Cross-firm social ties and merger outcomes
How does board type of the acquiring firm influence the cross-firm social tie
effect on merger cumulative abnormal returns?
Radboud University Faculty of Management
Nijmegen School of Management Supervisor: Dr. J. Qiu
Stephan de Waal s3044173 Master Economics
Specialization: Corporate Finance & control Abstract
This study provides an analysis on the cross-firm social tie effect on acquirer cumulative abnormal returns (CAR) and the influence of acquirer board type on this effect. Using a sample of 90
European within border merger or acquisition deals between 2002 and 2016 univariate as well as multivariate analysis is used to study the effects. The study provides evidence that all ties and educational ties have a negative effect on acquirer CAR for a three, five and seven-day event window around the merger announcement date. The study also finds a significant positive effect of strong ties on acquirer CAR for the five-day event window. Looking at the effect of board type on the cross-firm social tie effect the study finds indications, but no conclusive evidence that two-tier board acquirer firms are less affected by the negative cross-firm social tie effect compared to firms with a one-tier board.
Contents
1. Introduction ... 3
2. Literature review and development of hypotheses ... 5
2.1 Social tie effects on cumulative abnormal returns ... 5
2.1.1 Friendship vs. Advice ties ... 5
2.1.2 The effects of social ties on mergers and acquisitions ... 5
2.1.3 Measuring merger outcomes ... 9
2.2 Effect board type on social tie effect ... 9
2.2.1 Board structure ... 9
2.2.2 Board structure and the cross firm social tie effects on merger outcomes. ... 11
3. Methodology ... 13
3.1 Sample and data collection ... 13
3.2 Dependent variable ... 15 3.3 Independent variables. ... 15 3.4 Control variables ... 16 3.5 Research method ... 18 4. Results ... 20 4.1 Descriptive statistics ... 20 4.2 Regression results ... 22 4.2.1 Univariate analysis. ... 22
4.2.2 Multivariate OLS regression first hypothesis ... 23
4.2.3 Results interpretation first hypothesis ... 26
4.2.4 Results interpretation control variables ... 27
4.2.5 Multivariate OLS regression second hypothesis ... 28
4.2.6 Results interpretation second hypothesis ... 32
5. Discussion and Conclusion ... 35
6. Literature ... 37
1. Introduction
After the Enron scandal the importance of a good functioning corporate governance system became clear. One of the lessons learned from the Enron case was that the internal and external control of corporate management should be executed by financially independent individuals and institutions. The study of Bruynseels and Cardinaels (2014), indicated that not only financial dependence, but also social ties between corporate management and the internal or external auditor has a negative effect on oversight quality. These ties can arise from graduating from the same educational
institution (educational ties) or past employment at the same company (employment ties). Multiple studies support the claims of Bruynseels and Cardinaels (2014) that these social ties have a negative effect corporate-decision making and financial control (Menon and Williams, 2004; He et al., 2016; Guan, Su, Wu, Yang, 2016. However, there is little literature which discusses the effects of cross-firm social ties on economic-decision making.
According to Ishii, & Xuan (2014), social tie effects are not only present in the context of corporate control, but also in the context of mergers and acquisitions. Ishii & Xuan (2014), provide evidence that social ties between the acquirer and target board of directors has a negative effect on acquirer merger performance. This performance is measured by the cumulative abnormal returns (CAR) of the acquirer stock around the merger announcement date. However, this study is not supported with evidence of other studies and is based on a U.S dataset. Further research on the cross-firm social tie effects on merger performance is important because a merger or acquisition deal is one of the biggest economic events of a company which has major influence on shareholder value
(DePampillis, 2015).
Ideally, such biases stemming from social ties as reported by Ishii and Xuan (2014) should be diminished by solid corporate governance mechanisms. The supervisory board is one of the
mechanisms that play an important role in controlling corporate management. The main task of the supervisory board is to appoint, dismiss and monitor members of the management board
(Jungmann, 2006). They should therefore prevent the management board from making mistakes caused by social ties between the acquiring and target management boards. This means that potential negative effects of cross-firm social ties may depend on the performance of the supervisory board.
There are two main sets of legal rules on the supervision of the management board: one-tier boards and two-tier boards (Jungmann, 2006). In one-tier boards the board of directors consists of
executive board members and non-executive board members. In this case non-executive board members supervise the executive board members which are corporate management.
Hence, this one board contains both management and supervisory board. In a system with two-tiers, the executive and supervisory functions are separated in two distinct boards.
The separate supervisory board monitors management board. As there are both advantages and disadvantages associated with the board types, literature provides no unambiguous evidence regarding the effect of board type on supervisory performance (Jungmann, 2006; Millet-Reyes and Zhao, 2010). The most suitable board type therefore seems to be situation dependent. This study will add to this literature by studying the effect of board structure on a specific (mis-) behavior of management: bad merger outcomes caused by too strong social ties between acquirer management and target. This leads to the following central question:
How does board type of the acquiring firm influence the cross-firm social tie effect on merger cumulative abnormal returns?
This study contributes to existing literature in multiple ways. First, this study provides more evidence regarding the effect of cross-firm social ties between the acquiring and target company on acquirer CAR around the merger announcement date as it is the first research to study this effect in Europe. Second, this is the first study to investigate this effect on both one- and two-tier board companies. Comparing the cross-firm social tie effect on both board types contributes to corporate governance literature with respect to supervisory board performance.
This study provides practical insights for company shareholders by enhancing their understanding with respect to the cross-firm social tie effects on CAR. The study may not only help shareholders in understanding the effect of cross-firm social ties, but also provides insights on how the choice of board type influences this effect. This study uses a European sample of 90 within boarder mergers or acquisitions between 2002 and 2016. A European sample is used because both board systems are present within Europe, as opposed to the U.S which only allows the one-tie board system. The data will be collected from several databases. Merger announcements are extracted from Thomson One. Information regarding cross-firm social ties is derived from Boardex. Financial information is
retrieved from the Thomas Reuters Eikon database.
In order to be able to answer the research question first the effect of cross-firm social ties on acquirer CAR will be investigated using the methods and variables of Ishii, & Xuan, (2014). After that the effect of board type on the cross-firm social tie effect on CAR around the merger announcement date will be studied. The remainder of this thesis is as follows: In chapter two the theoretical
background and hypothesis development will be discussed. The third chapter will discuss the research sample and methodology. Chapter four will discuss the results of the research. Lastly, the conclusion and discussion can be found in chapter five.
2. Literature review and development of hypotheses
This chapter will discuss all relevant literature which help to develop the hypotheses of this thesis. It will also show how this study contributes to existing literature by adding necessary information to current insufficient literature. First, the concepts of social ties will be explained and existing research on social ties and CAR will be reviewed. Secondly, factors regarding cross-firm social ties that may affect CAR will be discusses. Finally, the literature on the effects of board structure, firm
performance and decision-making will be reviewed, leading us to the hypotheses on the effect of board structure on the cross-firm social tie effects on CAR will be outlined.
