Retained, hired or retired? An analysis of board composition and the
director labour market
Evidence from Mergers & Acquisitions
MASTER’S THESIS
Name: Roos van den Berg Student number: 10814914
Date: June 2018
MSc Finance, Corporate Finance Thesis supervisor: T. Jochem
Page | 2 Statement of originality
This document is written by Student Roos van den Berg 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.
Page | 3 Abstract
This thesis attempts to analyse board rigidity as well as board composition in the framework of mergers and acquisitions. Using a sample of U.S. firms between 1996 and 2013, I use propensity score matching and a sample of exogenously withdrawn bids to create two distinct benchmarks for typical board changes. By implementing a differences-in-differences
framework, it is found that mergers induce board changes and increase the likelihood of (target) directors being added to the acquiring board. However, no significant effect was found for departing directors. These results are partially driven by the complexity of the deal, as quantified by relative deal size. This suggests that larger firms require more directors to effectively monitor the post-merger firm. Using several logit regressions to compare hired, retained and departing directors I then find that age and busyness primarily drive the decision to hire a director. When restricting the analysis to selected and unselected target directors, a different picture emerges. In this case, it is often the CEO that is hired by the acquirer’s board of directors. This is consistent with prior literature that shows target CEOs are willing to exchange merger premium for post-merger control rights.
Keywords: Board of directors, Director labour market, Mergers & Acquisitions, Director selection
Page | 4 Table of contents Statement of originality ... 2 Abstract... 3 Table of contents ... 4 1. Introduction ... 6 2. Theoretical framework ... 9
2.1. Career consequences as a result of director’s corporate policies ... 9
2.2. Board changes: stability vs change Empirical evidence ... 10
2.2.1. Stable boards………...10
2.2.2. Dynamic boards………..11
2.3. The underlying rationale for board stability ... 12
2.4. The underlying rationale for board changes ... 14
2.5. Hypotheses ... 18
3. Data ... 19
3.1. Data description ... 19
3.2. Sample overview ... 20
4. Methodology ... 22
4.1. Benchmark of typical board changes in a non-merger scenario ... 22
4.1.1. Propensity-score matching ... 22
4.1.2. Withdrawn firms ... 22
4.2. Main methodology……….24
4.2.1. The effect of Mergers & Acquisitions on board dynamics ... 24
4.2.2. The director characteristics associated with director selection & retention ... 25
4.2.3. Future board seats of acquiring directors………..………26
5. Results & Discussion………...27
5.1. How stable are boards following a merger? ... 27
Page | 5
5.1.2. Exogenously Withdrawn bids ... 28
5.2. Which director characteristics are associated with director retention and selection? ... 30
5.2.1. New, retained and departing directors of the acquiring board... 30
5.2.2. Selected vs Unselected Target Directors………32
5.3. What happens to acquiring directors that leave the board? ... 33
5.4. Robustness checks ... 34
5.4.1. Pre- and Post-Sarbanes-Oxley Act…...34
5.4.2. Alternative specifications for board change ...35
6. Concluding remarks and limitations ... 35
Reference list ... 37
Figures ... 41
Tables ... 41
Appendix A: Variable definitions ... 56
Appendix B: Propensity Score Matching ... 59
Page | 6 1. Introduction
The size and composition of the board and its subsequent effect on firm value has been a hotly debated topic (Ahern & Dittmar, 2012; Becher, Walkling & Wilson, 2017). A board should maintain the right balance between skills, knowledge of the firm, experience and
independence in order to fulfil its monitoring duties in the best way possible. A lot of research has been conducted on the optimal composition of the board and how to maximise firm value. However, little is known about the frequency and the drivers of board changes or about the characteristics of hired, retained and retired directors. This is especially important given the very limited influence exerted by shareholders, activists and investors on the director selection process.
This topic has become more relevant in light of the recent campaigns for board refreshment and related corporate governance practices. Not just shareholders urge wider use of mandatory retirement ages but also governments, activists and regulators are increasingly stimulating greater diversity on all fronts. For instance, mandatory gender quota have been installed in Norway in 2006 to come into force in 2008 and many other countries have implemented advisory quota in order to stimulate diversity in the boardroom (Ahern & Dittmar, 2012).
In this thesis, I examine two opposing hypotheses as identified in the literature: stability and change (Becher et al., 2017; Cicero, Wintoki & Yang, 2013; Denis, Denis & Walker, 2015; Harford, 2003). The stability hypothesis posits that boards are rigid and barely respond to increasing pressures and investor scrutiny to increase diversity and independence. Transaction costs, entrenchment motives and pre-existing qualities of the board are all potential explanations of board rigidity. The change hypothesis, on the other hand states that boards are flexible and will respond to such pressures. For instance, changes could be due to an extensive negotiation process, desires for cognitive or demographic diversity, improved outside employment opportunities or simply because of CEO connections. In order to
examine these hypotheses, the question posed in this thesis is as follows: How stable are post-merger boards and which director characteristics influence the director selection and retention process?
These hypotheses are examined in the framework of mergers and acquisitions for several reasons. If boards merely change their boards for natural reasons, such as retirement or death, we would not expect to see an effect of a shock induced by a merger. Therefore, it allows for examining whether directors use mergers as an opportunity to refresh their board
Page | 7 and thus find support for either the change or stability hypothesis. Secondly, it allows for a comparison of the appointed director to a set of unselected directors. A merger essentially provides the acquiring board with a pool of target directors. By comparing the selected target director to the rest of the (unselected) target board, conclusions can be drawn on
characteristics that are valued in the director labour market. Finally, by looking at mergers, the methodology of using exogenously withdrawn bids can be exploited which mitigates endogeneity concerns to a greater extent.
To examine the hypotheses, an extensive dataset is constructed from Thomson One Mergers & Acquisitions, CRSP, Compustat, ISS Governance and BoardEx. In order to
examine board stability, a benchmark of board changes absent a shock is needed. Suggestions for such a benchmark in empirical literature are limited (Becher et al., 2017). In order to improve the validity of the study, two distinct benchmarks are used. First, I examine board changes around mergers by using a propensity score matched sample based on firm and board characteristics. Secondly, I use a sample based on exogenously withdrawn bids. The idea behind the second methodology is that boards that are willing to engage a merger tend to be more comparable. This methodology accounts for the potential unobservable characteristics that could simultaneously drive the decision to engage in a merger as well as to change board composition that propensity score matching fails to account for. In order to examine the second question of director characteristics, I use several logit regressions. First, I focus on the difference between hired, retained and fired acquiring directors after which I turn to the difference between selected and unselected target directors. Finally, I carry out an event study to provide an overview of the board seats held by acquiring directors of value-enhancing and value-destroying mergers two years after the merger.
