER EFFECT ... 28
6.2.VALUE OF ‘LOCAL’ INFORMATION ... 33
6.3.COST OF ‘LOCAL’ INFORMATION ... 41
7.
R
OBUSTNESS CHECKS... 43
8.
C
ONCLUSION AND DISCUSSION... 46
9.
A
PPENDICES... 49
1.
I
NTRODUCTIONAs technological improvements make our world smaller and smaller, the world’s economy becomes increasingly intertwined. Combined with the everlasting competitiveness of firms worldwide, cross-border mergers and acquisitions will likely to become even more important in the near future. Thomson Reuters reported a 1.3 trillion dollar total cross-border deal value last year; this represents 36.9% of the total global merger and acquisition deal value. Cross-border deals have not seen deal values this high since 2007. When involved in a cross-border deal, cultural- and financial differences between the two countries represent a barrier for the acquirer. This barrier could potentially be detrimental for the flow of information, which is crucial in the process, as this has a positive influence to decision making (Ishii and Xuan, 2014). This study seeks to find a solution for this barrier by introducing ‘local’ information in the form of a board member or C-suite executive that is native to the target’s country. This ‘local’ board member or C-suite executive has ‘local’ information as he or she knows the culture of the foreign target and thus knows how to mitigate the barriers. More specifically, this study is focused on the (abnormal) acquirer’s announcement returns during a cross-border merger or acquisition. Is there any value in having ‘local’ information? In addition to this, this study will also examine the cost of this ‘local’ information. Do ‘local’ board members or C-suite executives receive a relatively higher increase in compensation when involved in a cross-border deal?
First of all, this study seeks to investigate whether there is a so-called cross-border effect in cross-border mergers and acquisitions. The hypothesis follows from the findings of Moeller and Schlingemann (2005) and states that there is a negative cross-border effect. In other words: the acquirer’s announcement return is relatively lower when the merger or acquisition is a cross-border deal rather than a domestic deal.
Secondly, this study will focus on the value of ‘local’ information. This will be the primary focus of this paper. The hypothesis regarding this subject is based on the findings of Chatterjee, Lubatkin, Schweiger and Weber (1992) and Datta and Puia (1998). They find that an acquiring firm does not have a comprehensive understanding of the target’s country/market and that this lack of understanding is detrimental for the performance of the deal. The key assumption this study makes, is that a ‘local’ board member or C-suite executive has the ‘local’ understanding to mitigate this problem. The semi-strong version of the Efficient Market Hypothesis (Malkiel and Fama, 1970) is of
great importance in this study. As the EMH assumes, all available data is acknowledged in the share price of the company. Investors will thus value the ‘local’ information as they drive the price of a particular share. This study will make an attribution to the growing field of literature which focuses on the cultural side of mergers and acquisitions. More specifically, it will add to the field that addresses cultural differences in cross-border deals.
Last of all, the cost of this ‘local’ information will be examined. If the ‘local’ board member or C-suite executive actually has, or is perceived to have valuable ‘local’ information, the company will be willing to pay for this. The differential of the yearly compensation increase will be the main focus here. Put more simply: does the compensation of a ‘local’ board member or C-suite executive increase relatively more in the year of the deal than the compensation of a ‘non-local’ board member or C-suite executive? The hypothesis is that it will; ‘local’ information has value and will therefore also has a higher cost. To my best knowledge, this study will be the first study that links the nationality of a board member or C-suite executive to compensation data in the situation of a cross-border merger or acquisition.
The design of this study is subjective to endogeneity issues, primarily due to the possibility of reversed causality. Does the British firm expand to Belgium because it has a Belgian CEO or does the British firm have a Belgian CEO because it wishes to expand to Belgium? When testing for the value of ‘local’ information, the possible existence of endogeneity is acknowledged when drawing conclusions but at the same it is assumed to be not problematic enough to change the design of the study. However, when using person specific information to test for the cost of ‘local’ information, endogeneity is expected to bias the results. This study mitigates this problem by using a certain cutoff point in tenure of the board members and C-suite executives. The reason for this is that it is deemed unlikely that, for example, an Italian company hired a Spanish CFO because it wishes to expand to Spain five years later. The use of a specific cutoff point is not based on existing literature and is therefore of arbitrary nature. This study will thus use multiple different cutoff points as a form of a robustness check.
Data from 3,721 global mergers and acquisitions in the period 2010-2015 obtained from Orbis, Zephyr, Execucomp and Datastream will be merged to calculate the Cumulative Average Abnormal Returns by performing an event study as described by MacKinlay
(1997). This will then be used as the dependent variable in multiple univariate and multivariate OLS regressions. Depending on the hypothesis being tested, a combination of independent variables will be used to explain the relationship.
This study finds that there is a negative cross-border effect in mergers and acquisitions. This is in line with the findings of Moeller and Schlingemann (2005) and acts as a starting point for further research. The results regarding the value of the ‘local’ information are mixed: having ‘local’ information leads to relatively lower acquirer’s announcement returns, but as the distance increases between the acquirer and target, the value of ‘local’ information increases. There seems to be a threshold of about 13-15 thousand kilometers whereas firms are better off when having ‘local’ information. In addition to this, this study also examined what the effect is of having the same language in a cross-border merger or acquisition. The rationale behind this is that language is one of the main determinants of cultural barriers and that the relative value of ‘local’ information thus decreases when the language of the acquirer is identical to the target’s. Although the results are not highly significant, there is reason to believe that cross-border deals in countries with the same language perform better when there is no ‘local’ information. This study provides a couple of explanations for the surprising finding that ‘local’ information does not always is perceived as valuable. Examples of explanations are overconfidence by the company of having ‘local’ information and tunnel-vision resulting from having a foreign board member or C-suite executive. Last of all, this study finds that company compensate ‘local’ board members and C-suite executives relatively more richly compared to their ‘non-local’ peers.
This paper will have the following structure. First of all, it will introduce the concept of cross-border mergers and acquisitions and discuss the existing literature in section 2. The research question and derivation of the exact hypotheses will be discussed in section 3. The methodology used to examine these hypotheses will be discussed in section 4. In section 5, the collected dataset will be described and descriptive statistics will be given. The results of the statistical analyses will be presented and interpreted in section 6. To ensure a causal relationship, robustness checks will be presented in section 7. And last of all, the results will be discussed and summarized in section 8.
2.
L
ITERATURE REVIEWThis part will discuss the motives firms have when they expand abroad by merging with or acquiring a firm. Then the existing literature discussing the performance of cross-border mergers and acquisitions will be set out. After that, the positive and negative effect of cultural and financial differences will be discussed. Finally, the existing literature regarding the cost of ‘local’ information will be discussed.