2.1 Social tie effects on cumulative abnormal returns
2.1.1 Friendship vs. Advice ties
There are many ways in which two individuals can be connected to each other. One of the most visible and strongest connections is the connection through family. However, there are many other ways how individuals can be tied to each other. Literature distinguishes two main types of social connections, being friendship ties and advice ties (Carrol and Teo,1996; Gibbons, 2004 and Saint-Charles and Mangeau, 2009).
Friendship ties are formed by individuals who are connected by participating at the same leisure clubs, societies or charitable organizations (Carroll and Teo, 1996). Advice ties are considered when two individuals are connected in a professional context (Gibbons, 2004). Advice ties are divided in educational ties and employment ties. This study considers an educational tie if two individuals attended the same educational institution and an employment tie if two individuals used to work at the same company. Only little data on friendship ties is available. This thesis therefore will only look at the effect of advice ties on merger CAR.
2.1.2 The effects of social ties on mergers and acquisitions
Current studies suggest that social ties have both positive and negative effects on economic decision-making. They however show contradicting results regarding the net overall effect of the social ties. For instance, information asymmetry may be reduced as well as increased under the influence of social ties (Ishii, & Xuan, 2014). On the one hand, social ties help to lower the costs of gathering information and thereby improve information flows. On the other hand, increased social ties may result in less need for information gathering for monitoring purposes, thus actually resulting in increased information asymmetry (Guan, Su, Wu, Yang, 2016). Both sides will be explained in this section.
If there is a social connection between top management of the acquiring and target firm they are more likely to share the same background and interests.
It is also likely that if they attended the same education facility or used to work at the same company involved directors already know each other. According to homophily theory both parties will trust the other party more because of their social connections. Homophily is the principle that interaction between similar people occurs more often compared to more dissimilar people (Mc Pherson et al., 2001). In the context of educational ties Mc-Pherson (2001) for example suggests that individuals who selected the same schools are likely to have the same background and similar interests. Having the same background or beliefs makes people more comfortable with each other. This will increase the willingness to exchange information which will lower information gathering costs as well as information asymmetry (Ishii, & Xuan, 2014).
A decrease in information asymmetry will improve economic decision-making quality of the acquiring company. On the other hand, the level of trust between both parties will lower the requirements and critical analysis of the received information (Uzzi,1996). They for instance may decrease due diligence norms (Ishii, & Xuan, 2014). Due diligence is the critical review of records and facilities.
Parties not only review financials, but also conducts reviews on strategic operations and legal issues (DePamphilis,2015). Although due diligence is performed by both acquiring and target company, due diligence usually works to the advantage of acquirer. Uncovered issued as a result of due diligence will help the acquirer to negotiate a lower offer price (DePamphilis,2015). Therefore, the acquirer wants to spend as much time on due diligence as possible. This shows that as a result of sufficient due diligence a reduction of information asymmetry will reduce the risk of overpayment.
Although due diligence is supposed to be beneficial for the acquiring company, conducting due diligence is expensive as well as time consuming (DePamphilis,2015). As social ties increase the level of trust between both parties, it is more likely for the acquiring company management not wanting to spend a lot of money and time on due diligence as they (mis)judge the risk for unfavorable issues to be unlikely. For this reason, lower due diligence norms enhance information asymmetry and will lead to less optimal decision making of the acquiring firm.
As a result of information asymmetry, the acquirer does not exactly know the value of target’s current assets. In order to reduce the risk of overpayment the acquiring company will discount for uncertainty which lowers the offer price. The target company knows the value of its assets and will only accept the offer price if it least equals present value of its current assets (Balakrishnan and Koza, 1993). In other words, the information asymmetry only leads to overpayment risk for the
More information asymmetry increases the chance of insufficient discounts by the acquiring company and therefore also overpayment risk.
This “adverse selection” situation shows the importance of sufficient due diligence as it lowers information asymmetry and overpayment risk. As literature provides arguments for both increasing and decreasing effects of cross-firm social ties on information asymmetry, no clear expectations regarding social tie effects on overpayment can be made based on these arguments.
In the context of mergers and acquisitions the familiarity bias may also have an effect. The familiarity bias suggests that individuals have a preference regarding status quo choices, familiar goods or familiar people (Ishii, & Xuan, 2014). In other words, people have the tendency to stick to what they are familiar with. For example, investors of hedge funds prefer to invest in funds which are from the same city as they are located in (Sialm et al, 2014). This so-called home bias, which is part of the familiarity bias suggests a preference to invest in those geographical areas the investor is most familiar with.
Another example of the familiarity bias is the huge amount of retirement money employees invest in employer stock (Huberman, 2001). In the context of social ties the familiarity bias suggests that economic decision makers have the tendency to get embedded with someone they are already familiar with (Ishii, & Xuan, 2014). While, from a strategic or financial point of view an evolvement with another party which they are less familiar with may well have been the better choice. According to (DePamphilis, 2015) the acquirer must form selection criteria during the search and screening phase of an acquisition. These criteria can help to find the target which is most
undervalued as well as suitable to achieve the strategic goals of the merger. For example, only selecting target companies with low multiples will help to the acquirer to find currently undervalued target companies. Criteria regarding market segments, or product lines may help the acquirer to find the target company to achieve highest synergy effects (DePamphilis, 2015). Synergy is described as increased value as a result of combining two firms (DePamphilis, 2015).
The acquirer should therefore try to acquire the company which best fits these criteria. Ishii, & Xuan (2014) suggest that company management has the tendency to acquire the company they feel most familiar with rather than the company who best fits financial or strategic criteria. The acquirer may not put effort in finding better target options because of the preference for familiarity (Ishii, & Xuan, 2014).
This preference for familiarity may lead to mergers with only limited synergy effects which will lower acquirer post-merger expected profits. Less synergy effects will also lower the offer price as the acquiring company. It may also result in mergers with relatively less undervalued of even overvalued target companies.
The acquirer will not be able to buy the company below its current price, which makes the merger less profitable for the acquirer. Unlike the merger with an undervalued target company, the target company is not likely to negotiate a purchase price high above current market value. This way social ties can lower both acquiring and target company stock abnormal returns around the announcement period.
During the merger and acquisition process acquiring and target management needs to negotiate and discuss intensively to jointly search for solutions to problems (DePamphilis, 2015). This way acquirer and target board form a group during the negotiations. Social ties enhance the homogeneity of the group and therefore also the chance of groupthink (Asch, 1951). Groupthink refers to the desire within the group to agree with each other (Ishii, & Xuan, 2014). Group members have the tendency to conform to the social consensus even if this consensus is evidently wrong (Asch, 1951). Although this concept of groupthink may lead to a more efficient decision making as both parties will come to an agreement relatively quickly as there is less room for dissent and debate, it may also lead to less effective decision making as groupthink leads hastily made and ill-advised decision.
If the general consensus of the group becomes that the merger deal will be beneficial for both parties, groupthink may cause the group to neglect evidence that suggests the merger deal to be less beneficial than expected. For instance, information that indicates only limited cost reductions as a result of the merger. This way social ties may lead to suboptimal decision making.