I find that mergers do induce board changes in terms of overall change to the board and the addition of new directors. These results are supported both when using the sample of propensity score matched firms and the exogenously withdrawn bids. The effect is partially driven by the complexity of the deal, which suggests that the need for a larger board increases as the monitoring duties become more complex. No evidence is found for acquiring directors leaving the acquiring board. Also, when looking at new and leaving directors in absolute numbers, it seems that in general, board changes are rare. This discrepancy between economic and statistical significance could potentially explain the current inconsistency in academic literature. Then, when examining director characteristics of hired, retained and fired directors, evidence is found for age and busyness of directors as drivers of the director selection
Page | 8 different picture emerges. In this case, it is often the CEO that is hired by the acquirer’s board of directors. This is consistent with prior literature that shows target CEOs are willing to exchange merger premium for post-merger control rights (Wulf, 2004).
This study makes several contributions. First of all, limited empirical evidence exists on board changes in general, or in a merger setting in particular. Analysing board changes induced by mergers allows for a comparison of selected and unselected target directors. Moreover, the limited evidence that exists tends to focus on target director characteristics who are hired by the acquirer. However, this ignores the aspect of whether they are replacing existing acquirer directors or if they are added to the board. Perhaps acquirer directors will leave the board as a result of (forced) retirement or alternatively, because of enhanced employment opportunities. This study will aim to provide greater insight into the dynamics underlying board changes in merger settings and of board composition in general by additionally examining the characteristics of departing acquirer directors. In addition, this study contributes to the discussion on whether boards are rigid and aims to resolve the current inconsistency. Historically, there is more support for the stability hypothesis, but Becher et al. (2017) found the opposite result. I will contribute by employing a unique sample of failed bids which likely mitigates endogeneity more than by using a propensity score matched sample. To the best of my knowledge, this is the first study that employs a sample of exogenously withdrawn bids to examine board changes and composition. Finally, from a policy perspective it is important to know the underlying drivers of board changes in order to decide on the most effective policies.
This thesis is structured as follows. First the underlying theories and literature review will be outlined in Section 2, which will be concluded with the hypotheses. Then, in Section 3, a description of the data used will be given along with summary statistics. This section is followed by an extensive outline of the methodologies used in Section 4. Section 5 presents the results, including the robustness tests. Section 6 provides a conclusion and discusses several limitations as well as suggestions for future research.
Page | 9 2. Theoretical framework
Two main strands of literature relate to this work; it relates to studies that look at the director labour market and studies that look at board changes in a merger setting or in general. This thesis is mostly related to the studies conducted by Becher et al. (2017) and Harford and Schonlau (2013). In order to understand corporate governance, director quality and the choice of directors should be analysed simultaneously (Hermalin & Weisbach, 1998). By examining the demand and supply of director characteristics, this thesis contributes to the discussion regarding the dynamics between the director characteristics valued in the market and those valuable to the firm. Moreover, expanding on existing literature, I will examine whether merger experience and firm performance extend to post-merger benefits in the form of board seats.
The theoretical framework is structured as follows. First, the director labour market will be analysed. Second, the relationship between director characteristics and firm value as documented in prior literature will be presented. Thirdly, literature related to board changes will be analysed after which in section four, the underlying reasons will be explored. The theoretical framework will be concluded by several hypotheses related to the research question.
2.1 Career consequences as a result of director’s corporate policies
In order to understand the director characteristics valued in the labour market, it is important to analyse whether and how these traits translate into employment opportunities. There is a vast amount of literature related to the incentives of directors as well as employment
opportunities in the director labour market (Harford & Schonlau, 2013). Rather than focusing on specific director characteristics, this section aims to explain the role of the firm in
directors’ number of future board seats.
Harford and Schonlau (2013) found that directors involved in large acquisitions are significantly more likely to have a larger number of future board seats. This result holds for both acquiring and target directors and for both CEOs as well as directors in general. Interestingly, no distinction seemed to be made for value-enhancing or value-destroying acquisitions in the sense that directors were rewarded similarly in both cases. In addition, Harford (2003) showed that fewer board seats are awarded to directors that reject a takeover offer after poor stock performance.
Page | 10 corporate events or policies. Gilson (1990) studied the director labour market in firms that filed for bankruptcy or restructured their debt between 1970 and 1985 and subsequently found that only 46% of the incumbent directors remained on the board. Directors who resigned hold fewer board seats in the future than directors who decided to stay (Gilson, 1990). Relatedly, Kaplan and Reishus (1990) established that CEOs of dividend-reducing companies receive fewer future employment opportunities in the form of board seats.
Even though they did not examine a specific policy or corporate event, Brickley, Linck and Coles (1999) similarly found that better firm performance is positively related to CEO retention. In addition, evidence was presented for post-retirement career concerns of CEOs. They showed that CEO performance is strongly and positively related to the number of board seats after retirement. Thus, they established that post-retirement concerns can provide a source of incentives facing a current CEO. Poorly performing CEOs are not merely more likely to lose their current positions, but may also be disadvantaged post-retirement in the form of fewer awarded board seats.
Ferris, Jagannathan and Pritchard (2003) confirmed the positive relation between firm performance and board seats. However, rather than focusing on post-retirement board
opportunities, Ferris et al. (2003) looked at directors with multiple board seats. They found evidence for reputational effects in the director labour market. Specifically, directors who serve on larger boards and for larger firms are more likely to attract directorships.
2.2. Board changes: stability vs change Empirical evidence
As aforementioned, little empirical evidence exists regarding the frequency of board changes as well as the underlying reasons. However, several papers have addressed board dynamics, either directly or indirectly. In this section, two competing strands of literature will be discussed; those that argue boards are stable and studies which find that board changes are a relatively frequent occurrence. This section describes empirical evidence for observed board changes, whereas in section 2.4 and 2.5 the respective reasons for these changes will be outlined.
2.2.1. Stable boards
The first hypothesis is that board changes are a relatively infrequent occurrence (Becher et al., 2017; Cicero et al., 2013; Denis et al., 2015; Harford, 2003).
Page | 11 First, Harford (2003) examined the effect of a takeover bid on the subsequent effect on target directors using a sample of 1,091 directors from the Fortune 1000 firms that received a takeover bid between 1988 and 1991. His results were in line with the stability hypothesis, as it was found that the target director retention rate was low. In addition, he found that target directors hold fewer directorships after the merger than comparable directors in non-merging firms. However, his results were not fully compatible with the stable boards hypothesis as he also supported the view that directors currently serving the firm are rarely retained following a completed offer. In conclusion, Harford (2003) failed to find evidence for target director retention, but did find evidence for departure of acquiring directors.