2.1.MOTIVES OF CROSS-BORDER MERGERS AND ACQUISITIONS
As Martynova and Renneboog (2006) stated, there are multiple motives for a firm to pursue a merger or acquisition in general. These motives include, but are not limited by, a merger or acquisition being a major strategic tool to increase market power and market share, increase efficiency through the obtained synergies and the acquisition of complementary resources. The motives of cross-border acquisition are similar to the motives of a domestic acquisition except for the fact that a cross-border acquisition is primarily used to enter a new, foreign market in a faster manner. Erel, Liao and Weisbach (2012) conducted research on the determinants of cross-border mergers and acquisitions. They found that geography is an important determinant: e.g. it is more likely for a firm to acquire another firm in a neighboring country than to acquire a firm on the other side of the world (a cultural interpretation of this finding will be discussed more extensively in part 2.3. and distance will also be used as a control variable). Furthermore, they find that higher economic development and higher accounting standards in the home country increase the probability of a firm to be the acquirer in a cross-border merger or acquisition. In addition, Erel et al. (2012) find that exchange rates and stock performance tend to affect the propensity to engage in a cross-border merger or acquisition. An appreciated currency or a high performing stock market increases the probability of a firm to be the acquirer in a cross-border merger or acquisition. This is in line with the findings of Shleifer and Vishny (2003) who find that misevaluations in the stock market fuel high valued firms to purchase low valued firms.
2.2.CROSS-BORDER MERGERS AND ACQUISITION PERFORMANCE
The existing literature does not have a clear-cut answer to the question whether cross-border mergers and acquisitions under- or over perform domestic mergers and acquisitions. Doukas and Travlos (1988) find mixed results in their study. They find that shareholders of internationally expanding domestic firms have insignificant positive
returns at announcement. Shareholders of multinationals expanding in a country in which it is already active have insignificant negative returns. However, when these multinational firms expand in countries they are not yet active in, shareholders experience a positive and significant return. Cakici, Hessel and Tandon (1996) also find mixed results. They find that foreign acquirers experience a positive and significant abnormal return of two percent over an estimation period of [-10,+10] days. However, they also found that U.S. firms have no positive returns when acquiring a cross-border firm in the same time period. Chakrabarti, Gupta-Mukherjee and Jayaraman (2009) conduct research on the short- and long-term effect of cross-border mergers and acquisitions. They find that acquiring firms, on average, experience significant positive returns of 0.71% in a three-day period. They also find that cross-border acquisitions tend to perform better in the long-run when the cultural difference between acquirer and target is larger. On the other hand, there are also studies that find the opposite effect. Based on stock and operating performance measures, Moeller and Schlingemann (2005) find that cross-border acquisitions perform less than domestic acquisitions. The performance of cross-border mergers and acquisitions is still a topic that is up for debate. This research will contribute to this puzzle by finding a different explanation for the (lack of) success of a cross-border merger or acquisition.
2.3.CROSS-BORDER CULTURAL DIFFERENCES
Every country has its own culture. The Cambridge English Dictionary defines culture as “the way of life, especially the general customs and beliefs of a particular group of people at particular time”. Just like every country, every firm has its own culture as well. The culture of a firm reflects national culture, professional subculture and the firm’s own history (Hofstede, 1980). Hofstede (1980) also introduces the five main dimensions of national business culture: power-distance, uncertainty avoidance, individualism, masculinity and long termism. Every country scores differently on each of these dimensions and thus every country has a different business culture. For example: where the power-distance in Sweden is relatively low, businesses in France have a much greater power-distance. And where the Netherlands score low on masculinity, Germany, Switzerland and the United Kingdom all score relatively high.
In the existing literature there are two main and contrarian views on the effect of these cultural differences on merger and acquisition performance: the positive view suggests
that cultural differences have a positive effect on cross-border acquisition performance and the negative view suggests that cultural differences have a detrimental effect on performance. Both of these views will be discussed in this section.
Stahl and Voigt (2008) combined nine different analyses to create a meta-analysis on the acquirer’s announcement return in a merger or acquisition with cultural differences. They find that the weighted mean effect size is significantly positive at 0.08. In other words: the acquirer’s announcement returns in a merger or acquisition with cultural differences is 8% higher. Morosini, Shane and Singh (1998) find empirical support for the notion that cultural differences have a positive effect on cross-border acquisition’s performance. They especially find evidence for improved performance for the acquisitions where the cultural difference between acquirer and target is greater. Larsson and Finkelstein (1999) have conducted research with a resource-based view. They find that a merger or acquisition of a culturally different target could result in a competitive advantage by giving the acquirer access to the unique and valuable capabilities which are inherent to a different culture. Barkema and Vermeulen (1998) and Vermeulen and Barkema (2001) find that acquiring firms with cultural differences increases a firm’s ventures in the futures, increases innovation and boosts the development of new knowledge.
Krug and Nigh (2001) collected survey data from 284 top U.S. executives which were involved in cross-border merger and acquisitions. They found that more than 90% of these executives stated that cultural differences had a negative effect on the merger. Examples of the reasons these executives gave were: communication problems, difficulty developing trust, lack of sensitivity and management style differences. Vaara (2003) finds four specific ‘irrational’ characteristics that often have a negative effect in an effective cross-border organizational integration: inherent ambiguity, cultural confusion, organization hypocrisy and issue politicization. Cultural differences in a cross-border merger can also have a negative effect on shareholder gains (Chatterjee, Lubatkin, Schweiger and Weber, 1992). They state that the acquirer’s management should pay at least as much attention to the cultural fit as they do to the strategic fit in a premerger search process because the cultural differences are generally negative, large and ignored. Datta and Puia (1995) find that cultural differences results in the acquiring firm overpaying the cross-border target firm because they do not have an adequate understanding of the foreign market. They also find, similar to Ollie (1990) and Cartwright and Cooper (1992), that cross-border mergers and acquisitions experience
problems in post-acquisition consolidation. All of the above suggests that cultural differences have a detrimental effect on cross-border mergers and acquisitions. This research will contribute to the existing literature by conducting research on a possible solution to this problem: having ‘local’ information present in the form of a board member or a C-suite executive who is a native in the target’s country.
2.4.CROSS-BORDER FINANCIAL DIFFERENCES
Besides the cultural aspect, there is also a financial aspect in cross-border mergers and acquisitions. Angwin (2001) states that the Anglo-American system has the tendency to find all truth in numbers. Accountants therefore dominate the financial due diligence process in mergers and acquisitions. However, in other countries, this tendency to find truth in numbers might be less severe. This could potentially trigger uncertainty of the acquisitions performance when considering a deal with a company from a more exotic country. Moeller and Schlingemann (2005) find that cross-border acquisitions have announcement returns of approximately hundred basis points less than domestic acquisitions. They state that a potential source for this cross-border effect is the legal system in the target’s country. Because the legal system is a very country-specific, local knowledge could potentially be very valuable.