So far, the literature on social ties on merger and acquisition merger outcomes is still relatively small. Ishii & Xuan (2014) find a negative effect of educational and employment ties on both acquirer and target merger outcomes. On the other hand, Cai and Sevilir (2012) provide evidence that social ties have a positive effect on acquiring company merger outcomes if both acquirer and target share a board member. Cai and Sevilir (2012) argue that if the same board member represents both acquirer and target shareholders there no longer is a situation of information asymmetry. The decrease of overpayment risk will enhance acquirer merger outcomes.
2.1.3 Measuring merger outcomes
In this study merger outcomes are measured by the cumulative abnormal returns (CAR) around the merger announcement date. The CAR reflects how shareholders value the announced merger or acquisition. Previous studies already used CAR to measure the market reaction on merger announcements (Andrade and Stafford, 2004; Ishii and Xuan, 2014; Huang and Walkling, 1987; Wansley et al., 1983; Moeller et al., 2004; Singh and Montgomery, 1987). For a detailed description of CAR used in this study please see the methodology section.
Supported by the evidence of (Ishii, & Xuan, 2014) and the theoretical contributions regarding the effects of homophily, the familiarity bias and groupthink on economic decision making the following hypothesis can be formulated.
H1: Cross-firm social ties have a negative effect on the cumulative abnormal returns around the merger announcement date on our sample of European 2002-2016 data.
2.2 Effect board type on social tie effect
2.2.1 Board structure
Board structure is part of firm’s corporate governance measures (Millet‐Reyes and Zhao, 2010). Corporate governance is a set of mechanisms that helps to protect investors or other outside stakeholders from self-interested inside stakeholder behavior (La Porta et al., 2000). In other words, corporate governance has to make sure company management’s decision making will be based on the interests of all stakeholders (usually mostly shareholders). There are both external and internal corporate governance mechanisms (Millet‐Reyes and Zhao, 2010).
External mechanisms are mechanisms outside the firm such as rules and regulations, external auditors and financial analysts (Millet‐Reyes and Zhao, 2010). Internal mechanisms are mechanisms within the firms including compensation programs, the audit committee and also board structure (Millet‐Reyes and Zhao, 2010). Board structures represents how the board of directors are organized.
Literature describes the one-tier and the two-tier board as the most common legal forms of board structure a publicly listed firm can follow. Often firms often are not able to choose which legal system they want to use because national law usually obliges firms to implement one specific system (Jungmann, 2006). For instance, UK and US listed firms have to follow the one-tier board system. On the other hand, German and Finnish listed firms have to implement a two-tier system. Only a few countries such as France, Portugal and Belgium enable firms to choose which system they want to use. The main difference between both systems is the way corporate management is supervised (Jungmann, 2006).
In a one-tier board system the board of director contains both executive and non-executive directors. Executive directors are in charge of daily management and the non-executive directors have to supervise the work of the executive directors.
In case of a two-tier board system the board of directors only contains daily management and the supervisors form a separate board, the supervisory board (Millet‐Reyes and Zhao, 2010). Literature describes advantages and disadvantages of both board structure types.
One of the described main advantages of the one-tier board system is the greater flow of information between management and supervisors (Jungmann, 2006). Non-executive board members have direct access to information as they meet with company management on a daily basis. The direct involvement of the supervisors in corporate decision making enhances their knowledge of the day to day business of the firm (Lipton and Lorsch, 1992). This may help the non-executive board members to make better decisions (Jungmann, 2006). Another advantage of the one-tier board system is the ability to make decisions more rapidly. Non-executive members meet the executive members every day which enables quick decision making.
In theory, non-executive board members will receive the same information as executive board members (Jungmann, 2006). Hence, there will be no information asymmetry between daily management and its supervisors. However, the study of Turnbull (2000) suggests that in reality executive board members decide what information to share with non-executive board members (Turnbull, 2000). Therefore, non-executive board members are not guaranteed to receive
independent and unfiltered information (Jungmann, 2006). This also implies that the one-tier board system does not eliminate the information asymmetry between daily management and supervisors. Non-executive board member can try to decrease the information asymmetry by requesting
additional information (Jungmann, 2006).
In contrast to supervisors of a two-tier board system, in a one-tier board supervisors meet company management on a daily basis and are therefore directly involved in company’s decision-making process (Jungmann, 2006). Non-executive members should therefore be eager to gain enough information as they will be held responsible for bad decision making (Jungmann, 2006). However, according to Tungler (2000), non-executive board members may become reluctant to ask for the required information as it is likely for non-executive board members to develop feelings of respect or gratitude towards their other board colleagues. This makes is harder for them to step up to their daily colleagues and demand the required additional information to be able monitor the decision-making process sufficiently. According to Tungler (2000), like social ties, these feelings can lead to sociological biases which result in a lower monitoring quality (Tungler, 2000).
Two-tier board supervisors meet less often with company management and are not directly involved in the decision-making process (Jungmann, 2006). In this case, information asymmetry between management and supervisors is between boards instead of within boards. Because the supervisory is less involved with firm’s daily business and also meets less often than the non-executive board members of the one tier board, the initial information gap between daily management and supervisors is supposed to be bigger in a two-tier board system.
Although there still may be some issues regarding sociological biases as a result of collegiate feelings, this is supposed to be less problematic in case of separated boards (Block and Gerstner, 2016). This is less problematic because supervisors and daily management do not work with each other on daily basis. This reduces the risk of developing collegiate feeling which affect supervisor’s monitoring abilities.
The supervisory board also has greater power to pressure company management as they nominate and discharge the members of the management board (Jungmann, 2006).
The supervisory board also has the ability to intervene if firm’s interests are violated by company management. Therefore, the supervisory board will be less reluctant in requesting additional information (Tungler, 2000). This will reduce information asymmetry and increase supervision quality. In other words, the one-tier board system initially provides supervisors with more
information and therefore less information asymmetry. But due to sociological phenomena, two-tier board supervisors will make greater information collecting efforts. This reduces the information advantage of the one-tier board system. This shows great resemblance with the effects of social ties on information asymmetry. Whereas social ties may help to enhance information availability, but sociological phenomena decrease the effort on information collection.
Literature does not provide conclusive evidence on which legal system is better (Jungmann, 2006). For instance, Millet‐Reyes and Zhao (2010) found no significant differences in the performance of both stock and firm operations. It seems to be situation dependent if the independence advantages of the two-tier board outweigh the advantages of information availability and faster decision making of the one tier-board.
2.2.2 Board structure and the cross firm social tie effects on merger outcomes.
Cross-firm social ties are expected to have a negative effect on CAR (Ishii, & Xuan, 2014). This indicates that firm’s daily management makes sub-optimal decisions as a result of cross-firm social connections. This also implies that firm’s supervisors are not always able to prevent this sub optimal decision making from happening. Literature suggests that sociological phenomena such as
groupthink, homophily and the familiarity bias cause cross-firm social ties to have a negative effect on merger and acquisition decision making (Ishii & Xuan, 2014).