Although mostly consistent with the change hypothesis, Cicero et al. (2013) provided some evidence that changes have become less frequent since the Sarbanes-Oxley Act of 2002. Specifically, they found that firms became less likely to adjust their board structures to
decrease independence when firm fundamentals warranted such changes. This is somewhat counterintuitive in the sense that these regulations were expected to induce more frequent board changes in order to comply and avoid shareholder scrutiny. However, Cicero et al. (2013) found that particularly board changes that opposed the common perception of what constitutes a ‘good’ board structure became less frequent. Despite the focus of this thesis being changes at the firm-level, Cicero et al. (2013) suggested that at an aggregate industry level, regulation plays a role in the frequency of board changes. This is further supported by Kole and Lehn (1999), who used the Airline Deregulation Act of 1978 as a natural
experiment to find that in the event of deregulation, board size decreases, equity ownership becomes more concentrated and CEO pay increases. In fact, they found that the governance structures of airlines gravitate towards governance structures of non-regulated firms.
However, as this process of adaptation was found to slowly progress, they argued that it is costly to alter current board structures.
2.2.2. Dynamic boards
The second hypothesis refers to changes around mergers and acquisitions. Currently the empirical evidence regarding the director labour market is inconsistent in the sense that there has also been empirical evidence that documented frequent board changes. For instance, Denis and Sarin (1999) conducted a time-series analysis of ownership and board composition in individual firms and found that it fluctuates in firm-years. More specifically, by examining 583 firms over the ten-year period 1983-1992, they found that 12% of the firm-years were
Page | 12 characterised by large changes in either ownership structure or board composition. In
addition, they documented a relationship between changes in board structure and executive turnover, stock performance and threats to corporate control. They also found that changes in board structure usually follow after fundamental changes in the business environment.
Finally, Denis and Sarin (1999) showed that in 65% of their sample, at least one large change in ownership or board composition was observed. When restricting the analysis to changes in board composition, it was found that in 40% of the firm-years the board size changes.
Cicero et al. (2013) further extended the analysis as carried out by Denis and Sarin (1999) to a more recent time frame and concluded that the number of changes in board composition has increased and is negatively related to the level of CEO power. In particular, they found that over any two-year period between 1991 and 2003, 50% of the firms changed their board size. Although not directly comparable to Denis and Sarin (1999) due to different data sets and methodologies, it suggests that after 1992, board changes became more frequent. However, after the Sarbanes-Oxley Act in 2002, they found a decline in the frequency of board changes, which is again consistent with the findings by Harford (2003). The change hypothesis can be split up further into several smaller, testable hypotheses in order to examine the reason why the changes take place.
More recently, Becher et al. (2017) examined board changes after mergers and acquisitions and compared non-merging firm years to firms in which a merger took place. Additionally, they looked at the difference between non-merging firms and merging firms by employing propensity score methods. They found that boards change substantially.
Specifically, they found that the size or composition of the board changed in 84% of the merging years, as opposed to 68% in the absence of a merger.
2.3. The underlying rationale for board stability
There are various studies that explore reasons as to why relative stability around mergers should be expected. These can be divided up into three main categories: pre-existing qualities of the board, reputational & transaction costs and negotiations between the CEO and the board of directors.
2.3.1. Pre-existing qualities of the board
There are various reasons for this board rigidity, among which the pre-existing qualities of the board (Becher et al., 2017). In general, target firms tend to be smaller relative to the acquirer’s board. Therefore, the acquirer’s board may already possess the required characteristics to
Page | 13 adequately monitor the post-merger company.
For instance, Ahern and Dittmar (2012) observed a negative impact on firm value induced by mandated female board representation. One potential explanation for this result is that boards were already optimal and this sudden implementation of a new law brought about a negative shock to this existing equilibrium. However, there is no way of separating the gender effects from experience. In fact, Bennedsen, Nielsen, Pérez–González and Wolfenzon (2007) failed to find such negative value effects following director deaths, in which scenario an absence of similarly qualified candidates is unlikely.
2.3.2. Reputational and transaction costs
In the absence of transaction costs to altering board structure and assuming managerial entrenchment is not a concern, board structure should not be related to firm performance. It could also be a result of transaction costs such as reputational costs of firing directors or the search costs associated with identifying potential candidates (Coles, Daniel & Naveen, 2008). These costs could dissuade boards from changing its current composition or size (Becher et al., 2017; Coles et al., 2008; Denis et al., 2015). Transaction costs encompass all costs which directly or indirectly prevent value-enhancing adjustments to board size or independence. Based on this, Coles et al. (2008) argued that complex firms will likely have smaller boards than warranted in the optimal scenario. Complex firms should have larger boards due to the added experience additional board members bring along (Coles et al., 2008). Also, boards of R&D-intensive firms constitute of a smaller portion of insiders compared to the optimal (Coles et al., 2008). R&D-intensive firms tend to benefit more from firm-specific knowledge, so more insider directors are beneficial.
Mergers would likely increase the complexity of the firm, thereby warranting changes to both the size and independence of the board. If transaction costs ensure that these
adjustments occur gradually, rather than immediately, this would likely negatively affect firm value.
2.3.3. Negotiations between the CEO and the board of directors
Finally, board composition could be a result of negotiations between a CEO and the board. Hermalin and Weisbach (1998) derived a formal model which explains that the CEO exerts bargaining power over the board of directors. This bargaining power is subsequently derived from the CEOs perceived superior ability relative to successors. If the structure of boards is determined in a process at least partially controlled by the CEO, we would expect relatively
Page | 14 few changes surrounding mergers and acquisitions. In the literature, this phenomenon is also referred to as a ‘captive board’. It is inherently connected to the agency problem arising from the separation of ownership and control. However, this is based on the assumption that directors with bad monitoring abilities are not subsequently punished in the director labour market. For instance, it could damage their reputation if CEOs are entrenched. Kaplan and Reishus (1990) confirmed this hypothesis by showing that directors from poorly performing monitors, and therefore perhaps perceived to be bad monitors, are penalised through fewer future board seats. Nevertheless, these reputational concerns are unlikely to resolve all agency problems (Kaplan & Reishus, 1990). In addition, the model by Hermalin and Weisbach (1998) was empirically supported by Boone, Field, Karpoff and Raheja (2007). By using several measures of insider influence such as CEO tenure and ownership, Boone et al. (2007) showed that a CEO’s influence is negatively related to the proportion of independent directors serving on the board. Their results reflected the study conducted by Baker and Gompers (2003), who came to similar conclusions regarding the bargaining hypothesis.
Cicero et al. (2013) also argued that entrenchment costs likely explain rigid board structures. However, in addition they considered negotiation costs. Whereas the former stems from managers entrenching themselves in such a way that they are insulated from market pressures, the latter is derived from the ability of directors. Due to directors’ perceived ability, shareholders are more likely to trust directors to exert their influence. Thus, whilst both increase the bargaining power of management, the difference stems from the extent to which their influence is warranted by their qualities (Cicero et al., 2013).
2.4. The underlying rationale for board changes
2.4.1. Diversity
The first reason for board changes stems from the characteristics of target directors.
Irrespective of the qualities of acquiring directors, target directors may possess certain skills or demographic attributes which enhances their employability. Although often studied as one phenomenon, I would argue that this hypothesis can be broken up into two parts; the first being related to particular skills target directors possess (‘cognitive diversity’) whereas the second refers to demographic characteristics (‘demographic diversity’).