2.5.COST OF ‘LOCAL’ INFORMATION
Bliss and Rosen (2001) foundthat mergers and acquisitions have a positive net effect on the total compensation of a CEO in the period of 1986-1995. Grinstein and Hribar (2004) found that CEOs with more power to influences the decisions of the board, receive higher compensation. They also found a positive relationship between the effort a CEO puts in the deal and the compensation he or she receives. This can be translated into this study as follows: it is plausible that a ‘local’ board member or C-suite executive puts in more effort than a ‘non-local’ board member or C-suite executive because he or she has country-specific knowledge. When this is the case, he or she is expected to be compensated more richly when following the conclusions of Grinstein and Hribar (2004). To the best of my knowledge, there is no existing literature that focuses on the actual cost of a ‘local’ board member or C-suite executive. Therefore, this study will provide a first insight in this topic.
3.
R
ESEARCH QUESTION AND HYPOTHESESThe main focus of this study will be on what the effect of ‘local’ information is on the announcement returns of the acquirer in a cross-border merger or acquisition. To do so, the following research question must be answered: “What is the effect of ‘local’ information on the acquirer’s announcement return in a cross-border merger or acquisition in the period of 2010 until 2015?” Once again: ‘local’ information represents a board member or C-suite executive who has the same home country as the target’s country. So for example, a German board member of a Dutch company is considered to be ‘local’ if the Dutch company is acquiring a German firm. This paper will have three separate research questions that all have its own hypothesis. Each of those will be discussed in the following paragraphs.
First of all, this study will focus on the cross-border effect. In the existing literature there is still no consensus about the performance of cross-border mergers and acquisitions. The literature is divided into two camps that state cross-border deals under- or outperform domestic deals. This research will first focus on the question: “Is there a cross-border effect in mergers and acquisition deals in the period of 2010 until 2015?” The hypothesis is based on the findings of Moeller and Schlingemann (2005) and is as follows:
Hypothesis 1: there is a negative cross-border effect for mergers and acquisitions
Although there is no consensus on what the effect of cultural differences is on the performance of the deal, there seems to be consensus about the notion that culture does affect the outcome of the merger. The second hypothesis in this paper will build on the statements of Chatterjee et al. (1992) and Datta and Puia (1998). Their statements imply that an acquiring firm does not have a comprehensive understanding of the target’s country/market and that this lack of understanding is detrimental for the performance of the deal. This research will then make the assumption that a ‘local’ board member or C-suite executive has a better understanding of the country/market. Another assumption this research makes, is the notion of the semi-strong version of the Efficient Market Hypothesis of Malkiel and Fama (1970). This version of the Efficient Market Hypothesis states that all publicly available information will immediately be absorbed by market participants and will thus be reflected in a firm’s stock price. This leads to the research
question: “What is the effect of ‘local’ information on the acquirer’s announcement return in a cross-border merger or acquisition in the period of 2010 until 2015?” The hypothesis, which is the main focus of this research, will be as follows:
Hypothesis 2: there is a positive relationship between having ‘local’ information and the announcement returns of the acquirer in a cross-border merger or acquisition.
In addition, this research will also examine what the cost of this ‘local’ information is. As we all know, there is no such thing as free lunch. When this ‘local’ information is valuable to a firm, the willingness to pay more for this information becomes greater. The research question will be: “What is the relative cost of having ‘local’ information in the form of a board member or C-suite executive when pursuing a cross-border merger or acquisition?”. The hypothesis will be as follows:
Hypothesis 3: there is a positive relationship between being a ‘local’ board member or C-suite executive and annual compensation.
4.
M
ETHODOLOGYIn this part the methodology needed to answer the research questions will be discussed. First, the general event study methodology will be explained. Then, the application of this methodology on the hypotheses will be discussed. Finally, possible endogeneity issues will be discussed and solutions to these issues will be presented.
4.1.EVENT STUDY METHODOLOGY
As MacKinlay (1997) states, an event study measures the impact of a specific event on the value of a specific firm. The impact of the event will be incorporated into the stock price by the semi-strong version of the Efficient Market Hypothesis as discussed in section 3 (Malkiel and Fama, 1970). There is a clear-cut procedure which must be followed when conducting an event study. This procedure is as follows:
− Choose an event of interest
− Identify the event window; the period over which the prices will be examined − Identify the firms to investigated
− Define the estimation window − Calculate the abnormal returns
− Compute Cumulative Abnormal returns (CAR) and Cumulative Average Abnormal Returns (CAAR)
Figure 1 shows the timeline of an event study. Time t=0 represents the event of interest. The estimation window, which is sometimes referred to as L1, is defined as L1 = T1 – T0.
The event window is defined as L2 = T2 – T1. On occasion the post event window,
defined as L3 = T3 – T2, is included in the event window to increase the robustness of the
study (MacKinlay, 1997).
Figure 1
MacKinlay (1997): This event study timeline makes a distinction between the estimation window, the event window and the post event window.
4.2.APPLICATION OF EVENT STUDY METHODOLOGY
The event study methodology of MacKinlay (1977) will be used to test hypothesis (1) and (2). In this part, the application of this methodology will discussed step-by-step following the steps given in part 4.1.
The event of interest in this study is the announcement of a firm to merge or to acquire another firm. Thus, the announcement date is the event date, T0.
There are four different event windows in this study to examine what the announcement effect is on the returns of the acquirer. These event windows are [-10,+10], [-5,+5], [-2.+2] and [-1,+1]. Strictly speaking, only the event window [-1,+1] estimates the true announcement effect. However, it is possible that information leaks before the actual official announcement. Under the semi-strong Efficient Market Hypothesis (Malkiel and Fama, 1970), this information will be immediately incorporated in the stock price. When only examining the period [-1,+1], some of the effect will thus be lost. Hence, also longer event windows are used. The disadvantage of a longer event
window is that the results become less reliable due to other events that have an effect on the returns of that particular firm and thus bias the findings.
The investigated firms in this event study consist of all acquiring firms in the period of 2010 until 2015. The acquiring firms have to be listed on a stock exchange and have to have completed a merger or acquisition with a deal value equal to or greater than 20 million dollars. More information about the data and dataset construction can be found in section 5.1. of this paper.