In a one-tier board system, non-executive and executive board members are more connected to each other than the supervisory board and the management board in a two-tier board system. Therefore, supervisors in a one-tier board system are more likely to get affected by these sociological phenomena.
This is in line with the findings of Tungler (2000) which suggest that sociological phenomena affect the monitoring quality of non-executive board members. Two-tier board supervisors are therefore expected to be more independent and have better monitoring performance which may avoid a sub-optimal decision regarding mergers and acquisitions.
The supervisory board also has more tools to avoid a bad merger decision as it has the power to dismiss management board members and intervene when firm’s stakes are violated (Jungmann, 2006). Bad decision-making regarding mergers and acquisition are a typical example how firm’s interest gets violated tremendously. In this case the supervisory board can take over control and stop the merger deal. In a one-tier board the non-executive members are supposed know more about the firm and to get provided more information (Jungmann, 2006). This may enhance the quality of decision making of the board (Block and Gerstner, 2016).
Although it may not be ruled out that this will also lead to better decisions in the context of mergers and acquisitions, the two-tier board system seems to be better equipped to break through the sociological biases that arise from cross-firm social ties. There is no empirical evidence regarding the effect board structure on mergers outcomes. However, Faleye et al. (2011) provide circumstantial evidence as they show that more excessive management monitoring leads to less activity regarding mergers and acquisitions. This may indicate that better supervision prevents management from making non-optimal merger and acquisition deals.
Supported by the circumstantial evidence of Faleye et al. (2011) and the theoretical contributions regarding the effects of board structure on supervision quality the following hypothesis can be formulated:
H2: The negative cross-firm social tie effect on cumulative abnormal returns around the merger announcement date are smaller for two-tier board than one-tier board firms.
3. Methodology
This chapter will discuss the used research methodology and data sample. First the sample and data collection will be discussed. Secondly, used dependent, independent and control variables will be discussed and explained. Lastly, the research methods will be explained.
3.1 Sample and data collection
In order to be able to test the research question, data is retrieved from the Thomson one, Thomas Reuters Eikon and the Boardex databases. The sample consists of 90 unique European within boarder merger or acquisition deals. The deals took place in 14 different European countries in the period 2002-2016. A European sample is used because the previous study on social tie effects on CAR around the merger announcement date from Ishii and Xuan (2014) only focused on U.S merger deals. Using a European sample therefore contributes to current literature. A second advantage of the European sample is the diversity in board type choice among the different deals. This is necessary to be able to look at the interaction effect of board type on the social tie effects on CAR. The 2002-2016 period is chosen because this is the full period of available data regarding cross-firm social ties.
To start, all European merger or acquisition deals between 2002 and 2016 are retrieved from Thomson one. In order to achieve reliable results this study tries to follow the previous study of Ishii and Xuan (2014) as much as possible. For this reason, merger or acquisition deals must meet the same selection criteria in order to be included in the sample. First of all is it required that both acquirer and target are publicly traded in the same country. Cross-border merger and acquisition deals are less likely to be socially connected. Including cross-border deals would therefore lead to an unbalanced sample. Secondly, the acquisition has to be completed and lastly the acquirer must own the full 100% of the target company after acquisition. Based on these criteria 561 merger deals could be included in the sample. After this, corresponding board member information is extracted from Boardex.
The Boardex database contains information regarding board members past and current employment as well as education. This information is used to determine social ties between acquirer and target board members. Boardex however has no board information on all companies. All deals for which Boardex did not have information for both parties were excluded. This led to a final sample of 90 merger or acquisition deals. Collected information regarding the acquirer and target board members provided insights regarding the level of education and employment ties between the acquirer and target boards.
Data concerning the CAR and the control variables were extracted from Thomas Reuters Eikon. The CAR data consists of the stock and index returns (-200, -20) days before the announcement date and [-3, +3] around the announcement date. By this the study follows Ishii and Xuan (2014) who used the same estimation and event windows. This information was used to calculate abnormal returns in Stata. With respect to the acquirer company only two deals showed some missing data. In order to prevent the sample getting too small, concerning deals were not eliminated from the sample, but the missing values were given the sample’s mean value. Replacing missing data with the mean value is an acknowledged effective way to deal with missing data (Raaijmakers,1999). The target company only had four missing values regarding the CAR, but showed 58 missing values regarding some control variables. As a result, this study will not be able to include control variables in the target’s OLS regression for target CAR. With respect to the three missing CAR values, again the sample mean value will be given to the missing values.
Data regarding the board type is retrieved from several sources. Most information about the board types is derived from Boardex. Other information was found in literature which discussed which Board type is associated with which country. For example, UK only allows a one-tier board structure while Germany only allows a two-tier board (Jungmann, 2006). Thomson one also provides data regarding company’s board type.
3.2 Dependent variable
The dependent variable in this study are the cumulative abnormal returns (CAR) around the merger announcement date. Previous studies already used CAR to measure the market reaction on merger announcements (Andrade and Stafford, 2004; Ishii and Xuan, 2014; Huang and Walkling, 1987; Wansley et al., 1983; Moeller et al., 2004; Singh and Montgomery, 1987). For firm (i) and event date (t) abnormal returns can be calculated with the following equation (MacKinlay, 1997).
With ARit being abnormal returns, Rit being actual returns and - EXt(Rit) being normal returns. As the equation shows, expectations are made conditional on information factor Xt. Normal returns however do not only depend on Xt, but also on alpha (intercept) and beta (coefficient). These can be found by running a OLS regressions in STATA of each company over the full estimation window. Stock performance of the company related to the performance of Xt in the estimation window is used to calculate expected (normal) returns during the event window. In this research Xt is the index return at which the company was listed at time of the announcement date.
This study will calculate target and acquirer CAR on three different event windows: a three-day event window [-1, +1], which means that abnormal returns one day before to one day after the announcement date are cumulated, a five-day event window [-2, +2] and also a seven-day event window [-3, +3].
3.3 Independent variables.
The first independent variable of this study measures social ties between the acquirer and target board members (CONNECTION). These ties are divided in educational ties (CONNECTIONEDUC) and employment ties (CONNECTIONEMP). This study considers an educational tie if two individuals attended the same educational facility and an employment tie if two individuals both worked at the same firm in the past. If two individuals are connected through both education and employment the connection is described as a strong tie (STRONGTIES).
This study uses the same proxy for social connection as Ishii and Xuan (2014). If the acquirer has 8 board members and the target firm 6 this means that 48 (8x6) cross-firm social connections are possible. In case of 9 actual connection based on employment and educational ties this means that the average social connection between acquirer and target is 18,75%. For the univariate t-test analysis the sample will be divided in deals with an average connection above sample median and deals with an average connection below sample median. A value of 1 is given to merger deals with a high social connection and a value of 0 is assigned to deals with a low social connection.
The second independent variable is company’s board type (BOARDTYPE). A company with a one-tier board is given the value 0 and two-tier board companies will receive value 1.
3.4 Control variables
Control variables are included in the regression to exclude the effects of these variables on the dependent variable. This way the research will show unbiased effects of the independent variable on the dependent variable.