Cognitive diversity
Page | 15 greater need for cognitive diversity (Becher et al., 2017; Boone et al., 2007; Denis et al., 2015; Harford & Schonlau, 2013). A particularly important characteristic of directors is experience (Bennedsen et al., 2007; Denis et al., 2015; Harford & Schonlau, 2013). Becher et al. (2017) found empirical evidence for the period from 1992 to 2013 for two phenomena which they refer to as the need hypothesis and the bargaining power hypothesis. By analysing director characteristics in a merger setting as well as by using proxies to directly test for the need and bargaining hypothesis, they found that in merging years, directors seem to be added to the board for their particular skills, such as merger experience, CEO experience or board experience.
Denis et al. (2015) examined corporate spinoffs to provide further insight into the determinants of board structure and found evidence for the importance of industry expertise. In particular, when boards are formed from scratch, as is the case in a corporate spinoff, they found that these boards have more outsiders with relevant industry experience and are generally more independent.
Even though for the purpose of this study, experience and ability have been grouped together Harford and Schonlau (2013) found that experience is more important than ability in the case of acquisitions. In contrast, Brickley et al. (1999) showed that firms consider ability in selecting directors, thereby casting doubt on the claim that family, acquaintances and friends tend to be appointed as directors. Gilson (1990) further supports the need hypothesis by showing that directors who leave the board of a firm engaged in bankruptcy filing or significant debt restructuring receive fewer future employment opportunities. This shows that firms select directors based on performance and that their skills affect a director’s individual market valuation. In addition, Booth and Deli (1996) focused on perquisite consumption, CEO tenure as well as relationships between boards. They found that the number of directorships held is positively related to the CEO’s tenure. The rationale behind this is as follows. CEOs approaching retirement are likely to groom a successor and eventually pass on decision rights. As more decision rights are transferred, it becomes less costly to the firm for the CEO to serve on additional outside boards.
The need for skilled directors would arguably be more prominent in the case of
diversifying mergers as the current board may not possess the needed skills to monitor a more complex firm. In other words, selecting directors with the most desired qualities becomes more important as a firm become more complex (Becher et al., 2017; Boone et al. 2007). Boone et al. (2007) presented evidence from IPOs for the need hypothesis which they refer to as the scope of operations hypothesis. They tracked changes in board composition over a
10-Page | 16 year time period starting from a firm’s IPO. Using firm size, firm age and the number of business segments as proxies for the complexity of the firm’s operations, they concluded that it is positively related to both board size and the level of board independence. Based on these results, they concluded that as companies expand, the need for monitoring and director specialisation increases. Therefore, in a merger setting we would expect to see more frequent board changes.
Demographic diversity
The second reason, as mentioned before, stems from demographic diversity. A vast amount of empirical literature has emerged that examines the effect of demographic characteristics on firm value (Adams & Ferreira, 2009; Ahern & Dittmar, 2012; Bernile, Bhagwat & Yonker, 2018; Farrell & Hersch, 2005). Diversity encompasses many different aspects, such as gender, ethnicity and age of which gender has received the most attention in the policy debate (Adams & Ferreira, 2009; Ahern & Dittmar, 2012; Schwartz-Ziv, 2017). For instance, Ahern and Dittmar (2012) examined the corporate board changes and subsequent effects on operating performance following a change in the law that required 40% of Norwegian firms’ directors to be women. Related to this, Adams and Ferreira (2009) found that gender-diverse boards engage in greater monitoring, but the average effect of gender diversity on firm performance is negative. By focusing on Corporate Social Responsibility practices rather than firm value, Cronqvist and Yu (2017) demonstrated that the positive effects of gender diversity extend beyond those directly affecting performance.
In addition to gender diversity, ethnical diversity has also been proven to increase firm value (Carter, Simkins & Simpson, 2003). Carter et al. (2003) for instance combined gender and ethnical diversity and showed that both are associated with an increase in firm value. Taking the definition of diversity one step further by additionally including different educational backgrounds and financial expertise as components of diversity, Bernile et al. (2018) showed that greater board diversity reduces volatility and increases performance.
Demographic diversity is related to cognitive diversity in the sense that both originate from specific director characteristics. However, the question is whether directors are retained for their particular skills relevant to the more complex post-merger firm (‘cognitive diversity’) or alternatively for their more symbolic director-specific attributes in response to the
increasing societal and shareholder pressures (‘demographic diversity’).
Page | 17 2.4.2. Managerial objectives
This reason is not so much related to the demographic or professional characteristics of the target directors, but rather by managerial entrenchment. However, this hypothesis is two-sided as it can be driven by target or acquirer managerial objectives.
First of all, retention of target directors could be dominated by negotiations between the acquirer board and the target board (Becher et al., 2017). In addition to the name of the post-merger firm, the premium paid and the organisational structure after the merger, target board members may also seek to be hired onto the post-merger board. The likelihood of getting a director position increases in their bargaining power. Bargaining power can stem from a variety of characteristics, such as the complexity of the firm, anti-takeover defences or the value of the deal (Rajan & Zingales, 1998). Becher et al. (2017) found empirical evidence for the bargaining hypothesis. In addition, Hartzell, Ofek & Yermack (2004) found that target director retention is negatively related to received bonuses in a merger and acquisition setting whereas Wulf (2004) found that target CEOs exchange premium for post-merger control rights. This implies that target directors do have bargaining power to negotiate personal benefits.
In addition, it could be due to CEO entrenchment on the acquirer’s board. Social connections between the CEO of the acquirer and the target directors could induce the CEO to exploit the merger as an opportunity to get directors on the board that support his agenda (Becher et al., 2017; Fracassi & Tate, 2012).
2.4.3. Supply-side dynamics
The retention of target directors could also be due to improved acquirer directors’ outside employment opportunities or retirement (Brickley et al., 1999; Harford & Schonlau, 2013). In other words, target directors are filling the gap left by acquirer directors who have terminated their directorships in favour of retirement or more rewarding directorships. This hypothesis is not extensively documented in prior literature, but could drive the decision to hire target directors. Brickley et al. (1999) found that pre-retirement performance is negatively related to retention. Lee (2011) confirmed this result for the time period of 1989 to 1993 but failed to find such evidence for subsequent time intervals.