To estimate the normal returns of a firm, the actual returns will be calculated by the following simple equation:
!"# = %&
%&'(− 1 (1)
Where Rit is the return of stock i on day t, Pt is the price of stock i on day t and Pt-1 is the price of stock i on day t-1. The actual returns of the estimation period, which has an interval of [-120,-20], are used to calculate the normal returns of the firm. This can be done with a number of different models. Examples are the constant mean return model, the market return model, the market model and the Capital Asset Pricing Model (CAPM). The constant mean return model assumes that returns can differ by company, but are constant over time for that particular company. The market return model assumes that the returns of a company follow the market precisely. The market model builds on the market return model but assumes that each company has its own sensitivity to the benchmark market. This sensitivity (measured by β) is less than 1 if the company’s price increases (decreases) less than the market’s price and vice versa. The CAPM takes into account the asset’s sensitivity to non-diversifiable risk, the expected return of the market and the expected return of a risk-free asset. This study will use the market model because it allows every company to have its own sensitivity to its home market. In other words: when a French company acquires a Swiss company, the expected return of the French company will be calculated by measuring its sensitivity with respect to the CAC40. The target’s country will thus have no impact on the expected return. To calculate the expected return, the following formula will be used:
+[!"#] = /"!0#+ 2"# where + 2"# = 0 (2)
456 2"# = 78"9
/" = :;<=>"=?@A(CD,CF)
H=>"=?@A(CF)
This study will also use other models to calculate the expected normal returns as a robustness check. The results of these robustness check are discussed in section 7.
The estimation window for this research is determined at [-120,-20]. This is in line with the existing literature. The start of the estimation period at T0 = -120 is distant
enough to determine the ‘normal’ movement of the asset, while the end of the estimation period at T1 = -20 is still far enough from the event itself to prevent possible information
leaks to interfere with the estimation.
The abnormal return is then calculated for every day in the event window L2.
This is done by the following formula:
I!"# = !"#− +[!"#] (3)
Finally, the Cumulative Abnormal Returns (CAR) and Cumulative Average Abnormal Returns (CAAR) will be calculated. The CAR is the simple accumulation of the abnormal returns ARit in the event window. The CAAR is then the CAR divided by the number of observations in the event window. Both equations are given below:
JI!"(K(,KL) = KL I!"#
#MK( (4)
and
JII!(K(,KL) = ?N
DJI!"(K(,KL) (5)
4.3.ORDINARY LEAST SQUARES (OLS) ESTIMATIONS
The Cumulative Average Abnormal Return (CAAR) will eventually be used as the dependent variable in a univariate and multivariate Ordinary Least Squares regressions. Each of the three hypotheses have their own specific regressions, which are now discussed.
Hypothesis 1 seeks to test whether there is a negative relationship between a cross-border merger or acquisitions and the announcement returns of the acquirer. To test this, the following specification will be used:
JII!K(,KL = ON+ /NJ6PQQRP6ST6" + /9UPVW6PXQ" + 2" (6)
Where CAAR(T1,T2) is the Cumulative Average Abnormal Return of one of the four event windows as explained in part 4.2. Furthermore, Crossborderi is the main variable of interest and is a dummy variable that equals 1 if the merger or acquisition is cross-border and 0 if otherwise. The control variables are inspired by the research of Ishii and Xuan (2014) and consist out of the logarithm of the total assets, Tobin’s Q, cash flow divided by total assets, relative deal value (defined as deal value divided by total assets), a dummy variable for pure stock deals and a dummy variable for mergers and acquisitions in a related industry. All the variables are winsorized at the 5%-level when deemed appropriate. The coefficient of interest is /N. When this coefficient is significant and negative, we can
conclude that cross-border mergers and acquisitions underperform compared to domestic mergers and acquisitions.
Hypothesis 2 seeks to investigate whether there is a positive relationship between having ‘local’ information in the form of a board member or C-suite executive and the acquirer’s announcement returns. To test this, the following multivariate regression will be used:
JII!K(,KL = ON+ /NYPU5XSZ[[\"+ /9]^QW5VUT" + /_YPU5XSZ[[\ ∗ ]^QW5VUT" + /aJPVW6PXQ" + 2" (7)
Localdummyi is the main variable in this formula and is a dummy variable that equals 1 if there is at least one ‘local’ board member or C-suite executive and 0 otherwise. Distancei represents the distance between the acquirer’s and target’s country measured in thousands of kilometers (see section 4.5. for more information about the methodology used to measure distance). The rationale behind using a variable for distance is that countries close to each other are more likely to have the same cultural values and ‘local’ information would thus be less valuable. This variable is also interacted with Localdummyi. /N and /_ are the main coefficients of interest. Depending on the outcome of the tests
does not lead to a better performing merger or acquisition. Please keep in mind that only the data of cross-border mergers and acquisitions will be used in these particular regressions. Rather than controlling for distance, one could also control for the spoken language in the acquirer’s and target’s country. Having an identical language in a cross-border merger is valuable in the sense that in enhances the flow of information between the two parties. This, in turn, implies that having ‘local’ information is relatively less valuable compared to the situation where the languages are not identical. To test this, the following additional formula will be used:
JII!K(,KL = ON+ /NYPU5XSZ[[\"+ /9bSTVW^U5XX5VcZ5cT"+ /_YPU5XSZ[[\ ∗ bSTVW^U5XX5VcZ5cT" +
/aJPVW6PXQ"+ 2" (8)
Where Identicallanguagei is a dummy variable that equals 1 if the languages of the acquirer and target are identical and 0 otherwise.
Equation (7) can also be modified to exactly test what the marginal effect of additional ‘local’ board members or C-suite executives is. This results in the following equation:
JII!K(,KL = ON+ /NYPU5XdT6UTVW5cT"+ /9]^QW5VUT" +
/_YPU5XdT6UTVW5cT ∗ ]^QW5VUT"+ /aJPVW6PXQ"+
2" (9)
where Localpercentagei is the percentage of ‘local’ board member or C-suite executives. The specification and interpretation is identical to those of equation (7).
The third and final hypothesis seeks to investigate whether there is a positive relationship between being a ‘local’ board member or C-suite executive and compensation increase when the firm is pursuing a cross-border merger or acquisition. More specifically: does compensation increase more when a board member or C-suite executive provides the company of ‘local’ information (compared to a ‘non-local’ board member or C-suite executive)? To do this, the average normal compensation increase (decrease) and the increase (decrease) at the time of a deal have to be calculated. Once this is done, the following regression will be used:
ΔJP[dTVQ5W^PV ^VU6T5QT = ON+ /NYPU5X" + /9UPVW6PXQ" + 2" (10)
where the dependent variable is the proportional difference between a ‘normal’ increase (decrease) and the increase (decrease) after a deal. To illustrate this, imagine the following example. A ‘local’ CEO receives an average yearly 5% compensation increase over the years when there is no merger or acquisition. His fellow C-suite executives receive the same average 5% increase. However, in the year of a cross-border merger, the ‘local’ CEO receives an increase of 30% while other, ‘non-local’ executives only receive a 15% increase. The ΔCompensation increase for the ‘local’ CEO equals 500% (30% is 500% greater than 5%) while the ΔCompensation increase for the ‘non-local’ executives equals 300%. This methodology eliminates the differences between board members and C-suite executives that are present in the starting situation (when there is no merger or acquisition). Locali is a dummy that equals 1 if the board member or C-suite executive is ‘local’. The control variables are inspired by the research of Brick, Palmon and Wald (2006) and include tenure, Tobin’s Q and relative deal value. All of the financial control variables are the one year lagged values. The coefficient of interest in this equation is /N.