Including control variables therefore enhances research internal validity. This study follows the study of Ishii and Xuan (2014) by adding the same control variables to the regression.
The first control variable concerns the method of payment. According to Travlos (1987) the method of payment may signal investors if acquirer’s current stock price is over or undervalued. If
management of the acquirer company considers current stock to be undervalued it will be less likely to purchase the target company with stock. Investors therefore may interpret a stock offer as bad news as it implies current stock not to be undervalued (Myers and Majluf 1984). For this reason, a stock offer is associated with lower abnormal returns. To exclude this effect from the regression the control variable pure stock deal (PureStock) is added. This control variable is a dummy variable at which a pure stock deal will be given a value of 1 and any other method of payment will be given the value of 0.
The second control variable included in the research model is Tobin’s Q (Q). Q is the ratio of market value to the replacement costs of capital and is often used as a proxy of firm’s performance as it shows how management’s investments pay-off (Wernerfelt and Montgomery, 1988; Rousseau, 2002). High acquirer Q values are associated with higher acquirer CAR (Lang, Stulz and Walkling, 1989; Servaes, 1991). In this study Q is calculated with Eikon data. Q is calculated by dividing firm’s market capitalization + total liabilities divided by common stock + total liabilities
(finabase.blogspot.nl)
The third control variable is leverage (LEVERAGE). According to agency cost theory high leveraged firms are scrutinized and disciplined by their capital providers. This gives company management less freedom to make risky or investments (Maloney et al., 1993). For this reason, high leverage is associated with higher CAR (Maloney et al., 1993). In this study leverage is measured as total debt divided by total assets.
Size (SIZE) is the fourth control variable to be included in the research model. Overconfidence of large company managers increases the risk of overpayment for large companies (Malmendier and Tate, 2005). For this reason, the size of the target company is negatively associated with CAR
The study of Ishii and Xuan (2014) includes the dummy variable tender offer as control variable in their research model. Literature provides however inconsistent results regarding the effects of tender offers on CAR (Dodd and Ruback, 1977; Firth, 1980; Desai and Kim, 1980). If no clear effects are expected the variable is not useful as a control variable. A tender offer is an offer directly made to target company shareholders (DePamphillis, 2015). This means that acquirer management does not negotiate with target company management, but with target shareholders. As a result, tender offer deals are not suitable to look at cross-firm social tie effects. For this reason, tender offer deals are excluded from the sample.
The fourth control variable included in the research model is related industries (RELATEDIND). If the acquirer and target operate in the same or related industry higher synergy effects are expected (Blackburn et al., 1990). Because synergy effects are value enhancing, higher synergy effects are associated with higher CAR. This control variable is a dummy variable at which related merger deals will be given a value of 1 and unrelated deals a value of 0. This study considers a related deal when acquirer and target company share the same two-digit SIC code.
The fifth added control variable is the relative size (RELTRANSACTION) of the of the merger deal. The impact of a merger deal on target CAR depends on the relative size of the target to the acquirer company (Asguith et al., 1983). A relatively small deal will have a smaller influence on acquirer future performance. For this reason, relatively large merger deals will lead to larger CAR effect. In this study, the relative deal size is calculated by the transaction value divided by acquirer market capitalization.
The sixth control variable included in the research model is the level of company’s free cash flow
(RELCASHFLOW). Free cash flow is within company generated cash as a result of excess returns on
investments (Jensen, 1993). The free cash flow hypothesis suggests that managers will use this money to satisfy personal interest rather than shareholder interests (Jensen, 1993). The positive cash position enables managers to pursuit their personal goals as they become less dependent from external capital providers which scrutinize and discipline self-interested managers (Maloney et al., 1993). For this reason, high free cash flows are associated with lower CAR. In this study relative, free cash flows are calculated by firm’s operating income before depreciation divided by firm’s total assets.
Lastly, this study follows Ishii and Xuan (2014) by adding time and industry fixed effects to the research model. This way the study controls for potential systematic effects of time (YEAR) and industry (INDUSTRY) differences. Industries will be classified using the 12-industry classification of Fama and French.
3.5 Research method
The effect of board type on the cross-firm social tie effect on merger CAR is tested. In order to do so, first the cross-firm social tie effect on merger CAR is tested. This effect is tested in multiple ways. First a univariate analysis is performed. The sample is divided in firms with a high cross-firm social connection (average connection above sample median) and firms with a low cross-firm social connection (average connection below sample median). The mean CAR of the acquirer, target and combined firms are measured for both groups for the three [-1, +1], five [-2, +2] and seven [-3, +3] day event window. Mean differences between both group are tested for statistical significance performing both a t-test and a Wilcoxon rank-sum test.
The second way the relation between cross-firm social ties and merger CAR is tested is by
performing a multivariate ordinary least squares (OLS) regressions. This way control variables can be added which help to control for deal and firm characteristics which may influence CAR around the merger announcement date. With the regression, the total effect of cross-firm social ties on CAR can be measured.
Before the dependent, independent and control variables are added to the regression model the variables are tested for normality and outliers. The tests show normality for the dependent and independent variables. With respect to the control variables Q shows no normality. For this reason, Q is transformed to log Q. The tests also show an outlier between CAR and connection for the 3 and 5-day event window. The outlier is removed by winsorizing both variables on a 0.02 level.
Next, the variables are tested for autocorrelation using the Pearson correlation matrix. Variables are highly correlated when they show a correlation above 0.7 or below -0.7. Table 1 in the appendix shows the values of the correlation matrix. Only the different CAR event windows and different social ties have a critical correlation value. The values indicate that no relevant variables a
correlated. Therefore, all necessary variables can be included in the regression model. The variables are also tested for multicollinearity performing a variance inflation test (VIF). All VIF are below the critical value of 10 which indicates no multicollinearity. After this the variables are ready to be included in the regression models.