Page | 18 2.5. Hypotheses
In this section, the hypotheses resulting from the discussion above will be briefly summarised. In the first part of this thesis, the board changes subsequent to Mergers & Acquisitions will be examined. Prior research has predominantly found evidence for board changes rather than for stability. In addition, I argue that differences between studies are due to different samples and the extent to which endogeneity is accounted for. As Becher et al. (2017) pointed out, finding a benchmark against which board changes should be compared is problematic. By employing a sample of exogenously withdrawn bids, I hope to reconcile the inconsistencies in prior literature. For non-merging firms, it is more likely that directors leave for reasons unrelated to diversity or the bargaining process due to the absence of a pool of target directors. Therefore, it is hypothesised that they will be more likely to leave for retirement reasons. This leads to the following hypothesis. Based on prior empirical evidence, along with increasing pressures for cognitive and in particular, demographic diversity, the following hypothesis arises:
Hypothesis 1: Board changes occur in response to Mergers & Acquisitions
In the section following, the director characteristics associated with director retention and selection will be examined. Based on an increasing number of policies aimed at creating gender (Adams & Ferreira, 2009; Ahern & Dittmar, 2012; Schwartz-Ziv, 2017) and ethnical diversity (Bernile et al., 2018; Carter et al., 2003), it is expected that characteristics proxying for demographic diversity will positively affect the likelihood of director retention &
selection and thus negatively affect director departure. In addition, based on a variety of research conducted in this field, it is expected that cognitive diversity will positively and significantly affect director change (Becher et al., 2017; Bennedsen et al., 2007; Boone et al., 2007; Denis et al., 2015; Harford & Schonlau, 2013). This hypothesis can be summarised as follows.
Hypothesis 2: Director characteristics related to cognitive diversity as well as demographic diversity increase the likelihood of director retention & selection
Finally, I will provide some descriptive statistics for the number of board seats held by directors of acquiring firms and firms which experienced a withdrawn bid. It is expected that acquiring directors subsequently have a higher number of board seats as it was found that regardless of whether a merger is value-enhancing or value-destroying, directors are rewarded
Page | 19 in the labour market (Harford, 2003; Harford & Schonlau, 2013). Therefore, the final
hypothesis reads as follows.
Hypothesis 3: Given prior evidence that directors are rewarded for merger experience, it is expected that acquiring directors subsequently have a higher number of board seats.
3. Data
3.1. Data description
In order to construct the full dataset, all U.S. mergers and acquisitions announced and completed will be retrieved from Thomson ONE. More specifically, I obtain deal premium, attitude (hostile/friendly), percentage of shares acquired and sought, and method of payment. A complete overview of all the definitions of the variables used is provided in Appendix A. Due to the sample restriction imposed by ISS Governance, January 1, 1996 will be taken as the first date of the complete data set. In order to exclude all economically small or
insignificant transactions, it is required that both target’s and acquirer’s assets are worth more than 1 million dollars or that the total value of the transaction is larger than 1 million dollars (Bena & Li, 2014). Also, both firms have to be publicly traded, in order to observe director characteristics for both the target and the acquirer. Deals in which either the target or acquirer is incorporated outside the United States are excluded. In addition, acquirers and targets from the financial sectors, as identified by the SIC codes 6000-6999, are excluded from this
analysis due to their inherently different nature. An additional restriction is that data has to be available for the acquirer from CRSP, BoardEx, ISS Governance and Compustat. Finally, in line with Bena and Li (2014), only completed deals are included if at least 90% of the target is acquired and if prior to the merger, the percentage of common stock sought was more than 50%.
From Thomson ONE I also retrieve bids that were unsuccessful. To construct this control sample, withdrawal dates will be obtained. Similar to completed deals, only deals will be included with intended deal size larger than 1 million. Alternatively, target and acquirer’s assets should be larger than 1 million. In addition, to prevent deals from being included as both failed and completed, the restriction is imposed that a withdrawn bid cannot
subsequently be completed for the following 3 years.
Daily stock data is retrieved from CRSP, in order to calculate one-year returns as well as stock volatility. Stock volatility is calculated as the square root of the sum of squared daily
Page | 20 returns over the year, multiplied by 252 and divided by the number of trading days in order to account for differences in the number of trading days. Returns are winsorised at 1% and 99%. Firm-level variables are taken from Compustat. Governance data with respect to the board of directors is retrieved from ISS (formerly known as RiskMetrics) Governance and ISS
Governance Legacy (for the years prior to 2007). This dataset includes information on anti-takeover defences, such as golden parachutes, staggered boards and poison pills, as well as on general board characteristics, such as average director age. All variables are collected for both the target and the acquirer and for years prior, during and post the merger. This will allow for comparison of all variables pre- and post-merger.
Finally, this database is merged with BoardEx to obtain director data. Manipulating and merging in this database provides several difficulties. BoardEx only supplies the ISIN as a firm level identifier. ISIN refers to the securities of the firm. However, BoardEx supplies all ISIN codes per firm. In order to find the ISIN corresponding with the CUSIP, I added the restriction that the ISIN should start with the country code ‘US’. The several ISIN’s were separated using Excel. BoardEx provides data on current and historic directorships, which were merged in separately in order to have an extensive overview of directorships held both in the past and in the present. After having merged in the employment data, this is then matched on the director id supplied by BoardEx with data on education, characteristics and non-board roles. An additional challenge is posed by the way BoardEx describes the data. In order to generate the ‘same name’ variable, I separated the first name, middle name, last name and potential titles in order to isolate the surname. A similar problem was posed by the
information on committees. The committee names are merged in one cell by BoardEx with no particular order and need to be separated in order to create separate dummy variables for the different committees. Education data was merged in based on the date of having finished the degree being before the date of employment. The final sample consists of 1,302 deals, for which director information is available on 596 completed and 69 withdrawn deals.
3.2. Sample overview
Panel A of Table 1 shows deal characteristics for 596 completed and 69 withdrawn deals, respectively. The summary statistics show a slightly lower deal value as compared to Becher et al. (2017), which can be explained by the relatively conservative restriction imposed on minimum deal value ($1 million) and minimum target and acquirer asset value ($ 1 million). However, Harford and Schonlau (2013) found similar values for the dollars spent on
Page | 21 which could be explained by the fact that most studies use the Fama-French (1997)
classification, whereas I used the more simplistic SIC classification. All other variables are consistent with prior literature.
There are also several differences in the summary statistics for completed and withdrawn deals. First of all, the frequency of hostile deals is much higher for withdrawn deals. This is consistent with the finding that hostile deals have become less likely to succeed due to anti-takeover statutes and anti-takeover defences (Fos, 2016). Also, not surprisingly, the frequency of deals with multiple bidders is higher for withdrawn deals, which is consistent with lower deal success. In addition, it seems as if diversifying acquisitions are more likely to succeed. The deal value is also higher, which could point at financing problems for the acquirer as a reason for withdrawal. Finally, deal premia are slightly lower which could possibly be a cause of deal failure if target management feels the price is unfair.
[INSERT TABLE 1 HERE]
The firm statistics in panel B show that the target is usually smaller than the acquirer, both in terms of firm size and number of employees. They also have a higher debt/equity ratio and more R&D activity (Bena & Li, 2014). Withdrawn firms are also smaller than pre-merger acquirers. Cash reserves decrease in the post-merger firm, which could be due to relaxed financing constraints as a result of better access to capital (Erel et al., 2015). The differences in deal and firm characteristics for completed and withdrawn deals underlines the importance of identifying failed bids that were withdrawn for exogenous reasons. This will likely enhance the comparability and thus support the underlying assumption that absent the treatment, exogenously withdrawn and completed deals would have been identical.