If the test shows a positive and significant value for this coefficient, it can be concluded that a ‘local’ board member or C-suite executive has a relatively higher annual compensation compared to a ‘non-local’ board member or C-suite executive.
4.4.TESTING FOR SUBSAMPLE DIFFERENCES
In part 5.2., the coefficients of different subsamples will be tested to determine whether or not they statistically differ from each other. The model used for this test is the Welch’s t-test. The Welch’s t-test is a test that builds on the Student’s t-test and is used to test whether two means are equal when variance and sample size are unequal (Keller and Gaciua, 2012). Welch’s t-test is given by the following formula:
W = f(gfL h(L i(jiLhLL
where: Sl = h(L i(jhL L iL L h(L i( L i('(j hLL iL L iL'( (12)
where mN denotes the mean of the first coefficient, QN9 denotes the variance of the first
coefficient and n1 denotes the sample size of the first coefficient. The critical values of
the Welch’s t-test are identical to the critical values of the Student’s t-test when the degrees of freedom are very large.
4.5.MEASURING DISTANCE
This study uses the distance between the acquirer’s and target’s country as a sort of control variable. The rationale and assumption behind this is that countries close to each other are likely to have similar values as discussed by Hofstede (1980). This implies that having ‘local’ information is less valuable when countries are close to each other and thus more valuable when countries are far apart. This study follows the methodology of Uysal, Kedia and Panchapagesan (2008) regarding the determination of the distance between countries. To do this, the Haversine formula is used (see appendix A.1. for further details). Due to the availability of data, this study used the geographical center of an acquirer’s or target’s country to determine the distance instead of the exact location of the company. It is assumed that this will have a minimal impact because cultural values are assumed to be the same in the entire country.
4.6.ENDOGENEITY ISSUES
Endogeneity occurs when an explanatory variable is correlated with the error term (Stock and Watson, 2006). Endogeneity can be caused by a measurement error, autoregression with autocorrelated errors, reversed causality and omitted variable bias. This research is primarily subjective to two of the sources of endogeneity: reversed causality and omitted variable bias.
This research implicitly assumes that the composition of the board of directors and C-suite executives is a given. The choice to expand internationally by either merging with or acquiring of a firm is subsequent to the formation of management. When this assumption holds, endogeneity caused by reversed causality would not affect the
estimations. However, it is plausible that a firm explicitly seeks to find a foreign, ‘local’ board member or a C-suite executive because they wish to expand abroad. In that way, reversed causality will result into biased estimation results. This study acknowledges that reversed causality might be present, but due to the way the research is designed, the consequences are not expected to be problematic. More specifically, this study aims to find whether the public believes that having ‘local’ information is advantageous in a cross-border deal.
This endogeneity issue also arises when testing the third hypothesis. This hypothesis states that there is a positive relationship between being a ‘local’ board member or C-suite executive and annual compensation. To mitigate the endogeneity issues, only board members and C-suite executives with a long tenure are included in the final sample. The rationale behind this is as follows: it is unlikely that a firm hires a ‘local’ board member or C-suite executive multiple years before the merger or acquisition. This study has multiple cutoff points because the cutoff points are not based on existing literature and are thus of arbitrary nature. The length of minimum tenure lays between 3 and 6 years.
Omitted Variable Bias (OVB) is the bias caused by omitting a variable that actually has a correlation with the dependent variable. This study also acknowledges that OVB is plausible in this setting. However, the consequences are not expected to be problematic. The control variables are based on existing literature.
5.
D
ATA AND DESCRIPTIVE STATISTICS5.1.DATA AND DATASET CONSTRUCTION
The data used for this study can be divided into five parts: deal-data, financial data, personal data, distance data and compensation data. Each of these datasets will be discussed.
The Zephyr database contains data of public and private mergers and acquisitions globally. The data needed for this study consisted out of all deals in the time period of 2010 until 2015. This time period is chosen because it is most recent and because it excludes the financial crisis. Only the acquirer needed to be a listed company. The reason for this is twofold. First of all, as discussed in part 4.2., the Cumulative Average Abnormal Returns can only be calculated when there is a daily stock price available. In
other words, it is impossible to compute the abnormal of a company that has no publicly traded stock. The second reason is the availability of data. Listed firms are obliged to share much more information compared to private firms. This information is needed to compute the control variables and to identify which board members and C-suite executives are considered to be ‘local’. Furthermore, only completed mergers and acquisitions with a final stake higher than 50% and a deal value higher than 20 million dollars are used. The reason for this is that smaller deals are more easily subjective to outside factors compared to larger firms. And last of all, financial firms are taken out of the sample because these firms have the tendency to bias the coefficients. The Zephyr database also contains all the necessary data on the countries of the acquirer and target.
The Datastream database is used to obtain all the financial data of all the acquirers in the sample. This includes the beta at the time of the deal (with the local index as a benchmark) and returns.
The Orbis database provided all the necessary personal data considering the board members and C-suite executives. The most important data point of this database is the nationality of a particular board member or C-suite executive. Unfortunately, it was not possible to retrieve historical data from the Orbis database directly. However, there was an option to include previous board members and C-suite executives. This, combined with the appointment and resignation date, made it possible to calculate whether a particular person was active at the time of the deal. Another difficulty was the output of Orbis considering the nationality of the person of interest. When, for example, a board member is active in England and is from German and French descent, Orbis’ output was: “German; French”. Suppose this particular board member is involved in a deal with a German target. The formula that is supposed to compare the board member’s nationality with the target’s firm country, will result in no match (because ‘Germany’ is not equal to ‘Germany; France’. Therefore, only the first mentioned nationality is used. It is assumed that this will have a minimal impact on the results. When all nationalities of all active board members and C-suite executives were identified, the percentage of ‘locals’ had to be calculated. To do this, a matrix was created in which all the acquiring companies were on the y-axis and all the present nationalities were on the x-axis. The percentage of ‘locals’ was then calculated for every merger or acquisition.
The distance data, consisting out of the latitudes and longitudes of the acquirer’s and target’s country is retrieved by using the Google Maps API. These latitudes and longitudes are then used to calculate the geographical distance as explained in section 4.5.
The compensation data for testing hypothesis 3 is provided by the Execucomp database on WRDS. This database provides yearly compensation data for board members and C-suite executives (current and previous). However, Execucomp does not provide the nationality of the board member or C-suite executive. Therefore, this dataset had to be merged with the personal executive and board member data of Orbis. It was not possible to merge these two datasets on a unique code or ID number. Therefore, a new variable was created by combining the ISIN (International Securities Identification Number) with the last name of the board member or C-suite executive. This variable was used to merge the compensation data with the personal data. Then, the deal-data of Zephyr was used to find the moment in time when the deal took place. This moment in time is the starting point of the analysis. For every deal, it had to be defined whether or not a particular board member or C-suite executive was ‘local’. When all this has been done, the compensation increase (decrease) under ‘normal’ circumstances and the increase (decrease) during a merger or acquisition had to be calculated.