All regressions will be run for the three, five and seven-day invent window. The regression model with respect to the effect of cross-firm social ties on CAR is as follows:
𝐶𝐴𝑅 = 𝛽0+ 𝛽1𝐶𝑂𝑁𝑁𝐸𝐶𝑇𝐼𝑂𝑁 + 𝛽2 𝑆𝐼𝑍𝐸 + 𝛽3 𝑄 + 𝛽4 𝑆𝐼𝑍𝐸 + 𝛽5 𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽6 𝑅𝐸𝐿𝐶𝐴𝑆𝐻𝐹𝐿𝑂𝑊 + 𝛽7 𝑅𝐸𝐿𝑇𝑅𝐴𝑁𝑆𝐴𝐶𝑇𝐼𝑂𝑁 + 𝛽8 𝑅𝐸𝐿𝐴𝑇𝐸𝐷𝐼𝑁𝐷 + 𝜖
This model will be run with and without Industry and years fixed effects. Furthermore, the model will also be run replacing independent variable CONNECTION with CONNECTIONEDUC,
CONNECTIONEMP and STRONGTIES. These regressions will be run with industry and year fixed effects. After regressing the effect of cross-firm social ties on CAR the effect of board type on this effect can be tested. This effect will be measured in two ways. First, BOARDTYPE will be added to the first regression model. Now the effect of cross-firm social ties on CAR is controlled for board type. Changing t-values and significance levels will show the effect of board type on the social tie effect. The new model is as follows:
𝐶𝐴𝑅 = 𝛽0+ 𝛽1𝐶𝑂𝑁𝑁𝐸𝐶𝑇𝐼𝑂𝑁 + 𝛽2 𝑆𝐼𝑍𝐸 + 𝛽3 𝑄 + 𝛽4 𝑆𝐼𝑍𝐸 + 𝛽5 𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 +
𝛽6 𝑅𝐸𝐿𝐶𝐴𝑆𝐻𝐹𝐿𝑂𝑊 + 𝛽7 𝑅𝐸𝐿𝑇𝑅𝐴𝑁𝑆𝐴𝐶𝑇𝐼𝑂𝑁 + 𝛽8 𝑅𝐸𝐿𝐴𝑇𝐸𝐷𝐼𝑁𝐷 + 𝛽9 𝐵𝑂𝐴𝑅𝐷𝑇𝑌𝑃𝐸 + 𝜖
Like the first model the model will be run with and without fixed effects and with different independent variables. The second way the board type effect will be tested is by adding an interaction term. This changes the way the model interprets the variable coefficients. The
interaction term provides insights in what way the effect of cross-firm social ties on CAR depends on board type. Now the regression model is as follows:
𝐶𝐴𝑅 = 𝛽0+ 𝛽1𝐶𝑂𝑁𝑁𝐸𝐶𝑇𝐼𝑂𝑁 + 𝛽2 𝑆𝐼𝑍𝐸 + 𝛽3 𝑄 + 𝛽4 𝑆𝐼𝑍𝐸 + 𝛽5 𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽6 𝑅𝐸𝐿𝐶𝐴𝑆𝐻𝐹𝐿𝑂𝑊 + 𝛽7 𝑅𝐸𝐿𝑇𝑅𝐴𝑁𝑆𝐴𝐶𝑇𝐼𝑂𝑁 + 𝛽8 𝑅𝐸𝐿𝐴𝑇𝐸𝐷𝐼𝑁𝐷 + 𝛽9 𝐵𝑂𝐴𝑅𝐷𝑇𝑌𝑃𝐸 + 𝛽9 𝐶𝑂𝑁𝑁𝐸𝐶𝑇𝐼𝑂𝑁 ∗ 𝐵𝑂𝐴𝑅𝐷𝑇𝑌𝑃𝐸 + 𝜖
4. Results
4.1 Descriptive statistics
The final sample contains 90 within border merger or acquisitions between 2002 and 2016. The number of mergers or acquisitions per years are showed in table 1. The first column shows the numbers of the full sample and the second and third column show the numbers for the subsample which are based on the cross-firm average connection level are showed in the second and third column. The high social connection subsample contains all deals with an average connection above full sample median while the low social connection subsample contains all deals below full sample median. The sample contains no deals from 2004 and only 5.55 % of full sample deals took place between 2002 and 2006. From 2007 till 2016 deals are more evenly distributed. The Boardex dataset only provides little board information from 2002 till 2006 which reduced the number of useful sample deals for these years drastically. There are also no extreme differences between low and high social connection distributions.
The sample deals are categorized by industry using the 12 Fama-French industry classifications. Table 2 shows how sample deals are distributed over the different industry categories. The most frequent industry categories in the full sample are Finance and Other. Chemical products is only represented once in the sample. The subsamples, which are based on the level of cross-firm connection, are more or less similar distributed. Only the Other and Wholesale and industry show great differences between both subsamples.
Table 3 provides an overview of the full and subsamples summary statistics. Acquirer deal
characteristics show an average cross-firm board connection of 7.04%. The high social connection subsample has a mean average connection of 12.49% and the low social connection subsample has a mean average connection of 1.49%. Furthermore, the high social connection subsample has lower total assets, a higher Q, lower debt/assets and equal cash flow/assets compared to the low social connection subsample. The deal characteristic summary statistics shows a lower mean transaction value for the high social connection subsample, but a higher relative transaction value. The high social connection subsample also consists of less related deals and a higher percentage of one-tier board companies.
4.2 Regression results
4.2.1 Univariate analysis.
The effect of cross-firm social ties on the CAR around the announcement date is tested with both univariate and multivariate analyses. Table 4 shows CAR around the announcement date. Mean CAR for the three [-1, +1], five [-2, +2] and seven [-3, +3]-day event window are reported for the acquirer, target and the combined parties. Table 4 also presents mean CAR for the high and low cross-firm social connection subsample. Lastly, the table contains Wilcoxon and t-test results, which show if the differences in mean CAR between both subsamples are statistically significant. The results show a positive CAR for the full sample combined parties of around 7% for the three event windows. Target CAR are around 15% while acquirer CAR has small positive CAR for the three and five-day event window and small negative CAR for the seven-day event window.
The low social connection subsample shows higher CAR for the acquirer, target and combined parties in for all event windows. Largest mean differences are for the acquirer companies. High social connection mean CAR for the acquirer are respectively -0.97%, -0.75% and -2.50% for the three, five and seven-day event windows. This while low social connection mean CAR
are 2.54%, 1.91% and 1.37%.
The t-tests only show a significant mean difference for the acquirer seven-day event window at a 10% significance level. The Wilcoxon test shows no statistical evidence that the distributions of the CARs are different for both subsamples.
Hence based on the univariate analysis hypothesis 1: Cross-firm social ties have a negative effect on the cumulative abnormal returns around the merger announcement date, only holds for the
acquirer at the seven-day event window at a 10% significance level. The lack of significant results can be explained by the small sample. The results show higher mean CAR for the acquirer, but the mean differences are not significant as a result of the small sample size.
4.2.2 Multivariate OLS regression first hypothesis
The results of the multivariate OLS regressions for acquirer CAR for the three, five and seven-day event windows are displayed in table 5, 6 and 7. Columns 1 till 4 focuses on the effect of the average cross-firm social connection based on both educational and employment ties on acquirer CAR. These regressions are described as all ties Column five only looks at the effect of educational ties on acquirer CAR. Column six focuses on cross-firm past employment and column 7 only looks at the effect of strong ties, which means that acquirer and target directors are socially connected through both education and past employment. Standard errors are in brackets and Asterisks indicate the significance level at a 10%(*), 5%(**) and 1%(***) level.
The results in table 5 indicate that cross-firm social ties have a significant negative effect on acquirer CAR for the three-day event window. The regression without control variables shows a negative effect at a 5% significance level (z=-2.55, p<0.05). After adding the control variables, the negative effect is still significant at a 5% significance level (z=-2.37, p<0.05). Including year fixed effects reduces the significance level to 10% (z=-1.83, p<0.1). Including both year and Industry fixed effects shows a negative cross-firm social tie effect at a 5% significance level (z=-2.31, p<0.05). The results also show that educational ties have a significant negative effect on acquirer CAR at a 5%
significance level (z=-2.19, p<0.05). Employment ties and strong ties have no significant effect on acquirer CAR. With respect to the control variables, (RELATEDIND) has a significant positive effect on acquirer CAR at a 10% significance level (z=1.89, p<0.1). This effect is no longer significant after adding year and industry fixed effects.