The board and governance characteristics are shown in panel C and the director statistics are presented in panel D. When imposing the additional restriction that data is available on ISS Governance and BoardEx, fewer observations remain. Most notably, the board size is larger in the post-merger firm. In addition, the percentage of female directors is higher in the post-merger firm rather than in the target firm or the pre-merger acquirer.
Page | 22 4. Methodology
This section is structured as follows. First, the benchmark of typical board changes in the absence of a merger will be explained as well as how it can be used to mitigate endogeneity concerns. In section two, the main methodology will be outlined by sub question, which will be used to test the several hypotheses outlined in section 2.5.
4.1. Benchmark of typical board changes in a non-merger scenario
The current academic literature to date is divided with respect to what constitutes a good benchmark of typical board changes. For this reason, I will adopt two distinct approaches in order to examine the stability and change hypotheses.
4.1.1. Propensity-score matching
The first methodology is propensity-score matching. Propensity-score matching allows for taking into account varying time and industry trends, as well as firm-characteristics. Firm-years in which the acquirer engaged in a merger are matched to non-merger firm-Firm-years based on board and firm characteristics, the results of which can be found in Appendix B.
Importantly, the firms are matched based on pre-merger characteristics. Based on the same methodology as adopted by Becher et al. (2012), each treated observation is matched 10 times with replacement. This yields a sample of 584 treated merger firm-years and 2,599 non-merger firm years. The model upon which the propensity score matching is based is presented in Appendix B.
4.1.2. Withdrawn firms
This thesis fits in with a number of papers that use failed bids as a control group (Bena & Li, 2014; Liu, 2017; Malmendier, Opp & Saidi, 2016). In order to identify whether a deal is withdrawn, announcement and withdrawal dates will be retrieved from Thomson ONE. As mentioned before, prior literature fails to account for potential unobservable characteristics that may simultaneously drive the decision to engage in a merger as well as the impetus for subsequent board changes. In other words, rather than because of the merger itself, it could be the case that firms endogenously select into the treatment group. A way to mitigate this endogeneity is to conduct a quasi-experiment in which I examine a control sample of withdrawn deals that failed for reasons exogenous to the acquirer’s board or firm
Page | 23 exogeneity, based on the categorisation by Malmendier et al. (2016). A deal is classified as exogenously withdrawn if the deal failed as a result of regulatory intervention, competing bids, and adverse market or industry conditions (Bena & Li, 2014; Liu, 2014; Malmendier et al., 2016). To a lesser extent, it could also be argued that a deal is exogenously withdrawn if it is due to news about the target as this is not directly related to the acquirer’s board or firm characteristics. Yet I argue that it could be correlated with the acquirer’s ability to pick undervalued targets, so in order to be conservative, these reasons are excluded from the analysis. To determine whether the deal is withdrawn for exogenous reasons, I will read news articles retrieved from LexisNexis. Finally, in order to provide a comprehensive analysis, both completed and withdrawn deals are required to have data for three years prior and three years after the completion or withdrawal.
[INSERT TABLE 2 HERE]
Table 2 reports the summary statistics for the failure reasons as categorised by Malmendier et al. (2016). The reasons for withdrawal are divided into 12 categories. Despite the different sample periods, the distribution of reasons can largely be compared to Malmendier et al. (2016). ‘Alliance’ refers to those bids that were withdrawn because the acquirer and target entered into a different type of engagement other than a merger. ‘Bidder acquired’ and ‘Bidder problems’ both refer to the acquirer being the sole reason for the failed merger. The former refers to the situation in which the acquirer is being acquired by a third party, whereas the latter points at a variety of problems, the most common of which is financing issues or a sudden drop in the stock price. ‘Industry problems’ relates to issues with respect to a certain industry, such as a decline in demand within a certain sector or declining oil prices. ‘Market problems’ encompass more widespread issues, such as an economic downturn, most notably in 2008, or the terrorist attacks of 9/11. Two other categories are ‘competing bids’ and ‘management rejection’. The two are related in the sense that in both cases the deal does not go through due to target management rejecting the deal. However, in the former case it is explicitly due to the price or other terms being less favourable as compared to the competitor, whereas in the latter management rejects it for other reasons. ‘Shareholder rejection’ refers to deals in which not enough shares were tendered for the deal to go through or if a proxy vote was held. The failure reason is classified as ‘Management terms’ when target and acquirer management cannot agree on certain terms, such as who will become the new CEO of the company. Finally, it could be due to news related to target financials, for instance (‘Target
Page | 24 news (public)’ and ‘Target news (private)’). Finally, one of the most common reasons to why mergers fail to be completed is prevention by a regulatory organisation (‘Regulator’).
Originally, I collected data on failure reasons for a total of 132 withdrawn bids, which is consistent with the number found by Malmendier et al. (2016). However, after imposing the restriction of data availability from ISS Directors, a control sample of 69 failed mergers remains, which is approximately of similar size as the sample of Bena and Li (2014).
4.2. Main methodology
In what follows, I will explain the main methodology per sub question and hypothesis. First, I will explain the main methodology for whether mergers and acquisitions lead to changes in the board of directors. Secondly, I will go on to explain the approach to examining director characteristics associated with director retention, selection and departure. Finally, I will focus on departing acquiring directors to provide further insight into the board dynamics.
4.2.1. The effect of Mergers & Acquisitions on board dynamics
First, to clarify how often board changes take place in the context of mergers and acquisitions, I will use a differences-in-differences framework. As a control group, the sample of
propensity score matched firms and the sample of withdrawn bids will be used in two separate regressions. The regression can be specified as follows:
𝑦𝑖𝑗𝑡 = 𝛽0+ 𝛽1𝑀𝑒𝑟𝑔𝑒𝑟𝑖𝑗𝑡+ 𝛽2𝐴𝑓𝑡𝑒𝑟𝑖𝑗𝑡+ 𝛽3𝑀𝑒𝑟𝑔𝑒𝑟𝑥𝐴𝑓𝑡𝑒𝑟𝑖𝑗𝑡+ 𝛽4𝑍𝑖𝑗𝑡 + 𝜆𝑗+ 𝛾𝑡+ 𝜀𝑖𝑗𝑡
The dependent variable is yijt which is variable that proxies for board change, three different
dependent variables are used. First, yijt is an indicator variable equal to 1 if an acquirer
director leaves. A similar dependent variable is constructed for a director added to the board, either from a target or not. Finally, as a dependent variable, the change in overall board size is used.