After all the data was downloaded and constructed to fit the needs of this study, the separate datasets needed to be merged. A new variable had to be created because not all data sets had the same identifiers. Because a firm occasionally was involved than more than one deal in the chosen period of time, a new variable was created that consisted out of the announcement date combined with the ISIN of the acquirer. Based on this new and deal-specific identifier the deal data could be merged with financial-, personal-, distance- and compensation data.
5.2.DESCRIPTIVE STATISTICS
The final sample consists out of 3,721 deals in the period of 2010 until 2015. Table 1 shows the number of deals by year. This full sample can be divided into two subsamples: domestic deals and cross-border deals. Cross-border deals are deals where the acquirer’s country is different than the target’s country. There are a total of 2,075 domestic deals and 1,646 border deals. This means that more than 44% of all deals are cross-border. This strengthens the notion that cross-border deals receive relatively little attention in the existing literature. In both subsamples, deals are fairly equally divided over the years.
Table 1 - Number of acquisitions by year.
This table presents the number of acquisitions by year. Numbers for the full sample are presented first, followed by domestic and cross-border mergers or acquisitions.
Year Full sample Domestic Cross-border
Number Percent (%) Number Percent (%) Number Percent (%)
2010 619 16,6 327 15,8 292 17,7 2011 708 19,0 384 18,5 324 19,7 2012 662 17,8 367 17,7 295 17,9 2013 491 13,2 290 14,0 201 12,2 2014 590 15,9 337 16,2 253 15,4 2015 651 17,5 370 17,8 281 17,1 Total 3721 100 2075 100 1646 100
The cross-border subsamples can also be divided further into two new subsamples: ‘non-local’ board members or C-suite executives and ‘‘non-local’ board members and C-suite executives. Table 2 provides insight to these two new subsamples. There a total of 1,288 deals in which there is no ‘local’ board member or C-suite executive and there are 358 deals in which there is at least one ‘local’ board member or C-suite executive
An overview of the distances between the acquirer’s and target’s country are given in appendix A.2. Noticeable is the spike in cross-border mergers and acquisition with a distance ranging from 15 to 16 thousand kilometers. This spike is primarily created by deals between Australia – UK, Australia – USA and Taiwan – British Virgin Islands.
Table 2 - Number of acquisitions by year (specified for cross-border mergers and acquisitions).
This table presents the number of cross-border acquisitions by year. Numbers for the cross-border sample are presented first, followed by non-local board members and C-suite executives and local board members and C-suite executives. A board member or C-suite executive is considered to be local when his or her nationality is equal to the target’s country.
Year Cross-border Non-local board/C-suite Local board/C-suite
Number Percent (%) Number Percent (%) Number Percent (%)
2010 292 17,7 230 17,9 62 17,3 2011 324 19,7 249 19,3 75 20,9 2012 295 17,9 236 18,3 59 16,5 2013 201 12,2 163 12,7 38 10,6 2014 253 15,4 198 15,4 55 15,4 2015 281 17,1 212 16,5 69 19,3 Total 1646 100 1288 100 358 100
Table 3 shows the distribution of all deals over the different industries. The full sample is, equally to table 1, divided into domestic and cross-border deals. The three industries most frequently represented in the sample are ‘Manufacturing’, ‘Services’ and ‘Transportation, Communications, Electric, Gas and Sanitary services’. These three
industries cumulate to 79%, 75% and 83% for the full-, domestic- and cross-border sample respectively.
Table 4 shows the number of acquisitions per SIC industry, specified for deals with and without a ‘local’ board member or C-suite executive. There are a couple of notable facts to be found here: there are no cross-border deals with a ‘local’ board member or C-suite executive, ‘Mining’ and ‘Services’ are relatively popular in the ‘local’ deals while ‘Transportation, Communications, Electric, Gas and Sanitary services’ and ‘Wholesale trade’ are not.
Table 3 - Number of acquisitions by SIC industry.
This table presents the number of acquisitions by SIC industry. Numbers for the full sample are presented first, followed by domestic and cross-border mergers or acquisitions. The SIC industry is determined by using the two first digits of the SIC code.
SIC industry Number Full sample Percent Domestic Cross-border
(%) Number Percent (%) Number Percent (%)
Agriculture, Forestry and
Fishing 58 1,6 33 1,6 25 1,5
Mining 248 6,7 160 7,7 88 5,3
Construction 136 3,7 104 5,0 32 1,9
Manufacturing 1812 48,7 930 44,8 882 53,6
Transportation,
Communications, Electric, Gas
and Sanitary service 466 12,5 299 14,4 167 10,1
Wholesale Trade 168 4,5 81 3,9 87 5,3
Retail Trade 173 4,7 131 6,3 42 2,6
Services 655 17,6 334 16,1 321 19,5
Public Administration 4 0,1 2 0,1 2 0,1
Table 4 - Number of acquisitions by SIC industry (specified for cross-border mergers and acquisitions).
This table presents the number of cross-border acquisitions by SIC industry. Numbers for the cross-border sample are presented first, followed by non-local board members and C-suite executives and local board members and C-suite executives. A board member or C-suite executive is considered to be local when his or her nationality is equal to the target’s country. The SIC industry is determined by using the two first digits of the SIC code.
SIC industry Cross-border Non-local board/C-suite Local board/C-suite
Number Percent (%) Number Percent (%) Number Percent (%)
Agriculture, Forestry and Fishing 25 1,5 19 1,5 6 1,7
Mining 88 5,3 56 4,3 32 8,9
Construction 32 1,9 32 2,5 0 0,0
Manufacturing 882 53,6 703 54,6 179 50,0
Transportation,
Communications, Electric, Gas
and Sanitary service 167 10,1 140 10,9 27 7,5
Wholesale Trade 87 5,3 75 5,8 12 3,4
Retail Trade 42 2,6 31 2,4 11 3,1
Services 321 19,5 230 17,9 91 25,4
Public Administration 2 0,1 2 0,2 0 0,0
Total 1646 100 1288 100 358 100
Table 5 shows the summary statistics for the full sample, divided into the domestic and cross-border subsamples. The asterisks in the second to last column (Cross-border, Mean) denote whether the cross-border mean differ statistically from the domestic mean. Virtually all variables are statistically different at a significance level of 1%, except for the variable ‘Pure cash deal’, which is statistically different at a significance level of 10%. Once again, this shows that cross-border mergers and acquisition are different from domestic mergers and acquisitions. It can be seen that acquirers are much bigger in a cross-border deal compared to a domestic deal. Furthermore, deals are bigger in absolute terms, but smaller is relative terms.