The results in table 6 show that cross-firm social ties have a significant negative effect on acquirer CAR for the five-day event window. The regression without control variables shows a negative effect at a 5% significance level (z=-2.02, p<0.05). The regression with control variables, but without fixed effects has a negative effect at a 10% significance level (z=-1.76, p<0.10). After including year fixed effects, the negative effect is no longer significant. Including both year and Industry fixed effects the results show negative cross-firm social tie effects at a 5% significance level (z=-2.13, p<0.05).
Educational ties have a significant negative effect on acquirer CAR at a 5% significance level (z=-2.28, p<0.05). Employment ties has no significant effect on acquirer CAR. The results indicate that strong
With respect to the control variables, (RELTRANSACTION) has a significant negative effect on acquirer CAR at a 10% significance level (z=-1.76, p<0.1). This effect is no longer significant after including year and industry fixed effects. For the strong ties regression leverage has a positive effect on acquirer CAR at a 5% significance level (z=2.41, p<0.05).
The results in table 7 show that cross-firm social ties have a significant negative effect on acquirer CAR for the seven-day event window. The regression without control variables indicates a negative effect at a 5% significance level (z=-2.56, p<0.05). After adding the control variables, the negative effect is significant at a 5% significance level (z=-2.33, p<0.10). By including year fixed effects, the negative effect is significant at a 10% significance level (z=-1.93, p<0.1). Adding both year and industry fixed effects, cross-firm social ties have a negative effect at a 5% significance level (z=-2.63, p<0.05). Educational ties have a significant negative effect on acquirer CAR at a 1% significance level (z=-2.92, p<0.05). Employment ties and strong ties show no significant effect on acquirer CAR. With respect to the control variables, (RELTRANSACTION) has a significant negative effect on acquirer CAR at a 5% significance level (z=-2.00, p<0.05). This effect is no longer significant after including year and industry fixed effects. For the strong ties regression (LEVEREGE) has a positive effect on acquirer CAR at a 5% significance level (z=2.15, p<0.05).
4.2.3 Results interpretation first hypothesis
The first hypothesis: Cross-firm social ties have a negative effect on the cumulative abnormal returns around the merger announcement date is tested. The hypothesis holds if the results show a
significant negative effect of (CONNECTION) on acquirer CAR. Table 8 summarizes the relevant (CONNECTION) results from tables five, six and seven.
The results show that the hypothesis holds for the all ties regressions for all the event windows. Only the five-day event window shows no significant negative effect if year fixed effects, but no industry fixed effects are added to the regression. The three and seven-day event window results show a larger negative of all cross-firm social ties compared to the five-day event window. There is no theoretical explanation for this difference in effect. The results also show a lower negative effect after including year fixed effects to the regression. This means that year partly explains the found negative relation of all ties on acquirer CAR. This lower effect however diminishes when industry fixed effects are added. The results a supported by the univariate analysis, which showed a lower mean CAR for acquirers in the high connection subsample.
The hypothesis also holds for educational ties, but no longer holds with employment ties as the only measure of cross-firm social ties as the results show no significant effects on acquirer CAR. The results show small negative effects of employment ties on acquirer CAR. They are however not significant which can be explained by the small sample and effect size. This means that the significant all ties results mainly depend on educational ties.
The hypothesis also does not hold for strong ties as the only measure for cross-firm social ties. The results even show a positive significant effect on acquirer CAR for the five-day event window. Literature does not provide a theoretical foundation for this contradicting effect. The positive strong tie effect also contradicts the findings of Ishii and Xuan (2014) who found a significant negative effect of strong ties on acquirer CAR for the three-day event window.
The incomplete data coverage of the Boardex database may have led to biased results regarding the strong ties effect on acquirer CAR.
4.2.4 Results interpretation control variables
The results show only a few significant effects regarding the control variables. For (SIZE) a negative effect on acquirer CAR was expected. The regressions show no unambiguous effect direction. The five-day event window shows a consistent non-significant negative effect of size on acquirer CAR.
On the other hand, the three and seven-day event window show small positive as well as small negative effects of size depending on the regression. Although these effects are not in line with the theoretical expectations. The results are in line with the findings of Ishii and Xuan (2014) who only found a small non-significant negative effect of (SIZE). As the study of Ishii and Xuan (2014) uses the regression model it can be argued that (SIZE) has no significant effect on acquirer CAR for this particular research model.
Results show a consistent non-significant negative effect of (Q) on the acquirer CAR. This result may not be significant as a result of the small sample size. A negative effect of (Q) is not in line with theoretical expectations.
Theory however also suggests that the effect of (Q) on CAR does not depend on acquirer (Q), but on how it relates to the acquirer (Q). Following the study of Ishii and Xuan (2014), this study does not take target (Q) into account. For this reason, study results may deviate from theoretical
expectations. This study Ishii and Xuan (2014) showed a significant negative effect of (Q) at a 1% significance level. It can be argued that for this measure of (Q) in this research model (Q) is negatively associated with acquirer CAR.
(LEVEREGE) has a significant effect on strong ties for the five and seven-day event window. Although other regressions show no significant (LEVERAGE) effect, results show a consistent positive non-significant effect. The small sample size may have cause these effects not to be non-significant. The positive effect of (LEVERAGE) on acquirer CAR is supported by literature.
There are no significant results with respect to the effect of (RELCASHFLOW) on acquirer CAR. The effect direction also deviates among the different event windows. This is in contrast with literature which suggests a negative effect of (RELCASHFLOW) on acquirer CAR. The small and possible biased sample caused by limited data coverage of the Boardex database could explain the variant results. Another explanation could be that there is no significant effect for this variable in this particular research design. This theory is supported by similar non-significant findings of Ishii and Xuan (2014), who used the same regression model.
The (RELTRANSACTION) variable shows significant negative effects for the without fixed effects regression at the five and seven-event day window. The effect becomes non-significant, but still negative after including year and industry fixed effects.
It is plausible that both industry and time influence (RELTRANSACTION) which justifies controlling for both effects. The regarding (RELTRANSACTION) are again in line with the similar study of Ishii and Xuan (2014), who also found a negative non-significant effect.
Although the results show no significant effects of (RELATEDIND) on acquirer CAR, results do show positive non-significant results. The lack of significance may be caused by the small data sample. The results also suggest that the effect of (RELATEDIND) becomes less possible as the number of event days increase. A possible explanation could be that targets from related industries are better known by the investors. Investors will discount for the uncertainly of the relatively unknown target. Over time investors will get informed about the target, which reduces uncertainty.