Mergerijt is an indicator variable that is equal to 1 when a merger has taken place,
whereas it is 0 for a withdrawn bid/non-merging firm year. Afterijt indicates whether the year
is after the merger. The interaction term represents the key variable of interest and will show the effect of a merger on board dynamics. Specifically, the coefficient on the interaction term Merger x After should be positive (and significant) if acquiring boards exhibit significantly
Page | 25 more change than non-merging firms in the control group. Zijt represents a set of control
variables for firm and board characteristics (Ahern & Dittmar, 2012; Becher et al., 2017). As firm variables, return on assets, firm size, R&D, stock performance, stock volatility, the level of cash and the number of employees are included. As far as board characteristics, the average director age, board size, the fraction of independent directors on the board and poison pill are included. Finally, for some specifications time-fixed effects are included as well as industry fixed effects. The regressions are also repeated with firm-fixed effects in order to control for any firm characteristics not captured by the control variables.
4.2.2. The director characteristics associated with director selection & retention
After having examined whether and how frequently mergers and acquisitions induce changes to the board, I next turn to which director characteristics are decisive in the decisions related to director selection and retention. Several logit regressions will be employed, with the dependent variable being two indicator variables. The first one is equal to 1 if a director is accepted onto the acquirer board, whereas the second variable is an indicator equal to 1 if the acquirer director is retained. In other words, one dummy variable for selection and one dummy variable for retention is used. In line with characteristics likely to affect board composition as found by prior research, the regression specification can be formalised as follows (Ahern & Dittmar, 2012; Becher et al., 2017; Bernile et al., 2018; Fich & Shivdasan, 2006; Schwartz-Ziv, 2017):
Pr(Yijt = 1) = 𝛽0+ 𝛽1𝐴𝑔𝑒𝑖𝑗𝑡 + 𝛽2𝐹𝑒𝑚𝑎𝑙𝑒𝑖𝑗𝑡+ 𝛽3𝐼𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑗𝑡+
𝛽4𝐶𝐸𝑂 𝑖𝑗𝑡+ 𝛽5𝐶𝐹𝑂𝑖𝑗𝑡+ 𝛽6𝑀𝐵𝐴𝑖𝑗𝑡+ 𝛽7𝑆𝑎𝑚𝑒 𝑛𝑎𝑚𝑒𝑖𝑗𝑡+ 𝛽8𝑇𝑒𝑛𝑢𝑟𝑒 𝑖𝑗𝑡+
𝛽9𝑃ℎ𝐷𝑖𝑗𝑡+ 𝛽10𝐵𝑢𝑠𝑦 𝑑𝑖𝑟𝑒𝑐𝑡𝑜𝑟𝑖𝑗𝑡 + 𝛽11𝑁𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑡𝑦𝑖𝑗𝑡+
𝛽12𝐶𝑜𝑚𝑚𝑖𝑡𝑡𝑒𝑒𝑖𝑗𝑡+ 𝜆𝑖𝑗+ 𝜀𝑖𝑗𝑡
Using a wide variety of director characteristics, will allow for assessing which variables are most important in determining director retention and selection. A detailed description of the variable definitions can be found in Appendix A.
In addition, deal fixed effects will be included. Similar regressions will be performed in order to draw comparisons between unaffiliated directors, retained and unretained acquirer directors and between selected and unselected target directors. Mergers provide a unique setting to
Page | 26 compare the target director that was hired to the remaining target board. To better understand the changes induced by the mergers, the average director characteristics will be compared between retained target directors and unretained acquirer directors in a similar fashion to Ahern and Dittmar (2012), by performing t-tests on the differences of the means.
4.2.3. Future board seats of acquiring directors
Finally, in order to provide further insight into the post-merger employment consequences for acquiring directors, several descriptive statistics will be provided. Specifically, for each merger in the sample, the Cumulative Abnormal Return is calculated for the 21 days surrounding the merger (-10,10). The estimation period is (-250,15).
𝐶𝐴𝑅𝑖 = ∑ 𝐴𝑅𝑖𝑡 10
𝑡=−10
This gives an idea of whether the acquisition was seen as value-enhancing or value-destroying by the market. In order to calculate the abnormal returns, the adjusted market model was used. The number of board seats are calculated for the second year after the merger took place. A distinction is made based on CARs being positive or negative. Additionally, a comparison is made between directors involved in completed mergers and withdrawn bids to see whether directors are rewarded for acquisitions in general, irrespective of its effect on firm value (Harford & Schonlau, 2013).
Page | 27 5. Results & Discussion
5.1 How stable are boards following a merger?
Firstly, board stability will be examined using the two distinct benchmarks for typical board changes as defined in section 4.1. I will first discuss the propensity score matching after which the exogenously withdrawn bids methodology will be addressed. For the purpose of this analysis, the time frame is restricted to the years 1996 to 2013.
5.1.1. Propensity score matching
The results of the multivariate Differences-in-Differences using propensity score matching to match merger firm years to non-merger firm years are presented in table 3. This yields a sample of 3,183 observations of which 584 are firm-years and 2,599 are non-merger firm years. The reason the sample is smaller than the total mergers that occurred within the time frame (see Appendix C), is data availability. Especially data availability for director data as well as governance data is limited. For ISS Governance, data is restricted to S&P1500 firms.
[INSERT TABLE 3 HERE]
As mentioned in the methodology section, the coefficient on the interaction term Merger x After should be positive and significant to conclude that a merger leads to board changes. Table 3 shows 9 regressions with different dependent variables and varying controls. Model 1, 4, and 7 show basic specifications with no controls. Model 2, 5, and 8 only control for industry and time fixed effects. Model 3, 6, and 9 include a range of control variables, including firm and board characteristics.
Model 1 – 3 show the effect of a merger on board change as a whole. The coefficient is positive in all three cases, but it only exhibits significance in the most basic regression at a 5% level. The coefficient ranges from 0.13 to 0.24. This provides moderate support for board changes induced by mergers as compared to matched non-merging firm years. The first regression suggests that a merging firm adds on average 0.24 more directors to the board in the year after the merger. The regressions also show that ‘Merger’ on its own is insignificant, meaning that firms that engage in mergers are not inherently more likely to induce board changes in non-merging years.
The effect of mergers on director selection is presented in Model 4, 5 and 6. All three coefficients are significant to a 5% or a 1% level. The coefficient for the first two regressions
Page | 28 is approximately 0.09, meaning that a merging firm is 9% more likely to hire a new director in the year after the merger. The coefficient grows in magnitude and the significance
increases as more control variables are added.
Finally, in model 7, 8 and 9 the effect of mergers on director departure as compared to matched non-merging firm-years is shown. None of the regression coefficients are significant, although they do exhibit the expected positive effect. Yet, as significance is lacking, no conclusions can be drawn from these specifications.