Table 6 shows the summary statistics for the cross-border subsample, divided into ‘local’ and ‘non-local’ deals. Again, statistically significant differences between the means of ‘local’ and ‘non-local’ are denoted by the asterisks. The most notable difference is the difference in absolute deal value. Where ‘non-local’ cross-border deals have a mean deal value of $386 million, ‘local’ cross-border deals have a mean deal value of $871 million. This is 126% higher. However, due to the endogeneity issues as described in part 4.5., one must be careful drawing conclusions based on these statistics. This because a firm that is seeking to make a large (in absolute terms) cross-border deal is more likely to appoint a ‘local’ executive.
Table 5 - Summary statistics for control variables.
This table presents the summary statistics for the acquirers and deals in the full sample. Numbers for the full sample are presented first, followed by domestic and cross-border mergers and acquisitions. Market capitalization is defined as total number of shares outstanding multiplied by the share price. Tobin’s Q is defined as the equity market value over the equity book value. Log(assets) is the logarithm of total assets. Cash flow is defined as the cash flow per share multiplied by the amount of shares outstanding. All acquirer characteristics are defined at the announcement date. Relative deal value is defined as the deal value divided by the acquirer’s market capitalization. A deal is classified to be in a related industry when the two first digits of the SIC are equal. A pure stock (cash) deal is a deal where 100% of the acquisition is paid by stock (cash). All variables are winsorized at the 5% when deemed necessary. Asterisks denote when the sub-samples are statistically different (Welch’s t-test) at the 1% (***), 5% (**) or 10%(*) level.
Variable name Full sample Domestic Cross-border
Mean S.D. Mean S.D. Mean S.D.
Acquirer characteristics
Market capitalization ($ millions) 6714,43 16600,00 4157,46 11900,00 9892,55*** 20700,00
Tobin's Q 1,95 2,48 2,12 2,66 1,74*** 2,21
Log(assets) 14,03 2,00 13,58 1,93 14,61*** 1,94
Cash flow/assets 7,06 14,45 8,46 16,11 5,32*** 11,86
Deal characteristics
Deal value ($ millions) 361,12 1956,07 257,99 1076,65 491,13*** 2675,96
Relative deal value 0,21 0,34 0,24 0,36 0,17*** 0,31
Related industries 49,9% 50,0% 46,5% 49,9% 54,3%*** 49,8%
Pure stock deal 15,9% 36,6% 22,9% 42,1% 7,1%*** 25,7%
Pure cash deal 47,5% 49,9% 48,9% 50,0% 45,7%* 49,8%
Table 6 - Summary statistics for control variables (specified for cross-border mergers and acquisitions).
This table presents the summary statistics for the acquirers and deals in the cross-border sample. Numbers for the cross-border sample are presented first, followed by non-local board members and C-suite executives and local board members and C-suite executives. A board member or C-suite executive is considered to be local when his or her nationality is equal to the target’s country. All variables are defined and tested similar to as mentioned in table 5.
Variable name Cross-border
Non-local board/
C-suite Local board/C-suite
Mean S.D. Mean S.D. Mean S.D.
Acquirer characteristics
Market capitalization ($ millions) 9892,55 20700,00 8784,84 18900,00 13900,00*** 25500,00
Tobin's Q 1,74 2,21 1,71 2,19 1,85 2,29
Log(assets) 14,61 1,94 14,57 1,90 14,78 2,09
Cash flow to assets 5,32 11,86 6,04 12,37 2,72*** 9,39
Deal characteristics
Deal value ($ millions) 491,13 2675,96 385,54 2198,42 871,03** 3922,66
Relative deal value 0,17 0,31 0,16 0,30 0,20** 0,35
Related industries 54,3% 49,8% 53,6% 49,9% 57,0% 49,6%
Pure stock deal 7,1% 25,7% 5,6% 23,0% 12,6%*** 33,2%
Table 7 shows the summary statistics regarding the data obtained to test what the cost of ‘local’ information is. These summary statistics should be handled with care due to the possible endogeneity issues (as a result of reversed causality) as discussed in section 4.6. In the remainder of this, certain cut-off points will be used to mitigate the probability of endogeneity. But the summary statistics do give us some insight in the results. It shows that board members and C-suite executives receive an increase in compensation, regardless of the sort deal or board member/C-suite executive.
Table 7 - Summary statistics (regarding compensation data).
This table presents the summary statistics regarding the data of the acquirer’s board member or C-suite executive. ΔCompensation increase (in %) equals the increase/decrease in compensation in the year of the deal. The top part presents the numbers for the full sample first, followed by domestic and cross-border deals. The bottom part presents the subsample of cross-border deals first, followed by two new subsamples: without or with a ‘local’ board member or C-suite executive. A board member or C-suite executive is considered to be local when he or she is a native to the target’s company. Asterisks denote when the sub-samples are statistically different (Welch’s t-test) at the 1% (***), 5% (**) or 10%(*) level.
Variable name Full sample Domestic Cross-border
Mean S.D. Mean S.D. Mean S.D.
ΔCompensation increase (in %) 9,47 31,35 6,12 26,90 9,65 31,59
Localdummy 0,32 0,47 0,19 0,40 0,33 0,47
Tenure (in years) 6,59 2,49 6,46 2,18 6,59 2,51
Variable Cross-border
Non-local
board/C-suite Local board/C-suite
Mean S.D. Mean S.D. Mean S.D.
ΔCompensation increase (in %) 9,65 31,59 8,57 31,77 11,87 31,20
Localdummy 0,33 0,47 0,00 0,00 1,00*** 0,00
Tenure (in years) 6,59 2,51 6,49 2,48 6,81 2,58
6.
R
ESULTSIn this part, the results of the analysis will be discussed. This will be done discussing every hypothesis separately.
6.1.CROSS-BORDER EFFECT
The cross-border effect is term that Moeller and Schlingemann (2005) introduced when they found that cross-border acquisitions perform less than domestic acquisitions. The same paper of Moeller and Schlingemann (2005) led to the formation of the first hypothesis of this paper: there is a negative cross-border effect for mergers and acquisitions. In this section, the methodology described in section 4 is used to test this.