4.2.5 Multivariate OLS regression second hypothesis
The seconds hypothesis: The cross-firm negative social tie effect on post-announcement cumulative abnormal returns around the merger announcement date are smaller for two-tier board than one-tier board firms is tested by adding (BOARDTYPE) to the regression model. Tables 9, 10 and 11 show regression results after including (BOARDTYPE) to the model for the three event windows. Except for (BOARDTYPE) the regression models remain the same.
The results in table 9 show that (BOARDTYPE) has a positive non-significant effect on acquirer CAR for all regressions except for educational ties. In this model, there are no significant effects regarding the control variables in the three-day event window.
The regression without control variables shows a negative effect at a 5% significance level (z=-2.30, p<0.05). The regression with control variables, but without fixed effects has a negative effect at a 5% significance level (z=-2.33, p<0.05). After including year fixed effects the negative effect is significant at the 10% significance level (z=-1.82, p<0.1). By adding both year and Industry fixed effects the negative cross-firm social tie effect becomes significant a 5% level (z=-2.27, p<0.05).
Education as the only measure of social ties shows a negative effect on acquirer CAR at a 5% significance level (z=-2.14, p<0.05). Employment and strong ties have no significant effect on acquirer CAR.
The results in table 10, which shows the cross-firm social tie effects on acquirer CAR for the five-day event window, indicate a postive non-significant effect of (BOARDTYPE) on acquirer CAR. Only the regression model with educational ties as the dependent variable shows a non-significant negative effect. The control variables show no significant effect on acquirer CAR. The regression without control variables indicates a negative effect at a 10% significance level (z=-1.77, p<0.1).
After adding the control variables, the negative effect is still significant at a 10% significance level (z=-1.10, p<0.1). Only including year fixed effects shows a negative non-significant effect. Adding both year and Industry fixed effects a negative cross-firm social tie effect at a 5% significance level (z=-2.31, p<0.05) is showed. The results also show that educational ties have a significant negative effect on acquirer CAR at a 5% significance level (z=-2.10, p<0.05). Employment ties have no significant effect on acquirer CAR. The results indicate a positive effect of cross-firm strong ties on acquirer CAR at a 5% significance level (z=2.59, p<0.05).
For the seven-day event window results show mixed non-significant effects of (BOARDTYPE) on acquirer CAR. The regressions without fixed effects show non-significant positive effects while the regressions with fixed effects indicate non-significant negative effects of (BOARDTYPE). The regressions with only (CONNECTION) and (BOARDTYPE) included in the regressions model shows a negative significant effect of cross-firm social ties on acquirer CAR at a 5% significance level (z=-2.44, p<0.05).
After adding the control variables, the negative effect remains significant at a 5% significance level (z=-2.30, p<0.05). Adding year fixed effects reduces the negative effect to a significance level of 10% (z=-1.92, p<0.1). Including both year and Industry fixed effects shows a negative cross-firm social tie effect at a 5% significance level (z=-2.62, p<0.05). With respect to educational ties the results show a significant negative effect on acquirer CAR at a 1% significance level (z=-3.00, p<0.01). Employment ties and strong ties have no significant effect on acquirer CAR.
The control variable, (RELTRANSACTION) has a significant negative effect on acquirer CAR at a 5% significance level (z=2.01, p<0.05). This effect is no longer significant after including year and industry fixed effects to the regression model. The results also show positive significant effect of (LEVEREGE) at a 10% significance level for employment ties (z=1.81, p<0.1) and at a 5% significance level for strong ties (z=2.12, p<0.05).
The second way the hypothesis is tested is by looking at the interaction effect of (BOARDTYPE) and (CONNECTION). This approach directly shows how (BOARDTYPE) influences the effect of
(CONNECTION) on acquirer CAR. Table 12 tabulates results of the interaction effect. The results regarding the interaction effect between (BOARDTYPE) and the dependent variables on acquirer around merger announcement date CAR are displayed. Columns 1 through 4 focus on the interaction effect between (BOARDTYPE) and the average cross-firm social connection based on both
educational and employment. Column 5 only looks at the interaction effect between (BOARDTYPE) employment ties. Column 6 focuses on the interaction effect between (BOARDTYPE) and
employment ties. Lastly, column 7 looks at the interaction effect of (BOARDTYPE) and strong ties. A negative z-value indicates a dampening effect of (BOARDTYPE) on the cross-firm social tie effect on acquirer CAR. Robust standard errors are in brackets and Asterisks indicate the significance level at a 10%(*), 5%(**) and 1%(***) level.
The results in table 12 shows a significant interaction effect between (BOARDTYPE) and
(CONNECTION) for the seven-day event window (z=-1.66, p<0.1). All other interaction effects are non-significant.
4.2.6 Results interpretation second hypothesis
The seconds hypothesis: The cross-firm negative social tie effect on post-announcement cumulative abnormal returns around the merger announcement date are smaller for two-tier board than one-tier board firms is first tested by adding (BOARDTYPE) to the regression model.
Table 13 compares results of the regressions with and without (BOARDTYPE) for all different measures of (CONNECTION) and event windows.
The results of the regressions without (BOARDTYPE) showed significant negative effect of all ties and educational ties on acquirer CAR. After adding (BOARDTYPE) 13 out of 15 results show a less
negative z-value. This suggests that two-tier board companies are less affected by the negative effects of all ties and educational ties on CAR. The differences are however very small. Only for two regressions the significance level decreased after including (BOARDTYPE). For this reason, the second hypothesis does not hold based on these results. The effect of employment on acquirer CAR
becomes slightly more or less negative depending on the event window. Strong ties effects on acquirer CAR are almost similar for one and two-tier board firms. This can be explained by the fact that the supervisory board does not have to intervene if social tie effects are positive.
The second way the hypothesis is tested is by including an interaction term to the regression model. This test measures how (BOARDTYPE) influences the cross-firm social tie effect. For this reason, this test is the most accurate way to test the second hypothesis. The results of the first hypothesis showed significant negative effects of cross-firm social ties on acquirer CAR for all the event windows. Only employment and strong ties showed no significant negative effects.
If the initial effect is negative a negative interaction result indicates that the effect becomes less negative under the influence of the (BOARDTYPE) variable. The second hypothesis therefore holds if the interaction term is negative and statistically significant.
Results of the all ties and educational ties column are negative for all event windows. This implies (BOARDTYPE) has a dampening effect on the negative cross-firm social tie effect on acquirer CAR. As two-tier board companies are given a value of 1 and one-tier board values a value of 0 the results shows that two-tier boards reduce the negative cross-firm social tie effect. The negative interaction results are however not statistically significant.
Only the interaction term included in the seven-day event window regression without control variables has a significant dampening effect at a significance level of 10% (z=-1.66, p<0.1). For this reason, the second hypothesis can only be accepted for this particular regression.
The results of the second test are in line with the results of the first test. Results of both tests can be used as an indication for less negative cross-firm social tie effects on two-tier board firms. Results are however not strong enough to accept the hypothesis. For the five-day event window, social tie effects became less significant in the first test and the interaction term was significant for the all ties regression with no control variables. As this is the regression with the lowest predictive power, which is showed by R2 in the table, the hypothesis also does not hold for the five-day event window.