[INSERT FIGURE 1 HERE]
These results are supported by figure 1, in which it is clear that director selection and board changes jump at the merger year, whereas director departure shows a more gradual incline. Whereas director selection seems to revert back to its pre-merger level, the effect on director departure is more permanent. It could potentially be the case that target directors are hired after a merger. However, rather than firing an existing director, the board grows in size in order to monitor the more complex firm effectively. Thus, the optimal board size could increase after a merger. This will be tested in more detail in section 5.1.2. Another potential reason is that directors are not fired immediately, but in the years during the aftermath of the merger when the effects on firm value become clear. In this case, we would expect to see a greater effect for value-destroying acquisitions. In these cases, it would be more likely that directors are forced to leave the firm. This effect will be examined in section 5.3.
5.1.2. Exogenously Withdrawn bids
Due to data availability for some of the variables, the specification differs from propensity score matching as far as the control variables. However, this is unlikely to substantially affect the results, considering that firm fixed effects are included in the regression.
For this sample, the same regressions were carried out which are summarised in Table 4. In model 1-3, the coefficient is larger than in the Propensity Score Matched sample and it is consistent across all three specifications, in magnitude as well as in significance. The
significance level has increased to 1% for all three models. This provides additional and stronger evidence that mergers do indeed lead to board changes.
Page | 29 The results for director selection are again significant at either a 5% significance level (Model 4) or a 1% significance level (Model 5 and 6). The coefficient is quantitatively unchanged as compared to the third specification in Table 3. When examining the effect for departing directors, the coefficients are again around 0.04 and insignificant. There does not seem to be any evidence that directors leave in the year after the merger.
The benchmark of exogenously withdrawn bids is arguably more credible due to its identification strategy. Propensity score matching fails to account for unobservable
characteristics that simultaneously drive the decision to engage in a merger and to induce board changes. By only including bids that are withdrawn as control firms, the underlying assumption is that boards are similar in the respect that both were willing to engage in a merger. Only exogenously withdrawn bids were included, in order to exclude cases in which the board of directors purposely impeded the merger or was simply correlated with the probability of failure.
The results so far seem to be more consistent with the change hypothesis rather than the stability hypothesis. Based on the stability hypothesis, we would not expect to see any response to an exogenous shock. In that case, changes would be induced by more natural causes such as death or retirement which would not be more likely after a merger. The fact that mergers have a positive and significant effect on both board changes as a whole as well as newly added directors, suggests that firms use the event to refresh or increase their board size. This need for additional board members could increase with the size of the firm. Hence it could be the case that these results are partially driven by the complexity of the deal (Becher et al., 2017; Boone et al. 2007). Therefore, I calculated relative deal size and ran separate regressions based on the median. The results of these tests are found in Table 5. For sake of space, only the regressions with all controls are included. From Table 5 it is clear that the effect remains positive for board change both for mergers with a relative deal size above the median and below. However, the effect is greater and more significant for relatively more complex deals. For new directors, the effect is even more striking. For deals below the median, the effect has become insignificant. The results for departing directors are
qualitatively and quantitatively unchanged. Overall, the results are partially consistent with the change hypothesis and seem to be, at least partially, driven by the complexity of the deal. This is consistent with the findings by Boone et al. (2007) but inconsistent with the strong interpretation of the stable board hypothesis (Becher et al., 2017; Cicero et al., 2013; Denis et
Page | 30 al., 2015; Harford, 2003). In addition, it is inconsistent with the findings from Harford (2003) as they found evidence for departing acquiring directors.
[INSERT TABLE 5 HERE]
5.2 Which director characteristics are associated with director retention and selection? As mentioned before, a lot of inconsistency remains within academic literature as far as how rigid board structures are. Even though the previous analysis provides moderate evidence that mergers do increase the likelihood of board changes and new directors, nothing is known about the absolute numbers of board changes. Therefore, in this section, first univariate results will be presented after which director characteristics will be analysed in a multivariate setting. This will not only provide information on how frequent board changes occur but also on the underlying reasons for such board changes.
5.2.1. New, retained and departing directors of the acquiring board
In Table 6, descriptive statistics for acquiring directors are shown for the two years prior to the two years after the merger in order to account for changes in response to the impending merger. Within the dataset, only 44 cases of new directors were documented and a mere 47 leaving directors. 986 directors are retained following a merger. Thus, even though a merger increases the likelihood of board changes as compared to a control group of propensity-score matched firms and exogenously withdrawn bids, the changes are still marginal in absolute terms.
Table 6 also shows director characteristics related to both cognitive diversity as well as demographic diversity. The means are shown for new directors, retained directors as well as leaving directors. The t-statistics are presented in parentheses and are calculated based on the assumption of unequal variances.
[INSERT TABLE 6 HERE]
There is a significant difference of approximately 2.08 years between new and retained directors and a difference of 3.58 between new and departing directors. New directors are on average younger than retained or departing directors, whereas no such significant difference exists between retained and departing directors. A similar conclusion can be drawn from tenure, though the result is not as striking. Departing directors have a shorter tenure (10.45)
Page | 31 than retained directors (15.46), which is significant at the 10% level. Combined with the results from section 5.1, this suggests that mergers do induce board changes, but that board composition is driven by characteristics we would expect to drive the decision, such as retirement and tenure to a lesser extent.
New directors are also less likely to have completed an MBA degree than departing directors as well as retained directors. However, a greater proportion of new directors has served on a committee. These director specific skills could potentially be seen as valuable by the acquiring firm and hence influence the hiring decision. Finally, new directors are more likely to have 3 or more outside directorships as compared to departing directors and are therefore more frequently classified as busy directors. Current academic literature is divided as to whether busy directors are value-enhancing (Cashman, Gillan, & Jun, 2012; Ferris et al., 2003; Fich & Shivdasani, 2006). On the one hand it certifies their experience and abilities, but on the other hand, busyness may increase inefficiency and thus detract from firm value
(Cashman et al., 2012). The results are consistent with Ferris et al. (2003), which proved that busy directors are rewarded for multiple board seats. In conclusion, the univariate results show that age, busyness, MBA and committee could potentially drive the decision to hire and fire certain directors. These findings are partially consistent with the second hypothesis, which stated that director characteristics related to cognitive diversity as well as demographic diversity increase the likelihood of director retention & selection. It was found that these characteristics affect director selection, but no such result was found for director retention. In order to see if these results hold, a multivariate framework is adopted in Table 7.
[INSERT TABLE 7 HERE]
Table 7 shows logit regressions for director characteristics that would affect the likelihood of director selection and retention. The independent variables are the same as in table 6. Deal fixed effects are included in column 3 and 4. The univariate results for director age are confirmed by the logit model. It has a negative effect (-0.1219) on director selection and a positive effect on director departure (0.0420), though smaller in absolute magnitude. For a one-year increase in age, the odds of being hired decrease by a factor of -0.12.
In addition, it is confirmed that busyness positively affects the hiring decision whereas it decreases the likelihood of being made redundant. This reinforces the initial finding that busyness is seen as a desirable director characteristic. The effects of committee and MBA on board composition are no longer significant. PhD is significant at the 10% level and is shown