Figure 2 shows the Cumulative Abnormal Returns for domestic and cross-border mergers and acquisition in the event window of [-10,+10] days. This visualization gives a first insight to the cross-border effect in the total sample. It is clear that there is a positive announcement effect for the acquirer for both domestic and cross-border deals. This is in line with existing literature (Masulis, Wang and Xie, 2007; Officer, Poulsen and Stegemoller, 2009; Moeller, Schlingemann and Stulz, 2007; Uysal, Kedia and Panchapagesan, 2008). It seems that domestic deals outperform cross-border deals from day -3 on. This could imply two things. First of all, it suggests that cross-border effect does exist, this is in line with hypothesis 1 and the findings of Moeller and Schlingemann (2005). Secondly, the fact that the CAR of domestic and cross-border deals is virtually equal from day -10 to -4, but suddenly differs from day -3 on, suggests that there is more information leakage in domestic deals. A possible explanation for this is related to the behavioral finance phenomenon of home bias: investors have a strong preference to focus primarily on domestic companies (Coval and Moskowitz, 1999). This implies that a shareholder of a particular company is more likely to acknowledge any deal information in a domestic deal compared to a cross-border deal. This because of the simple fact that the investor focuses on the acquirer and target in a domestic deal while only focusing on the acquirer in a cross-border deal. So any leaked information by the cross-border target is more likely to remain unnoticed. Figure 2 gives a first insight in the existence of a cross-border effect. However, one should be careful with drawing conclusions based on this figure solely.
Figure 2 - Cumulative Abnormal Return over time.
This figure presents the Cumulative Abnormal Return (CAR) over time for all domestic and cross-border mergers and acquisitions. The CAR is calculated as described in section 4.2.
Appendix A.3. presents the Cumulative Average Abnormal Return (CAAR) for the full sample and for the domestic and cross-border subsamples. At the event window of [-1,+1], there is CAAR of 0.62%, 0.70% and 0.52% for the full-, domestic- and cross-border sample respectively. Furthermore, the cross-cross-border CAAR is statistically smaller than the domestic CAAR. The same can be found for event windows [-2,+2], [-5,+5] and [-10,+10]. This is statistical evidence in favor of the first hypothesis: there is a negative cross-border effect for mergers and acquisitions.
In addition to the Welch’s t-test of appendix A.3, this study also performed univariate and multivariate OLS regressions to test the first hypothesis. Table 8 shows the OLS estimation results of equation (6):
JII!K(,KL = ON+ /NJ6PQQRP6ST6" + /9UPVW6PXQ" + 2" (6)
The dependent variable, Cumulative Average Abnormal Return (CAAR), is specified by four different event windows: [-1,+1], [-2,+2], [-5,+5] and [-10,+10]. The CAAR is given in percentages. Specification (1) through (4) present the results for the univariate regression where the CAAR is regressed against the coefficient of interest: Crossborderi. This independent variable is a dummy variable that equals 1 when the merger or acquisition is a cross-border deal. Similar to the results in appendix A.3, all coefficients are negative and significant at either the 5% or 1% level. Economically, this means that
during the [-1.+1] event window, the share of an acquirer in cross-border deal underperformed an acquirer in a domestic deal with 0.179% per day. This might not seem as much, but extrapolating this return to a 250-day year shows that the cross-border deal underperforms by 56.38% on year basis. Of course, this effect will be short-lived, but the significance is clear. As expected, the coefficient has a downward trend as the event window becomes larger. This is the result of the announcement effect averaging out as the number of days increase. The constant gives an interpretation of the acquirer’s announcement returns regardless the target’s country. The positive and significant constants imply a positive announcement return for acquirers in a merger or acquisition.
Specification (5) through (8) present the results for the multivariate OLS regressions where control variables for assets, Tobin’s Q, cash flow to assets, relative deal value, pure stock or cash deal and related industries are added. With the addition of the control variables, the significance of the coefficient of interest disappears.
The absolute results of specification (1) through (4) are in line with the findings of Chakrabarti, Gupta-Mukherjee and Jayaraman (2009) and Cakici, Hessel and Tandon (1996). They found that cross-border mergers and acquisitions led to positive announcement returns. Please note that the net return (as in table 8) of a cross-border deal is still positive, despite the negative cross-border coefficient. The findings on the coefficient of interest are also in line with the findings of Moeller and Schlingemann (2005), who found that cross-border deals underperform compared to domestic deals. However, one should be careful with drawing conclusions because the findings were not significant when controlled for acquirer and deal characteristics.
Table 8 - Regression analysis on cross-border effect.
This table presents the OLS regressions for the acquirer’s Cumulative Average Abnormal Return (CAAR). The CAAR is given in percentages. The notation [-T1,+T2] denotes a
specific event window. Specification (1) through (4) show the univariate regressions with only the coefficient of interest. Specification (5) through (8) show multivariate regressions, identical to respectively (1) through (4) except for the addition of control variables. The main coefficient of interest is Crossborder. This dummy variable equals 1 when a deal is a cross-border deal and equals 0 otherwise. Market capitalization is defined as total number of shares outstanding multiplied by the share price. Tobin’s Q is defined as the equity market value over the equity book value. Log(assets) is the logarithm of total assets. Cash flow is defined as the cash flow per share multiplied by the amount of shares outstanding. All acquirer characteristics are defined at the announcement date. Relative deal value is defined as the deal value divided by the acquirer’s market capitalization. A deal is classified to be in a related industry when the two first digits of the SIC are equal. A pure stock (cash) deal is a deal where 100% of the acquisition is paid by stock (cash). All variables are winsorized at the 5% when deemed necessary. P-values are given in parentheses and asterisks denote statistical significance at the 1% (***), 5% (**) or 10%(*) level.
CAAR
[-1,+1] [-2,+2] CAAR [-5,+5] CAAR [-10,+10] CAAR [-1,+1] CAAR CAAR [-2,+2] CAAR [-5,+5] [-10,+10] CAAR
Independent var. (1) (2) (3) (4) (5) (6) (7) (8) Crossborder -0.179*** -0.103** -0.121*** -0.079*** 0.057 0.055 0.008 0.010 (0.00) (0.01) (0.00) (0.00) (0.34) (0.23) (0.80) (0.65) Log(assets) -0.183*** -0.129*** -0.080*** -0.051*** (0.00) (0.00) (0.00) (0.00) Tobin's Q 0.003 0.005 0.008 0.009* (0.81) (0.67) (0.27) (0.07)
Cash flow to assets -0.008*** -0.007*** -0.005*** -0.002***
(0.00) (0.00) (0.00) (0.00)
Relative deal value 0.189* 0.140* 0.112** 0.080**
(0.06) (0.07) (0.04) (0.02)
Pure stock deal 0.389*** 0.262*** 0.233*** 0.143***
(0.00) (0.00) (0.00) (0.00)
Pure cash deal 0.142** 0.077 0.059* 0.020
(0.02) (0.10) (0.08) (0.37) Related industries -0.107* -0.074* -0.042 -0.021 (0.05) (0.08) (0.16) (0.28) Constant 0.697*** 0.493*** 0.365*** 0.202*** 3.100*** 2.201*** 1.380*** 0.832*** (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) N 3718 3718 3718 3718 3413 3413 3413 3413 Adj. R-sq. 0.003 0.001 0.004 0.004 0.069 0.060 0.060 0.058
